ASM Handbook, Volume 15, Casting (2008)

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

W

ASM Handbook Volume 15 Casting Prepared under the direction of the ASM International Handbook Committee

Editorial Committee Srinath Viswanathan, University of Alabama, Chair Diran Apelian, Worcester Polytechnic Institute Raymond J. Donahue, Mercury Marine Babu DasGupta, National Science Foundation Michael Gywn, ATI Inc John L. Jorstad, J.L.J. Technologies, Inc Raymond W. Monroe, Steel Founders’ Society of America Mahi Sahoo, CANMET Materials Technology Laboratory Thomas E. Prucha, American Foundry Society Daniel Twarog, North American Die Casting Association Steve Lampman, Project Editor Charles Moosbrugger, Editor Eileen DeGuire, Editor Madrid Tramble, Senior Production Coordinator Ann Britton, Editorial Assistant Diane Whitelaw, Production Coordinator Kathryn Muldoon, Production Assistant Scott D. Henry, Senior Product Manager Bonnie R. Sanders, Manager of Production

Editorial Assistance Elizabeth Marquard Heather Lampman WPR Indexing Service

Materials Park, Ohio 44073-0002 www.asminternational.org

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Copyright # 2008 by ASM InternationalW All rights reserved No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the written permission of the copyright owner. First printing, December 2008

This book is a collective effort involving hundreds of technical specialists. It brings together a wealth of information from worldwide sources to help scientists, engineers, and technicians solve current and long-range problems. Great care is taken in the compilation and production of this Volume, but it should be made clear that NO WARRANTIES, EXPRESS OR IMPLIED, INCLUDING, WITHOUT LIMITATION, WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, ARE GIVEN IN CONNECTION WITH THIS PUBLICATION. Although this information is believed to be accurate by ASM, ASM cannot guarantee that favorable results will be obtained from the use of this publication alone. This publication is intended for use by persons having technical skill, at their sole discretion and risk. Since the conditions of product or material use are outside of ASM’s control, ASM assumes no liability or obligation in connection with any use of this information. No claim of any kind, whether as to products or information in this publication, and whether or not based on negligence, shall be greater in amount than the purchase price of this product or publication in respect of which damages are claimed. THE REMEDY HEREBY PROVIDED SHALL BE THE EXCLUSIVE AND SOLE REMEDY OF BUYER, AND IN NO EVENT SHALL EITHER PARTY BE LIABLE FOR SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES WHETHER OR NOT CAUSED BY OR RESULTING FROM THE NEGLIGENCE OF SUCH PARTY. As with any material, evaluation of the material under end-use conditions prior to specification is essential. Therefore, specific testing under actual conditions is recommended. Nothing contained in this book shall be construed as a grant of any right of manufacture, sale, use, or reproduction, in connection with any method, process, apparatus, product, composition, or system, whether or not covered by letters patent, copyright, or trademark, and nothing contained in this book shall be construed as a defense against any alleged infringement of letters patent, copyright, or trademark, or as a defense against liability for such infringement. Comments, criticisms, and suggestions are invited, and should be forwarded to ASM International. Library of Congress Cataloging-in-Publication Data ASM International ASM Handbook Includes bibliographical references and indexes Contents: v.1. Properties and selection—irons, steels, and high-performance alloys—v.2. Properties and selection—nonferrous alloys and special-purpose materials—[etc.]—v.21. Composites 1. Metals—Handbooks, manuals, etc. 2. Metal-work—Handbooks, manuals, etc. I. ASM International. Handbook Committee. II. Metals Handbook. TA459.M43 1990 620.1’6 90-115 SAN: 204-7586 ISBN-13: 978-0-87170-711-6 ISBN-10: 0-87170-711-X

ASM InternationalW Materials Park, OH 44073-0002 www.asminternational.org Printed in the United States of America Multiple copy reprints of individual articles are available from Technical Department, ASM International.

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Dedicated to the memory of JOSEPH R. DAVIS (1954–2007) Editor, Metals Handbook, 1982–1991

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Foreword

In this revision of ASM HandbookW Volume 15 on casting, ASM International is indebted to the volunteer efforts of the Volume 15 Editorial Committee and over 150 participants who helped as authors or reviewers. Their professional commitment and efforts represent a continuing devotion to the practice of metalcasting and the publication of peer consensus information on it. Special thanks are extended to Srinath Viswanathan, University of Alabama, for recruiting an outstanding Editorial Committee with Diran Apelian, Worcester Polytechnic Institute; Babu DasGupta, National Science Foundation, Raymond J. Donahue, Mercury Marine; Michael Gywn, ATI Inc.; John L. Jorstad, J.L.J. Technologies, Inc.; Raymond W. Monroe, Steel Founders’ Society of America; Thomas E. Prucha, American Foundry Society; Kumar Sadayappan and Mahi Sahoo, CANMET Materials Technology Laboratory; Edward S. Szekeres, Casting Consultants Incorporated; and Daniel Twarog, North American Die Casting Association. We thank them and the other contributors for this publication.

Dianne Chong President ASM International Stanley C. Theobald Managing Director ASM International

iv

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Policy on Units of Measure

By a resolution of its Board of Trustees, ASM International has adopted the practice of publishing data in both metric and customary U.S. units of measure. In preparing this Handbook, the editors have attempted to present data in metric units based primarily on Syste`me International d’Unite´s (SI), with secondary mention of the corresponding values in customary U.S. units. The decision to use SI as the primary system of units was based on the aforementioned resolution of the Board of Trustees and the widespread use of metric units throughout the world. For the most part, numerical engineering data in the text and in tables are presented in SI-based units with the customary U.S. equivalents in parentheses (text) or adjoining columns (tables). For example, pressure, stress, and strength are shown both in SI units, which are pascals (Pa) with a suitable prefix, and in customary U.S. units, which are pounds per square inch (psi). To save space, large values of psi have been converted to kips per square inch (ksi), where 1 ksi = 1000 psi. The metric tonne (kg  103) has sometimes been shown in megagrams (Mg). Some strictly scientific data are presented in SI units only. To clarify some illustrations, only one set of units is presented on artwork. References in the accompanying text to data in the illustrations are presented in both SI-based and customary U.S. units. On graphs and charts, grids corresponding to SI-based units usually appear along the left and bottom edges. Where appropriate, corresponding customary U.S. units appear along the top and right edges. Data pertaining to a specification published by a specification-writing group may be given in only the units used in that specification or in dual units, depending on the nature of the data. For example, the typical yield strength of steel sheet made to a specification written in customary U.S.

units would be presented in dual units, but the sheet thickness specified that specification might be presented only in inches. Data obtained according to standardized test methods for which the standard recommends a particular system of units are presented in the units of that system. Wherever feasible, equivalent units are also presented. Some statistical data may also be presented in only the original units used in the analysis. Conversions and rounding have been done in accordance with IEE/ ASTM SI-10, with attention given to the number of significant digits in the original data. For example, an annealing temperature of 1570  F contains three significant digits. In this case, the equivalent temperature would be given as 855  C; the exact conversion to 854.44  C would not be appropriate. For an invariant physical phenomenon that occurs at a precise temperature (such as the melting of pure silver), it would be appropriate to report the temperature as 961.93  C or 1763.5  F. In some instances (especially in tables and data compilations), temperature values in  C and  F are alternatives rather than conversions. The policy of units of measure in this Handbook contains several exceptions to strict conformance to IEEE/ASTM SI-10; in each instance, the exception has been made in an effort to improve the clarity of the Handbook. The most notable exception is the use of g/cm3 rather than kg/m3 as the unit of measure for density (mass per unit volume). SI practice requires that only one virgule (diagonal) appear in units formed by combination of several basic units. Therefore, all of the units preceding the virgule are in the numerator and all units following the virgule are in the denominator of the expression; no parentheses are required to prevent ambiguity.

v

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Preface

From the preceding Volume 5 (1970) of the 8th Edition Metals Handbook and the 9th Edition Metals Handbook Volume 15 (1988) on casting, this Handbook provides an update on the continuing advances in casting technologies and applications. Casting, as both a science and practical tool in art and technology, is enormously varied in scope. It is impossible to capture the full scope of casting technology in one volume.

on sand casting is expanded relative to the previous edition with the intention of providing a reference that may be helpful as a communication tool between product designers and metalcasters in developing successful and economical products. This Volume consists of 18 sections. The first section introduces the historical development of metal casting, as well as to the advantages of castings over parts produced by other manufacturing processes, their applications, and the current market size of the industry. This includes an article on “Metalcasting Technology and the Purchasing Process” written by Al Spada and the technical staff of the American Foundry Society. Then, the principles and practice of melt processing are described in the next three sections followed by a section on principles of solidification including nucleation kinetics, fundamentals of growth, transformation behavior, and microstructure development. Solid-state processing of casting, such as heat treat treatment and hot isostatic pressing, are also introduced. This is followed by a section on the “Modeling and Analysis of Casting Processes.”

The main focus of this Volume is on the products and processes of foundry (shape) casting, although primary (ingot or continuous casting) of steel and aluminum are also covered. In addition, continuous casting of copper is described in an article, as copper continuous casting was a precursor to steel and aluminum continuous casting in some respects. Some of the articles on melt processing, such as the articles on “Electric Arc Furnace Melting” and “Steel Melt Processing,” also briefly describe primary production of cast metal. Shape casting of metal is dominated by cast iron, which constitutes just over 70% of the worldwide production of castings on a tonnage basis (See Table 2 in the first article “History and Trends of Metal Casting”). This is followed by steel, copper-alloy, and aluminum-alloy castings, which make up about 25% on the worldwide tonnage of casting production. Magnesium and zinc are on the order of 1% or less. The dominance of just a few alloys in shape casting is due to the fact that successful and economic shape casting typically involves alloy compositions near a eutectic. The lower melting points and narrower freezing range of near-eutectic compositions promote better castability.

Like the previous edition, traditional subjects such as patterns, molding and casting processes, foundry equipment, and processing considerations are extensively covered in the next sections. As noted, coverage on sand casting has been consolidated and expanded. For example, the major method of shell molding is described in an article—based on an update of a still largely valid 1970 handbook (Volume 5) article. New updates are also provided on processes growing in use, such as squeeze casting, lost-foam casting, semisolid metal forming, and low-pressure casting. The latter is particularly important in producing quality products, as described by John Campbell in the article “Filling and Feeding Concepts.”

Since the 1988 edition of Volume15, several developments have occurred (see Table 5 in the first article “History and Trends of Metal Casting”). Of these developments, computer technology continues to shorten development time and help simulate the casting process. Automation and robotic technology also has improved the productivity and process control of casting. In terms of processes, semisolid processing, squeeze casting, lost-foam, vacuum molding, and various dies casting technologies continue to improve and finds new applications. These important topics are updated in this Volume. For nonferrous alloys, high-pressure die casting of aluminum is a major area of expansion and update in this volume.

Finally, the last five sections describe the major types of cast alloys in term of processing and the properties and characteristics of cast ferrous and nonferrous alloys. Emphasis is placed on cast iron, cast steel, aluminum, copper, and zinc. The last section covers the quality aspects of cast products and the processing of castings. It is hoped that this Handbook is a useful work of peer-consensus reference information for the producers, designers, and buyers of castings. Many thanks are extended to all the contributors and the editors who worked on this Volume. This publication would not have been possible without their commitment and effort.

In addition, coverage on sand casting is expanded and consolidated in this Volume with major articles on “Green Sand Molding,” No-Bake Sand Molding,” and “Shell Molding and Shell Coremaking.” Bonded sand mold casting, although well-established for many years, is the most widely used method of casting on a tonnage basis. Improvement in methods and materials continue to provide better yields, productivity, and product quality. The sand system is also a major factor in the economics of large-volume, production casting. Coverage

Srinath Viswanathan University of Alabama Chair, Volume 15 Editorial Committee

vi

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Officers and Trustees of ASM International (2007–2008)

Dianne Chong President and Trustee The Boeing Company Roger J. Fabian Vice President and Trustee Bodycote Thermal Processing Lawrence C. Wagner Immediate Past President and Trustee Texas Instruments Inc (retired) Paul L. Huber Treasurer and Trustee Seco/Warwick Corporation Stanley C. Theobald Secretary and Managing Director ASM International

Trustees

Sue S. Baik-Kromalic Honda of America Christopher C. Berndt Swinburne University of Technology Brady G. Butler University of Utah Danniel P. Dennies The Boeing Company Leigh C. Duren Worcester Polytechnic Institute Pradeep Goyal Pradeep Metals Ltd.

Digby D. Macdonald Penn State University Subhash Mahajan Arizona State University Charles A. Parker Honeywell Aerospace Megan M. Reynolds Washington State University Mark F. Smith Sandia National Laboratories Jon D. Tirpak ATI

Members of the ASM Handbook Committee (2007–2008) Larry D. Hanke (Chair 2006–; Member 1994–) Materials Evaluation and Engineering Inc. Kent L. Johnson (Vice Chair 2006–; Member 1999–) Engineering Systems Inc. Viola L. Acoff (2005–) University of Alabama David E. Alman (2002–2008) National Energy Technology Laboratory Tim Cheek (2004–) DELTA (v) Forensic Engineering Lichun Leigh Chen (2002–) Technical Materials Incorporated Sarup K. Chopra (2007–) Nook Industries Craig Clauser (2005–) Craig Clauser Engineering Consulting Incorporated Craig V. Darragh (1989–)

The Timken Company Jon L. Dossett (2006–) Consultant David U. Furrer (2006–) Rolls-Royce Corporation Lee Gearhart (2005–2008) Moog Inc. Ernest W. Klechka (2006–) Lloyd Register Capstone Inc. Alan T. Male (2003–) University of Kentucky William L. Mankins (1989–) Metallurgical Services Inc. Dana J. Medlin (2005–2008) South Dakota School of Mines and Technology Joseph W. Newkirk (2005–) University of Missouri-Rolla Cory J. Padfield (2006–) Hyundai America Technical Center Inc.

Toby V. Padfield (2004–) ZF Sachs Automotive of America Charles A. Parker (2007–) Honeywell Aerospace Elwin L. Rooy (2007–) Elwin Rooy & Associates Karl P. Staudhammer (1997–) Los Alamos National Laboratory Kenneth B. Tator (1991–) KTA-Tator Inc. George F. Vander Voort (1997–) Buehler Ltd.

Chairs of the ASM Handbook Committee J.F. Harper (1923–1926) (Member 1923–1926) W.J. Merten (1927–1930) (Member 1923–1933) L.B. Case (1931–1933) (Member 1927–1933) C.H. Herty, Jr. (1934–1936) (Member 1930–1936) J.P. Gill (1937) (Member 1934–1937) R.L. Dowdell (1938–1939) (Member 1935–1939) G.V. Luerssen (1943–1947) (Member 1942–1947) J.B. Johnson (1948–1951) (Member 1944–1951) E.O. Dixon (1952–1954) (Member 1947–1955) N.E. Promisel (1955–1961) (Member 1954–1963)

R.W.E. Leiter (1962–1963) (Member 1955–1958, 1960–1964) D.J. Wright (1964–1965) (Member 1959–1967) J.D. Graham (1966–1968) (Member 1961–1970) W.A. Stadtler (1969–1972) (Member 1962–1972) G.J. Shubat (1973–1975) (Member 1966–1975) R. Ward (1976–1978) (Member 1972–1978) G.N. Maniar (1979–1980) (Member 1974–1980) M.G.H. Wells (1981) (Member 1976–1981) J.L. McCall (1982) (Member 1977–1982) L.J. Korb (1983) (Member 1978–1983)

vii

T.D. Cooper (1984–1986) (Member D.D. Huffman (1986–1990) (Member D.L. Olson (1990–1992) (Member 1989–1992) R.J. Austin (1992–1994) (Member W.L. Mankins (1994–1997) (Member M.M. Gauthier (1997–1998) (Member C.V. Darragh (1999–2002) (Member Henry E. Fairman (2002–2004) (Member Jeffrey A. Hawk (2004–2006) (Member Larry D. Hanke (2006–2008) (Member

1981–1986) 1982–) 1982–1988, 1984–1985) 1989–) 1990–2000) 1989–) 1993–) 1997–) 1994–)

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Authors and Contributors

University of Birmingham, UK Kevin R. Anderson Mercury Marine

Paul G Campbell Metaullics Systems Division Pyrotek, Inc.

Princewill N. Anyalebechi Padnos College of Engineering & Computing, Grand Valley State University

Robert D Carnahan Thixomat Inc.

Diran Apelian Metal Processing Institute, Worcester Polytechnic Institute

Sujoy Chaudhury Honeywell International, Inc. K. K. Chawla University of Alabama At Birmingham

Alan P. Druschitz University of Alabama at Birmingham Cor van Ettinger Gieterij Doesburg Henry “Ed” Fairman Cincinnati Metallurgical Consultants Edward W. Flynn General Motors

Leigh Chen Technical Materials Incorporated

Jim Frost ACIPCO (American Cast Iron Pipe Company)

Henry Bakemeyer Die Casting Design and Consulting

Yeou-Li Chu Ryobi Die Casting

Richard Fruehan Carnegie Mellon University

Bryan Baker Vulcan Engineering Co.

Craig D. Clauser Craig Clauser Engineering Consulting Incorporated

Rafael Gallo Foseco Metallurgical

Lars Arnberg Norwegian University Of Science & Technology

Steve L. Cockroft The University of British Columbia

Charles-Andre´ Gandin Centre de Mise en Forme des Mate´riaux (CEMEF)

Chris Cooper Vulcan Engineering Co.

George Giles Morganite Crucible Inc

James D. Cotton The Boeing Co.

Peter C. Glaws The Timken Company

Randall Counselman Oldenburg Group Incorporated

Martin E. Glicksman University of Florida

Joe Bigelow CONTECH U. S. LLC.

Jonathan A. Dantzig University of Illinois-Urbana

George Goodrich Stork Climax Research Services

Ron Bird Stainless Foundry & Engineering

Atef Daoud Central Metallurgical Research and Development Institute, Cairo, Egypt

Frank E. Goodwin International Lead Zinc

Stewart A Ballantyne Allvac Allegheny Technologies Company Scott A. Balliett Latrobe Specialty Steel Company Christoph Beckermann University Of Iowa, The Michel Bellet Ecole des Mines de Paris, Centre de Mise en Forme des Mate´riaux (CEMEF)

Malcolm Blair Steel Founders Society of America William Boettinger NIST James A. Brock Sr. Alcoa Harold D. Brody University Of Connecticut Craig C. Brown Stork Technimet, Inc. Zach Brown CONTECH U.S. LLC Andreas Buhrig-Polackzek Gieberei-Institut, Germany Willam A. Butler General Motors (Retired) John Campbell

Craig V. Darragh The Timken Company Babu DasGupta National Science Foundation Bob Dawson Honeywell Aerospace MCOE Raymond Decker Thixomat Incorporated Raymond J Donahue Mercury Marine Daniel Dos Santos Gießerei Institut der RWTH Aachen Jon Dossett Metallurgical Consultant Jean Marie Drezet Ecole Polytechnique Fe´de´rale de Lausanne

viii

Lee Gouwens CMI Novacast Inc. Ronald Graham ATI Wah Chang Patrick Grant Oxford University Daniel E.Groteke Q.C. Designs Inc. Mustafa Guclu Stanley Associates Inc. Richard Gundlach Stork Climax Research Services Nikhil Gupta Polytechnic University Mike Gwyn ATI, Inc. John Hall

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

Hall Permanent Mold Machines Qingyou Han Purdue University Larry D. Hanke Materials Evaluation and Engineering Inc Justin Heimsch Madison Kipp Corporation

www.asminternational.org

Makhlouf M. Makhlouf Metal Processing Institute, Worcester Polytechnic Institute William L. Mankins Metallurgical Services Incorporated Stephen J. Mashl Bodycote HIP Andover

Intermet-PCPC Robert Pischel FOSECO Metallurgical Inc. David R. Poirier University of Arizona David Poweleit Steel Founders’ Society of America

Juan Carlos Heinrich The University of New Mexico

Ragnvald H. Mathiesen Norwegian University of Science and Technology

Gregg Holman Wah Chang

Robert J. McInerney BuhlerPrince

Thomas E. Prucha American Foundry Society

Robert A. Horton Precision Metalsmiths Incorporated

Scott McIntyre Grede Foundrie

Donald J. Hurtuk H.C. Starck Inc.

Brad McKee Deloro Stellite Inc.

Peter Quested Industry and Innovation Division, National Physical Laboratory, UK

Alain Jacot Laboratoire de Simulation des Mate´riaux

Harold T. Michels Copper Development Association

John L. Jorstad J.L.J. Technologies, Inc.

Tony Midea Foseco Metallurgical Inc.

Ursula R. Kattner N I S T/Metallurgy Division

Paul H Mikkola Metal Casting Technology Incorporated

Seymour Katz S. Katz Associates

Asbjrn Mo SINTEF Materials and Chemistry

Jay S. Keist Worcester Polytechnic Institute

Raymond W. Monroe Steel Founders’ Society of America

Graham Keough Consarc Corporation

Y.V. Murty Cellular Materials International, Inc.

Matthew Krane Purdue University

Laurentiu Nastac Concurrent Technologies Corporation

Mary Beth Krusiak Sand Technology Co. LLC

David V. Neff Metaullics Systems Division Pyrotek, Inc.

T.A. Kuhn Alcoa Primary Metals

Charles Nelson Morris Bean and Company

Wilfried Kurz Ecole Polytechnique Federale De Lausanne (EPFL) Switzerland

Itsuo Ohnaka Osaka Sangyo University, Japan

Selcuk Kuyucak CANMET Materials Technology Laboratory Vic La Fay Hill & Griffith Company Robin A. Lampson Retech Systems LLC

Cory Padfield American Axle & Manufacturing Qingyue Pan SPX Corporation

William L. Powell ThyssenKrupp-Waupaca (Retired)

Wayne Rasmussen Consultant Martin Reeves Striko-Dynarad Allen Richardson SPX Corporation Pradeep Rohatgi University of Wisconsin-Milwaukee Elwin L. Rooy Elwin Rooy and Associates Gary Ruff SMW Automotive Corporation Kumar (Muthukumarasamy) Sadayappan CANMET Materials Technology Laboratory Mahi Sahoo CANMET Materials Technology Laboratory Kenneth Savage Magnesium Elektron Gary Scholl Metal Casting Technology Incorporated Ted Schorn Enkei America, Inc. Benjamin F. Schultz University of Wisconsin-Milwakee David Schwam Case Werstern Reserve University

Charles A. Parker Honeywell Aerospace

William D. Scott AAA Alchemy

David Lee Howmet Corp.

Kent Peaslee Missouri University of Science and Technology (Missouri S&T)

Ben Q. Li University of Michigan Dearborn

Pehlke, Robert D. Univ of Michigan

Sumanth Shankar McMaster University

Alan Luo General Motors R&D Center

John H. Perepezko University of Wisconsin, Madison

Geoffrey K. Sigworth Alcoa Primary Metals

Dan Maas Ex-One

Ralph Perkul Foundry Solutions and Design LLC

Steve Sikkenga Cannon-Muskegon Corporation

James F. Major Alcan International Ltd

Gordon H. Peters

Brian V. Smith

ix

John Serra Carpenter Brothers Inc.

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

General Motors Corporation, Powertrain Division Jerry Sokolowski University of Windsor Al Spada American Foundry Society Karl Staudhammer Los Alamos National Laboratory Doru M. Stefanescu The Ohio State University Ingo Steinbach RWTH-Aachen, ACCESS e. V. Karthik Subbiah SPX Product Solutions Group Erin H. Sunseri Iowa State University Tony Suschil Foseco Inc Richard Sutherlin ATI Wah Chang Edward S.Szekeres Casting Consultants Incorporated Brian G. Thomas

www.asminternational.org

University of Illinois Michael L Tims Concurrent Technologies Corporation David W. Tripp TIMET

Srinath Viswanathan University of Alabama Birmingham Robert C. Voigt Penn State University Vaughan R. Voller University of Minnesota

Rohit Trivedi AMES Laboratory Paul Trojan University of Michigan-Dearborn John Troxler Rolls Royce

Tom Waring ME Elecmetal Sufei Wei Centrifugal Casting Machine Company Incorporated David White Schaefer Furnace

Nao Tsumagari American Showa, Inc. Daniel Twarog North American Die Casting Association Derek E. Tyler Olin Metals Thin Strip (retired) Don Tyler General Aluminum Manufacturing Company Stephen Udvardy North American Die Casting Association Juan J. Valencia Concurrent Technologies Corporation

x

Lawrence Whiting CANMET Materials Technology Laboratory Richard Williams FOSECO Metallurgical Inc. Greg G.Woycik Woycik Consulting Oscar Yu RTI International Metals, Inc. Keith Zhang Teck Cominco Metals Ltd.

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Contents

Applications and Specification Guidelines. . . . . . . . . . . . . . . . . . . 1 Raymond W. Monroe, Steel Founders’ Society of America; Thomas E. Prucha, American Foundry Society; Daniel Twarog, North American Die Casting Association History and Trends of Metal Casting. . . . . . . . . . . . . . History of Metal Casting . . . . . . . . . . . . . . . . . . . . Metal Casting Methods. . . . . . . . . . . . . . . . . . . . . . Production Trends . . . . . . . . . . . . . . . . . . . . . . . . . Market Trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . Metalcasting Technology and the Purchasing Process . . . . The Purchasing Process . . . . . . . . . . . . . . . . . . . . . Define Requirements and Develop a Purchasing Plan. Requesting and Evaluating Quotations/Bids . . . . . . . Select a Supplier and Negotiate Contract Terms . . . . Contract Fulfillment and Continuous Improvement . . Considerations in Purchasing Metal Castings . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

Principles of Liquid Metal Processing . . . . . . . . . . . . Chemical Thermodynamics and Kinetics . . . . . . . . . . . . Chemical Thermodynamics . . . . . . . . . . . . . . . . . . Chemical Kinetics . . . . . . . . . . . . . . . . . . . . . . . . Kinetics of Alloys Additions . . . . . . . . . . . . . . . . . Thermodynamic Properties of Iron-Base Alloys . . . . . . . Purification of Ferrous Melts. . . . . . . . . . . . . . . . . Thermodynamics of Ferrous Systems . . . . . . . . . . . Multicomponent Iron-Carbon Systems . . . . . . . . . . Structural Diagrams . . . . . . . . . . . . . . . . . . . . . . . Thermodynamic Properties of Aluminum-Base and Copper-Base Alloys . . . . . . . . . . . . . . . . . . . . Activities and Thermal Properties . . . . . . . . . . . . . Ineraction Coefficients . . . . . . . . . . . . . . . . . . . . . Thermal Properties for Hypothetical Standard State . Phase Diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . Gases in Metals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cast Iron. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Aluminum and Its Alloys . . . . . . . . . . . . . . . . . . . Copper and Its Alloys . . . . . . . . . . . . . . . . . . . . . Magnesium and Its Alloys . . . . . . . . . . . . . . . . . . Overcoming Gas Porosity . . . . . . . . . . . . . . . . . . . Inclusion-Forming Reactions . . . . . . . . . . . . . . . . . . . . Inclusion Types . . . . . . . . . . . . . . . . . . . . . . . . . . Physical Chemistry . . . . . . . . . . . . . . . . . . . . . . . Control of Inclusions . . . . . . . . . . . . . . . . . . . . . . Inclusions in Ferrous Alloys . . . . . . . . . . . . . . . . . Inclusions in Nonferrous Alloys . . . . . . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

29 31 31 35 35 41 41 45 50 51

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

56 56 56 58 58 64 64 67 70 71 73 74 74 75 76 77 81

Melting and Remelting . . . . . . . . . . Electric Arc Furnace Melting . . . . . . . Power Supply . . . . . . . . . . . . . . Arc Furnace Components . . . . . . Weighing and Chemical Analysis Scrap and Alloy Storage . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

85 87 87 88 91 91

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . .

Acid Melting Practice . . . . . . . . . . . . . . . . . Basic Melting Practice . . . . . . . . . . . . . . . . . Newer Technologies. . . . . . . . . . . . . . . . . . . Cupola Furnaces . . . . . . . . . . . . . . . . . . . . . . . . . Progressive Refinements in Cupola Equipment Cupola Construction and Operation . . . . . . . . Charge Materials . . . . . . . . . . . . . . . . . . . . . Cupola Control Principles . . . . . . . . . . . . . . . Control Tests and Analyses. . . . . . . . . . . . . . Specialized Cupolas . . . . . . . . . . . . . . . . . . . Induction Furnaces . . . . . . . . . . . . . . . . . . . . . . . Types of Furnaces . . . . . . . . . . . . . . . . . . . . Electromagnetic Stirring . . . . . . . . . . . . . . . . Power Supplies . . . . . . . . . . . . . . . . . . . . . . Water Cooling Systems . . . . . . . . . . . . . . . . Lining Material . . . . . . . . . . . . . . . . . . . . . . Melting Operations . . . . . . . . . . . . . . . . . . . Vacuum Induction Melting . . . . . . . . . . . . . . . . . Process Description . . . . . . . . . . . . . . . . . . . VIM Metallurgy . . . . . . . . . . . . . . . . . . . . . Production of Nonferrous Materials . . . . . . . . Electroslag Remelting . . . . . . . . . . . . . . . . . . . . . ESR Furnaces . . . . . . . . . . . . . . . . . . . . . . . Ingot Solidification. . . . . . . . . . . . . . . . . . . . Process Variations . . . . . . . . . . . . . . . . . . . . Steel ESR . . . . . . . . . . . . . . . . . . . . . . . . . . Superalloy ESR . . . . . . . . . . . . . . . . . . . . . . Vacuum Arc Remelting . . . . . . . . . . . . . . . . . . . . Process Description . . . . . . . . . . . . . . . . . . . Solidification . . . . . . . . . . . . . . . . . . . . . . . . Process Variables. . . . . . . . . . . . . . . . . . . . . Superalloy VAR . . . . . . . . . . . . . . . . . . . . . Titanium VAR . . . . . . . . . . . . . . . . . . . . . . Skull Melting . . . . . . . . . . . . . . . . . . . . . . . . . . . Vacuum Arc Skull Melting . . . . . . . . . . . . . . Furnaces . . . . . . . . . . . . . . . . . . . . . . . . . . . Induction Skull Melting . . . . . . . . . . . . . . . . Electron Beam Melting . . . . . . . . . . . . . . . . . . . . Process Description . . . . . . . . . . . . . . . . . . . Drip Melting . . . . . . . . . . . . . . . . . . . . . . . . Cold Hearth Melting . . . . . . . . . . . . . . . . . . Plasma Melting and Heating . . . . . . . . . . . . . . . . Plasma Torches . . . . . . . . . . . . . . . . . . . . . . Furnace Equipment . . . . . . . . . . . . . . . . . . . Atmosphere Control . . . . . . . . . . . . . . . . . . . Melting Processes . . . . . . . . . . . . . . . . . . . . Crucible Furnaces . . . . . . . . . . . . . . . . . . . . . . . . Applications . . . . . . . . . . . . . . . . . . . . . . . . Furnace Types . . . . . . . . . . . . . . . . . . . . . . . Design and Operation. . . . . . . . . . . . . . . . . . Reverberatory and Stack Furnaces . . . . . . . . . . . . Furnace Types . . . . . . . . . . . . . . . . . . . . . . . Reverberatory Furnace Practice . . . . . . . . . . . Molten Metal Circulation . . . . . . . . . . . . . . .

3 4 8 9 14 16 16 17 21 22 22 22

xi

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . .

91 95 98 99 99 100 103 106 106 107 108 109 109 111 111 113 116 116 119 122 124 125 128 129 129 131 132 132 134 135 137 137 139 139 139 140 142 142 143 145 149 149 149 151 151 151 155 155 156 158 160 161 163 165

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Molten Metal Processing and Handling . . . . . . . . . . . . . Transfer and Treatment of Molten Metal—An Introduction . Launders. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tundishes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Molten-Metal Pumps . . . . . . . . . . . . . . . . . . . . . . . . Pouring and Dosing . . . . . . . . . . . . . . . . . . . . . . . . . Degassing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Aluminum Foundry Degassing . . . . . . . . . . . . . . . . . Degassing of Magnesium . . . . . . . . . . . . . . . . . . . . . Degassing of Copper Alloys . . . . . . . . . . . . . . . . . . . Molten-Metal Filtration . . . . . . . . . . . . . . . . . . . . . . . . . . Sources of Inclusion. . . . . . . . . . . . . . . . . . . . . . . . . Removal of Inclusions . . . . . . . . . . . . . . . . . . . . . . . Filter Characteristics . . . . . . . . . . . . . . . . . . . . . . . . Types of Molten-Metal Filters . . . . . . . . . . . . . . . . . . In-Mold Filtration . . . . . . . . . . . . . . . . . . . . . . . . . . In-Furnace Filtration of Molten Aluminum . . . . . . . . . Filtered Metal Quality . . . . . . . . . . . . . . . . . . . . . . . Steel Melt Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . Furnaces and Refractories . . . . . . . . . . . . . . . . . . . . . Converter Metallurgy . . . . . . . . . . . . . . . . . . . . . . . . Ladle Metallurgy . . . . . . . . . . . . . . . . . . . . . . . . . . . Aluminum Fluxes and Fluxing Practice . . . . . . . . . . . . . . . Aluminum Fluxing. . . . . . . . . . . . . . . . . . . . . . . . . . Categories of Fluxes . . . . . . . . . . . . . . . . . . . . . . . . Foundry Practices . . . . . . . . . . . . . . . . . . . . . . . . . . Modification of Aluminum-Silicon Alloys . . . . . . . . . . . . . Heat Treated Castings . . . . . . . . . . . . . . . . . . . . . . . Porosity and Shrinkage Formation . . . . . . . . . . . . . . . Modification Mechanisms . . . . . . . . . . . . . . . . . . . . . Eutectic Nucleation . . . . . . . . . . . . . . . . . . . . . . . . . Eutectic Grain Structure . . . . . . . . . . . . . . . . . . . . . . Recommended Foundry Practices . . . . . . . . . . . . . . . Growth of Silicon Eutectic . . . . . . . . . . . . . . . . . . . . Effect of Other Elements . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Grain Refinement of Aluminum Casting Alloys . . . . . . . . . Measurement of Grain Size. . . . . . . . . . . . . . . . . . . . Mechanisms of Grain Refinement . . . . . . . . . . . . . . . Boron as a Grain Refiner . . . . . . . . . . . . . . . . . . . . . Best Practices for Grain Refinement. . . . . . . . . . . . . . Benefits of Grain Refinement . . . . . . . . . . . . . . . . . . Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . Refinement of the Primary Silicon Phase in Hypereutectic Aluminum-Silicon Alloys . . . . . . . . . . . . . . . . . . . . . Importance of Primary Silicon Refinement . . . . . . . . . Accomplishing Primary Silicon Refinement . . . . . . . . Principles of Solidification . . . . . . . . . . . . . . . . Diran Apelian, Worcester Polytechnic Institute Thermodynamics and Phase Diagrams . . . . . . . . Thermodynamics . . . . . . . . . . . . . . . . . . . . Phase Diagrams . . . . . . . . . . . . . . . . . . . . . Solidification . . . . . . . . . . . . . . . . . . . . . . . Nucleation Kinetics and Grain Refinement . . . . . Thermodynamics of Solidification . . . . . . . . Nucleation Phenomena . . . . . . . . . . . . . . . . Inoculation Practice . . . . . . . . . . . . . . . . . . Grain Refinement Models . . . . . . . . . . . . . . Master Alloy Processing . . . . . . . . . . . . . . . Transport Phenomena During Solidification . . . . . Fundamentals . . . . . . . . . . . . . . . . . . . . . . Transport and Microstructure . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . Plane Front Solidification . . . . . . . . . . . . . . . . . Transients and Steady State . . . . . . . . . . . . Morphological Stability . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

171 173 174 174 175 178 185 185 188 189 194 194 196 197 199 199 201 204 206 206 208 212 230 231 233 235 240 241 242 243 244 245 247 249 249 250 255 255 255 257 258 260 260

Rapid Solidification Effects. . . . . . . . . . . . . . . . . . . . . . Solidification Microstructure Selection Maps . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Non-Plane Front Solidification . . . . . . . . . . . . . . . . . . . . . . . Cellular Microstructures . . . . . . . . . . . . . . . . . . . . . . . . Dendrite Growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nonplanar Eutectic Growth. . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Eutectic Solidification—An Introduction . . . . . . . . . . . . . . . . Binary Eutectic Phase Diagram . . . . . . . . . . . . . . . . . . . Eutectic Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solidification of Eutectic Alloys—Aluminum-Silicon . . . . . . . The Aluminum-Silicon Equilibrium Phase Diagram . . . . . Classification of the Aluminum-Silicon Eutectic Structure Microstructure of the Aluminum-Silicon Eutectic . . . . . . Theories of Solidification of the Aluminum-Silicon Eutectic . . . . . . . . . . . . . . . . . . Theories of Chemical Modification of the Aluminum-Silicon Eutectic . . . . . . . . . . . . . . . . . . Solidification of Eutectic Alloys—Cast Iron. . . . . . . . . . . . . . Structure of Liquid Iron-Carbon Alloys . . . . . . . . . . . . . Nucleation of Eutectic in Cast Iron . . . . . . . . . . . . . . . . Nucleation of Primary Phases in Cast Iron . . . . . . . . . . . Growth of Eutectic in Cast Iron. . . . . . . . . . . . . . . . . . . Cooling Curve Analysis . . . . . . . . . . . . . . . . . . . . . . . . Peritectic Solidification . . . . . . . . . . . . . . . . . . . . . . . . . . . . Peritectic Solidification. . . . . . . . . . . . . . . . . . . . . . . . . Peritectic Reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . Peritectic Transformation and Direct Precipitation . . . . . . Primary Metastable Precipitation of Beta . . . . . . . . . . . . Peritectic Cascades . . . . . . . . . . . . . . . . . . . . . . . . . . . Peritectic Transformations in Multicomponent Systems . . Microsegregation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solidification Models . . . . . . . . . . . . . . . . . . . . . . . . . . Binary Isomorphous Systems, k > 1: Titanium-Molybdenum . . . . . . . . . . . . . . . . . . . . . Binary Eutectic Systems, k > 1: Hypoeutectic Aluminum-Copper Alloys . . . . . . . . . . . . . . . . . . . Binary Eutectic Systems: Hypoeutectic Aluminum-Silicon Alloys . . . . . . . . . . . . . . . . . . . Binary Peritectic Systems: Copper-Zinc a-Brass . . . . . . . Multicomponent, Multiphase Eutectic Systems: Hypoeutectic Al-Si-Cu-Mg Alloys . . . . . . . . . . . . . . . Multicomponent Multiphase Steel: Fe-C-Cr . . . . . . . . . . Multicomponent, Multiphase Nickel-Base Superalloy. . . . Macrosegregation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Macrosegregation Induced by Flow of the Liquid . . . . . . Macrosegregation Induced by Movement of the Solid . . . Interpretation and use of Coolng Curves—Thermal Analysis . . Quantitative Thermal Analysis . . . . . . . . . . . . . . . . . . . Deviation from Equilibrium . . . . . . . . . . . . . . . . . . . . . Differential Thermal Analysis . . . . . . . . . . . . . . . . . . . . Determining Liquidus and Solidus Temperature . . . . . . . X-Ray Imaging of Solidification Processes and Microstructure Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X-Ray Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . In-House Laboratory X-Ray Imaging of Solidification Processes . . . . . . . . . . . . . . . . . . . . . X-Ray Imaging Studies of Solidification Processes with Synchrotron Radiation . . . . . . . . . . . . . . . . . . . . Further Reading. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Oxide Film Defect—The Bifilm. . . . . . . . . . . . . . . . . . . Bifilms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Entrainment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Furling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Unfurling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reliability and Reproducibility of Mechanical Properties .

. . . . . 263 . . . . . 263 . . . . . 263

. . . . . . . . . . . . 267 . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

269 269 271 272 276 276 278 280 281 285 288 288 290 292 293 294 295

xii

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

296 297 297 299 300 301 304 305 307 307 307 312 312 313 313

. . . 313 . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

315 317 317 317 320 320 327 330 330 331 332 333 334 335 338 338

. . . 339 . . . 340 . . . 344 . . . 344 . . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

345 345 345 348 348 351 353 353 354 354 355

. . . 357 . . . 357 . . . 357 . . . . . . . .

. . . . . . . .

. . . . . . . .

358 360 362 362 363 364 364 367

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Central Role of Bifilm Defects in Fracture . . . . . . . . . . . Shrinkage Porosity and Gas Porosity. . . . . . . . . . . . . . . . . . . Shrinkage Porosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gas Porosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Factors Affecting Porosity Formation . . . . . . . . . . . . . . . Castability—Fluidity and Hot Tearing. . . . . . . . . . . . . . . . . . Fluidity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hot Tearing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Semisolid Metal Processing . . . . . . . . . . . . . . . . . . . . . . . . . Flow Behavior of SSM. . . . . . . . . . . . . . . . . . . . . . . . . Semisolid Microstructural Formation . . . . . . . . . . . . . . . Semisolid Processes: Thixocasting versus Rheocasting . . . Spray Casting. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Atomization and the Droplet Spray . . . . . . . . . . . . . . . . Deposition and Grain Mulitplication . . . . . . . . . . . . . . . Microsegregation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Macrosegregation and Coarsening . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rapid Solidification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microstructural Modification . . . . . . . . . . . . . . . . . . . . . Solidification During Casting of Metal-Matrix Composites . . . Incorporation of Reinforcements . . . . . . . . . . . . . . . . . . Reinforcement-Metal Wettability . . . . . . . . . . . . . . . . . . Solidification Processing of MMC . . . . . . . . . . . . . . . . . Solidification Fundamentals . . . . . . . . . . . . . . . . . . . . . Modeling of Particle-Pushing Phenomenon . . . . . . . . . . . Nanocomposites. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solidification Research in Microgravity. . . . . . . . . . . . . . . . . Availability of Reduced Gavity: Past, Present, and Future Solidification Research in Microgravity . . . . . . . . . . . . . Key Solidification Experiments . . . . . . . . . . . . . . . . . . . The Future of Microgravity Solidification . . . . . . . . . . . . Homogenization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sources of Inhomogeneity. . . . . . . . . . . . . . . . . . . . . . . Benefits of Homogenization Treatment. . . . . . . . . . . . . . Characterization Methods . . . . . . . . . . . . . . . . . . . . . . . Computational Modeling . . . . . . . . . . . . . . . . . . . . . . . Heat Treatment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Designations of Heat Treatment . . . . . . . . . . . . . . . . . . Solution Heat Treatment. . . . . . . . . . . . . . . . . . . . . . . . Quenching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Aging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Emerginhg Heat Treating Technologies . . . . . . . . . . . . . Microstructural Changes due to Heat Treatment of Cast Alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hot Isostatic Pressing of Castings. . . . . . . . . . . . . . . . . . . . . History of HIP. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reasons for Using HIP . . . . . . . . . . . . . . . . . . . . . . . . . Overview of a HIP System . . . . . . . . . . . . . . . . . . . . . . Effect of HIP on Mechanical Properties . . . . . . . . . . . . . Effect of Inclusions on As-HIPed Properties . . . . . . . . . . Effect of HIP on the Shape and Structure of Castings . . . Problems Encountered in HIP . . . . . . . . . . . . . . . . . . . . Economics of HIP . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Modeling and Analysis of Casting Processes. . . . . . . . Srinath Viswanathan, University of Alabama Numerical Methods for Casting Applications . . . . . . . . . Fundamentals of CFD . . . . . . . . . . . . . . . . . . . . . Numerical Solution of the Fluid-Flow Equations . . . Grid Generation for Complex Geometries. . . . . . . . Casting Applications . . . . . . . . . . . . . . . . . . . . . . Modeling of Transport Phenomena and Electromagnetics Conservation of Thermal Energy . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

368 370 370 372 373 375 375 376 379 379 380 381 382 382 383 384 385 385 385 386 386 390 391 391 391 392 394 395 395 398 398 399 399 400 402 402 402 402 403 404 404 404 405 405 406

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

406 408 408 409 409 411 412 413 414 415 415

. . . . . . . 417 . . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

419 419 420 420 422 425 425

Conservation of Solute . . . . . . . . . . . . . . . . . . . . . . . . . Equations Cast in Advection-Diffusion Form . . . . . . . . . Basic Equations for Mass and Momentum Conservation . Modeling the Momentum in the Solid + Liquid Mushy Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Modeling Turbulence . . . . . . . . . . . . . . . . . . . . . . . . . . Phenomena Restraining or Driving Flow . . . . . . . . . . . . Dealing with Fronts and Free Surfaces . . . . . . . . . . . . . . Simulation of Electromagnetic Effects in Castings. . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Direct Modeling of Structure Formation . . . . . . . . . . . . . . . . Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Direct Microstructure Simulation Using the Phase Field Method. . . . . . . . . . . . . . . . . . . . . . . . Direct Grain Structure Simulation Using the Cellular Automation Method . . . . . . . . . . . . . . . . . Coupling of Direct Structure Simulation at Macroscopic Scale . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Modeling of Microsegregation and Macrosegregation . . . . . . . Microsegregation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microsegregation with Diffusion in the Solid . . . . . . . . . Continuum Model of Macrosegregation . . . . . . . . . . . . . Modeling of Stress, Distortion, and Hot Tearing . . . . . . . . . . Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . Thermomechanical Coupling . . . . . . . . . . . . . . . . . . . . . Numerical Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . Model Validation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Example Applications . . . . . . . . . . . . . . . . . . . . . . . . . Hot-Tearing Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Practical Issues In Computer Simulation of Casting Processes . Setting Simulation Objectives, Modeling, and Selection of Simulation Code . . . . . . . . . . . . . . . . . . . . . . . . Modeling of Shape and Phenomena . . . . . . . . . . . . . . . . Initial and Boundary Conditions and Physical Properties . Enmeshing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Evaluation of Simulation Results . . . . . . . . . . . . . . . . . . Estimation of Structure and Properties . . . . . . . . . . . . . . Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thermophysical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . Sources and Availability of Reliable Data . . . . . . . . . . . Limitations and Warning on the Use of Data . . . . . . . . . Methods to Determine Thermophysical Properties . . . . . . Specific Heat Capacity and Enthalpy of Transformation . . Enthalpy of Melting, Solidus, and Liquidus Temperatures Coefficient of Thermal Expansion . . . . . . . . . . . . . . . . . Density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Surface Tension. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Viscosity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electrical and Thermal Conductivity . . . . . . . . . . . . . . . Emissivity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Typical Themophysical Properties Ranges of Some Cast Alloys . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Principles and Practices of Shape Casting . . . . . . . Shape Casting Processes—An Introduction . . . . . . . . Patterns and Patternmaking . . . . . . . . . . . . . . . . . . . Types of Patterns . . . . . . . . . . . . . . . . . . . . . . . Additional Features of Patterns . . . . . . . . . . . . . Pattern Allowances . . . . . . . . . . . . . . . . . . . . . Pattern Materials . . . . . . . . . . . . . . . . . . . . . . . Selection of Pattern Type . . . . . . . . . . . . . . . . . Trends in Pattern Design and Manufacture . . . . . Casting Practice—Guidelines for Effective Production of Reliable Castings . . . . . . . . . . . . . . . . . . . . .

xiii

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . 425 . . . 427 . . . 427 . . . . . . . .

. . . . . . . .

. . . . . . . .

428 428 429 429 431 432 435 435

. . . 436 . . . 438 . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

441 443 445 445 445 446 449 449 451 451 453 454 456 458 462

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

462 462 463 464 465 466 466 468 468 468 468 470 470 471 471 471 474 474 474

. . . 477 . . . 477 . . . . . . . . .

. . . . . . . . .

. . . . . . . . .

483 485 488 488 490 491 492 494 495

. . . . . . . . . 497

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Rule 1: Good-Quality Melt . . . . . . . . . . . . . . . . Rule 2: Liquid Front Damage . . . . . . . . . . . . . . Rule 3: Liquid Front Stop. . . . . . . . . . . . . . . . . Rule 4: Bubble Damage . . . . . . . . . . . . . . . . . . Rule 5: Core Blows . . . . . . . . . . . . . . . . . . . . . Rule 6: Shrinkage Damage . . . . . . . . . . . . . . . . Rule 7: Convection Damage . . . . . . . . . . . . . . . Rule 8: Segregation . . . . . . . . . . . . . . . . . . . . . Rule 9: Residual Stress. . . . . . . . . . . . . . . . . . . Rule 10: Location Points . . . . . . . . . . . . . . . . . Filling and Feeding System Concepts . . . . . . . . . . . . Entrainment Defects in Gravity-Poured Castings . Traditional Filling Designs . . . . . . . . . . . . . . . . Prepriming Techniques . . . . . . . . . . . . . . . . . . . Naturally Pressurized Filling System . . . . . . . . . Filters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Performance of Alternative Filling Systems . . . . Recommended Foundry Practice . . . . . . . . . . . . Processing and Finishing of Castings . . . . . . . . . . . . Shakeout and Core Knockout . . . . . . . . . . . . . . Cleaning Operations. . . . . . . . . . . . . . . . . . . . . Automating Gate Removal and Grinding . . . . . . Flame Cutting . . . . . . . . . . . . . . . . . . . . . . . . . Blast Cleaning of Castings . . . . . . . . . . . . . . . . Inspection . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

Expandable-Mold Casting Processes with Permanent Patterns. . . . . . . . . . . . . . . . . . . John L. Jorstad, J.L.J. Technologies, Inc Introduction—Expendable Mold Processes with Permanent Patterns . . . . . . . . . . . . . Aggregates and Binders for Expendable Molds Mold Media . . . . . . . . . . . . . . . . . . . . . Foundry Sands. . . . . . . . . . . . . . . . . . . . Clays for Green Sand Molding . . . . . . . . Inorganic Binders . . . . . . . . . . . . . . . . . Organic Binders. . . . . . . . . . . . . . . . . . . Other Expendable Mold Media . . . . . . . . Green Sand Molding . . . . . . . . . . . . . . . . . . . Sand Molding . . . . . . . . . . . . . . . . . . . . System Control and Formulation . . . . . . . Influence of Molding Equipment . . . . . . . Maintaining Sand System Quality . . . . . . Molding Methods. . . . . . . . . . . . . . . . . . Green Sand Media Preparation . . . . . . . . Molding Problems . . . . . . . . . . . . . . . . . Mold Finishing . . . . . . . . . . . . . . . . . . . Shakeout. . . . . . . . . . . . . . . . . . . . . . . . Sand/Casting Recovery. . . . . . . . . . . . . . Sand Reclamation . . . . . . . . . . . . . . . . . No-Bake Sand Molding . . . . . . . . . . . . . . . . . No-Bake Sand Processing . . . . . . . . . . . . CO2 -Cured Sodium Silicate . . . . . . . . . . Air-Setting Sodium Silicates . . . . . . . . . . Air-Setting Organic Binders . . . . . . . . . . Gas-Cured Organic Binders. . . . . . . . . . . Sand Reclamation and Reuse . . . . . . . . . Coremaking . . . . . . . . . . . . . . . . . . . . . . . . . Binders for Sand Coremaking . . . . . . . . . Basic Principles of Coremaking. . . . . . . . Core Sands and Additives. . . . . . . . . . . . Compaction. . . . . . . . . . . . . . . . . . . . . . Curing of Compacted Cores . . . . . . . . . . Core Coatings (Core Washes) . . . . . . . . . Assembly and Core Setting . . . . . . . . . . . Design Considerations . . . . . . . . . . . . . . Coring of Tortuous Passages . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

497 497 499 500 501 501 502 503 504 505 506 506 506 507 508 511 511 512 513 513 516 517 519 519 520

. . . . . . . . . . . . . . 523 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

525 528 528 529 535 538 539 546 549 549 551 552 553 554 558 560 560 561 562 564 567 567 569 572 573 576 577 581 581 582 583 584 585 587 588 589 592

Green Sand versus Dry Sand Cores . . . . . Cores in Permanent Mold Castings . . . . . Cores in Investment Castings . . . . . . . . . Designing to Eliminate Cores . . . . . . . . . Coring versus Drilling . . . . . . . . . . . . . . Shell Molding and Shell Coremaking . . . . . . . Applicability . . . . . . . . . . . . . . . . . . . . . Properties and Selection of Shell Sands . . Resins . . . . . . . . . . . . . . . . . . . . . . . . . Preparation of Resin-Sand Mixtures . . . . . Sand Reclamation . . . . . . . . . . . . . . . . . Control Testing of Resin-Sand Properties . Patterns and Core Boxes. . . . . . . . . . . . . Production of Molds and Cores . . . . . . . . Mold Assembly . . . . . . . . . . . . . . . . . . . Casting Process . . . . . . . . . . . . . . . . . . . Dimensional Accuracy . . . . . . . . . . . . . . Causes and Prevention of Casting Defects Comparisons with Green Sand Molding . . Slurry Molding . . . . . . . . . . . . . . . . . . . . . . . Plaster Molding . . . . . . . . . . . . . . . . . . . Applications . . . . . . . . . . . . . . . . . . . Plaster Mold Compositions . . . . . . . . . Patterns and Core Boxes . . . . . . . . . . . Mold-Drying Equipment . . . . . . . . . . . Metals Cast in Plaster Molds . . . . . . . . Conventional Plaster Mold Casting. . . . Match Plate Pattern Molding . . . . . . . . The Antioch Process . . . . . . . . . . . . . . Foamed Plaster Molding Process . . . . . Ceramic Molding . . . . . . . . . . . . . . . . . . Applications . . . . . . . . . . . . . . . . . . . Shaw Process . . . . . . . . . . . . . . . . . . . Unicast Process . . . . . . . . . . . . . . . . . No-Bond Sand Molding. . . . . . . . . . . . . . . . . Magnetic Molding . . . . . . . . . . . . . . . . . Vacuum (V-Process) Molding . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Expandable-Mold Casting Processes with Expendable Patterns . . . . . . . . . . . . . . . . . . . John L. Jorstad, J.L.J. Technologies, Inc Introduction—Expendable Mold Processes with Expendable Patterns . . . . . . . . . . . . . . . . . Lost Foam Casting . . . . . . . . . . . . . . . . . . . . . Process Technique . . . . . . . . . . . . . . . . . . Processing Parameters . . . . . . . . . . . . . . . Advantages of Lost Foam Casting . . . . . . . Casting Quality . . . . . . . . . . . . . . . . . . . . Investment Casting . . . . . . . . . . . . . . . . . . . . . Pattern Materials . . . . . . . . . . . . . . . . . . . Patternmaking . . . . . . . . . . . . . . . . . . . . . Pattern and Cluster Assembly . . . . . . . . . . Ceramic Shell Mold Manufacture . . . . . . . Manufacture of Ceramic Cores . . . . . . . . . Pattern Removal . . . . . . . . . . . . . . . . . . . Mold Firing and Burnout . . . . . . . . . . . . . Melting and Casting. . . . . . . . . . . . . . . . . Postcasting Operations . . . . . . . . . . . . . . . Inspection and Testing . . . . . . . . . . . . . . . Design Advantages of Investment Castings . Design Recommendations . . . . . . . . . . . . . Applications . . . . . . . . . . . . . . . . . . . . . . Special Investment Casting Processes. . . . . Replicast Molding. . . . . . . . . . . . . . . . . . . . . . Process Details . . . . . . . . . . . . . . . . . . . . Process Capabilities . . . . . . . . . . . . . . . . .

xiv

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

593 593 594 595 596 598 599 599 600 601 603 603 604 604 607 610 611 614 615 617 617 617 617 618 618 618 619 620 620 621 622 622 622 624 628 628 628

. . . . . . . . . . . . . 635 . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

637 640 640 642 645 645 646 646 648 650 651 654 654 654 655 656 656 657 657 657 657 662 662 663

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

Centrifugal Casting . . . . . . . . . . . . . . . . . John L. Jorstad, J.L.J. Technologies, Inc Centrifugal Casting . . . . . . . . . . . . . . . . . . Centrifugal Casting Methods . . . . . . . . Equipment . . . . . . . . . . . . . . . . . . . . Molds. . . . . . . . . . . . . . . . . . . . . . . . Defects in Centrifugal Castings . . . . . . Applications . . . . . . . . . . . . . . . . . . . Horizontal Centrifugal Casting . . . . . . . . . . Equipment . . . . . . . . . . . . . . . . . . . . Casting Process . . . . . . . . . . . . . . . . . Applications . . . . . . . . . . . . . . . . . . . Vertical Centrifugal Casting. . . . . . . . . . . . Mold Design . . . . . . . . . . . . . . . . . . . Process Details . . . . . . . . . . . . . . . . .

www.asminternational.org

. . . . . . . . . . . . . . . . 665 . . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

Permanent Mold and Semipermanent Mold Processes . . . . . . . . . . . . . . . . . . . . . . . . John L. Jorstad, J.L.J. Technologies, Inc Permanent Mold Casting . . . . . . . . . . . . . . . . . . Gravity Casting Methods . . . . . . . . . . . . . . Advantages and Disadvantages . . . . . . . . . . Applications . . . . . . . . . . . . . . . . . . . . . . . Casting Design . . . . . . . . . . . . . . . . . . . . . Mold Design . . . . . . . . . . . . . . . . . . . . . . . Mold Materials . . . . . . . . . . . . . . . . . . . . . Mold Coatings. . . . . . . . . . . . . . . . . . . . . . Mold Life . . . . . . . . . . . . . . . . . . . . . . . . . Mold Temperature . . . . . . . . . . . . . . . . . . . Control of Mold Temperature . . . . . . . . . . . Dimensional Accuracy . . . . . . . . . . . . . . . . Surface Finish . . . . . . . . . . . . . . . . . . . . . . Low-Pressure Die Casting . . . . . . . . . . . . . . . . . Conventional Low-Pressure Casting . . . . . . . Counterpressure Casting . . . . . . . . . . . . . . . Vacuum Riserless/Pressure Riserless Casting Low Pressure Counter Gravity Casting . . . . . . . . Mold Filling . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

Die Casting Tooling . . . . . . . . . . . . . . . . . . . . . . . Die Casting Machines and Dies. . . . . . . . . . . . Product Design and the Process. . . . . . . . . . . . Tooling and Metal Flow . . . . . . . . . . . . . . . . . Heat Removal . . . . . . . . . . . . . . . . . . . . . . . . Materials Selection for Tooling . . . . . . . . . . . . Automation in High-Pressure Die Casting . . . . . . . . Die Casting Cycle . . . . . . . . . . . . . . . . . . . . . Automatic Pouring . . . . . . . . . . . . . . . . . . . . . Injection Process . . . . . . . . . . . . . . . . . . . . . . Solidification and Heat Removal . . . . . . . . . . . Casting Extraction and Insert Loading . . . . . . . Automatic Spray . . . . . . . . . . . . . . . . . . . . . . Finishing Automation . . . . . . . . . . . . . . . . . . . Die Casting Internal Quality Check Automation

667 667 669 670 670 671 674 674 675 678 680 680 684

. . . . . . . . . . . . 687 . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

High-Pressure Die Casting. . . . . . . . . . . . . . . . . . . . . . . . . John L. Jorstad, J.L.J. Technologies, Inc High-Pressure Die Casting . . . . . . . . . . . . . . . . . . . . . . . . . . Die Casting Alloys and Processes . . . . . . . . . . . . . . . . . Advantages of High-Pressure Die Casting. . . . . . . . . . . . Disadvantages of High-Pressure Die Casting. . . . . . . . . . Product Design for the Process . . . . . . . . . . . . . . . . . . . Metal Injection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hot Chamber Die Casting . . . . . . . . . . . . . . . . . . . . . . . . . . Melting Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Injection Components. . . . . . . . . . . . . . . . . . . . . . . . . . Distinctions Between Hot and Cold Chamber Processes . . Gate and Runner Design. . . . . . . . . . . . . . . . . . . . . . . . Temperature Control . . . . . . . . . . . . . . . . . . . . . . . . . . Ejection and Postprocessing . . . . . . . . . . . . . . . . . . . . . Cold Chamber Die Casting . . . . . . . . . . . . . . . . . . . . . . . . . Machine Components . . . . . . . . . . . . . . . . . . . . . . . . . . Process Parameters. . . . . . . . . . . . . . . . . . . . . . . . . . . . Squeeze Casting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Squeeze Casting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Casting and Tooling Design for Squeeze Casting . . . . . . Factors Affecting Squeeze-Cast Products . . . . . . . . . . . . Squeeze-Cast Applications . . . . . . . . . . . . . . . . . . . . . . Commonly Used Alloys, Properties, and Microstructures . Vacuum High-Pressure Die Casting . . . . . . . . . . . . . . . . . . . Conventional Die Casting . . . . . . . . . . . . . . . . . . . . . . . Vacuum Die Casting . . . . . . . . . . . . . . . . . . . . . . . . . . High-Vacuum Die Casting . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

689 690 690 691 691 692 693 695 697 697 698 698 699 700 700 704 706 709 709

. . . 713 . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . .

715 715 716 717 717 717 719 719 719 720 721 721 722 724 724 725 727 727 727 728 729 730 732 732 732 732

xv

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

Semisolid Casting . . . . . . . . . . . . . . . . . . . . . . . . . John L. Jorstad, J.L.J. Technologies, Inc Semisolid Casting — Introduction and Fundamentals . Advantages of SSM Processing . . . . . . . . . . . . . SSM Processing Routes . . . . . . . . . . . . . . . . . . Thixocasting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bar-Making Processes . . . . . . . . . . . . . . . . . . . Billet Sawing and Volume Control . . . . . . . . . . Reheating and Billet Handling. . . . . . . . . . . . . . Rheological Tests . . . . . . . . . . . . . . . . . . . . . . Basics of Semisolid Tooling Design. . . . . . . . . . Casting Process and Parameters. . . . . . . . . . . . . Rheocasting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thixomolding. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Process Description . . . . . . . . . . . . . . . . . . . . . Emerging Developments . . . . . . . . . . . . . . . . . . Mechanical Properties and Microstructure . . . . .

. . . . . . . . . . . . . . . .

Cast Irons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thomas E. Prucha, American Foundry Society Introduction to Cast Irons . . . . . . . . . . . . . . . . . . . Classification of Cast Irons . . . . . . . . . . . . . . . Composition of Cast Irons . . . . . . . . . . . . . . . Solidification of Cast Irons . . . . . . . . . . . . . . . Graphite Morphologies . . . . . . . . . . . . . . . . . . Matrix Structure Development. . . . . . . . . . . . . Physical Properties of Cast Irons . . . . . . . . . . . Thermal Properties. . . . . . . . . . . . . . . . . . . . . Conductive Properties. . . . . . . . . . . . . . . . . . . Magnetic Properties . . . . . . . . . . . . . . . . . . . . Acoustic Properties . . . . . . . . . . . . . . . . . . . . Heat Treatment . . . . . . . . . . . . . . . . . . . . . . . Welding of Cast Irons . . . . . . . . . . . . . . . . . . Machining and Grinding . . . . . . . . . . . . . . . . . Coatings . . . . . . . . . . . . . . . . . . . . . . . . . . . . Materials Selection . . . . . . . . . . . . . . . . . . . . Cast Iron Foundry Practices . . . . . . . . . . . . . . . . . . Cast Iron Melting Practice . . . . . . . . . . . . . . . Gray Iron Foundry Practice. . . . . . . . . . . . . . . Ductile Iron Foundry Practice . . . . . . . . . . . . . Compacted Graphite Iron Foundry Practice. . . . Malleable Iron Foundry Practice . . . . . . . . . . . Foundry Practice for High-Nickel Ductile Irons Foundry Practice High-Silicon Ductile Irons . . . Foundry Practice for High-Silicon Gray Irons . . Foundry Practice for High-Alloy White Irons . . Gray Iron Castings . . . . . . . . . . . . . . . . . . . . . . . . Classes of Gray Iron . . . . . . . . . . . . . . . . . . . Applications . . . . . . . . . . . . . . . . . . . . . . . . . Castability. . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

734 734 735 737 740 742 747 747 748 749 751 753 754 755 757

. . . . . . . . . 759 . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

761 761 762 764 764 766 767 768 769 770 773 773 777 777 778 780

. . . . . . . . . . 783 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

785 785 788 791 792 794 794 795 797 800 801 802 805 806 809 809 812 812 815 820 826 829 830 830 830 831 835 835 835 835

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

Microstructure . . . . . . . . . . . . . . . . . . . . . . . . . . Section Sensitivity . . . . . . . . . . . . . . . . . . . . . . . Prevailing Sections. . . . . . . . . . . . . . . . . . . . . . . Test Bar Properties . . . . . . . . . . . . . . . . . . . . . . Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fatigue Limit in Reversed Bending . . . . . . . . . . . Pressure Tightness . . . . . . . . . . . . . . . . . . . . . . . Impact Resistance . . . . . . . . . . . . . . . . . . . . . . . Machinability . . . . . . . . . . . . . . . . . . . . . . . . . . Wear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Resistance to Wear . . . . . . . . . . . . . . . . . . . . . . Elevated-Temperature Properties . . . . . . . . . . . . . Dimensional Stability . . . . . . . . . . . . . . . . . . . . . Effect of Shakeout Practice. . . . . . . . . . . . . . . . . Alloying to Modify As-Cast Properties. . . . . . . . . Heat Treatment . . . . . . . . . . . . . . . . . . . . . . . . . Physical Properties. . . . . . . . . . . . . . . . . . . . . . . Ductile Iron Castings . . . . . . . . . . . . . . . . . . . . . . . . Classes and Grades of Ductile Iron . . . . . . . . . . . Factors Affecting Mechanical Properties. . . . . . . . Hardness Properties . . . . . . . . . . . . . . . . . . . . . . Tensile Properties . . . . . . . . . . . . . . . . . . . . . . . Shear and Torsional Properties . . . . . . . . . . . . . . Compressive Properties . . . . . . . . . . . . . . . . . . . Fatigue Properties . . . . . . . . . . . . . . . . . . . . . . . Fracture Toughness . . . . . . . . . . . . . . . . . . . . . . Physical Properties. . . . . . . . . . . . . . . . . . . . . . . Austempered Ducfile Iron (ADI) . . . . . . . . . . . . . Compacted Graphite Iron Castings . . . . . . . . . . . . . . . Graphite Morphology . . . . . . . . . . . . . . . . . . . . . Chemical Composition . . . . . . . . . . . . . . . . . . . . Castability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mechanical Properties and Corrosion Resistance . . Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . Malleable Iron Castings. . . . . . . . . . . . . . . . . . . . . . . Melting Practices . . . . . . . . . . . . . . . . . . . . . . . . Microstructure . . . . . . . . . . . . . . . . . . . . . . . . . . Current Production Technologies . . . . . . . . . . . . . Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . Ferritic Malleable Iron . . . . . . . . . . . . . . . . . . . . Pearlitic and Martensitic Malleable Iron . . . . . . . . White Iron and High-Alloyed Iron Castings. . . . . . . . . Nickel-Chromium White Irons . . . . . . . . . . . . . . High-Chromium White Irons. . . . . . . . . . . . . . . . High-Alloy Graphitic Irons . . . . . . . . . . . . . . . . . . . . High-Silicon Irons for High-Temperature Service . Austenitic Nickel-Alloyed Gray and Ductile Irons . High-Silicon Irons for Corrosion Resistance . . . . .

www.asminternational.org

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Steel Castings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Raymond W. Monroe, Steel Founders’ Society of America; Thomas E. Prucha, American Foundry Society Steel Ingot Casting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pouring Method. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ingot Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hot Tops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ingot Molds and Stools . . . . . . . . . . . . . . . . . . . . . . . Ingot Solidification. . . . . . . . . . . . . . . . . . . . . . . . . . . Ingot Stripping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solidification Simulation. . . . . . . . . . . . . . . . . . . . . . . Steel Continuous Casting. . . . . . . . . . . . . . . . . . . . . . . . . . Historical Aspects of Continuous Casting of Steel . . . . . General Description of the Process . . . . . . . . . . . . . . . Plant Layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Near-Net Shape Casting . . . . . . . . . . . . . . . . . . . . . . . Tundish Metallurgy . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

836 837 838 839 840 841 843 844 846 846 848 848 849 849 850 851 851 853 ... 856 857 859 859 861 861 862 865 867 869 872 872 872 873 874 879 884 884 885 887 888 889 890 896 896 899 904 904 906 907

. . . . 909 . . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

911 911 912 913 913 914 916 916 918 918 918 919 919 920

Pouring Stream Protection . . . . . . . . . . . . . . . . . . Molds for Continuous Casting of Steel. . . . . . . . . . Productivity Improvements . . . . . . . . . . . . . . . . . . Quality Improvements . . . . . . . . . . . . . . . . . . . . . Horizontal Continuous Casting . . . . . . . . . . . . . . . Future Developments . . . . . . . . . . . . . . . . . . . . . . Shape Casting of Steel . . . . . . . . . . . . . . . . . . . . . . . . Castability of Steel. . . . . . . . . . . . . . . . . . . . . . . . Melting Furnaces . . . . . . . . . . . . . . . . . . . . . . . . . Metlting Practice . . . . . . . . . . . . . . . . . . . . . . . . . Pouring Practice . . . . . . . . . . . . . . . . . . . . . . . . . Foundry Factors in Design . . . . . . . . . . . . . . . . . . Design Factors. . . . . . . . . . . . . . . . . . . . . . . . . . . Avoidance of Hot Spots . . . . . . . . . . . . . . . . . . . . Thin-Wall Steel Castings . . . . . . . . . . . . . . . . . . . Tolerances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Size of Cores . . . . . . . . . . . . . . . . . . . . . . . . . . . Casting in Graphite Molds . . . . . . . . . . . . . . . . . . Steel Castings Properties . . . . . . . . . . . . . . . . . . . . . . . Carbon Steels . . . . . . . . . . . . . . . . . . . . . . . . . . . Low-Alloy Steels . . . . . . . . . . . . . . . . . . . . . . . . . Wear-Resistant Steels . . . . . . . . . . . . . . . . . . . . . . Corrosion-Resistant Steels. . . . . . . . . . . . . . . . . . . Heat-Resistant Steels . . . . . . . . . . . . . . . . . . . . . . Common Alloys . . . . . . . . . . . . . . . . . . . . . . . . . Carbon and Low-Alloy Mechanical Properties . . . . Mechanical Properties of Wear-Resisting Steels . . . Mechanical Properties of Corrosion-Resisting Steels Mechanical Properties of Heat-Resisting Steels . . . . Ferrite in Cast Stainless Steels . . . . . . . . . . . . . . . Carbon and Low-Alloy Heat Treatments . . . . . . . . High-Alloy Heat Treatment. . . . . . . . . . . . . . . . . . Corrosion-Resistant Applications . . . . . . . . . . . . . . Heat-Resistant Applications . . . . . . . . . . . . . . . . . Selection and Evaluation of Steel Castings . . . . . . . . . . Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . . Nondestructive Examination . . . . . . . . . . . . . . . . . Purchasing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Research and Development . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

921 922 923 924 924 924 926 926 928 930 933 934 936 938 943 946 947 948 949 949 950 950 951 952 953 953 960 961 966 968 970 971 972 973 975 975 980 982 986 986

Casting of Nonferrous Alloys . . . . . . . . . . . . . . . . . . . . . Nonferrous Casting — An Introduction . . . . . . . . . . . . . . . Nickel Alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Titanium Alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lead. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dross, Melt Loss, and Fluxing of Light Alloy Melts. . . . . . Dross Formation . . . . . . . . . . . . . . . . . . . . . . . . . . . Influence of Melter Type on Dross Generation . . . . . . Influence of Charge Materials on Melt Loss . . . . . . . . Influence of Operating Practices on Melt Loss . . . . . . Economic Implications of Dross . . . . . . . . . . . . . . . . In-Plant Enhancement or Recovery of Dross Metallics . Ways to Reduce Dross Formation . . . . . . . . . . . . . . . Aluminum Alloy Ingot Casting and Continuous Processes. . Ingot Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Molten-Metal Processing . . . . . . . . . . . . . . . . . . . . . Ingot Casting Processes . . . . . . . . . . . . . . . . . . . . . . Continuous Processes . . . . . . . . . . . . . . . . . . . . . . . . Solidification in the DC Process . . . . . . . . . . . . . . . . Postsolidification Processes . . . . . . . . . . . . . . . . . . . . Safety. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Aluminum Shape Casting . . . . . . . . . . . . . . . . . . . . . . . . Melting and Melt Handling . . . . . . . . . . . . . . . . . . . . Casting Process Selection . . . . . . . . . . . . . . . . . . . . . Pressure Die Casting of Aluminum . . . . . . . . . . . . . . Premium Engineered Castings . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . .

987 989 989 989 990 992 992 993 994 994 998 998 999 1001 1001 1001 1002 1004 1006 1007 1007 1009 1009 1010 1013 1016

xvi

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

Copper Continuous Casting . . . . . . . . . . . . . . History . . . . . . . . . . . . . . . . . . . . . . . . . Vertical Continuous Casting . . . . . . . . . . Horizontal Continuous Casting . . . . . . . . Strip Casting . . . . . . . . . . . . . . . . . . . . . Wheel Casting . . . . . . . . . . . . . . . . . . . . Ohno Continuous Casting Process . . . . . . Casting of Copper and Copper Alloys . . . . . . . Castability. . . . . . . . . . . . . . . . . . . . . . . Melting and Melt Control . . . . . . . . . . . . Fluxing of Copper Alloys . . . . . . . . . . . . Degassing of Copper Alloys . . . . . . . . . . Deoxidation of Copper Alloys . . . . . . . . . Grain Refining of Copper Alloys . . . . . . . Filtration of Copper Alloys . . . . . . . . . . . Melt Treatments for Group I to III Alloys Production of Copper Alloy Castings . . . . Casting Process Selection . . . . . . . . . . . . Gating . . . . . . . . . . . . . . . . . . . . . . . . . Feeding . . . . . . . . . . . . . . . . . . . . . . . . Casting of Zinc Alloys . . . . . . . . . . . . . . . . . Control of Alloy Composition . . . . . . . . . Melt Processing . . . . . . . . . . . . . . . . . . . Die Casting of Zinc Alloys . . . . . . . . . . . Other Casting Processes for Zinc Alloys. .

www.asminternational.org

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

Nonferrous Alloy Castings . . . . . . . . . . . . . . . . . . . . . . . Mahi Sahoo, CANMET Materials Technology Laboratory Aluminum and Aluminum Alloy Castings . . . . . . . . . . . . . Chemical Compositions . . . . . . . . . . . . . . . . . . . . . . Effects of Alloying and Impurity Elements . . . . . . . . . Structure Control . . . . . . . . . . . . . . . . . . . . . . . . . . . Modification and Refinement of Aluminum-Silicon Alloys . . . . . . . . . . . . . . . . . Foundry Alloys for Specific Casting Applications . . . . Heat Treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . Welding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Properties of Aluminum Casting Alloys . . . . . . . . . . . Copper and Copper Alloy Castings . . . . . . . . . . . . . . . . . . Alloying Additions. . . . . . . . . . . . . . . . . . . . . . . . . . Alloying Systems and Specifications . . . . . . . . . . . . . Physical Metallurgy . . . . . . . . . . . . . . . . . . . . . . . . . Types of Copper Castings. . . . . . . . . . . . . . . . . . . . . Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Heat Treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . Welding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cupronickels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zinc and Zinc Alloy Castings . . . . . . . . . . . . . . . . . . . . . Alloying Additions. . . . . . . . . . . . . . . . . . . . . . . . . . Physical Metallurgy . . . . . . . . . . . . . . . . . . . . . . . . . Die Casting and Aging . . . . . . . . . . . . . . . . . . . . . . . Specialty Alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . Finishing of Die Castings . . . . . . . . . . . . . . . . . . . . . Metalworking and Machining . . . . . . . . . . . . . . . . . . Joining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Applications for Zinc Die Castings . . . . . . . . . . . . . . Magnesium and Magnesium Alloys . . . . . . . . . . . . . . . . . Magnesium Alloys . . . . . . . . . . . . . . . . . . . . . . . . . . General Applications . . . . . . . . . . . . . . . . . . . . . . . . Melting Furnaces and Auxiliary Pouring Equipment . . Melting Procedures and Process Parameters . . . . . . . . Sand Casting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Investment Casting. . . . . . . . . . . . . . . . . . . . . . . . . . Permanent Mold Casting . . . . . . . . . . . . . . . . . . . . . Die Casting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

1019 1019 1020 1022 1023 1024 1024 1026 1026 1027 1029 1032 1035 1037 1037 1038 1039 1040 1041 1044 1049 1049 1049 1051 1054

. . . . 1057 . . . .

. . . .

. . . .

. . . .

1059 1059 1059 1063

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1066 1069 1071 1077 1077 1085 1085 1087 1088 1091 1091 1091 1091 1091 1095 1095 1096 1097 1097 1098 1098 1099 1097 1099 1100 1102 1102 1103 1106 1109 1109 1109

Direct Chill Casting . . . . . . . . . . . . . . . . . . . . . High-Temperature Alloys . . . . . . . . . . . . . . . . . Welding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cobalt and Cobalt Alloy Castings. . . . . . . . . . . . . . . Physical Metallurgy . . . . . . . . . . . . . . . . . . . . . Foundry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Postcasting Treatment . . . . . . . . . . . . . . . . . . . Applications . . . . . . . . . . . . . . . . . . . . . . . . . . Nickel and Nickel Alloy Castings. . . . . . . . . . . . . . . Compositions . . . . . . . . . . . . . . . . . . . . . . . . . Structure and Property Correlations . . . . . . . . . . Melting Practice and Metal Treatment . . . . . . . . Foundry Practice . . . . . . . . . . . . . . . . . . . . . . . Pouring Practice . . . . . . . . . . . . . . . . . . . . . . . Gating Systems . . . . . . . . . . . . . . . . . . . . . . . . Risers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Welding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Heat Treatment . . . . . . . . . . . . . . . . . . . . . . . . Specific Applications . . . . . . . . . . . . . . . . . . . . Titanium and Titanium Alloy Castings . . . . . . . . . . . Historical Perspective of Casting Technology . . . Physical Metallurgy . . . . . . . . . . . . . . . . . . . . . Alloy Ti-6Al-4V . . . . . . . . . . . . . . . . . . . . . . . Casting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Postcasting Practice . . . . . . . . . . . . . . . . . . . . . Product Applications . . . . . . . . . . . . . . . . . . . . Zirconium and Zirconium Alloy Castings . . . . . . . . . Zirconium Reactivity Considerations . . . . . . . . . Melting Processes . . . . . . . . . . . . . . . . . . . . . . Casting Processes. . . . . . . . . . . . . . . . . . . . . . . Processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . Physical Properties of Zirconium Castings . . . . . Impact Properties of Zirconium Castings . . . . . . Synthesis and Processing of Cast Metal-Matrix Composites and Their Applications . . . . . . . . . . Classification of MMCs . . . . . . . . . . . . . . . . . . Solidification Processing of MMCs and Possible Effects of Reinforcement on Solidification. . Property Motivation for Using MMCs . . . . . . . . Components of MMCs Currently in Use. . . . . . . Development of Select Cast MMCs . . . . . . . . . . Research Imperatives . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1112 1112 1112 1114 1114 1116 1117 1117 1119 1119 1121 1123 1124 1124 1124 1125 1125 1125 1126 1128 1129 1129 1132 1134 1137 1138 1143 1143 1144 1144 1145 1147 1147

. . . . . . . . 1149 . . . . . . . . 1149 . . . . .

. . . . .

. . . . .

Quality Assurance, Nondestructive Evaluation, and Failure Analysis. . . . . . . . . . . . . . . . . . . . . . . Babu DasGupta, National Science Foundation Approaches to Measurement of Metal Quality. . . . . . Metal Cleanliness Assessment Techniques . . . . . . . . Nondestructive Testing of Components . . . . . . . . . . . . . . Liquid Penetrant Inspection. . . . . . . . . . . . . . . . . . . Radiographic Inspection . . . . . . . . . . . . . . . . . . . . . Fluoroscopic Inspection and Automated Defect Recognition . . . . . . . . . . . . . . . . . . . . . Ultrasonic Inspection . . . . . . . . . . . . . . . . . . . . . . . Eddy Current Inspection . . . . . . . . . . . . . . . . . . . . . Process-Controlled Resonant Testing (Contributed by Quasar International, Inc.) . . . . . . . . . . . . . . . . Leak Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electrical Conductivity Measurement . . . . . . . . . . . . Casting Fracture Characteristics . . . . . . . . . . . . . . . . . . . General Fracture Examination . . . . . . . . . . . . . . . . . Fracture Features . . . . . . . . . . . . . . . . . . . . . . . . . . Casting Failure Analysis Techniques and Case Studies . . . Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Background Information . . . . . . . . . . . . . . . . . . . . .

xvii

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

1150 1155 1156 1160 1162

. . . . . 1165 . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

1167 1167 1174 1174 1174

. . . . . 1174 . . . . . 1174 . . . . . 1175 . . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

1175 1177 1177 1178 1178 1179 1183 1183 1183

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

Planning and Sample Selection . . . . . . . . . . . . . . . . Preliminary Examination . . . . . . . . . . . . . . . . . . . . Material Evaluation . . . . . . . . . . . . . . . . . . . . . . . . Data Analysis and Report Preparation . . . . . . . . . . . Case Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Common Defects in Various Casting Processes . . . . . . . . Case Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Repair Welding of Castings . . . . . . . . . . . . . . . . . . . . . . Repair Welding of Ferrous Castings. . . . . . . . . . . . . Repair Welding of Nonferrous Castings . . . . . . . . . . Quality Planning Tools and Procedures . . . . . . . . . . . . . . Advanced Product Quality Planning (APQP). . . . . . . Maintenance, Repair, Alterations, and Storage of Patterns and Tooling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

www.asminternational.org

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

1184 1184 1184 1187 1187 1192 1199 1203 1203 1205 1207 1207

. . . . . 1213

Ownership . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Maintenance Responsibility. . . . . . . . . . . . . . . . . . . . . Common Failure Mechanisms . . . . . . . . . . . . . . . . . . . Maintenance Interval . . . . . . . . . . . . . . . . . . . . . . . . . Maintenance Records . . . . . . . . . . . . . . . . . . . . . . . . . Preventive Maintenance Program. . . . . . . . . . . . . . . . . Repair . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Alterations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Storage—Physical Protection, Cataloging, and Tracking. Reference Information. . . Metric Conversion Guide. . Abbrieviations & Symbols . Index. . . . . . . . . . . . . . . .

xviii

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

1213 1213 1213 1213 1214 1214 1214 1214 1215

. . . .

. . . .

. . . .

1217 1219 1222 1225

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 3-15 DOI: 10.1361/asmhba0005186

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

History and Trends of Metal Casting Thomas E. Prucha, American Foundry Society Daniel Twarog, North American Die Casting Association Raymond W. Monroe, Steel Founders’ Society of America

CASTING is one of the oldest manufacturing methods known to humankind and a very direct method of producing metal parts. It involves pouring molten metal into a cavity that is close to the final dimensions of the finished form, and it is used to manufacture many types of complex components of any metal, ranging in weight from less than an ounce to single parts weighing several hundred tons. Typical uses of cast components include infrastructure and structural components, water distribution systems (pipes, pumps, and valves), automotive Table 1

components (engine blocks, brakes, steering and suspension components, etc.), prosthetics, jewelry, and gas turbine engine hardware. In one study (Ref 1), castings were estimated to be used in 90% or more of all manufactured goods and in all capital goods machinery used in manufacturing. Some of the major market areas of castings are listed in Table 1 (Ref 2). Economically, casting processes are capable of producing highly reliable, cost-effective components ranging from low-volume, singlepart prototype production runs to economies

of scale for millions of parts. In terms of component design, casting also permits the formation of streamlined, intricate, integral parts with varied shape and size capabilities for designers. Metal casting is very flexible in terms of configuration design, and if a pattern can be made for a part, it can be cast. For example, the flexibility of metal casting, particularly sand molding, may permit the use of difficult design techniques, such as undercuts and curved or reflex contours that are not possible with other high-production processes. Tapered

      

 Other major markets include machine tools, construction

Major markets for metal castings

Ferrous castings Gray iron  Ingot molds  Construction castings  Motor vehicles  Farm equipment  Engines  Refrigeration and heating  Construction machinery  Valves  Soil pipe  Pumps and compressors  Pressure pipe  Other major markets include machine tools, mechanical power transmission equipment, hardware, home appliances, mining machinery, and oil and natural gas pumping and processing equipment(a) Malleable iron  Motor vehicles  Valves and fittings  Construction machinery  Railroad equipment  Engines  Mining equipment  Hardware  Other major markets include heating and refrigeration, motors and generators, fasteners, ordnance, chains, machine tools, and general industrial machinery(a) Ductile iron  Pressure pipe  Motor vehicles

Farm machinery Engines Pumps and compressors Valves and fittings Metalworking machinery Construction machinery Other major markets include textile machinery, woodworking and paper machinery, mechanical power and transmission equipment, motors and generators, refrigeration and heating equipment, and air conditioning(a)

Steel  Railroad equipment  Construction equipment  Mining machinery  Valves and fittings  General and special industrial machinery  Motor vehicles  Metalworking machinery  Other major markets include steel manufacturing, spring goods, heating and air conditioning, recreation equipment, industrial materials handling equipment, ships and boats, and aircraft and aerospace(a) Nonferrous castings Aluminum  Auto and light truck  Aircraft and aerospace  Other transportation  Engines  Household appliances  Office machinery  Power tools  Refrigeration, heating, and air conditioning

equipment, mining equipment, farm machinery, electronic and communication equipment, power systems, and motors and generators(a) Magnesium Power tools Sporting goods Anodes Automotive Other major markets include office machinery, health care, and aircraft and aerospace(a)

    

Copper base Valves and fittings Plumbing brass goods Electrical equipment Pumps and compressors Power transmission equipment General machinery Transportation equipment Other major markets include chemical processing, utilities, desalination, and petroleum refining(a)

       

Zinc Automotive Building hardware Electrical components Machinery Household appliances Other major markets include scientific instruments, medical equipment, radio and television equipment, audio components, toys, and sporting goods(a)

     

(a)Ranked in approximate order of tonnage shipped, but in some cases, the total of other major markets is larger as a whole than the individual markets listed. Source: Ref 2

4 / Applications and Specification Guidelines sections with thickened areas for bosses and generous fillets are routine. The inherent design freedom of metal casting also allows the designer to combine what would otherwise be several parts of a fabrication into a single piece. This is significant when exact alignment must be held, as in high-speed machinery, machine tool parts, or engine end plates and housings that carry shafts. Various surface finishes on a part are also possible, ranging from a rougher as-cast surface of sand-molded casting to the smoother surfaces obtained through shell molding, investment casting, or other casting methods. In general, casting can provide various functional advantages for component design that include:  Design both internal and external contours

independently to almost any requirement

 Place metal in exact locations where it is

needed for rigidity, wear, corrosion, or maximum endurance under dynamic stress  Produce a complex part as a single, dependable unit  Readily achieve an attractive appearance This article briefly introduces the history of metal casting, general capabilities and methods of casting, and industry trends. In terms of value and volume, metal casting ranks second

only to steel rolling in the metal-producing industry. Metal casting is one of the ten largest industries when rated on a value-added basis, and worldwide casting shipments have continued to increase in recent years. Production statistics for 2005 are summarized in Table 2 (Ref 3), with comparison to 2004 production statistics from Modern Casting’s “Annual Census of World Casting Production.”

History of Metal Casting The casting of metal is a prehistoric technology. Exactly when the casting of metals began is not known, but it appears relatively late in the archaeological record. Archaeologists give the name Chalcolithic to the period in which metals were first being mastered, and they date this period, which immediately preceded the Bronze Age, very approximately to between 5000 and 3000 B.C. (Table 3, Ref 4). There were many earlier fire-using technologies, collectively called pyrotechnology, that provided a basis for the development of metal casting. Among these were the heat treatment of stone to make it more workable, the burning of lime to make plaster, and the firing of clay to produce ceramics. At first, it did not include

smelting, because the metal of the earliest castings appears to have been native. Cast iron appeared in China in approximately 600 B.C. Its use was not limited to strictly practical applications, and there are many examples of Chinese cast iron statuary. Most Chinese cast irons were unusually high in phosphorus and, because coal was often used in smelting, high in sulfur as well. These irons therefore have melting points that are similar to those of bronze, and when molten are unusually fluid. The iron castings, like the Chinese cast bronzes, are often remarked upon for the thinness of their wall sections. From these early beginnings, development and application of metal casting continued (Table 4, Ref 5). During the wars of medieval times, metalcasters produced cannons. In peace, they recast the metal into bells. In the Middle Ages, church bells were cast by priests, abbots, or bishops who also were trained metal founders. As the metal was melted, the brethren stood around the furnace intoning psalms and prayers. The molten metal was then blessed, and divine protection was asked for the bell, which usually bore the name of a saint (Ref 5). During the 1800s, there were significant advances in technology. Metallurgy also began

Table 2 World casting production for 2004–2005 40th census of world casting production for 2005, metric tons Country

Austria Belgium Brazil Canada China Croatia Czech Republic Denmark Finland France Germany Great Britain Hungary India Italy Japan Korea Lithuania Mexico Netherlands(a) Norway Poland Portugal Romania Russia Slovenia(a) Slovakia South Africa Spain Sweden Switzerland Taiwan Thailand Turkey Ukraine (i) United States Totals

Gray iron

47,500 70,700(a) 2,460,000(b)(d) 483,000(a)(b)(d) 12,303,963 25,306 269,493 46,932 47,817 896,500 2,469,058 531,000 35,307 4,116,000 923,700 2,782,509 960,100 14,300 600,000 63,100 16,348 420,700 27,285 81,578 3,480,000 75,900 41,520 160,000 491,600 171,700(b) 26,400 827,932 170,000 567,000 626,610 4,457,905 40,788,763

Ductile iron

131,000 15,600(a)(b) ... ... 5,838,753 16,252 49,524 40,333 60,123 957,700 1,487,234 362,000 32,200 618,000 512,200 1,919,435 565,200 200 270,000 78,241 49,207 112,500 69,792 10,500 720,000 27,800(b) 8,760 65,000 551,600 83,300 40,400 220,413 70,000 327,000 40,000 4,241,088 19,591,355

Malleable iron

17,500 ... ... ... 514,129 50 5,435 0 0 1,400 52,713 12,000 10 56,000 8,400 57,851 46,500 ... 2,000 6,209 0 21,740 0 926 280,000 ... ... 10,000 17,400 ... ... ... 30,000 8,000 10,000 75,296 1,233,559

Steel

18,000 42,100 293,891 117,600(a) 3,224,374 2,219 109,680 0 18,532 113,700 200,409 112,000 5,385 805,000 73,324 276,589 149,600 30 75,000 438 3,723 59,560 12,096 37,900 1,200,000 22,700 4,300 136,000 83,200 23,400 ... 73,619 30,000 125,000 266,060 1,287,295 9,002,724

Copper base

5,459 ... ... 18,585(a) 416,097 1,120 1,724 1,126 3,999 26,264 84,386 15,000 2,110 0 16,800 97,794 23,200 5 180,000 ... 4,667 6,300 8,700 2,700 160,000 6,727(f) 2,160 14,900 7,886 11,200 2,689 41,959 28,600 16,000 11,000 292,113 1,511,270

Aluminum

103,347 19,428(c) 240,457(e) 324,973 1,886,000(c) 15,038 80,252(c) 4,000 6,565 318,445(c) 727,139 206,000(c) 75,720 516,000 857,500 1,477,349(c) 144,300 56 660,000 ... 22,926 174,300 20,250 20,000 920,000 30,183 26,260 56,000 139,312(c) 44,400 17,785 250,366 100,000 66,500 20,500 2,080,174 11,718,025

Magnesium

6,580 ... ... ... ... 0 ... 0 0 ... 27,283 ... 3,050 0 12,000 ... 10,300(f) ... 500 ... 0 30 150 10 70,000 ... ... 0 ... 1,700 ... 6,427 ... 500 ... 100,697 239,227

Zinc

12,997 ... ... ... 237,889 502 2,656 1 434 25,066 56,439 21,000 3,050 0 70,000 36,216 ... ... 350 ... 0 8,600 900 700 20,000 ... 1,800 2,600 16,051 5,900 1,616 69,316 16,900 12,700 ... 312,978 936,661

Other nonferrous

... ... ... ... ... 1,300 279 5.022 0 3,461 3,346 2,000 0 0 67,600 8,102 0 ... ... ... ... 770 120 5 50,000 ... 5 0 433 ... ... 4,418 ... ... ... 48,987 195,848

Total

342,383 124,507(a) 2,968,600 863,373(a) 24,421,205 61,787 519,043 97,414 137,470 2,342,536 5,108,007 1,261,000 156,832 6,111,000 2,541,524 6,655,845 1,899,200 14,591 1,787,850 147,988(g) 96,871 804,500 139,293 154,319 7,620,000(h) 163,310 84,805 444,500 1,307,482 341,600 88,890 1,494,450 445,500 1,122,700 974,170 12,896,533 85,741,078

Totals for 2004

325,205 124,507 2,829,916 863,373 22,420,452 56,449 522,400 95,949 133,325 2,465,617 4,984,473 1,273,000 164,892 4,623,000 2,441,282 6,386,449 1,857,300 18,715 2,185,200 147,988 86,757 804,500 122,750 207,023 6,300,000 163,310 4,800 444,500 1,309,249 321,100 104,077 1,451,768 235,850 982,000 974,170 12,314,121 79,745,467

Note: “. . .” indicates unreported data. (a) 2004 tonnage. (b) Includes malleable iron. (c) Includes magnesium. (d) Includes ductile iron. (e) Includes all nonferrous. (f) Includes zinc. (g) Includes only ferrous. (h) Includes 720,000 tons of special iron. (i) 2002 tonnage. Source: Modern Casting

History and Trends of Metal Casting / 5 Table 3

Chronological list of developments in the use of materials

Date

Development

9000 B.C. 6500 B.C. 5000–3000 B.C.

3000–1500 B.C.

3000–2500 B.C. 2500 B.C. 2400–2200 B.C. 2000 B.C. 1500 B.C.

Earliest metal objects of wrought native copper Earliest life-size statues, of plaster Chalcolithic period: melting of copper; experimentation with smelting Bronze Age: arsenical copper and tin bronze alloys Lost-wax casting of small objects Granulation of gold and silver and their alloys Copper statue of Pharoah Pepi I Bronze Age Iron Age (wrought iron)

Location

Near East Jordan Near East

Near East

Near East Near East Egypt Far East Near East

Date

700–600 B.C. 600 B.C. 224 B.C.

Development

Etruscan dust granulation Cast iron Colossus of Rhodes destroyed 200–300 A.D. Use of mercury in gilding (amalgam gilding) 1200–1450 A.D. Introduction of cast iron (exact date and place unknown) Circa 1122 A.D. Theophilus’s On Divers Arts, the first monograph on metalworking written by a craftsman 1252 A.D. Diabutsu (Great Buddha) cast at Kamakura Circa 1400 A.D. Great Bell of Beijing cast 16th century Sand introduced as mold material

Location

Italy China Greece

Date

Development

1709

Cast iron produced with coke as fuel, Coalbrookdale Boring mill or cannon developed Great Bell of the Kremlin cast Cast steel developed by Benjamin Huntsman Cast iron used as architectural material, Ironbridge Gorge Zinc statuary Electrodeposition of copper

1715 Roman world Europe

1735 1740

Germany

1779

Japan

1826 1838

China France

1884

Electrolytic refining of aluminum

Location

England

Switzerland Russia England England

France Russia, England United States, France

Source: Ref 4

Table 4

Timeline of casting developments to 1899

B.C. 9000 B.C.—Earliest metal objects of wrought native copper are produced in the Near East. 3200 B.C.—The oldest casting in existence, a copper frog, is cast in Mesopotamia. 3000 B.C.—Early metalcasters cast bronze tools and weapons in permanent stone molds. 3000–2500 B.C.—Small objects are cast via lost-wax (investment casting) process in the Near East. 1500 B.C.—Wrought iron is discovered in the Near East. 600 B.C.—The first cast iron object, a 270 kg (600 lb) tripod, is cast by the Chinese. 233 B.C—Iron plowshares are cast. 200 B.C.—Oldest iron castings still in existence are produced during the Han Dynasty. A.D. 500—Cast crucible steel is produced in India. 1200s—Loam or sweep molding is used by European metalcasters to cast bells for cathedrals. 1252—The colossal statue, the Great Buddha at Kamakura, Japan, is cast in high-lead tin bronze. The project began in the 700s, and its head alone weighed 140 tons. 1313—The first cannon is cast in bronze by a monk in the city of Ghent. 1400s—During the siege of Constantinople, heavy guns are cast from bronze “on the spot,” virtually under the walls of the besieged city. Movable, cast lead type for printing presses revolutionizes the world’s methods of communication. 1480—Vannoccio Biringuccio (1480–1539), the first true metalcaster and the “father of the metal casting industry,” is born. The founder of the Vatican, his De La Pirotechnia is the first written account of proper metal casting practice. 1500s—Sand is introduced as a molding material in France. 1600s 1612—Mined from under the sea, seacoal is mentioned for the first time by German metalcaster and inventor Simon Sturtevant. 1619—North America’s first iron furnace is built at Falling Creek, Va., on a branch of the James River, 100 km (60 miles) from the Jamestown colony. Three years later, Native Americans destroy it during a raid. 1645—Earliest recorded use of the term foundry appears in the Oxford English Dictionary in its variant “founderie.” 1646—America’s first iron metal casting facility (and second industrial plant), Saugus Iron Works, near Boston, pours the first American metal casting, the Saugus pot. The Saugus River site was selected by Richard Leader and was built to produce iron products for Massachusetts and England. 1661—First U.S. copper deposits are discovered by Gov. Winthrop in Middletown, Conn. Source: Ref 5; see also Selected References

1700s 1709—Two important developments by Abraham Darby from Coalbrookdale, England, improve casting methods. He developed the first true metal casting facility flask to modernize molding practices (which had been carried out in pits on the floor by use of pattern boards tied together or in crude box frames). He would later initiate the use of coke as a furnace fuel for iron production. 1722—A.F. de Reamur, recognized as the world’s first metallurgical chemist, develops malleable iron, known today as European whiteheart malleable. 1750—Benjamin Huntsman reinvents the cast crucible steel process in England, a process that disappeared after first being developed in India. The English parliament prohibits the refining of pig iron or the casting of iron in the American colonies, contributing to the start of the American Revolution. 1756—Coalbrookdale’s Richard Reynolds oversees the invention of cast iron tram-road rails, replacing wooden rails. 1775—Revolutionary patriot Paul Revere, who operated a bell-and-fittings foundry in Boston, rides from Boston to Lexington warning colonists of the British invasion. 1776—Metalcasters Charles Carroll, James Smith, George Taylor, James Wilson, George Ross, Philip Livingston, and Stephen Hopkins sign the American Declaration of Independence. 1779—First iron bridge ever erected (above England’s Severn River) was cast and constructed at Coalbrookdale Works. 1794—John Wilkinson of England invents the first metalclad cupola furnace, using a steam engine to provide the air blast. 1797—First cast plow in the United States is invented by Charles Newbold of Sauk, N.J. 1800s 1809—Centrifugal casting is developed by A.G. Eckhardt of Soho, England. 1815—The cupola furnace is introduced to the United States (Baltimore). 1817—The first iron water line in the United States, 120 m (400 ft) long, is laid in Philadelphia. 1818—The first U.S. cast steel is produced by the crucible process at historic Valley Forge Foundry. 1825—Aluminum, the most abundant metal in the Earth’s crust, is isolated from aluminum chloride by Denmark’s Hans Oerstad. 1830s—Seth Boyden of Newark, N.J., produces the first blackheart malleable iron in the United States. 1832—Nickel-bronze is produced commercially in England. 1837—The first reliable molding machine on the market is made and used by the S. Jarvis Adams Co., Pittsburgh. 1847—Cast steel guns are made by Krupp Works in Germany. Asa Whitney, Philadelphia, obtains a patent on

a process for annealing chilled-iron car wheels cast with chilled tread and flange. 1847—John Deere commissions Jones and Quiggs Steel Works, Pittsburgh, to cast and roll a steel plow, which it accomplishes at one-half the previous cost of the product. 1849—A manually operated die casting machine is patented to supply rapidly cast lead type for newspapers. 1850—Drop-bottom cupola is developed. 1863—Metallography is developed by Henry C. Sorby of Sheffield, England, enabling metalcasters to polish, etch, and microscopically examine metal surfaces for physical analysis. 1867—James Nasmythe, inventor of the steam hammer, develops a gear-tilted safety ladle to prevent pouring accidents. 1870—Sandblasting is developed for large castings by R.E. Tilghman of Philadelphia. 1874—The Colliau cupola, the first commercially made cupola in America, is introduced. 1876—The first authenticated aluminum castings were produced by Col. William Frishmuth at his Philadelphia foundry. Assembled to produce an engineer’s transit, these castings were made from $1/oz chemically produced aluminum. Manganese-bronze patent is granted to Parsons in England. Tobin Bronze begins developing manganesebronze in the United States. 1880–1887—W.W. Sly, Cleveland, develops the first casting cleaning mill, greatly reducing hand-chipping and grinding to allow a custom-finished product. 1884—The first architectural application of aluminum, a cast aluminum pyramid that was produced by Frishmuth, is mounted on the tip of the Washington Monument. 1886—Charles M. Hall, a 22 year old student at Oberlin College, discovers a process of aluminum reduction through electrolysis. The invention replaced chemical reduction and lowered the metal cost (from $15/lb in 1884 to $0.50/lb in 1890), spurring a new industry of aluminum applications. 1887—Eli Millett invents a core oven for drying small cores in individual drawers. 1890—The first motor-driven mold conveyor is installed, integrating molding, pouring, and cooling operations. 1897—Iowa dentist B.F. Philbrook adapts the lost-wax investment casting process for producing dental inlays, the first nonart application of the process in the modern metal casting age. 1898—Poulson and Hargraves (U.K.) produce the first sand molds bonded with sodium silicate. Germany’s Imperial Navy recommends copper-nickel alloys containing 4–45% Ni for saltwater piping system. 1899—Electric arc furnace, developed by France’s Paul Heroult, begins commercial production.

6 / Applications and Specification Guidelines to achieve prominence in 1889 when nickel was alloyed to make steel stronger. In terms of casting, centrifugal casting was developed by A.G. Eckhardt of Soho, England, in 1809. This method involves the pouring of molten metal into a rapidly rotating metallic mold. The method was soon adopted by the pipe foundries and was first used in Baltimore, Maryland, in 1848. Sir Henry Bessemer, famed for his converter, used centrifugal casting to remove gases and was the first to pour two or more metals into a single rotating mold. The centrifugal casting of steel was first attempted in 1898 at the plant of the American Steel Foundries in St. Louis, Missouri. Railroad car wheels were spun cast in 1901 at a rotation speed of 620 rpm. Following the early development of the centrifugal method, a permanent mold method known as slush casting was introduced. Slush casting is a process in which molten metal is poured into a split metal mold (generally made of bronze) until the mold is filled; then, the mold is immediately inverted, and the metal that is still liquid is allowed to run out. The time required for this casting operation is sufficient to freeze a metal shell in the mold, corresponding to the shape of the cavity wall. The thickness of the wall of the casting depends on the time interval between the filling and the inverting of the mold, as well as on the chemical and physical properties of the alloy and the temperature and composition of the mold. Usually, lead and zinc alloy castings are

Table 5

produced by slush casting. The process is limited to the production of hollow castings and was used to produce lamp bases (Ref 6). Manually operated die casting machines were patented as early as 1849 (Sturgiss) and 1852 (Barr) in an effort to satisfy the insatiable demands of a growing reading public by way of rapidly cast lead type. These early inventions led to Ottmar Mergenthaler’s Linotype, an automatic casting machine in which molten lead is forced by piston stroke into a metal mold. The first die casting machine bearing the Linotype name was patented in 1905 by H.H. Doehler. Two years later came E.B. Wagner’s casting machine, a prototype of the now familiar hot chamber die casting machine. It was first used on a large scale during World War I for binocular and gas mask parts. Zinc alloys were used for die casting as early as 1907 but were not competitive until Price & Anderson developed the Zamak die casting alloy in 1929. In 1851, James Bogardus’s factory in Chicago, which was constructed with cast iron supports, opened the way for what many art historians considered to be America’s only original contribution to the arts of the world— the skyscraper. One of the oldest casting techniques, investment (lost-wax) casting, was also rediscovered in 1897 by B.F. Philbrook of Iowa, who used it to cast dental inlays. Industry paid little attention to this sophisticated process until the urgent military demands of World War I overtaxed the machine tool industry. Shortcuts were then needed to provide finished tools and

precision parts, avoiding time-consuming machining, welding, and assembly. Other efforts at this time included the first fully automated plant in the United States (one of the first in the world), which was a Rockford, Illinois, foundry that cast hand grenades for the U.S. Army in 1918. The first separation of aluminum occurred in 1825, but important engineering applications did not occur until the cost of aluminum steadily declined as a result of smelting process improvements and scale of operations through the end of the 19th century. Aside from additions for aluminum-bronze, zinc-aluminum alloys, and the deoxidation of steel, castings were the first important commercial market for aluminum. The process was simple, low cost, and addressed the limited range of product interests. At first, applications were limited to curiosities such as house numbers, hand mirrors, combs, brushes, tie clasps, cuff links, hat pins, and decorative lamp housings that emphasized the low density, silvery finish, and novelty of the new metal. Lightweight, corrosion-resistant and heat-conductive aluminum cookware could be produced using the same patterns used to cast iron and brass pots, pans, and kettles (Ref 7).

The 20th Century Casting technology during the 20th century advanced in many ways (Table 5), with “more casting progress since World War II than in the previous 3000 years” (Ref 8). For more than

Timeline of casting developments in the 20th century

1900s 1900—Brinnell hardness test machines introduced. Aluminum-bronze in regular production in the United States. Early 1900s—First patent for low-pressure permanent mold casting process issued to England’s E.H. Lake. 1901—American Steel Foundries (St. Louis) produces the first centrifugal cast rail wheels. 1903—The Wright Brothers’ first successful machinepowered aircraft contains a cast aluminum block and crankcase (together weighing 69 kg, or 152 lb), produced either at Miami Brass Foundry or the Buckeye Iron and Brass Works. 1905—H.H. Doehler patents the die casting machine. 1906—The first electric arc furnace is installed in the United States at Halcomb Steel Co., Syracuse, N.Y. First lowfrequency induction furnace is installed at Henry Diston & Sons, Tacony, Pa. 1907—Alfred Wilm discovers that the properties of cast aluminum alloys can be enhanced through heat treating and artificial aging. 1908—Stockham Homogenous Sand Mixer Co., Piqua and Newark, Ohio, releases the sand cutter. 1910s 1910—Matchplates are developed, fostering the viability of jolt-squeeze machines. 1911—Metallurgical microscope is obtainable. First electric arc furnace for metal casting is installed at Treadwell Engineering Co., Easton, Pa. 1912—The first muller with individually mounted revolving mullers of varying weights is marketed by Peter L. Simpson. Sand slinger is invented by E.O. Beardsley & W.F. Piper, Oregon Works.

1915—Experimentation begins with bentonite, a colloidal clay of unusually high green and dry strength. Ajax Metal Co., Philadelphia, installs first low-frequency induction furnace for nonferrous melting. 1916—Dr. Edwin Northrup, Princeton University, invents the coreless induction furnace. 1917—Alcoa completes a great deal of early development work in aluminum as World War I generates a great demand for high-integrity castings for aircraft engines. 1918—The first fully automated metal casting facility in Rockford, Ill., casts hand grenade hulks for the U.S. Army. 1920s 1921—Modification of the silicon structure in aluminum begins as Pacz discovers that adding metallic sodium to molten aluminum just prior to pouring greatly improves ductility. Copper-silicon alloys are prepared in Germany as a substitute for tin-bronzes. 1924—Henry Ford sets a production record of 1 million autos in 132 working days. Automotive manufacturing will grow to consume one-third of casting demand in the United States. 1925—X-ray radiography is established as a tool for checking casting quality. By 1940, all military aircraft castings require x-ray inspection prior to acceptance. American Brass, Waterbury, Conn., installs first mediumfrequency induction furnace in the United States. 1928—Alcoa develops the first aluminum vehicle wheel, a sand-cast 355 alloy designed for truck trailers. 1930s 1930—First high-frequency coreless electric induction furnace is installed at Lebanon Steel Foundry, Lebanon, Pa. Spectrography is pioneered by University of Michigan (continued)

Source: Ref 5; see also Selected References

professors for metal analysis. Davenport and Bain develop the austempering process for iron castings. 1937—Applied Research Laboratories’ founder Maurice Hasler produces the first grating spectrograph for the Geological Survey of California. Spectrometers begin finding their way into foundries by the late 1940s, replacing the previous practice of metallurgists estimating chemical compositions with a spectroscope and welder’s arc. The austempered microstructure in cast iron is recognized. 1940s 1940—Chvorinov develops the relationship between solidification time and casting geometry. Early 1940s—Statistical process control is first employed as a quality control tool in U.S. machine shops, principally to control dimensional tolerances. Inoculation of gray iron becomes common as high-quality cast irons replace scarce steel. 1941—U.S. Lt. Col. W.C. Bliss tells the American Foundrymen’s Association St. Louis Chapter that “the side which maintains the larger production of war goods is going to win the war.” The War Production Board reports later that each U.S. soldier requires 2200 kg (4900 lb) of steel compared to 40 kg (90 lb) in World War I. 1942—The use of synthetic sands increases as a replacement for many war materials. 1943—Keith Millis, a 28 year old metallurgist working at the International Nickel Company searching for a replacement to chrome due to interrupted supply, discovers that magnesium alloy in molten iron produces a spheroidal graphite structure. In 1949, he, Albert Gagnebin, and Norman Pilling would receive a U.S. patent on ductile iron production via magnesium treatment.

History and Trends of Metal Casting / 7 Table 5 (continued) 1940s 1944—The first heat-reactive, chemically cured binder is developed by Germany’s Johannes Croning to rapidly produce mortar and artillery shells for Axis troops during World War II. Two years after the war, his shell process is discovered among other inventions in the German patent office and made public. Croning is recognized with an American Foundrymen’s Society (AFS) Gold Medal in 1957 for his invention. 1946—Allied investigators uncover German foundry research on high-temperature alloys. Having “heard” of the Croning discovery (prior to the release of the report), Ford’s Ed Ensign and E.I. Valyi, Navy Bureau of Ships, attempted to replicate the process and produced shellmolded castings at the Midwest Foundry, Coldwater, Mich. Late 1940s—Thermal sand reclamation is applied to core sands and, to a limited degree, clay-bonded sands. 1948—The first nonlaboratory ductile iron casting is produced at Jamestown Malleable Iron Co., Jamestown, N.Y., as a 168 cm (66 in.) test bar is poured. Industry’s first ductile iron pipe is cast at Lynchburg Foundry, Lynchburg, Va. 1949—Keel blocks, diesel engine parts, a pressure cylinder, an 20 cm (8 in.) cube, and two cylinder liners become the first commercial castings of ductile iron at CooperBessemer, Grove City, Pa. Development of magnesiumferrosilicon makes ductile iron treatment far easier. Buffalo Pipe & Foundry Co., Tonawanda, N.Y., is the first U.S. operation to pour castings using Croning’s shell process. 1950s Early 1950s—Experimentation in high-pressure molding begins as metalcasters begin to increase the air pressure in air squeeze-molding machines to increase mold hardness (density). Fast-drying core oils are introduced. The pneumatic scrubber is developed to reclaim clay-bonded sands. Several wet reclamation systems also are in operation. 1951—Ford Motor Co., Dearborn, Mich., converts 100% of its crankshaft production to ductile iron. 1952—D-process is developed for making shell molds with fine sand and fast-dry oil by Harry Dietert. Sodiumsilicate/CO2 system is introduced. 1953—Hotbox system of making and curing cores in one operation is developed, eliminating the need for dielectric drying ovens. 1954—The CO2 process, a novel mold and coremaking process, is introduced from Germany. Working closely with General Motors, B&P develops a method for coating individual sand particles with resin binder. It also introduces a coreblower capable of producing cores with resin-coated shell sand—a modification to the Croning process. 1955—Ductile iron pipe is introduced to the marketplace by People’s Gas of Chicago, the first to install ductile iron gas mains. Mid-1950s—The squeeze-casting process originates in Russia. 1956—The first Betatron is installed in a U.S. foundry at ESCO Corp., Portland, Ore., for radiography of heavy steel castings. 1957—The vertically parted flaskless green sand molding production machine is invented by Vagn Aage Jeppesen, a 40 year old professor at the Technical University of Denmark. He was granted a patent in 1959, which was purchased by Dansk Industri Syndikat in 1961. 1958—Harold F. Shroyer obtains a patent for the full mold process, a process developed by artists in which simple patterns and gating systems are carved from expanded polystyrene and placed into a green sand mold. The process, known today as lost-foam casting (using loose, unbonded sand), is patented a short time later. Phenolic and furan acid-catalyzed no-bake binder systems are introduced. Ductile iron desulfurization via shaking ladles is developed in Sweden. 1959—General Electric uses the transient heat-transfer digital computer program and successfully applies the finite difference method to heavy steel casting production.

1960s 1960—Furan hotbox binders are developed for core production. Deep bed filters are used commercially for aluminum casting at Alcoa and British Aluminum in the United Kingdom. Compactibility and methylene blue clay tests are developed for green sand control. 1961—Alcohol-borne shell coating process is introduced (warm-coated). 1962—New CO2 sand testing method is introduced for sands bonded with sodium silicate and cured with CO2. Beardsley & Piper’s Al Hunter, Bob Lund, and Angello Bisinello develop the first automated green sand molding machine. In their design, the cope and drag are side-blown simultaneously and then hydraulically squeezed. The birth of automated matchplate molding reportedly improved metal casting productivity by levels as much as 60% in a short amount of time. Phenolic hotbox binders are introduced. 1963—Shell flake resin is introduced, eliminating the need for solvents. 1964—Dell & Christ’s paper on mold inoculation spurs the development of many of today’s forms of mold and late stream inoculation. The first vertically parted green sand machine (max. 240 molds/h) is delivered to United Danish Iron Foundries in Frederiksvaerk, Denmark. Early adopters report man-hour per ton improvements on the order of 50%. 1965—Oil urethane no-bake binder systems are used for cores and molds. General Electric’s Jim Henzel and Jack Keverian predict freezing patterns in large steel castings via computer. Cast metal-matrix composites are first poured at International Nickel Co., Sterling Forest, N.Y. 1968—The coldbox process is introduced by Larry Toriello and Janis Robins and introduced to the metal casting industry by Ashland Chemical Co. for high-production coremaking. Germany’s Daimler-Benz foundry in Mannheim is the first to run the process for automotive parts. John Deere Silvas Foundry, Moline, Ill., is the first to use the process for mass production in North America. 1969—The Chevrolet Vega is introduced by General Motors, featuring the first all-aluminum block with no cast iron cylinder liners. A total of 2.5 million blocks were produced during the vehicle life cycle. Late-1960s—Scanning electron microscope is invented in England. Thermal analysis begins to be used in iron foundries for the rapid determination of carbon equivalent and phosphorus contents, making it possible to study the transformation of an alloy during cooling. Manganese Bronze & Brass Co. and J. Stone & Co. join to promote nickel-aluminum bronze propellers. 1970s 1970—The sodium-silicate/ester catalyzed no-bake binder system is introduced for cores and molds. Safety-critical ductile iron steering knuckles are introduced on Chevrolet’s Cadillac. A new phenolic urethane no-bake process is introduced by Ashland Chemical, replacing oil sand dump box operations and significantly reducing energy requirements for core/mold production throughout the 1970s. Diran Apelian’s doctoral work at the Massachusetts Institute of Technology (MIT) leads to the development of foam filters for metal casting by Olin Metals. Commercial ceramic foam filters will be in use in metal casting facilities by 1974. 1971—The vacuum-forming or V-process molding method of using unbonded sand with a vacuum is developed in Japan. MIT doctoral candidate David Spencer performs experiments leading to semisolid molding, a process in which a partially frozen metal (fluidity similar to machine oil) can be poured into a die cavity. Following advancement in metal slurry consistencies with Professor Mert Flemings, Newton Diecasting Co., New Haven, Conn., produces the first semisolid castings. 1972—A 0.5 kg (1 lb) crankshaft for a refrigerator compressor produced at Wagner Castings (designed and engineered by Tecumseh) becomes the first productionvolume austempered ductile iron (ADI) component. Wagner also produces the first as-cast ductile iron ductile rods for passenger cars. The Canada Center for Mineral and Energy Technology (CANMET) uses real-time (continued)

Source: Ref 5; see also Selected References

1970s radiography to study the flow of steel in molds. Hitchiner Manufacturing, Milford, N.H., patents countergravity (vacuum) casting process. 1973—The first U.S. foundry argon oxygen decarburization unit is installed at ESCO Corp. 1974—Fiat introduces the in-mold process for ductile iron treatment. The phenolic urethane no-bake binder system is introduced for mold production. Mid-1970s—Alcoa and Union Carbide commence rotary degassing for wrought aluminum. Reading Foundry Products would apply this technology to aluminum foundries in the mid-1980s. Digital codes are developed to simulate solidification and fluid flow analysis. Ultrasonic verification of ductile iron nodularity is developed. Metal casting facilities examine new beneficial reuse routes for spent foundry sand, leading to applications such as cement and paving products, bricks, and flowable fill. 1976—Foote Mineral Co. and the British Cast Iron Research Association (U.K.) develop compacted graphite iron. Acidslag cupola practices plus external desulfurization with CaC2 begin to replace basic slag cupolas. 1977—General Motors installs ADI rear differential sets in passenger cars. The alumina phosphate no-bake binder system, an inorganic nonsilicate binder, is introduced for mold production. 1978—The furan/SO2 binder system is developed for core and mold making. Polyurethane no-bake binder system is introduced for aluminum applications. 1980s Early 1980s—Tundish ladle is embraced by industry as favored practice of nodularizing ductile iron. 3-D relational parameters are developed for Computer-aided design solid models. 1981—High-production lost-foam casting begins at General Motors’ Massena, N.Y., plant for aluminum cylinder heads. 1982—Warmbox binder system is introduced. 1983—Air impulse molding process is developed. Free radical cure/SO2 binder system is introduced. 1984—Charles Hull applies for a patent on stereolithography process. Other rapid prototyping techniques emerge shortly after. Phenolic ester no-bake binder is introduced. Thermal analysis makes breakthrough in molten aluminum processing for determination of grain refinement and silicon modification. 1985—Phenolic ester coldbox binder is developed. New automaker Saturn makes a strategic decision to select lostfoam process for its aluminum cylinder blocks and heads and ductile iron crankshafts and differential cases. Mid-1980s—Computer solidification software is commercialized. Amine recycling is introduced to enhance the environmental benefits of the amine-cured coldbox process. Ube Machinery introduces first squeeze-casting equipment. Aikoh’s (Japan) flux injection technology is initiated into U.S. aluminum foundry market. Late 1980s—3-D visualization techniques are developed. CaO/CaF2 desulfurization of cupola-melted ductile base iron begins to replace CaC2 method. Lanxide, Newark, Del., develops pressureless metal infiltration process for particulate-reinforced metal bodies. Magnesium wire injection method for ductile iron treatment is first tested. Casting solidification modeling software gains acceptance, allowing metal casting facilities to optimize quality, production, and cost prior to actually pouring a casting. 1988—Rapid prototyping and Computer-aided design/ computer-aided manufacturing technologies combine in a breakthrough to shorten tooling development time. Ford adapts Cosworth process precision sand casting process for high production. Metaullics Systems combines flux injection/rotary degassing technologies for aluminum processing. 1989—IMI (Yorkshire, U.K.) begins experimenting with bismuth as a lead substitute in copper alloys. 1990s 1990—Equipment for semisolid casting is introduced by Alumax Engineered Materials and Buhler, Inc. Foseco patents a direct-pour system that permits casting production without conventional gating/risering. Major automotive application comes in 1995 with CMI International’s upper intake

8 / Applications and Specification Guidelines Table 5

(continued)

manifold. Precision sand casting and casting quality for engine blocks are improved in mass production by major automotive companies with the Cosworth and Zeus processes for aluminum and the Loramendi Key Core process for precision sand-cast iron applications. 1991—“Dry ice” CO2 process is developed for cleaning coreboxes and foundry tooling. A noncontact gage for accurate dimensional analysis of lost-foam patterns and sand cores is developed through the AFS Lost-Foam Consortium. Eight years later, the consortium develops an instrument to measure the gas permeability of lost-foam pattern coatings (which controls flow of metal and has a dominant effect on casting quality). 1993—First foundry application of a plasma ladle refiner (melting and refining in one vessel) occurs at Maynard Steel Casting Co., Milwaukee. 1994—Use of low-expansion synthetic mullite sand for lost foam is patented by Brunswick Corp., Lake Forest, Ill., to enable precision casting of large components. Delphi Chassis Systems and Casting Technology Co. begin a program on squeeze-cast aluminum front knuckle, the

first high-volume production (1.5 million cars) two-cavity squeeze-cast aluminum conversion of its kind. Mid-1990s—Microstructure simulation is developed, contributing to better understanding of metallurgy effects and the prediction and control of mechanical properties in castings. Semisolid casting makes commercial inroads and market penetration. 1996—Cast metal-matrix composites (brake rotors) are used for the first time in production model automobiles, the Lotus Elise. Environmentally friendly (fluorine-free) fluxes are developed at the Worcester Polytechnic Institute and are commercialized. General Motors Corp. introduces GMBond, a water-soluble biopolymer-based core sand binder that is nontoxic and recyclable. 1997—AFS Consortium research at CANMET, Ottawa, Canada, results in the development and commercialization of lead-free copper alloys using bismuth and selenium. Following tests at Germany’s Aachen University, American Cast Iron Pipe Co., Birmingham, Ala., builds a continuously operated electric arc furnace for cast iron production.

Late 1990s—Stress and distortion simulation introduces the benefits of controlling casting distortion, reducing residual stresses, eliminating hot tears and cracking, minimizing mold distortion, and increasing mold life. 2000s 2001—The National Aeronautics and Space Administration (NASA) and the Dept. of Energy/Office of Industrial Technologies released a physics-based software tool to accurately predict the filling of expanded polystyrene patterns and sand cores as numerous variables are changed. Mercury Marine installs North America’s first pressurized lost-foam casting line at its new facility in Fond du Lac, Wis. 2002—Math-based modeling is used to develop a new sand core blowing simulation capability for process modeling software. 2003—The AFS Magnesium Division completes a research project proving that magnesium could be cast via the lost-foam casting process. Similar trials confirmed the castability of magnesium via the V-process.

phenolics led the way to urea and the dielectric process and then to furans and urea-free resins. The continued development of binders for the production of chemically bonded cores and molds is being directed toward increasing productivity as well as achieving the dimensional repeatability necessary to meet the new challenges of net shape and near-net shape casting requirements (Table 6). Many patterns were made of epoxy resins and polyurethane and other expendables such as polystyrene. Current understanding of sand casting now addresses the fundamentals of clay mineralogy, sand preparation, sand compaction, and the physical properties of molding and core sands. A better understanding of molding sand technology has resulted in sands of a higher degree of uniformity being prepared for the repetitive green sand (clay-bonded) molding sand. This high degree of achievement could only be possible with the great advances in sand testing produced by the foremost researchers and developers of sand testing instrumentation. Permanent-mold methods, which preceded the loam mold and the sand mold by centuries, also progressed for die casting (high pressure, low pressure, and gravity), centrifugal casting (vertical and horizontal), and hybrid processes such as squeeze casting and semisolid metal casting. Other contributions to the continued advancement of metal casting include:

 Cast metal-matrix composites  Argon oxygen decarburization for steel refining

Source: Ref 5; see also Selected References

Table 6 Development of core and mold processes Process

Core oil Shell: liquid and flake Silicate/carbon dioxide Airset oils Phenolic acid-catalyzed no-bake Furan acid-catalyzed no-bake Furan hot box Phenolic hot box Oil urethane no-bake Phenolic/urethane/amine cold box Silicate ester-catalyzed no-bake Phenolic urethane no-bake Alumina phosphate no-bake Furan sulfur dioxide Polyol urethane no-bake Warm box Free radical cure sulfur dioxide Phenolic ester no-bake Phenolic ester cold box

Approximate time of introduction

1950 1950 1952 1953 1958 1958 1960 1962 1965 1968 1970 1974 1977 1978 1978 1982 1983 1984 1985

Source: Ref 9

400 years, foundry processes and materials often relied on the methods developed by Vannoccio Biringuccio, the 16th century “father of the foundry industry,” who recommended using the dregs of beer vats and human urine as binders for molding sand, both of which were in use well into the 20th century. Until the 1920s, sand testing consisted of squeezing a handful of sand to judge its ability to compact and stick together. Early in that period, a sand research committee of the American Foundrymen’s Society began to develop sand test methods. By 1924, standards were established that covered the various properties of molding sands. Since World War II, experimentation also accelerated in organic and chemical sand binders for the thermosetting of molds and cores. Beginning with the Croning process (shell process),

 Foundry automation  A scientific approach to the gating and riser   

ing of castings using computer simulation of solidification Evaporative (lost)-foam casting and semisolid casting Computer-aided design and manufacture of castings Integrated foundry systems The melting of metals using the plasma arc cupola

Metal Casting Methods Metal casting is unique among metal forming processes, with a variety of casting processes (Table 7) (with Fig. 1, Ref 10). Very broadly, casting processes fall into two groups (Fig. 2): expendable-mold and permanent-mold processes. Each manufacturing process has the capability of producing a part with a certain range of tolerance, surface finish, and complexity. Manufacturing costs depend on the required dimensional tolerance, surface detail, and the quantity of parts required. For each process, there is a minimum batch size below which it is not economical to go because of costs of tooling, fixtures, and equipment. Also related to part cost is the production rate or the cycle time, the time required to produce one part. The most commonly used casting processes are evaluated with respect to these characteristics in Table 8 (Ref 11). Virtually any metal that can be melted can and is being cast. Choice of process and mold material (Table 9) is greatly influenced by the melting point of the alloy. A vital issue is avoidance of voids (due to insufficient feeding of melt) and porosity (due to dendritic solidification and the presence of gases). Success depends, to a large extent, on the skill of the mold designer, but there are very strong influences of the metal (alloy) and mold material. Even with the best design, solidification on the mold walls closes passage of melt at a certain distance, and this leads to a minimum allowable section thickness. There are large differences in the fluidity of alloys, and therefore, minimum allowable section thickness depends on the alloy being cast. Lower minimum thickness can be allowed with zinc, aluminum, and cast

History and Trends of Metal Casting / 9 Table 7

General characteristics of casting processes

See Fig. 1 for shape descriptions Casting process Characteristic

Part Material (casting) Porosity and voids(a) Shape(b) See Fig. 1 for shapes

Green sand

Resin-bonded sand

Plaster

Lost foam

All C–E All

All D–E All

Zn to Cu D–E All

Al to cast iron C–E All

All E All

Minimum section, mm (in.) Minimum core diameter, mm (in.) Surface detail(a)

0.01–300,000 (0.02–660,000) 3–6 (0.12–2.4) 4–6 (0.16–0.24) C

0.01–100 (0.02–220) 2–4 (0.08–0.16) 3–6 (0.12–0.24) B

0.01–1000 (0.02–2200) 1 (0.04) 10 (0.4) A

0.01–100 (0.02–220) 2–4 (0.08–0.16) 4–6 (0.16–0.24) C

Cost Equipment(a) Die (or pattern)(a) Labor(a) Finishing(a)

C–E C–E A–C A–C

C B–C C B–D

C–E C–E A–B C–D

Production Operator skill(a) Lead time Rates (piece/h
mold) Minimum quantity

A–C Days 1–20 1–100

C Weeks 5–50 100

A–B Days 1–10 10

Size, kg (lb)

Permanent mold

Investment

Die

Zn to Cu A–C Not T3, 5, F5

0.01–100 (0.02–220) 1 (0.04) 0.5–1 (0.02–0.04) A

Zn to cast iron B–C Not T3, 5, F5 with solid core 0.1–100 (0.2–220) 2–4 (0.08–0.16) 4–6 (0.16–0.24) B–C

B–C B–C C C–D

C–E B–C A–B C–D

B B C B–D

A A E C–E

C Weeks–months 1–20 500

A–B Hours–weeks 1–1000 10–1000

C Weeks 5–50 1000

C–D Weeks–months 20–200 100,000

<0.01–50 (0.02–110) 0.5–1 (0.02–0.04) 3 (Zn: 0.8) (0.12; Zn: 0.03) A–B

(a) Comparative ratings, with A indicating the highest value of the variable, E the lowest (e.g., investment casting gives very low porosity, produces excellent surface detail, involves moderate to low equipment cost, medium to high pattern cost, high labor cost, medium to low finishing cost, and high operator skill. It can be used for low or high production rates and requires a minimum quantity of 10 to 1000 to justify the cost of the pattern mold). (b) From Fig. 1. Source: Adapted from Ref 10

iron than with steel. Streamlined flow also is helpful; thus, sharp corners are avoided. Thicker sections solidify last and must be fed adequately. Metals and alloys have different castability characteristics based on their combination of liquid-metal properties and solidification characteristics that promotes accurate and sound castings. Castability factors include fluidity, shrinkage, and resistance to hot cracking:  Shrinkage is whether the alloy has a high

tendency to form shrinkage porosity.

 Resistance to hot cracking is the ability of

the alloy to withstand stresses developed by the contraction while cooling from the hotshort temperature range.  Fluidity is the ability of the liquid alloy to flow readily in the mold and to fill thin sections. Fluidity is usually evaluated by pouring the metal into a narrow spiral mold cavity and measuring the length of penetration of the metal before it solidifies. Fluidity is inversely related to the temperature range over which solidification occurs. Thus, pure metals and eutectic alloys, with short freezing range, have greater fluidity Because the viscosity of molten metals is very low, fluidity is determined more by the solidification dynamics of the metal and the mold. A high surface tension of the liquid metal reduces fluidity, and oxide films forming on the surface of the molten metal thus have a significantly adverse effect on fluidity. Insoluble inclusions also reduce fluidity. Castability ratings from the three factors are often described on a scale of 1 to 5 for each of the important factors. Table 10 shows typical

applications for metal alloys and relative ratings for castability, weldability, and machinability. As an example of a castability comparison in terms of fluidity, Fig. 3 shows three fluidity spiral samples of gray, compacted graphite, and ductile (nodular) iron. They were poured from the same base metal at the same temperature and have comparable chemical compositions. They differ only in their melt treatment. Gray (flake graphite) iron shows the best fluidity, and ductile iron shows the worst. Compacted graphite iron occupies an intermediate position. However, because compacted irons are stronger than gray iron with the same carbon equivalent (CE), a higher CE can be used to obtain the same strength, allowing greater fluidity and easier pouring of thin sections.

Production Trends The use of metal castings is pervasive throughout the economies of all developed countries, both as components in finished manufactured goods and as finished durable goods (Ref 1). As indicated earlier, castings are used in 90% of all manufactured goods and in all capital goods machinery used in manufacturing. They are also extensively used in transportation, building construction, municipal water and sewer systems, oil and gas pipelines, and a wide variety of other applications. In the broadest sense, foundries are categorized into two general groups: ferrous foundries and nonferrous foundries. For marketing purposes, foundries are also categorized according to the nature of their operations as being either captive (producing castings for their own use)

or jobbing (producing castings for sale). The market is sometimes further broken down by major casting processes when they can be readily identified or are particularly significant. This is done more often in the case of nonferrous castings; in this case, the product is usually categorized as being produced in sand, permanent mold, or die cast. Because 90% or more of all iron and steel castings are produced in some form of sand medium, distinguishing the process used is not quite as significant. Ferrous castings shipments are usually classified by market category. For example, iron castings are generally categorized as engineered (designed for specific, differentiated customers) and nonengineered (produced in large volumes of interchangeable units, usually consisting of ingot molds and pressure and soil pipe). The diversity among the various foundries makes it difficult to determine the exact structure of the industry. For example, it is not unusual for a single operating foundry to produce a variety of metals and alloys, both ferrous and nonferrous, in the same plant. Some also use a variety of processes in their operations. Many aluminum foundries, for instance, use both sand and permanent-mold processes, and some even produce die castings in the same facility. In addition, the foundry industry consists of a variety of large and small facilities. Table 11 gives recent census results on the number of casting facilities in major industrialized countries. Ferrous castings easily make up the largest production sector of the foundry industry in terms of volume, with gray iron outdistancing the other major metals constituting this group (Table 2). In comparison, iron and steel

10 / Applications and Specification Guidelines Increasing spatial complexity

Type (abbreviation)

0 Uniform cross section

1 Change at end

2 Change at center

3 Spatial curvature

4 Closed one end

5 Closed both ends

6 Transverse element

7 Irregular (complex)

Round (R)

Bar (B)

Section, open (S) Semiclosed (SS)

Tube (T)

Flat (F)

Spherical (Sp)

Fig. 1

The choice of possible manufacturing methods is aided by classifying shapes according to their geometric features. Source: Ref 10

Shape-casting processes

Expendable mold

Permanent pattern

Sand molding Clay Green Skin-dry Dry Carbon dioxide Cement Oil Resin Cold box Hot box Shell Flaskless

Fig. 2

Permanent mold

Expendable pattern

Slurry molding Plaster Ceramic

Semipermanent (sand cores)

Gravity Slush

Permanent (metal cores)

Low pressure

Hot chamber Sand (full-mold)

Slurry Lost wax Centrifuged

True centrifugal (outer mold)

Semicentrifugal (complete mold)

General classification of shape-casting processes. Source Ref 10

Centrifuged (complete mold)

Squeeze

Die

Cold chamber

foundries in North America shipped an average of 15.5 million tons of castings annually in the years from 1947 to 1986. The following decade (1967 to 1976) saw a significant upturn as the average annual shipments of iron and steel castings rose to slightly more than 17.5 million tons. Iron casting shipments have averaged approximately 14 million tons per year during the last four decades. From 1977 to 1986, ferrous casting shipments from American foundries fell to approximately 9.5 million tons by 1982 (Fig. 4). In this same time period, casting production around the world has shown a significant decline, generally reflecting the state of manufacturing. World production of ferrous castings fell some 27%, from 64.6 million tons in 1979 to 47.3 million tons in 1982 (Table 12). Nonferrous shipments followed a similar path in this time frame (Fig. 5). Recent Production Trends (1995 to 2005). Currently, metal casting production is dominated by China, along with large growth rates

History and Trends of Metal Casting / 11 Table 8

Casting processes and their attributes Surface roughness

Process

Dimensional accuracy

Complexity

Production rate

Production run

Relative cost

Size (projected area)

Pressure die casting Centrifugal casting Sand casting Shell mold casting Investment casting Powder metallurgy

L M H L L L

H M M H H H

H M M H H H

H/M L L H/M L H/M

H M/L H/M/L H/M H/M/L H

H H/M H/M/L H/M H/M H/M

M/L H/M/L H/M/L M/L M/L L

Key: H M

>250 >63 and <250

High Medium

>100 >10 and <100

>5000 >100 and <5000

High Medium

>0.5 >0.02 and <0.5

L Units

<63 min.

<0.005 >0.005 and <0.05 >0.05 in.

Low ...

<10 Parts/h

<100 Parts

Low ...

<0.02 m2

Source: Adapted from Ref 11

Table 9

Process and mold material options for casting

Process

Sand casting Investment casting Die casting

Cast iron

Carbon steel

Alloy steel

Stainless steel

Aluminum and aluminum alloys

Copper and copper alloys

Zinc and zinc alloys

Magnesium and magnesium alloys

Titanium and titanium alloys

Nickel and nickel alloys

Refractory metals

A B

A A

A A

A A

A A

A A

B B

A B

B B

A A

B B

C

C

C

C

A

B

A

A

C

C

C

A, normal practice; B, less-common practice; C, not applicable. Source: Adapted from Ref 12

Table 10 Typical applications for castings and general rating of castability, machinability, and weldability Note that these are general overall ratings for a class of alloys. Within each class, different alloys may vary considerably with respect to specific castability characteristics of resistance to hot cracking, fluidity, and shrinkage tendency.

Table 11 Number of operating metal casting facilities by nation in 2005

Casting alloy type

Country

Applications

Aluminum Copper Gray iron Magnesium Malleable iron Nickel Nodular iron Steel (carbon and low alloy) Steel (high alloy) White cast iron Zinc

Castability Machinability Weldability

Pistons, clutch housings, exhaust manifolds Pumps, valves, marine propellors Engine blocks, gears, brake disks and drums, machine bases Crankcases, transmission housings Farm and construction machinery, heavy-duty bearings, railroad rolling stock Gas-turbine blades, pump and valve components for chemical plants Crankshafts, heavy-duty gears Die blocks, heavy-duty gear blanks, railroad wheels

1–2 1–3 1 1–2 2

1–2 1–2 2 1 2

3 3 5 2 5

3

3

3

2 3

2 2–3

5 1

Gas-turbine housings, pump and valve components, rockcrusher jaws Ball mill liners, shotblasting nozzles, railroad brake shoes, crushers Door handles, radiator grilles, carburetor bodies

3

3

1

2

4

4

1

1

5

1, excellent; 2, good; 3, fair; 4, very poor; 5, difficult. Source: Ref 13

FG iron

CG iron

SG iron

10 cm

Fig. 3

Spiral fluidity samples poured from gray (flake graphite), compacted graphite, and ductile (spheroidal graphite) iron. Source: Ref 14

Austria Belgium Brazil Canada(a) China Croatia Czech Republic Denmark Finland France Germany Great Britain Hungary India Italy Japan Korea Lithuania Mexico Netherlands(a)(b) Norway Poland Portugal Romania Russia(e) Slovenia(e) Slovakia South Africa Spain Sweden Switzerland Taiwan(a) Thailand Turkey Ukraine(e) United States

Iron

Steel

Nonferrous

Total

12 18 467 60 17,000 16 106 10 16 115 212 218 41 ... 180 457 499 8 380 16 7 173(c) 50 58 ... 16 12 95 67 36 19 520 230 857 400 619

4 9 139 24 4,700 5 32 0 6 38 53 46 27 ... 27 78 140 3 10 0 3 36(c) 10 41 ... 4 7(f) 42 35 15 0 45 26 73 233 262

29 9 658 66 4,300 23 56 11 15 352 365 216 143 ... 870 1,173 208 5 250 5 11 245(d) 38 65 ... 33 32 119 57 84 32 350 220 361 437 1,499

45 36 1,264 150 26,000 44 194 21 37 505 630 480 211 4,500 1,077 1,708 847 16 640 21 21 454 98 164 1,900 53 51 256 159 135 51 915 476 1,291 960 2,380

Note: “. . .”indicates unreported data.(a) 2004 data. (b) Includes only members of AVNeG (General Assoc. of Dutea Foundries). (c) Includes approximately 24 metal casting facilities making both iron and steel. (d) Includes approximately 40 metal casting facilities working with both ferrous and nonferrous metals. (e) 2002 data. (f) Includes investment casting facilities. Source: Modern Casting

12 / Applications and Specification Guidelines 25 000

20 000 Thousands of tons

Table 12 Ferrous casting production from 1982 to 1986

Gray iron Ductile iron Malleable iron Steel Total

15 000

Figures given in thousands of metric tons Year Country

United States Japan West Germany France United Kingdom Brazil Canada(a) Korea Taiwan

10 000

5 000

0 78

Fig. 4

79

80

81

82 83 Year

84

85

86

87

1982

1983

1984

1985

1986

8,651,455 5,735,001 3,501,299 2,230,264 1,621,700 1,207,700 671,913 639,000 401,990

9,313,866 5,515,999 3,311,600 2,014,387 1,549,100 986,142 781,722 762,000 492,150

10,856,200 5,128,896 3,387,300 1,855,634 1,490,500 ... 981,730 805,000 599,122

9,991,393 5,240,261 3,499,900 1,902,608 ... 1,407,324 856,077 848,000 637,660

8,564,725 4,860,591 3,452,000 1,797,960 1,222,000 1,543,147 865,777 934,000 778,930

(a) Canada’s tonnage is estimated to be 75% of actual. Source: Ref 15

U.S. ferrous casting shipments from 1978 to 1987. Source: Ref 2

24 2500

21 Aluminum base Copper base

18

Zinc base Total

Castings, millions of tons

Thousands of tons

2000

China

1500

1000

500

U.S.

15 12

Russia

9

Japan

6 0 77

78

79

80

81 82 Year

83

84

85

3

86

Germany

India

0 1998

U.S. nonferrous casting shipments from 1977 to 1986. Source: Ref 2

in India (Fig. 6). The market for castings is likely to shift more to Asia and, to a lesser extent, to Central and Eastern Europe. However, despite the contraction of industry production bases in the West, high-quality manufacturing should continue to meet the needs of home markets and the demands of more productivity, energy efficiency, and environmental sustainability. German metal casting sites are reported to have the highest productivity in 2005, followed by the United States and France (Fig. 7). From 1995 to 2005, global output in the foundry industry grew by some 25%, topping 85 million tons. This is due in part to strong growth in its traditional markets but also to increasing use of these materials in a wide variety of other applications. These include the use of cast iron products in wind farms and large motors for power generation and ships, as well as the increasing popularity of light materials in automotive engineering, especially aluminum. During this period, output of cast steel continued to grow at an average pace, although there was something of a downward trend until 2004. Cast steel tends to be used less often in automotive and mechanical engineering than in

1999

2000

2001

2002

2003

2004

2005

6.0 5.5 5.0

Castings, millions of tons

Fig. 5

4.6 4.0 India

3.5

Brazil

3.0

France

2.5 Italy

2.0 1.5

Korea

1.0 1998

Fig. 6

1999

2000

2001

2002

2003

2004

Tonnage of major casting-producing nations from 1998 to 2005. Source: Ref 3

2005

History and Trends of Metal Casting / 13

8,108

Productivity per metal casting site

5,419

Germany (630 sites)

4,639 France (505 sites) Russia (1900 sites)

4,010 3,097 2,360 2,348 2,242 1,358 939 0

Fig. 7

U.S. (2380 sites)

Japan (1708 sites)

Italy (1077 sites) Brazil (1264 sites) Korea (847 sites)

India (4500 sites)

China (24,000 sites) 1

2

3

4

5

6

7

8

9

Productivity per metal casting site in 2005 for major industrial nations. See Table 11 for breakdown of site count. Source: Ref 3

1995 20%

United States

17%

China

16%

Commonwealth of Independent State

10%

Japan

5%

India

6%

Germany

26%

Other countries

30%

China

14%

United States

8%

India

8%

Japan

7%

Russia

7%

Germany

26%

Other countries

2005

Fig. 8

National shares in cast iron production. Source: Ref 3

power plant and factory construction. It benefited from robust growth in its main markets in 2005.

Cast Iron. China dominates cast iron production (Fig. 8). In terms of volume, cast iron and cast steel account for the bulk of foundry

industry production, comprising approximately 80% of total output. Global production has increased since 1995 by some 15%, with Asia recording the fastest rate of growth: by 2005 the region was already responsible for over half of world output (Fig. 8). Conversely, production in North and South America and Europe declined slightly. However, the situation varies considerably from country to country. In North America, production of cast iron in the United States and Canada fell by approximately 15%, likewise reducing their share of the world market (Fig. 8). Production volume has increased again since 2002, although it still remains below the 1995 figure. Even so, the United States continues to be the world’s second-largest producer. By contrast, Latin American countries such as Brazil and Mexico boosted their output by over 80%. From 1995 to 2005, China doubled its share in the global market. In Japan, on the other hand, production volume dropped by nearly 10%. Growing at a fast pace, output in India rose by approximately 80%, surpassing Japanese production for the first time in 2005. Together, these three countries account for approximately 90% of Asian production of cast iron. Nonferrous Castings. Nonferrous metal casting have sustained an upward trend. Global output of nonferrous metal casting products grew by some 75% from 1995 to 2005 (Fig. 9). Although production in regions around the globe gained ground, growth in North and South America (50%) lagged behind the rest of the world, where the volume of production more than doubled. In the Western hemisphere, the United States accounted for a good two-thirds of production, its 17% rate of growth pulling down the regional average. Copper and copper alloys as well as zinc products contracted by 6 and 20%, respectively, whereas output of all other materials expanded. Particularly notable is the robust increase in aluminum production in Canada, where Alcan Inc., one of the world’s top aluminum producers, boosted capacity. In Mexico, copper output increased tenfold, evidently due to the transfer of production from its northern neighbor. The volume of production in Asia increased by over 130%, driven first and foremost by rapid expansion in China, where output nearly tripled, and India, which likewise succeeded in significantly widening its share of the world market. Although production in Japan also rose, the country’s relative importance in the world market diminished. Aluminum continues to dominate nonferrous metal casting. Worldwide, aluminum accounts for approximately 80% of nonferrous metal casting, trailed by copper and copper alloys with 10%; the share of zinc is slightly over 6%. Production of aluminum cast parts doubled during the period under review, while other metals posted total gains of “only” approximately 10%. This is due primarily to greater use of aluminum in the automotive sector. Owing to its low specific weight, aluminum is an increasingly common

14 / Applications and Specification Guidelines In million tons 14 12 10 8 6 4 2 0 1995

2000

2001

2002

2003

2004

2005

Asia Europe North and South America

Fig. 9

Trends in production of nonferrous metal castings. Source: Ref 3

1995 30%

United States

19%

Japan

9%

China

8%

Germany

8%

Italy Commonwealth of Independent State

3% 1%

India

22%

Other countries

18%

United States

16%

China

13%

Japan

8%

Russia

7%

Italy

6%

Germany

4%

India

28%

Other countries

2005

Fig. 10

National shares in production of aluminum castings. Source: Ref 3

material in lightweight engineering. In the aviation industry, too, it continues its rapid ascent. The relative share of the traditional industrial nations in global aluminum production declined markedly from 1995 to 2005 (Fig. 10). While the United States and Japan increased their production volumes by approximately 30%, China’s output practically tripled. India, starting from a very slender base, also expanded its share massively. By 2005, China’s

production volume had nearly reached that of the United States, which it is likely to over-take during the next two years.

Market Trends The two largest sectors of metal castings are the automotive industry and machinery manufacturers. In the automotive industry, the

world’s major industrial economies account for approximately 85% of global car production. However, vehicle output in China and the nations of Central and Eastern Europe has been growing at a much faster pace. The next ten years are likely to see a continuation of this trend. In 2015, the world’s traditional car-making nations will probably be responsible for just 70% of global production; China’s share is likely to increase from approximately 10% to a more than 15% (Ref 16). Another important market for castings is the machinery/instrumentation sector, with products ranging from massive hubs, housings, and shafts for power generators to miniscule components for semiconductors. The world’s two leading makers of machines are the United States and Japan, which both lost ground from 2000 to 2005. In context, German manufacturers widened their share of the world market (Fig. 11). The Chinese sector is also expected to grow from its fourth place ranking. Although China remains one of the world’s biggest importers of plant and machinery, it is increasing its presence as an exporter as well. In 2005, China’s machine-tools sector (with annual sales of approximately £5 billion) exceeded that of Italy. Even with a distinct slowdown in the growth of imports, foreign suppliers meet 60% of China’s total demand for machine tools.

REFERENCES 1. “Competitive Assessment of the U.S. Foundry Industry,” USITC Publication 1582, U.S. Department of Commerce, Sept 1984, p xiii 2. D.P. Kanicki, Casting Advantages, Applications, and Market Size, Casting, Vol 15, Metals Handbook, 9th ed., ASM International, 1988, p 37–46 3. Annual Census of World Casting Production, Mod. Cast., Dec 2005 and 2006 4. M. Goodway, History of Casting, Casting, Vol 15, Metals Handbook, 9th ed., ASM International, 1988, p 15–23 5. Timeline of Casting Technology, Mod. Cast. Cast Expo Issue, May 2005 6. G.L. Werley, Slush Casting, Forging and Casting, Vol 5, Metals Handbook, 8th ed., American Society for Metals, 1970, p 334 7. E. Rooy, A History of Aluminum Foundry Alloy Development, AFS International Conference on Structural Aluminum Castings: Recent Advancements for the Design and Manufacturing of Reliable Structural Aluminum Castings, Nov 2–4, 2003 (Orlando, FI), p 87–98 8. B.L. Simpson, in History of the Metalcasting Industry, American Foundrymen’s Society, 1948 9. E.L. Kotzin, Development of Foundry Technology in the United States, Casting, Vol 15, Metals Handbook, 9th ed., ASM International, 1988, p 15–23

History and Trends of Metal Casting / 15 SELECTED REFERENCES (1980–1990)  AFS/CMI Conference on Green Sand TechUnited States



Japan



Germany



China Italy

  

France Great Britain

 

South Korea billion

0

50

100

150

200

250

300

350

2000 2005

Fig. 11



Global engineering sales of machinery/machine-tool sector. Source. Ref 17

10. J.A. Schey, Introduction to Manufacturing Processes, 2nd ed., McGraw-Hill, 1987 11. E.B. Magrab, Integrated Product and Process Design and Development, CRC Press, Inc., 1997 12. G. Boothroyd, P. Dewhurst, and W. Knight, Product Design for Manufacture and Assembly, Marcel Dekker, Inc., 1994 13. S. Kalpakjian, Manufacturing Processes for Engineering Materials, 2nd ed., Addison-Wesley, 1991, p 262 14. D.M. Stefanescu, R. Hummer, and E. Nechtelberger, Compacted Graphite Irons, Casting, Vol 15, ASM Handbook, ASM International, 1988 15. Census of World Casting Production, Mod. Cast., American Foundrymen’s Society, 1987 16. Foundry Industry, IKB Deutsche Industriebank, June 2007 17. Verband Deutscher Maschinen und Anlagenbau, German Engineering Federation SELECTED REFERENCES (published after 1990)  A Tribute to Keith Millis and the Unveiling of

Ductile Iron 50 Years Ago, Mod. Cast., Oct 1998

 J.R. Bodine, 1999 Charles Edgar Hoyt Memo-

  



  

 

rial Lecture: From a Monument to the Vega: The Journey of the Aluminum Casting Industry, Mod. Cast., May 1999, p 60–63 R.M. Cooper, The History of the Soho Foundry, GEC Rev., Vol 11 (No. 2), 1996, p 114–119 J. Gerin Sylvia, Cast Metals Technology, AFS, 1990 F. Habashi, Milestones in the History of Metallurgy, BUMA-V: Fifth International Conference on the Beginnings of the Use of Metals and Alloys, April 21–24, 2002 (Gyeongju, Korea), p 53–60 E.L. Kotzin, 1997 Hoyt Memorial Lecture: Lost Wax to Lost Foam: Reflections on the Past, Present and Future, Mod. Cast., July 1997, p. 42–45 Modern Casting’s A Look Back at the 20th Century Series, 2000–2001 Proceedings of 1992 Coreless Induction Melting Conference, AFS, 1992 N. Wukovich, A Short History of the Steel Foundry, Trans. Am. Foundrymen’s Soc., Vol 104, April 20–23, 1996 (Philadelphia, PA), p 643–649 1991 AFS International Sand Reclamation Conference, AFS, 1991 1995 Lost Foam Technology and Applications Conference Proceedings, AFS, 1995

   

nology—Productivity for the ’80s, AFS, 1983 N.N. Bubtsov, History of Foundry Practice in USSR, Moscow, 1962; trans., Washington, D.C., 1975 Casting, Vol 15, Metals Handbook, 9th ed., ASM International, 1988 J. Foster, “The Iron Age Moulds from Gussage All Saints,” Occasional Paper 12, British Museum, London, 1980 Foundry Trade J., Sept 11, 1986 High-Pressure Molding, 2nd ed., AFS, 1987 E.L. Kotzin, Metalcasting & Molding Processes, AFS, 1981 J.P. LaRue, Basic Metalcasting, AFS, 1989 C.S. Smith, The Early History of Casting, Molds, and the Science of Solidification, A Search for Structure: Selected Essays on Science, Art, and History, MIT Press, Cambridge, MA, and London, 1981, p 127–173 R.F. Tylecote, History of Metallurgy, The Metals Society, London, 1976 R.F. Tylecote, The Early History of Metallurgy in Europe, Longman, London, 1987 C.F. Walton and T.J. Opar, Iron Castings Handbook, Iron Castings Society, Inc., 1981 T.A. Wertime and J.D. Muhly, The Coming of the Age of Iron, Yale University Press, New Haven, 1980 T.A. Wertime and S.F. Wertime, Early Pyrotechnology: The Evolution of the First FireUsing Industries, Washington, D.C., 1982

SELECTED REFERENCES (published prior to 1980)  L. Aitchison, A History of Metals, New York,

1960

 Aluminum: Design and Application, Vol II,

American Society for Metals, 1967

 J.A. Gitzen, Progress in the Development    

and Utilization of Additives in the Foundry Industry, AFS, 1957 P. Knauth, The Metalsmiths, New York, 1974 C.A. Sanders and D.C. Gould, History Cast in Metal, AFS, 1976 B.L. Simpson, Development of the Metal Castings Industry, American Foundrymen’s Association, Chicago, 1948 B.L. Simpson, History of the Metalcasting Industry, AFS, 1948

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 16-28 DOI: 10.1361/asmhba0005188

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Metalcasting Technology and the Purchasing Process American Foundry Society Technical Department

CAST METAL COMPONENTS are not items that are commonly bought out of catalogs; most often they are tailor-made objects, designed and produced to the specific needs of a particular customer. As such, cast components cannot and should not be treated as low-value, high-volume secondary commodity items, where price is the sole purchase criterion. Cast components should be considered highvalue manufacturing products, where the metalcasting facility is an extension of the buyer’s manufacturing capability, involving long-term relationships and open, on-going communication at all levels. As high-value items, the purchasing criteria for cast components include a balance of quality, cost, service, and delivery. Many original equipment manufacturing (OEM) component buyers are recognizing that a carefully considered purchasing policy is the key to bestvalue procurement of metal castings. The policy should define the buyer-seller relationship, based on the following principles:








Long-term relationships based on shared

benefits and best value for all parties


Qualified cast component suppliers who

meet production quality standards, deliver on schedule, provide expertise and knowledge to support component design and improvement, and provide best-value pricing
Open and continuous communication at executive, purchasing, and technical levels to support concurrent engineering/design, design for manufacturability, and continuous improvement Over the last 20 years, significant change has occurred in the industrial purchasing environment—the old purchasing rules and situations no longer apply. Today’s (2008) manufactured product market is marked by six major changes that impact both the buyers and the sellers:
Globalization with new competition and new

markets: Today (2008), information, goods, and services move around the globe at an unprecedented speed. The OEM purchasing







managers can find sources for manufactured products in Malaysia, Mexico, or Poland as easily as they find a source in Ohio, Georgia, or California. The flip side is that a producer in Wisconsin can market components in Europe, Japan, and Brazil. Products can be delivered the next day from around the world. Best-value purchasing: The competitive market is driving everyone to focus on providing the best value to the final customer. Best value means that buying decisions are made not just on cost but also on a balance of quality, service, price, and availability/delivery. Recognizing how subcontractors impact final cost and quality, smart OEM manufacturers purchase their components and supplies from vendors who consistently can meet the best-value requirements. Value-added services or “one-stop shopping” for castings is common. Outsourcing: In the old days, many OEM manufacturers used internal production divisions to supply components. Today (2008), more OEM companies are recognizing that it is smarter to outsource their components through strategic partnerships with sources that have the cutting-edge expertise and advanced technology to provide a best-value product. Supplier management and quality assurance: When the quality and cost of a manufacturer’s end product depend critically on components from an outside supplier, the manufacturer will take a supply-management approach to working with the source. In supply management, buyers work with all their purchasing sources, developing long-term relationships, providing oversight on quality assurance, and working cooperatively with the sources in product-improvement teams. Continuous improvement: Nothing stands still for long anymore. In a competitive market, if a company does not improve its product and services, its competitors will improve theirs and win the contract. The supplier must work with the customer to find

and exploit opportunities to improve quality, reduce cost, and cut cycle/production time. That kind of innovative effort applies to working on current production components as well as on new product designs.
Information availability and velocity: The internet and electronic commerce have changed the way that information moves and circulates. Both buyers and sellers are expanding their electronic commerce reach in locating and assessing sources and opportunities for purchases/sales. Electronic marketplaces serve as locations for posting requests for quotes, advertising production capabilities, and finding matches between requirements and services. In a similar manner, suppliers are posting information about their capabilities and expertise on the internet, providing instant information to potential customers. Electronic commerce extends beyond the marketing function. All types of information (orders, contracts, technical, and financial) move electronically rather than by paper and mail. The speed of information movement is measured in minutes rather than days, cutting cycle time, reducing data errors, and overcoming geographic separation between buyers and suppliers. The metalcasting market is not isolated from these changes. Cast component buyers and sellers recognize that they must respond to the new market imperatives if they are to stay in business and remain competitive.

The Purchasing Process Purchasing is a critical process in any business. A company purchases raw materials, subcomponents, and support services that are essential to the production and delivery of the finished product to their customer. There is tremendous value in developing a purchasing process that is tailored to a the, specific needs of a business and that obtains components and services from outside sources in an effective, efficient, and timely manner.

Metalcasting Technology and the Purchasing Process / 17

Fig. 1

Four basic steps of purchasing process

Whether it is a product or service, the purchasing process consists of four basic steps (Fig. 1): 1. Define requirements and develop a purchasing plan. 2. Request and evaluate bids from potential sources. 3. Select a source and negotiate contract terms. 4. Carry out the contract and pursue continuous improvement. These basic steps of the purchasing process are described in the next four sections of this article. It is also important to remember that purchasing does not end with the selection of a source and contract agreement. The process is not complete until the required product/service is received and meets the requirements for quality, cost delivery, and service. Purchasing continues through product delivery (Fig. 1). The following sections provide detailed guidance on purchasing cast components, going through each step and describing specific issues and approaches that have proved useful in purchasing castings.

Define Requirements and Develop a Purchasing Plan The first step in any purchasing process is to define the product requirements and develop a purchasing plan. For cast components, product requirements are based on technical performance and targets for quality, quantity, schedule and cost. With this information, a complete and accurate request for quotation can be drafted and distributed to metalcasting facilities for consideration. To be a smart buyer, one needs more than just a requirements list. A purchasing plan that outlines and defines the following is also required:
Objectives and issues in the purchase
Purchase requirements
How sources will be identified, screened,

and selected
How the purchase will be contracted and managed for quality, cost, and schedule This plan may be simple and straightforward for small, low-value purchases; it may be very detailed and extensive for a high-value, critical purchase.

Request a Quotation Purchasing cast components is generally not a simple matter of transferring pattern numbers and quantity desired from the engineering bill of materials to a purchase order. Careful consideration of technical quality, quantity, and schedule factors is necessary to ensure that all pertinent and important information is supplied to the metalcasting facility. When necessary information is not provided, this slows the procurement and results in an inaccurate cost estimate or in a cast component that does not fully meet requirements. When possible, the metalcasting facility should be supplied with Computer-aided design files or blueprints for both the as-cast and the machined part. If the metalcasting facility is to provide the most suitable and useful cast components, it is important that the buyer indicate the areas to be machined and the locating points for machining. In addition, the buyer must know and understand the critical factors in specifying cast components and how those factors impact production and costs in the metalcasting facility. A request-for-quotation (RFQ) form is used to good advantage by many metalcasting facilities and cast component buyers. It provides a convenient way to develop the component requirements in a comprehensive and detailed approach. A model RFQ form is provided in Fig. 2. The RFQ form supplies the metalcasting facility with complete information on component requirements and enables the metalcasting facility to make an accurate casting price quotation for the most suitable casting for a particular application.

The Purchasing Plan With the detailed requirements for the casting defined in a draft RFQ, the buyer can develop a purchasing plan (Fig. 3.) that specifies:
Major purchasing objectives
Purchasing strategy
Deliverables expected from the metalcast-

ing facility at each stage of the purchasing process
Desired purchaser-metalcasting facility working relationship to ensure that quality, schedule, and cost objectives are met
Specific selection criteria
Purchasing tasks, responsibilities, and schedule

Purchasing Objectives. The buyer must clearly define the overall requirements and priorities in purchasing the cast component. Those requirements must be described in hard numbers that are clearly understandable and easy to communicate. The overall requirements package will include the engineering requirements as well as cost targets, trade-off considerations, and the company supply-chain management strategy. Each metalcasting purchase will have different answers to the following questions:
What are the quality and performance require-

ments for the cast component?


How many cast components will be purchas-

ed over a certain period?


What is the cost target for the cast compo-

nents?


What is the delivery schedule for the cast com-

ponents?


How critical is the casting to the project/busi-

ness in terms of performance, risk, schedule, and cost?
How will the different objectives (quality, performance, cost, and schedule) be balanced and prioritized? This requires a trade-off analysis (either informal or formal).
How does the cast metal component purchase fit into the overall supply-chain strategy for the company? The first priority is to meet the quality, quantity, service, schedule, and cost requirements for a given cast part. A cast component order is seldom a one-time purchase for a company. More often, cast components are integral to a company product line, and the purchase must be considered in the context of a company overall supply-chain strategy. A corporate supplychain strategy considers how to maximize the value and efficiency of all purchases through coordination and integration, rather than simply maximizing the value of individual purchases. Purchasing Strategy. Purchasing is nothing less than asset management, where the assets are suppliers and the goal is to maximize performance and value while minimizing risk across the casting purchase process. As with any investment, the buyer should develop a strategy that will meet objectives with a balance between risk and return. For a purchasing strategy, the buyer must ask the following questions in light of the particular requirements for a cast component:
How much benefit is there in finding the best

possible source?


What is the level of risk in purchasing from

a nonperforming source?


What are the cost drivers for the component,

and where can the metalcasting facility have a positive or negative impact on those drivers?
What investment of time and resources is necessary to identify a suitable source versus a best-possible source?
How does the purchase of a given cast metal component fit into the overall purchasing

18 / Applications and Specification Guidelines

Fig. 2

Casting request-for-quotation form

Metalcasting Technology and the Purchasing Process / 19

Fig. 3

Basic elements of a purchasing plan and strategy

strategy for castings within a division or company?
How will the company work with the supplier to maximize value during all stages of the purchase cycle?

cast components will include more than just the receipt of qualified castings at the shipping dock. It also should include details for technical partnership services during the entire purchasing process that include:

Answers to these questions will provide a framework for determining the value of a well-qualified metalcasting facility source to the success of a given project/program as well as the overall supply-chain management effort. In that framework, the purchasing agent then must consider the strategic questions for a particular purchase:


Initial design services by the metalcasting


Will the cast part be purchased from a single

Working Relationship Requirements. The working relationship between the customer and the metalcasting facility is a critical factor in defining a purchasing strategy that will provide best value for the casting purchase. The buyer must recognize from the first step that a cooperative team effort between the production design engineer and the metalcasting facility engineer is the key to a cast component design and production effort that meets the desired quality, cost, and schedule requiements. The buyer should work with the production design engineer during the initial purchase plan to define the goals, tasks, level of effort, and schedule required for a successful buyer/metalcasting facility working plan. The plan must consider consultation, technical services, and information exchange at each stage of the purchasing process. This plan should include:












source or multiple sources? What levels of metalcasting facility capability and production experience are necessary to successfully produce and deliver the casting on time and at cost? Do current casting suppliers have the capability and capacity to meet the new purchasing requirement? How will the performance of the supplier be evaluated in terms of quality, schedule, and cost? How will the supplier be integrated into the initial design process and future productimprovement efforts? What is the plan for developing and maintaining an open exchange of information between the purchaser and the supplier? How will the benefits, risks, and costs of the purchase be shared between the purchaser and the supplier? What kind of contract (firm-fixed price, cost plus fee, contract with performance/penalty fees) will best meet the purchase objectives and the overall purchasing strategy?

Deliverables Expected. If cast components were simple commodities that were bought off the shelf, then a contract would require only the delivery of a finished qualified component at a certain location, on a specific schedule, in a known quantity, and for a given price. However, each casting is unique, and each metalcasting facility has different capabilities, skills, and experience to bring to the table. Engineering service and supplier partnerships are critical factors in the design, production, and delivery of reliable, quality cast parts. For that reason, a deliverables list in a contract for

Fig. 4

Selection criteria of a purchasing plan

facility


Scheduled information exchange on quality
Schedule and cost achievements and issues
Quality planning and assurance during

production


Quality and cost-improvement initiatives

with shared benefits during production


Initial casting design (for performance and

manufacturability)


Tooling design and construction
Casting development and qualification
Component production with value enhan-

cement Selection Criteria. A purchasing plan always will include the criteria for evaluating, prioritizing, and selecting the bids from the responding metalcasting facilities. The criteria for qualifying and/or selecting a metalcasting facility and its bid for a specific casting fall into five categories (Fig. 4):
Production facilities (capabilities, capacities,

and expertise)


Direct experience and past performance with

the required alloy, similarly sized and

complex parts, and the desired/ recommended casting method
Production organization and capabilities (design and engineering support, testing and evaluation capabilities)
Quality-assurance procedures and capabilities
Quality, delivery, and cost (goals and reliability) Based on the purchasing objectives, the defined strategy, and the detailed deliverables and requirements, the purchasing plan will specify the criteria in detail along with weighting factors for evaluating the bids. Depending on the complexity, critical value, and risk for a given cast part, the evaluation approach for bids may take two general forms:
For low-risk castings, qualify bidders as go/no

go against the different criteria, eliminate unqualified bidders, and then rank qualified bidders by quoted cost.
For complex, critical components, grade bidders against the designated criteria (not qualified, qualified, or exceptionally qualified), use weighting factors for the different criteria, and then rank bidders high/low by summed and weighted criteria. The OEM buyer has a responsibility to assess the capability of a metalcasting facility to meet the purchase requirement. Assessing the capability of a metalcasting facility to produce high-strength structural cast components is difficult when the metalcasting facility is new to the buyer and does not have a track record based on previous purchases. A critical review of a metalcasting facility facilities, organizational structure, and process documentation is necessary to judge its capabilities. The capability survey is usually combined with qualitycontrol and business assessments before any effort to use the metalcasting facility as a supplier occurs. A questionnaire to address these issues is given in Fig. 5.

20 / Applications and Specification Guidelines The reviewer/assessor should write a capability report on the facility and the type of castings it is capable of providing. In considering and evaluating the five selection factors, the assessor should use the following guidelines to focus inquiries and determine the capabilities of the metalcasting facility. Metalcasting facilities making structural castings must have the proper equipment to attain the necessary quality. The production capability (and available capacity) of the facility should be consistent with the production quantity and rate required by the contract. A capability assessment should be made before giving any metalcasting facility a purchase order. Costs and schedules may be adversely affected if the capability assessment is not performed conscientiously. A user survey team approach often is necessary; the team usually is composed of engineering, quality-control, and purchasing personnel familiar with metalcasting facility operations. The survey usually is not repeated unless there are changes in the metalcasting facility, personnel, or organization or if a significant time has elapsed since the metalcasting facility last supplied components to the user. Metalcasting facility records should be available for similar castings that were successfully produced to comparable requirements. This demonstrated experience is an important factor in qualifying and selecting a metalcasting facility that can meet the quality and production requirements for cast metal parts. Metalcasting facility records should include mechanical test results and photographs or drawings of cast components. These results should be verified by periodic destructive tests. Separate test bar and appendage data are not adequate proof of capability. Metalcasting facilities that have not had experience producing the size, complexity, or quality of the required castings will need time to develop this expertise. Often, metalcasting facilities without experience are not aware of the requirements to cast a particular type of component. To gain the experience needed, they should start with small, less complex parts. Complex, high-performance metal castings must have high radiographic quality to be capable of exhibiting the necessary mechanical properties. Experience, combined with the quality-assurance equipment and personnel of the metalcasting facility, indicates if the metalcasting facility can achieve high-quality cast components. The ability of the metalcasting facility to hold tight dimensions regarding the location of critical features and the location of contoured mold lines is important. Special equipment, techniques, and procedures to mold, trim, heat treat, straighten, machine, and inspect are necessary to accomplish this on a production basis. The organizational structure of a metalcasting facility varies to some degree with the size of the facility. However, the organizational structure of most metalcasting facilities is basically the same:

senior management, production, quality assurance, marketing/sales, and maintenance/operations. The larger facilities usually have a greater depth of personnel in each department. Smaller metalcasting facilities are under direct day-byday control of fewer people. Sometimes, only the owner can make decisions. Large organizations tend to delegate authority, and the owner or chief executive officer is involved only in business decisions on an administrative level. Metalcasting facilities that produce quality cast parts need the technical capability to develop metallurgical and dimensional procedures and to solve metallurgical and dimensional problems as they occur. A degreed metallurgist with several years of on-site experience should be a permanent part of the organizational structure. Each cast component is produced in a manner that is specific to that part. Casting procedures must be adequately documented, and methods should be in place to monitor and record the actual casting process. The details of each operation should be maintained in a manner that permits them to be retrieved easily each time production is required. Photographs can aid considerably in the process description. Details on rigging, chilling, coring, molding and mold assembly, melting, pouring, processing, welding, heat treatment, straightening, and final inspection should be recorded during processing and should be available for review. A quality-control department should be staffed to implement the necessary inspections and testing throughout the entire metalcasting facility process (e.g., Fig. 6), from the receipt of raw material through the shipment of the finished cast component. There should be a production-control department that has an efficient system of processing each purchase order, from receipt of the order to final shipment, and that can retrieve daily input from the floor regarding the production status of each order. The reliability of a metalcasting facility in meeting customer requirements is best determined by direct experience and demonstrated reliability on previous orders. If a history has not been directly established, references and verified performance on other contracts should be provided. However, performance must be considered at two levels: a demonstrated record of performance across all orders as well as the capability and expertise to meet the performance, cost, and schedule requirements of a specific cast component in terms of the required alloy, complexity, molding technique, and quality control. Purchasing Tasks, Responsibilities, and Schedule. The final step in developing the purchasing plan is to lay out the specific tasks, responsibilities, and schedules that must be achieved to meet the required production and delivery times. Schedule layout is an essential element to integrating and coordinating the various tasks that must be accomplished in purchasing metal castings. One of the largest frustrations felt by a casting user is when a metalcasting facility does not meet expectations. It is important to

accurately and completely specify requirements, because inferior specifications will lead to poor performance or unnecessary costs due to overspecification (Table 1). Ideally, cast components are ordered after the metalcasting facility quotes on the job and all specification requirements are given in the purchase order (PO) for accurate pricing. If the metalcasting facility has taken exception to any of the specifications, these must be noted in the PO.

Fig. 5

Questionnaire for foundries—short form

Fig. 6

Gas testing for quality assurance

Metalcasting Technology and the Purchasing Process / 21 Table 1 Sample cost differences attributed to overspecification Overspecified procedure

Additional costs, $/lb

Melting Heat treatment for riser removal Weld repair, preheat Increased cracking susceptibility

0.05 0.08 0.04 0.04

Total additional cost

0.21

Guidelines on Specifications for Accurate Pricing. The following is a guide for cast component purchasers to navigate their way through the pitfalls of over- or underspecification and toward accurate pricing to receive the most from their cast parts. For new jobs (first-timeordered castings), the following information must be included in the RFQ and PO. Parent number and engineering drawing with the correct revision for both casting and machining drawings must be supplied. The drawings must be legible, with tolerances clearly indicated. The referenced drawing number and revision in the PO must coincide with the actual number and revision of the drawing. Quantity, price, and delivery must be clearly defined, with no misunderstandings. For example, an order for 5000 ductile iron automotive connecting rods can read, “5000 castings— 1000 pieces by October 1999, 500/month thereafter.” Castings can be supplied in various conditions, such as blasted (shot or grit), ground, machined (rough or finished), heat treated, pickled, and/or painted. They can be shipped in open or closed boxes, crates and metal tubs, or individually packaged for protection. For every metalcasting specification, there is a particular grade or class of alloy. Examples of terminology in material specifications are as follows:
In low-alloy steel, the grade “low” alloy

refers to the minimum mechanical properties required. For example, ASTM A148 grade 105-85 means that the test bars must meet or exceed 105,000 psi tensile strength and 85,000 psi yield strength. Although not directly noted by the grade, there also are minimum values for ductility, as measured by the percent elongation and reduction of area from a tensile test.
In ductile iron castings, the tensile yield and elongation minimum requirements are specified in the grade. For example, the ductile iron specification ASTM A 536 class 7050-05 means that the minimum requirements are 70,000 psi tensile strength, 50,000 psi yield strength, and 5% elongation.
In gray iron castings, the class refers to the minimum tensile strength and the size test coupon from which properties are determined. For example, ASTM A48 class 3DB requires a 30,000 psi tensile strength determined from a 30.5 mm (1.2 in.) diameter test bar. Class A requires properties to be determined from a 22.4 mm (0.88 in.) test bar.


Aluminum cast components, similar to cast

iron, can be supplied in the as-cast or heat treated condition. The most critical aspect of ordering cast components is quality requirements. How does the casting purchaser assure that the quality requirement is met? It is difficult to assign quality requirements to the casting, especially nondestructive testing (NDT), but without sufficient knowledge of existing NDT methods and specifications, the result is either an overly conservative approach or one that is too stringent. Therefore, some knowledge of the test methods and quality levels is an asset for casting purchasers to ensure that they receive what they need. In addition to the standard requirements, the purchaser often has supplementary requirements that are not covered under the designated specification. For example, a steel casting may require good weldability, and as such, the composition may be adjusted to ensure the carbon equivalent is below 0.5. On the other hand, an application in which the casting is subjected to impact loading at colder temperatures may require that Charpy V-notch tests be performed at the metalcasting facility. A critical aluminum alloy casting may require traceability to the heat from which it was produced. This can be accomplished by identification of the casting by serial number. Stainless steel castings, which require superior corrosion properties, can have a maximum allowed carbon content. Certifications are required only when they are critical to the cast component performance. Common specifications are chemistry (usually obtained from an analysis of the heat but also from test bars). Tensile properties also are commonly specified from a test bar but less commonly from a test block of equivalent section thickness as the casting. Other specifications may include liquid penetrant, radiography, ultrasonic, and magnetic particle testing. It is sometimes necessary to record and certify heat treatment data. In more critical applications, the actual heat treatment chart may be required. Expectations for the sample casting must be clearly defined. Are they required to be liquid penetrant inspected? Does each component require an x-ray? Is 100% layout required? Is written sample approval required prior to production? Are castings subject to third-party inspection? If sample requirements are not listed, the metalcasting facility must assume that sample approval is not required prior to production. For repeat orders with no changes, it may be sufficient to reference the requirements from the original PO. For repeat orders with an engineering drawing revision, this must be noted clearly on the PO, with the new drawing attached.

Requesting and Evaluating Quotations/Bids The second step of the purchasing process is requesting and evaluating quotations/bids from

metalcasting facilities. This section describes how to locate and identify potential metalcasting facilities, request and receive quotations, and then evaluate and rank the offer for best value. Locate and Identify Suitable Metalcasting Facilities. There are many ways to locate suitable metalcasting facility suppliers: current casting suppliers, names selected from a buyer’s guide, advertising material and magazines, and recommendations from other buyers and engineering staff. In all cases, the buyer must actively seek reliable information about sources of supply and try to match metalcasting facility capabilities and capacity to purchasing needs. Replies should be studied carefully because they will provide the guidelines for future purchasing decisions. For example, the buyer will know whether the supplier can produce the grade of material required in sufficient quantity. Attitudes toward quality can be assessed from the percentage of total production and qualitycontrol personnel directly involved in the quality-control program. The buyer should recognize that the strong emphasis of a metalcasting facility on quality control indicates it is probably producing high-grade cast components. The price structure will reflect this commitment to quality. If the component is very complex, look for a supplier who specializes in this type of cast component. Discuss capacity, to avoid disappointment after the order is placed. Buyers would do well to make buying decisions on the realities of the situation: price, quality, and inspection standards required for any given set of physical and mechanical properties. Perhaps the most important point, when all other conditions have been satisfied, is to look at the current pattern of production at the metalcasting facility. Important factors can affect prices and quality of the cast part, such as whether or not the metalcasting facility is high volume, casts small components, is a jobbing shop, and so on. An initial survey of casting sources will provide a list of candidate suppliers, based on capabilities and expertise. With a source list, the next purchasing step is to request bids. Maintaining a Good Relationship. The buyer-supplier relationship does not end when the appropriate metalcaster is found. A close, mutually beneficial relationship must be established between the casting user and the metalcasting facility to produce acceptable, cost-effective cast components. The metalcasting facility must be encouraged to learn more about the casting application, including high-stress or critical sections. With this knowledge, it can suggest additional safeguards for good performance. The metalcasting facility also can suggest various quality requirements to assure that purchasers receive what they want. In some situations, a redesigned casting can have greater integrity at the critical sections. The casting purchaser (and user) also is encouraged to learn more about castings and processes, including newer technologies for producing high-quality

22 / Applications and Specification Guidelines cast parts. Cleaner casting practices allow weight savings in many types of cast components without sacrificing quality. Send Out the Bid Package and Review Quotations. The bid package consists of an RFQ form along with supporting technical drawings/information. The bid package should include contact information for the cast component buyer and the production designer as well as the closing dates for quotations. Requests for quotations may be distributed by both paper and electronic means, such as fax and e-mail. During the quotation process, metalcasting facility engineers often call the buyer and/or the designer for additional information on technical, quality, quantity, cost, and schedule and service issues. The buyer and the design engineer should be prepared to discuss these issues with the responding metalcasting facilities. When quotations are received, they should be reviewed for completeness and responsiveness. Incomplete quotations should be returned to the sender for review and resubmission. Grade and Rank the Bids/Quotations. When all the quotations have been received, they must be graded against the selection criteria defined in the purchasing plan. This may be an informal process or a detailed review and evaluation of the bids, based on the size, value, complexity, and risk involved in the component purchases. Evaluation of the supplied quotations consists of applying the selection/criteria defined in the original purchasing plan:
Assess the quote against the specific quality,

quantity, schedule, technical service, and cost targets for the casting.
Determine the overall reliability and value of the individual bids against the supplychain management strategy.
Rank/classify the quotations in terms of the best value.

Select a Supplier and Negotiate Contract Terms After the initial bids have been received, evaluated, and ranked, the third step is selecting the qualifying and best-value bids. If the component is well defined in terms of simple shape/complexity and of relatively low risk, the quotation may be sufficient for closing the contract and moving into production. However, for the more complex, high-value cast components, additional engineering design and contract negotiation may be required to reach final agreement. Specifically, it may be necessary to:
Perform a final review and refinement of the

design for manufacturability with the casting engineer
Finalize technical, quality, and schedule requirements
Review and resolve nontechnical contractual issues

In contract negotiations, all parties should consider the following factors:


Find a better way to phrase existing clauses
Improve the agreement by adding clarifying

clauses


Level and scope of technical service, includ








ing design consultation, production, quality control, and product improvement Detailed instructions and guidance on specifications and quality-assurance requirements Assigned responsibilities for and integration of supporting engineering activities among the various production activities. These include the patternmaker, metalcasting facility, machine shop, and heat treater. Cost of quality assessment (rejects, delays, rework, etc.) Scheduling and lead-time planning Product liability

After these factors are considered, the next steps are to review, negotiate, and resolve cost details, performance clauses, and conditions of sale and then finalize and sign the contract. Figure 7 is a sample template on conditions of sale that are considered in casting purchase contracts. Terms and Conditions for Sale Agreements in the Metalcasting Industry. Over the years, there have been constant inquiries about what constitutes typical terms and conditions of sale in the metalcasting industry. There is no association-endorsed standard agreement. However, the North American Die Casting Association compiled terms and conditions of sale information from dozens of metalcasting operations around the country in 2002. Each agreement was separated into various clauses to constitute a comprehensive agreement between the seller and the buyer. If there was a significant redundancy from one agreement to the next, then the exact text for a particular passage was not repeated. One section identifies the common points for each clause that creates the agreement. In addition, the percent of agreements that contain each common point is reported. Finally, a list of atypical clauses is discussed in the section about unique or special terms. This section should be given special attention, since several adept clauses may protect the interest of a foundry. The Report on Terms and Condition of Sale Agreements in the Die Casting Industry—2002 is meant to provide the foundry with information on various terms and conditions currently being supported by the metalcasting industry. (American Foundry Society (AFS), American Metalcasting Association (AMC), Advanced Technology Institute (ATI), Defense Logistics Agency (DLA), North American Die Casting Association (NADCA), Non-Ferrous Founders Society (NFFS), and Steel Founders’ Society of America (SFSA) neither endorse nor recommend the adaptation of any of the terms or conditions documented in this report, which is available through NADCA at http://www.diecasting.org/publications/default.htm.) These terms are available electronically for cutting and pasting to help those purchasing castings to:

The principal sections include:






















Alloy and tolerances Cancellation Clerical errors Credit in returns Delivery Design changes Dispute resolution Engineering disclaimer Errors in quantity Inserts Overage allowances Packaging Patents Patterns, foundry, tooling, and dies Payment and credit terms Preliminary agreement Pricing Samples Taxes Unique or special terms Warranty

Contract Fulfillment and Continuous Improvement A signed contract is not the completion of the purchasing process. The contract must be completed by two actions: on-time delivery of a fully qualified product at the agreed cost by the metalcasting facility, and prompt and timely payment by the buyer. During the life of the contract, all parties have a valuable stake in continuous improvement in quality, delivery, and best value. Specific issues that must be considered during the production cycle include:
Open and timely communication (by visits,

phone calls, e-mail, etc.) between customer and supplier on project status, shortfalls, potential problems, forecasted changes, and opportunities for value improvement
Monitoring of quality, delivery, cost, and performance by both the supplier and the customer
Identification of opportunities to improve quality and increase value to the final customer by the customer and supplier

Considerations in Purchasing Metal Castings The metalcasting process requires designers to consider a specific list or requirements prior to purchasing that they may not otherwise be familiar with. The following is a list of the most significant considerations when attempting to determine the overall cost and design requirements of a metal casting.

Metalcasting Technology and the Purchasing Process / 23

Fig. 7

Conditions of sale—sample contract

24 / Applications and Specification Guidelines Production Cost Factors. Efficient and effective purchasing of cast components requires the use of a considerable amount of specialized information. With the increased emphasis on value-analysis techniques, purchasing agents must evaluate the many important factors that directly influence the total cost of a component part. The total cost factors are usually grouped under the general headings of quality, service, price, and delivery. Each buyer will determine the relative importance of these factors, based on particular circumstances. Since a universal formula is not possible, the following information will assist purchasing managers by covering the important criteria that affect the cost of a cast component. It is important that the casting supplier be included during the early stages of the design and purchase planning. Regardless of the type of metalcasting facility, dollar costs will break down into three main categories: materials, direct labor, and expenses. Materials may be subdivided into raw materials (primary metal and alloying constituents) and process materials (coke, sand, molding materials, bonding agents, shot, paint, etc.). Automation has markedly improved productivity in the metalcasting facility industry, but labor remains a high percentage of costs. Expenses are extremely variable, depending on the metalcasting facility facilities and the cost-accounting procedures. Although there is no magic formula that can be universally applied to metalcasting facility costs, it is possible for the buyer to develop a basic understanding of the cost factors for any metalcasting facility or particular cast component. Numerous factors influence the total cost of a finished cast component. These are considered in two groups: factors directly affecting the cost of a cast component and indirect cost factors for a finished casting. Dimensional tolerances and allowances are also discussed, since they have an important bearing on casting costs. Size, weight, and complexity of a metal casting are generally the most important cost factors, because they directly influence the materials and labor required for its production. Weight of the metal in a cast part also is important, but it no longer dominates the cost, as was true at one time. Overall size of a casting also is important, because several tons of mold material and other expendables are involved in the casting process for each ton of cast components produced. A large, thin casting that must be made in a large mold can be much more expensive than a heavier, simpler part made in a small mold. When designing a cast component, complexity is also an important factor. In general, the more complex a design, the more difficult it will be to manufacture. This may lead to higher costs. However, this is not an ironclad rule. With careful design, complexity can be molded into the metal casting without increased production cost (Fig. 8). To minimize the cost of complex designs, the engineer should call on the metalcasting facility to make manufacturing

suggestions at the early design stage. The metalcasting facility can help to minimize costs and problems with such things as draft, fillet radius, parting line locations, coreprints, wall thickness, and surface finish. The type of pattern equipment provided is an important factor in the total cost of producing a metal casting. The term pattern equipment includes coreboxes, if they are necessary, and any jigs or fixtures used for checking or assembling cores or for checking the dimensions of the part. Note that part quality is only as good as the pattern quality. Geometric complexity of a design raises costs because the mold must be much larger in order to contain external coring to form complex outer configurations. Cored castings are more costly. Cores must be made ahead of time, transported to the molding department, and placed carefully in the mold. However, equally important is the extra cleaning and trimming that is usually necessary on cored cast components. At the same time, the cores make the metalcasting process unique since they allow for the creation of internal cavities while the external surfaces of the part are produced. Wall thicknesses are an important design feature in metal castings, with the part configuration and alloy selection playing a major role in determining what wall thicknesses can be cast. As an example, in aluminum castings, the normal minimum wall thicknesses will range from 3.0 to 6.4 mm (0.12 to 0.250 in.). Depending on the part size and the area involved, thinner wall sections, down to 2.0 mm (0.080 in.), are possible over limited areas (65 cm2, or 10 in.2). Type of Metal. The cost of a casting will vary in three ways, based on the alloy used:
Inherent cost of the alloy: Cost depends on

both the baseline cost of the metal along with the alloying elements in the alloy.
Purity of the alloy: Premium prices must be paid by the metalcasting facilities to obtain alloys with fewer impurities. The lower the impurity level requirements of the alloy, the fewer the gating and risering returns allowed in melt charges.
Castability of the alloy. In steel and aluminum alloys, various alloy compositions have differing levels of fluidity, weldability, and castability. With reduced castability, additional metal must be poured into the mold in the form of flow channels and feed heads to assure a sound casting. For such parts, the weight of the actual metal in the finished cast component may be less than half of the total weight poured. This total-metalpoured factor is especially important when considering alloys with costly alloying elements, since the alloying elements must be added to all of the metal poured into a mold. Of these three considerations, castability of the alloy has the most effect on the cost of a cast component.

Fig. 8

Comparison of (a) fabricated agricultural component and (b) redesign to a one-piece casting. Metalcasting allows production of geometrically complex shapes, consolidating multiple welded components into one casting. This casting provided a cost-savings while improving functionality.

Dimensional Allowances and Tolerances. Because dimensional allowances and tolerances affect both the cost and delivery of metal castings, it is important that part buyers have a working knowledge of the major factors influencing them. Realistic tolerance requirements are of prime importance in minimizing overall costs. Tolerances that are more rigid than necessary increase metalcasting facility production costs and lengthen delivery schedules. Those that are too loose result in castings that require more expensive machining and usually have a heavier section size than required. The following is a brief review of the more important dimensional allowances and tolerances that should be considered. Shrinkage Allowance. When metal is poured into and takes shape in the mold, the metal is still at or above its liquidus temperature. As the metal solidifies and cools to room temperature, the metal contracts. Thus, to obtain a casting of desired size, the pattern is made slightly oversized. This addition to the dimensions is called a patternmaker’s shrinkage allowance. Another type of shrinkage that occurs while the liquid metal solidifies in the mold is called solidification shrinkage. This aspect of metal solidification is corrected by feed heads, risers, or shrink-bobs, which are attached to the casting when they are poured. This is a separate phenomenon and should not be confused with pattern shrinkage allowance. Draft Allowance. This is the term describing taper on the vertical faces of a pattern that allows the pattern to be removed without binding or tearing the mold. Draft will normally vary between 1 and 3 of taper. Small or machinedrawn patterns require minimum draft. Large patterns, inside pockets on patterns, and handdrawn patterns generally require a greater draft. Under ideal conditions, some patterns as deep as 150 mm (6 in or more may be drawn with almost no draft, using special techniques. Maximum draft should be provided where it does not interfere with desired casting geometry.

Metalcasting Technology and the Purchasing Process / 25 The metalcasting facility should be consulted so that draft allowance may be minimized or even eliminated entirely on surfaces where it must be machined off. For a no-draft condition, a loose piece in the pattern or corebox can be used. This will increase the cost of the part due to an expensive pattern and higher labor costs in molding and coremaking. Alternatively, the buyer should consider the lost foam process. Cast tooling or datum points should be fixed on designated no-draft surfaces. Machining Allowances. Additional stock that is added to the surface of a metal casting so that it can be finished to size by machining is called machine-finish allowance. Thickness of this allowance will depend on size and type of the cast component, the type of surface, how it is to be machined, and how accurately the casting can be made. Large castings and those made on a low-volume basis will generally have larger machining allowances than smaller castings or components produced in large quantities. With newly designed cast parts, a full machining allowance is usually added. After some experience is gained with the pattern, machining allowances may be reduced to the minimum necessary. To avoid local surface hard spots, a ground finish can be applied directly to a cast surface, with just enough stock so that the surface will clean up well. A general guide to machining allowances is given in Table 2. Distortion Allowance. When one part of a casting cools more rapidly than another, the part may become distorted. The first line of defense is to pay attention to design details, such as avoiding sharp changes in adjacent sections. When this is not possible, distortion often can be minimized by special metalcasting facility practice to control heat extraction or by followup heat treatment. In some cases, the pattern may be intentionally distorted so that the casting will be true. This is called distortion allowance. Tolerances. Because castings are individually designed and process details are engineered for the characteristics of each pattern, casting requirements such as surface smoothness, dimensional tolerances, or machining allowances cannot be stated in a general specification. These requirements usually are established through mutual agreement between the casting user and the metalcasting facility. Ordinarily, variation in casting dimensions can be held to within half of the patternmaker’s Table 2

Machining allowance guide On bore

mm

in.

mm

Up to 305 330–610 635–1067 1092–1524 1549–2032 2057–3048 Over 3048

Up to 12 13–24 25–42 43–60 61–80 81–120 Over 120

3.2 4.8 6.3 7.9 4.8 11.1

in.

Outside surface mm

1=8 2.4 3=16 3.2 1=4 4.8 5=16 6.3 3=16 7.9 7=16 9.5 Special instructions


Use the proper shrink factor in pattern











Guide to pattern machine-finish allowance Pattern size

shrinkage allowance. On a simple 305 mm (12 in.) casting dimension, this would be a tolerance of  þ2.1 mm (þ 0.0825 in.). Closer tolerances often are obtained on repetitive production where corrective adjustments can be made. For such cast parts, a tolerance of 1.3 mm in a 203 to 305 mm (þ0.05 in. in an 8 to 12 in.) dimension may be con sidered normal. Special procedures have been developed to produce some cast components with dimensions as close as 0.13 mm (0.005 in.). However, note that tolerances do not vary proportionally with the component dimensions. Dimensions within each half of the mold are generally more accurate than those across the parting plane or those related to a cored surface. The type of pattern equipment used directly affects dimensional tolerances. This is particularly true when machine molding or coremaking methods are employed. Metal patterns and coreboxes that are machined to size often will provide closer tolerances than will cast-to-size metal or wooden patterns. High-strength cast components frequently demand tight dimensional tolerances or thin-wall sections. In either, higher production costs are incurred if a systems approach is not used by the OEM to work with the metalcasting facility to recognize the true requirements of the component. Frequently, components are overspecified. In other cases, the design can be altered to take advantage of the capabilities of the many casting processes that are commercially available to meet true design requirements. The factors that most affect dimensional tolerances are:

in.

3=32 1=8 3=16 1=4 5=16 3=8

design. This is less of a problem on smaller castings. For larger castings, the metalcasting facility must have experience with the type of part required so it can make proper judgments about what shrinkage factors to use. Variability in the pattern equipment and coreboxes from nominal blueprint dimensions. To minimize this, quality tooling is required. Variability in the molds and cores themselves due to the people, equipment, and materials used in their production. Quality metalcasting facilities are aware of this and have controls and gages in place to minimize the effect. Variability in cleaning and finishing operations performed on the part. The metalcasting facility will perform inspections after this process to assure that finishing operations are done properly. Distortions due to the heat treatment of the castings. The metalcasting facility should plan to use heat treatment fixtures and special quenchants where the potential for this problem exists.

The quality of both the tooling and the production equipment is of great importance in dimensional tolerances. Even if only a small quantity of castings is required, the tooling

must be of prime quality if the dimensional tolerances are to be satisfied. Designers must be aware that there are additional costs incurred when tight tolerances are specified. Linear tolerances often must be increased for larger casting sizes. Critical dimensions should he identified to the metalcasting facility so that proper tooling will be built. Cast Surface Finish. There is no standard finish specification for cast components. Although root mean square values for cast surfaces are sometimes given on drawings, they are of questionable value because there is no dependable method of quantitative surface roughness measurement for unmachined cast surfaces. However, sample castings of the type required provide a satisfactory way of designating desired surface finish and observing what can be produced. In some casting applications, the degree of roughness of the cast surface is critical to successful use. Dimensions that are held within close tolerances may be ineffective if the surface is not smooth enough for accurate measurement or proper positioning in a fixture. While a textured surface is desirable or even necessary for some finishing or coating procedures, a surface that is too rough can be detrimental to finish quality and may require extra finishing, with increased costs. Although a smooth surface on a cast part often is considered an aspect of its quality, this is not an accurate indicator of the overall quality of the casting. Surface quality can involve several characteristics that may not be evident or even measurable. Superficial roughness on the casting can be caused by duplicating the rough surface of a pattern in metal or by an improperly made mold. However, most of the surface roughness of iron castings is the direct result of the highenergy abrasive cleaning methods in common use. Accurately evaluating a cast component surface roughness is difficult. Instruments and methods that are used successfully for measuring the roughness of machined surfaces are unsuitable for cast component surfaces. A sand cast surface is not composed of regularly spaced surface features, as found on a machined surface. The size distribution of the sand mix produces a much more complex and richer pattern of surface features. A finish that varies from location to location on a casting (due to inherent variations in sand compaction around a complex engineered geometry) further complicates the problem of measurement. The surface texture or roughness is generally specified by the term root mean square (rms) or arithmetic average (AA). The rms yields approximately 11% higher values than AA, but the difference is not significant. The surface finish may be specified as 300 rms maximum, 115 AA maximum, or C50 maximum on a visual comparator. Finer finish requirements will require special molding media or secondary operations,

26 / Applications and Specification Guidelines such as grinding or machining. This will add to the cost. Measurement should be made by surface comparator. The profilometer probe diameter acts as a filter for the input data. Fine surface areas should be identified and specified on drawings. Order Quantity. An important factor in selecting appropriate tooling is order quantity. Even with a given pattern, the order quantity may have an important influence on casting cost. One reason for this is that there are a number of standard operations in the metalcasting facility that must be performed for each part order, regardless of the quantity involved. Some of these are:
Quoting, acknowledgment, and recording of

the order


Scheduling production and issuing shop

work orders


Retrieving pattern equipment from storage

and inspecting and transporting it to production departments
Setting up tooling for production (i.e., laying out or attaching to machines and lining up proper flasks)
If cores are required, delivering core equipment to coreroom, making cores, and transporting them to the molding line
Shipping and billing Efficiency of production operations is also greatly influenced by quantity. When an order is issued subject to release, the total quantity is important in the selection of metalcasting facility tooling, but each release is normally considered as a separate order by the metalcasting facility, since each production order requires separate processing. The purchasing manager can determine the most economical lot size for cast parts when he knows what his casting requirements will be for some time in advance. Due to usual billing practices, less than a one-month supply is seldom practical. Savings realized by larger releases should be weighed against inventory costs, which include investment, space, and records. Possible changes in product demand and casting design or prices may not be equated directly in numbers but are incorporated as a matter of judgment by experienced buyers. The production quantity and the production rate are major factors in selecting the casting process and the type of mold pattern/tooling for the casting run. Various metalcasting processes have different production rate capabilities, different tooling costs, and different design change flexibility. Sand cast components often are preferred, because design changes can be made with relative ease. Two factors play important roles in tool materials selection. The first factor is the total quantity of cast parts needed. The second factor depends on the delivery period involved. Quality wood tools will generally produce up to 100 parts within approximately a year. Over a longer period of time, the tool may distort and

produce castings with dimensional discrepancies. Wood and plastic tooling should be usable for up to 500 parts in a moderate time frame (one to five years) with periodic refurbishment. Whenever small lot quantities are required for more than three years, metal or plastic tooling should be used to assure only minor refurbishment costs. Metal tooling is recommended when more than 2000 castings are required annually. Additional Operations. Often, the metalcasting facility can perform additional production operations that may increase the direct cost of the casting but will reduce the total cost of the finished product. Heat treating, painting or coating, and inspections that are more sophisticated and testing procedures are typical of additional operations that may be efficiently performed by the metalcasting facility. For quantity production runs, castings may be justified or targeted in a gaging fixture to provide accurate locating points for subsequent machining operations. This assures that the correct amount of stock is available to be removed on each critical finish surface. Technical Services. Close cooperation through concurrent engineering between the metalcasting facility and its customer is the approach required to yield the greatest benefit for buyer and supplier. However, for the more complex engineered cast components, an additional amount of service or technical assistance may be required from the metalcasting facility. This commonly occurs in development of new products. Assistance with component design, information on properties and metallurgy, troubleshooting, pattern assistance, or pattern models and experimental castings may be provided by the metalcasting facility. These services may be billed to the customer at cost or included in the price of the parts. However, in either case, quotations from a metalcasting facility that stands ready to provide its customers with this assistance cannot be compared directly with quotations from a metalcasting facility that does not. The latter type of metalcasting facility should not be ignored, because it can be a good, economical source for castings. This is particularly true where special problems or requirements do not exist and where additional services are not necessary. When a complex casting will be produced in large quantities, a model casting may be made to assist in planning production equipment. The model will assist in establishing the most efficient metalcasting method by providing an opportunity to establish critical factors such as the parting line and core prints on the actual shape. This also can be of assistance in planning for subsequent processes, such as machining. Types of Price Quotations. At one time, metal was expensive compared with labor, and castings were commonly sold by the pound. This was a simple buying method and worked well because the weight of a cast part was a more important part of total cost than was the

workman’s time to make it. This is no longer true. With the tremendous increase in cost of labor and mechanized equipment in a modern metalcasting facility, the value of the metal in most castings has become a secondary factor. Accurate cost accounting must include cost of the worker’s time, value of materials used, and overhead costs. This provides price quotations that are based on cost per casting and a ready reference for the purchaser to arrive at unit cost. The cost-per-piece can easily be converted to cost-per-pound, but such a figure cannot be justly compared to other castings that may be ordered in different quantities or require more or less man-hours to be produced. Some metalcasting facilities, however, still retain the price-per-pound method of quoting for specific types of parts. Thus, equivalent quotations based on price-per-pound and price-per-piece may show variations when compared for individual castings. This is particularly true when the quotation is based on an estimated casting weight. Variations of actual from estimated weight still affect the cost more on a price-per-pound basis than on a price-perpiece basis. Lead Time and First-Article Delivery. Essentially, the metalcasting process involves pouring molten metal into a cavity close to the final dimensions of a desired component. It is the most direct and simplest forming method available for metals and is often an essential production factor in rapidly manufacturing finished metal parts. Lead time is commonly defined as the amount of time between contract agreement and delivery of the first article, either prototype or initial production. Lead time is reduced with metalcasting compared to other manufacturing methods by eliminating and/or shortening the time required for tool production, parts ordering and delivery, assembly, finishing, and machining. In the metalcasting arena, new rapid prototyping (RP) tools (Fig. 9) are being developed to produce patterns and tools for castings. The RP tools can dramatically reduce lead times and production costs required for patterns and tools, particularly for designs that have complex detail that would be difficult to produce by conventional machining. An example of an investment casting produced with rapid tooling is shown in Fig. 9. Quality Assurance and Specifications. Quality assurance has benefits for both the buyer and the seller of metal castings. The buyer is assured of a cast component that meets performance requirements and the designated specifications. The metalcasting facility has confidence in its production process, the documentation of the process methods, and the application of quality standards in its day-to-day operations. All parties benefit because a high-quality product is delivered on schedule without cost overruns. Standard specifications for various types and grades of alloys and castings are promulgated by several national organizations and

Metalcasting Technology and the Purchasing Process / 27

Fig. 10

Fig. 9

Example of rapid prototyping process with investment casting. (a) Valve body prototype prepared for investment casting manufacturing without hard tooling. (b) Soft tooling for production investment casting mold. (c) Final casting. With rapid tooling, metal casting has developed to the point of producing castings in days, with no tooling.

government agencies. The most commonly used specifications for alloys and metal castings in the United States and Canada are those of the ASTM International, which is composed of many volunteer committees that write and maintain the standard specifications. Each committee is assigned jurisdiction over standards for one type of material, application, or testing methodology. The Society of Automotive Engineers (SAE) also publishes commonly used specifications for cast components. Other organizations, such as the American Society of Mechanical Engineers and the American Petroleum Institute, have specifications concerning castings for specific applications, but the material portion of these specifications typically refers to an appropriate ASTM International specification. The engineering departments of some companies that design and use a large number of castings write their own specifications. Some casting applications require an unusual combination of

Indentation profiles of three types of hardness testing methods

properties that would not be provided by cast parts made to standard specifications. Types of Specifications. There are several methods of specifying cast components. Specifications can be written to designate the properties of the metal that are specifically needed (strength, hardness, finish, accuracy, or special characteristics, such as resistance to heat or corrosion). In addition, there are many casting applications in which the ordinary properties of the designated type of alloy are quite adequate. In purchasing metal castings for noncritical applications, there may be an advantage in economy and availability to ignore standard metal specifications. The purchase order simply could designate the generic type of alloy by name (gray iron or malleable iron). Other requirements, such as for dimensional tolerances, may be added. When the mechanical properties of the metal are important, there are two commonly used methods to designate this requirement. The more frequently used procedure for engineering metal castings uses a standard test bar that is poured along with the specified lot of parts. The required minimum tensile properties should be obtained in testing a sample from this test bar. Most ASTM International specifications apply this method to qualify the metal used to pour the casting. The actual properties of the metal in the part will depend on its characteristics and the metal cooling rate in the various sections of the casting. The relation between properties in the casting and the test bar will depend on their relative cooling rates. For example, the properties of gray iron are especially sensitive to the section thickness in which they are cast, so a series of standard test bars of increasing size have been established specifically for gray iron and related to various casting thicknesses. One of these standard test bars, or, if necessary, a larger- or smaller-sized one, should be selected to approximate the cooling rate in the critical sections of the cast component. Malleable iron is not poured into heavy sections, and since all malleable castings are heat treated, a single-sized test bar is used. Ductile iron has some variation in properties with metal section when not heat treated.

However, a single-sized test bar is satisfactory, except for very large castings. In another approach, a commonly used method of specifying iron for castings produced in larger quantities is based on the microstructure and Brinell hardness of the metal in the part itself at a designated location. This method was originally developed for automotive and agricultural implement castings but has proven satisfactory for many other types of iron castings produced in quantity. When the microstructure of the iron is normal, there is a good general relationship between Brinell hardness and tensile properties; however, this relationship is not part of the specification, nor are the tensile properties of the iron stipulated. The suitability of a casting for its intended use as produced in a particular type of iron at a specified hardness level must be established for each application. This is typically done by trial use or by simulated load tests. Some castings purchased under SAE specifications for critical applications are tested for a specified Brinell hardness (Fig. 10). Standard Specifications. Many casting purchases are referenced to standard specifications, primarily to those of ASTM International and SAE. In addition, there are also cast parts purchased without metal specifications other than simply stating the generic type of alloy wanted. This purchasing procedure is satisfactory when the requirements of the casting are quite ordinary and typical for that type of alloy. Some metalcasting facilities will not accept orders that require metal property tests but specialize in providing castings with other specified requirements. Where metal properties are to be specified, the use of a standard specification provides the buyer with a maximum number of sources and possible economic advantages. The use of a familiar specification saves time and reduces the possibility of mistakes. Even when purchasing cast parts with unusual requirements, it is preferable to use an appropriate standard specification as a base and then add additional requirements. Of course, the additional requirements should be clearly stated and given with the reference number of the standard specification on the purchase order.

28 / Applications and Specification Guidelines Quality-Management Systems. A buyer may ask a caster, “How good is your quality, and how do you control it?” The caster has two possible replies: (1) The caster can provide a detailed description of the quality-management system and wait for the buyer to review and qualify the methods and procedures, or (2) the reply can be, “My quality system has been reviewed and certified against a recognized quality-management standard. Here is the certification.” A certified quality-management program has two basic features:


Definition and documentation of the policies

and procedures required that meet the specified quality standards
Application and practice of the quality procedures on a day-to-day basis, supported by traceable records There are several quality-management standards and certifications that are widely accepted in the manufacturing community and are applicable to the metalcasting market: ISO 9000, QS 9000, AS 9000, and NQS 9000.

SELECTED REFERENCES
Cost-Effective Casting Design: A Designer’s

and Purchaser’s Reference, American Foundry Society, Inc.
Engineered Casting Solutions, American Foundry Society, Inc., www.castsolutions. com
Principles of Purchasing Castings, American Foundry Society, Inc.
Website for casting buyers and designers: www.metalcastingdesign.com

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 31-40 DOI: 10.1361/asmhba0005189

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Chemical Thermodynamics and Kinetics Revised by Seymour Katz

TWO SUBJECTS that underlie all foundry processes are chemical thermodynamics and chemical kinetics. Relationships derived from these two disciplines of physical chemistry are extremely useful in providing a quantitative framework for explaining and predicting chemical behavior. This includes such presolidification chemical processes as preparing liquid metal for casting (going back to the winning of ores, if desired), melt purification from scrap recycling, and alloy additions. Chemical processes are also involved in making molds and cores, the casting and solidification processes, the microstructural development of alloys, and the phase changes that occur after casting (for example, by heat treatment). This article introduces the fundamental concepts of chemical thermodynamics and chemical kinetics in describing presolidification phenomena. Thermodynamics defines the most stable chemical system (equilibrium system) that can be produced, given a set of starting conditions (compositions, temperature, pressure, and so on). An example is the phase diagram. It gives the stable phases that coexist at equilibrium, given the composition, temperature, and pressure. Chemical thermodynamics is only concerned with the initial and final states of a system, while chemical kinetics is concerned with how and at what rate the system is transformed from the initial to the final state. The practical importance of chemical kinetics implies that systems often fail to reach the most stable condition and that the path taken plays a major role in determining the speed of the reaction. In general, kinetics assumes its greatest importance when atomic movement is restricted, such as at low temperatures or in solids. Thermodynamics most often applies when atoms have high mobility, as in gases and liquids and at high temperatures. Another factor that determines the relative applicability of thermodynamics and kinetics is time. The less time available for a process, the more likely that kinetics will govern the chemical state of the system. The thermodynamics and kinetics are fundamental in the composition control of liquid metal. Composition control consists of two areas: alloy addition and melt purification. Both are of vital importance in producing quality

castings. The chemical and physical processes involved in alloy addition do not vary much with base metal; therefore, this subject can be treated generically in this article. On the other hand, melt purification processes tend to be material-specific and vary according to base metal. For example, metal purification may involve processing of scrap, which is a common source material used for iron and aluminum castings. By its nature, scrap is of uncertain composition, both with respect to concentrations of desirable alloy elements and the presence of deleterious elements. Thus, despite all precautions in melting, adjustments in melt composition may be required. Following the introduction to chemical thermodynamics and kinetics in this article, the next two articles provide pertinent thermochemical data for ferrous and nonferrous (aluminum and copper) casting alloys. Casting defects with origins in the liquid state (gas defects and inclusions) are covered in the articles “Gases in Metals” and “Inclusion-Forming Reactions.” Nomenclature of symbols used in physical chemistry is given in Table 1.

mass. Common units of H are joules/mole, calories/gram, or Btu/pound. Thus, the total enthalpy of ni moles of material i is: ðHi Þtotal ¼ ni Hi

(Eq 1)

For a mixture of materials, the total enthalpy is the sum of the individual enthalpies: Htotal ¼

X

ni Hi

(Eq 2)

i

The heat capacity, CP, given in joules/mole  K, and so on is a measure of how H varies with temperature and is defined as: CP ¼

  @H @T P

(Eq 3)

where the subscript P indicates that the values refer to constant pressure conditions. Enthalpy and heat capacity govern the thermal state of a system and the heat generated or absorbed by chemical reactions. These variables are particularly important for establishing the energy requirements for processes, for determining rates of solidification, and for heat treatment processes. Enthalpy and heat capacity data are tabulated in Ref 1.

Chemical Thermodynamics

Gibbs Free Energy

There are three thermodynamic laws that provide the basis for the relationships that describe chemical systems. These laws hold that perpetual motion and absolute zero temperature can never be achieved. The relationship between the thermodynamic laws and, for example, phase diagrams is not immediately apparent. However, the connection exists and is testimony to a great deal of brilliant theoretical work carried out over the last 100 years. For metallurgical systems, the most important thermodynamic variables are enthalpy and Gibbs free energy. The former governs the disposition of heat, and the latter governs chemical equilibrium.

Conditions for Equilibrium. Chemical equilibrium is a condition in which the chemical components of a system have no tendency to change. In mechanical systems, equilibrium is understood as the condition in which the potential energy is at a minimum. For chemical systems at constant temperature and pressure, an analogous statement for chemical equilibrium is that the Gibbs free energy is at a minimum, or equivalently from calculus:

Enthalpy and Heat Capacity The heat content of a material is termed its enthalpy, H. Because the value of the enthalpy is proportional to the amount of material present, H is usually tabulated as enthalpy per unit

dGðP; T; ni . . .Þ ¼ 0

(Eq 4)

For systems in which chemical change can occur, dG can be expressed in partial differential form: 

   @G @G dT þ @T P;ni ... @P T;nj    @G dP þ dni þ    @ni T;P;nj 

dG ¼

(Eq 5)

where ni stands for moles of component i. The term ð@G=@ni ÞT;P;nj ;    is called the partial

32 / Principles of Liquid Metal Processing Table 1 Physical chemistry nomenclature ai aoi A A0 Ci Cp Cs CE dp D DR eii E fi F* F Gi DG Gi Goi xs G Goi Goi ð%iÞ Goi ðXi Þ h hi Hi Hio H i

Activity of component i relative to pure material standard state Activity of component i in the pure material standard state Interfacial surface area Surface area of an alloy particle Concentration of component i Heat capacity Sulfide capacity Carbon equivalent Particle diameter Diffusion coefficient Desulfurization ratio Interaction coefficient for 1 wt% standard state Cell potential Activity coefficient for 1 wt% standard state Integral free energy of solution Faraday’s constant Gibbs free energy of component i Integral molar free energy Partial molar free energy of component i Molar free energy of component i Excess partial molar free energy of component i Free energy of component i in solution at the 1 wt% standard state Free energy of solution from pure material to 1 wt% standard state Free energy of solution from the pure material to the hypothetical pure material standard state Convective heat transfer coefficient Activity of component i relative to the 1 wt% standard state Molar enthalpy of component i Standard enthalpy of formation Partial molar enthalpy of component i

molar free energy, Gi , or the chemical potential. It is clear from Eq 4 and 5 that, at equilibrium and at constant temperature and pressure, the following relationship exists: 0 ¼ Gi dni þ Gj dnj þ . . .

(Eq 6)

When the change in composition is dictated by a chemical reaction, stoichiometry provides a relationship between dni, dnj, and so on, and Eq 6 can suitably be transformed to: 0 ¼ ni Gi þ nj Gj þ . . .

(Eq 7)

where ni, nj, and so on are the stoichiometric coefficients in the chemical reaction equation. For products and reactants, the stoichiometric coefficients are positive and negative numbers, respectively. As an example, for the chemical reaction: 4Al þ 3O2 ¼ 2Al2 O3

(Eq 8)

Equation 7 becomes: 0 ¼ 2GAl2 O3  4GAl  3GO2

(Eq 9)

Tabulation of Thermodynamic Data. To make thermodynamic calculations convenient, a system was developed for tabulating thermochemical data so that relatively small amounts of data could describe a wide range of chemical conditions. This was done by defining special

xs

H i J k, k0 , kB KA kH km KmN K m,m0 Mi ni Ni pi poi P P S, P M

Q Q r, r0 R S Sc Sr S i xs S i Sio

Excess partial molar enthalpy of component i Mass flux Rate constants Constant related to the absorption of an element on a surface Constant related to hydrogen solubility in iron Mass transfer coefficient Liquid-phase mass transfer coefficient Equilibrium constant Solubility factors giving change in carbon solubility in Fe-C-X from unit mass addition of a third element Molecular weight of component i Moles of component i Mole fraction of component i Pressure of component i in the gas phase Pressure of component i in its standard state Constant pressure conditions Partition coefficient. Ratio alloy element concentration between liquid and solid phases in equilibrium. The superscripts S and M refer to stable and metastable products of Fe-C solidification, respectively. Heat flux Gas flow rate Radius of a rod Universal gas constant Ratio of surface area to volume Ratio carbon concentration: iron alloy melt to eutectic Rectified saturation degree, that is, weight fraction Fe-C eutectic formed on solidification of hypoeutectic iron Partial molar entropy of component i Excess partial molar entropy of component i Molar entropy of component i in the pure material standard state

conditions called standard states. Thus, for a chemical compound i that is an ideal gas at pressure pi,Gi is now written: Gi ¼ Goi þ RT  ln

  pi poi

(Eq 10)

where Goi is the free energy per mole of i in its standard state and poi is the pressure of i in its standard state. The second term on the right-hand side of Eq 10 is the free energy change to take a mole of i from its standard state poi to pi. For gases, the usual standard state is poi ¼ 1 atm. Thus, poi is usually omitted in writing, Eq 10. Both Gi and Goi are absolute values that cannot be measured. Changes in these variables can be measured from an arbitrary baseline. By convention, the baseline is taken as the elements in their standard states for which ðGoi Þelements ¼ 0 is defined at all temperatures. To indicate that the values are relative to a baseline, Eq 6, 7, and 10 are rewritten by replacing Gi and Goi with Gi and Goi , respectively. Values of Goi for many compounds (called the standard free energy of formation) and phase changes have been tabulated (Ref 1, 2). Because Goi is a function of temperature, the most common tabulations are based on another thermodynamic definition of Goi , that is: Goi ¼ Hio  T Sio

(Eq 11)

t t tm T V V0 V0 w W Xi

Time Celsius temperature Melting time Kelvin temperature Gas flow rate Gas volume per tonne of metal Volume of an alloy particle Mass fraction: desulfurization to iron Weight of melt Mole fraction of component i

Greek symbols a Phase designation aij Alpha activity coefficient function for Darken’s quadratic formalism gi Activity coefficient of component i relative to pure material standard state io Henry’s law constant d Boundary layer thickness Interaction coefficient; pure material standard eji state y Fraction of surface sites covered by surface-active atoms k Empirical constant indicating iron compositions that have the same microstructure at a given cooling rate l Particle shape factor ni Stoichiometric coefficient for component i in a reaction r Density Superscripts B e, equil f ‘ s S *

Bulk property Equilibrium condition Final condition Liquidus concentration Solidus concentration Surface condition Interface condition

where Hio and Sio are the standard enthalpy and the entropy of formation, respectively. The tabulations provide Hio and Sio data. The value of Goi is calculated for any temperature using Eq 11. Equilibrium Constant. Substituting Eq 10 into Eq 7 gives the most important relationship governing reaction equilibrium: X

ni  Goi ¼ RT

i

X

ln pni i

(Eq 12)

i

Here, the thermodynamic data on the left are related to actual pressures (concentrations) on the right. Equation 12 is ordinarily written in abbreviated form as: Go ¼ RT  ln K

(Eq 13)

where K is the equilibrium constant. The form of K is illustrated in the following. Given the gas phase chemical reaction: 2H2 þ O2 ¼ 2H2 O

(Eq 14)

the equilibrium constant is: K14 ¼

p2H2 O p2H2 pO2

(Eq 15)

Activity and Condensed Phase Equilibrium. Metallurgical applications are primarily concerned with condensed phases (solids and liquids), which do not behave in the manner of an ideal gas. That is, for a condensed

Chemical Thermodynamics and Kinetics / 33 material, Goi does not vary with the logarithm of the concentration as may be inferred from Eq 12, where, for an ideal gas, Goi varies directly with ln pi. To preserve the useful equilibrium constant expression for nonideal, condensed phases, Eq 10 and 12 are retained in form but rewritten to enable the use of data expressing nonideality. Thus: ! ai aoj

(Eq 16)

ln ani i

(Eq 17)

Gi ¼ Goi þ RT  ln X

ni Goi ¼ RT

i

X i

The difference between Eq 16 and 10 and Eq 17 and 12 is that a term ai, called activity, replaces the pressure term. The activity is a function of concentration but usually not a simple function of concentration. Analogous to the case with gases, the activity in the standard state aoi is taken as unity. The equilibrium constant is now defined in terms of activity: ln K ¼

X

ln ani i

(Eq 18)

i

Activity Coefficients—Departure from Ideality. Common expressions for activity are: ai ¼ gi Xi

(Eq 19)

hi ¼ fi ð%iÞ

(Eq 20)

where the concentration terms Xi and (%i) are the mole fraction and weight percent, respectively, of the solute i, and gi and fi are the activity coefficients that measure departure from ideality, that is, generally gi and fi 6¼ 1. The terms ai and hi are often called the Raoultian activity and the Henrian activity, respectively, to distinguish the standard state being used. The Raoultian activity is usually used in conjunction with the pure material standard state, while the Henrian activity is used with the 1 wt% standard state (to be explained). The latter standard state and Eq 20 are used to describe dilute solutions and have been extensively applied to ferrous systems, especially steel. 1.0 A

T = 1873 K

Activity, ai

0.6

aNi aFe

0.4

γ⬚

Ni

0.2 B 0 0 (Fe)

Activity coefficient, γi

Raoult’s law

0.8

Henry’s law 0.2

0.4

0.6

0.8

1.0 (Ni)

Mole fraction, XNi

Fig. 1

Activities of iron and nickel in their binary solutions at 1873 K. Source: Ref 1, 3

The pure material standard state and Eq 20 are used in most other cases. The two standard states, their activities, and their activity coefficients are illustrated with Fig. 1, which gives the activity composition relationships for the liquid iron-nickel system. The point “A” represents the pure material standard state for nickel (pure nickel in its most stable form—liquid—at the temperature of concern, 1873 K). Activity, given on the ordinate, is based on Eq 19. The ideal solution line (Raoult’s law line), where aNi = XNi, is labeled. The change in the partial molar free energy of nickel in transferring nickel to solution is given by Eq 21, which was obtained by substituting Eq 19 into Eq 16: GNi ¼ GoNi þ RT  ln XNi þ RT  ln gNi

(Eq 21)

In this case, Goi ¼ 0 because the standard state and the baseline conditions are identical. The term RT  ln XNi is the change in GNi ideal for an ideal solution GNi . The term RT  ln gNi provides the correction for the real system. As seen, gNi < 1. Thus, GNi is more negative ideal than GNi . Larger negative free energy values indicate greater stability. Thus, real solutions of nickel in iron are more stable than ideal solutions of nickel in iron. A common characteristic of the activity of a component in solution is the relatively linear portion of the curve as Xi ! 0. This limiting slope, designated goi (Fig. 1), is known as the Henry’s law constant. The 1 wt% solution standard state is marked point “B” in Fig. 1. It falls within or near the range in which the activity coefficient is a constant ðgoi Þ. By defining the standard state at point “B”, fi = 1 in the linear region and conveniently, hi = (%i). The expression that corresponds to Eq 21 for the 1 wt% standard state is: ^o þ RT  lnð%NiÞ GNi ¼G Ni þ RT  lnðfNi Þ

(Eq 22)

^o to indiIn Eq 22, GoNi is designated G Ni cate that it refers to the 1 wt% standard state. Clearly, for any iron-nickel composition, GNi is the same whether it is calculated by Eq 21 or 22. Therefore, the choice is one of ^o 6¼ 0. It is the convenience. In this case, G i change in free energy in taking pure liquid nickel to point “B” on the Henry’s law line. The difference in Goi between standard states A and B is:   ^o ¼ Go þ RT  ln aB G i i aA   Ms loi ¼ Goi þ RT  ln Ms þ 99Mi

(Eq 23)

where the activities for composition A and B are Raoultian, and Ms and Mi are the molecular weights, respectively, of solvent and solute. The equilibrium constant expressions (Eq 17) for the pure material and 1 wt% standard states are: X i

ni Goi ¼ RT

X

lnðgi Xi Þni

(Eq 24)

X

^o ¼ RT ni G i

X

ln½fl ð%iÞni

(Eq 25)

i

o

Very often Gi for components in the equilibrium constant are expressed in different standard states. There is no problem with this as long as appropriate values Goi or ^o are used. G i Tabulations of Activity Coefficient Data. Tabulations of Goi data were already discussed. From Eq 23 to 25, it is seen that data for goi , gi, and fi are also needed. Tabulations of goi and ^o for elements in solution in liquid iron, G i cobalt, and copper are given in Ref 2. Similar data for aluminum are included in the article “Thermodynamic Properties of AluminumBase and Copper-Base Alloys” in this Section. Data for gi and fi for iron, copper, and aluminum multicomponent alloys are available in Ref 4 to 6. These data are based on a formalism (Ref 7) that is discussed in the article “Thermodynamic Properties of Aluminum-Base and Copper-Base Alloys” in this Section. Another formalism (Ref 8, 9), found in the article “Thermodynamic Properties of Iron-Base Alloys” in this Section, is used to describe the activities of carbon and silicon in multicomponent ferrous alloys. Most of the tabulated values were obtained from studies in dilute solution (Henry’s law region). From Fig. 1, it can be seen that the shapes of the activity curves are complex. Therefore, data for the Henry’s law region have only limited applicability outside the region. To describe Gi over the entire composition range, tables giving Gi as a function of composition are used. These data are usually available only for binary alloy systems. Tables for important aluminum and copper binary systems are given in the article “Thermodynamic Properties of Aluminum-Base and CopperBase Alloys” in this Section. Comparable data for iron are available in Ref 10. Because Gi is temperature dependent, the data are commonly given as H i and S i , where Gi is obtained by: Gi ¼ H i  T  S i

(Eq 26)

Examples of the use of Eq 1 to 26 are found in other articles in this Section.

Phase Diagrams Phase diagrams describe the phases that coexist at equilibrium. Because chemical equilibrium is involved, phase diagrams have a thermodynamic basis. The complexity and variety of relationships demonstrated by phase diagrams place a detailed description beyond the scope of this article. A qualitative demonstration of the interrelationship between phase diagrams and thermodynamics is given. For a multiphase system, a condition of equilibrium is that the partial molar free energy or chemical potential of each component must be the same in all phases (a, b, . . .):

34 / Principles of Liquid Metal Processing a

b

Gi ¼ Gi

(Eq 27)

Thus, for example, the chemical potential of carbon in carbon-saturated iron is the same as graphite, despite there being less than 5 wt% C in saturated iron. A similar situation exists with respect to silicon/aluminum-silicon alloy equilibrium, where the chemical potential of pure 0

silicon is matched by less than 15 wt% Si in solution. Phase diagrams describe stability relationships for compounds and solutions (solid and liquid). To understand the relationships based on thermodynamics, it is necessary to compare the phase diagram (temperature-composition plot) with the equivalent free energy-composition plot. To construct the free energy-composition plot, it must be recognized that for a binary solution the integral molar free energy is:

Free energy (F *), cal

G ¼ X1  G1 þ X2  G2 −500

−1000

F  ¼ G  ðX1  G1 þ X2  G2 Þ

(ΔG1 – ΔG1⬚)

−1500

−2000

(Eq 28)

Substituting Eq 16 into Eq 28 and assuming ideal solutions (ai = Xi) yields F*, the free energy change in forming 1 mole of an ideal, binary solution from components in the pure standard state (taken as the pure liquid in this case): ¼ RT ðX1  ln X1 þ X2  ln X2 Þ (ΔG2 – ΔG2⬚) 0

0.75

0.50

0.25

1.0

Mole fraction (X2)

Fig. 2

Free energy of formation of an ideal solution from liquid components in the pure material standard state. Also illustrated is the graphic method of tangents to determine the partial molar free energies of solution.

(Eq 29)

A plot of F* as a function of X2 is given in Fig. 2. Clearly, F* < 0 with a minimum at X2 = 0.5. For 6 1), the curve can be higher nonideal solutions (gi ¼ or lower than the ideal line, and it can be asymmetrical. It can be shown that the intercepts of a tangent line drawn at any point on the curve will give ðG1  G1 Þ and ðG2  G2 Þ (Ref 1). These are the partial molar free energies of solution from the liquid standard state.

0

For metallic solutions, Gi ¼ 0 ; therefore, GI is obtained directly from this plot. The free energy of a phase of definite composition is represented as a point on the free energy-composition diagram. Clearly, if such a phase were in equilibrium with a solution, the equilibrium solution and the compound would have to lie along the same tangent line in order for Eq 27 to hold. This condition is illustrated in Fig. 3(a). Figure 3(b) shows the phase diagram that results from the thermodynamic constraints. In Fig. 3(a), F* < 0 for pure solid phase 1 because the liquid was chosen as the standard state, and at the temperature chosen, the liquid is less stable than the solid. The equilibrium between a liquid and a primary solid solution is illustrated in Fig. 4. From Fig. 4(a), the respective equilibrium concentrations lie along a common tangent, thus satisfying the equilibrium conditions expressed by Eq 27. The phase diagram that results from these thermodynamic conditions is given in Fig. 4(b). For small values of X2, either solution can theoretically exist. The solid phase is expected to be present, however, because it is more stable (lower F*). If, by virtue of such factors as rapid cooling, the liquid phase is retained in a region where the solid is more stable, a metastable condition would exist. This is not uncommon in systems of metallurgical interest. Common examples are cementite (Fe3C), which is metastable with respect to iron and graphite, and the glass (supercooled liquid) condition of slags, which is metastable with respect to crystalline compounds.

Solid Solution

B

Composition of solution in equilibrium with pure solid

Solid + solution

α solid solution

Free energy (F *), cal

Free energy (F *), cal

0 A

Solution

Liquid solution

(a)

Temperature (T )

(a)

Temperature (T )

This line corresponds to temperature of free energy chart above

T1 L

α+L

β+L

α

0 0

(b)

Fig. 3

1.0 Mole fraction (X2) Relationship between (a) free energy-composition and (b) temperature-composition diagrams for the equilibrium between a solution and a pure solid component. Source: Ref 11

β

α+β

(b)

Fig. 4

1 Mole fraction (X2)

Relationship between (a) free energycomposition and (b) temperaturecomposition diagrams for the equilibrium between a liquid and a primary solid solution. Source: Ref 12

Chemical Thermodynamics and Kinetics / 35

Chemical Kinetics The rates of chemical processes depend so much on local conditions that it is difficult to describe the processes in the universal terms used in thermodynamics. The rates of chemical change in solid cast metals are usually governed by solid diffusion. This subject is examined in the Section “Principles of Solidification” in this Volume. For processes involving fluids at high temperatures, the rates of chemical reactions are inherently rapid and usually not rate limiting. The processes that generally limit the rate are nucleation of the product phase and interphase mass transport (moving material between immiscible phases).

Nucleation The problem of nucleation is universal to the kinetics of solidification, and the subject is treated more extensively in the article “Nucleation Kinetics” in this Volume. However, a few general comments are in order. Nucleation of a phase in an inclusion-free metal requires supersaturation with respect to the material being precipitated. For this condition, the chemical potential of the material in the melt exceeds that in the precipitate, which is the requirement for spontaneous reaction. Initially, the diameter of the precipitated phase is in the atomic range. Thermodynamics dictates that the activity of particles increases with decreasing radius of curvature. Therefore, the activity of atomic-sized particles is very high. This in turn requires very high melt supersaturation to meet the requirement for spontaneous nucleation. This barrier to the generation of nuclei from a homogeneous melt, imposed by thermodynamics, makes it unlikely that appreciable nucleation occurs by this mechanism. Instead, nucleation is initiated on particles purposely introduced to the melt before solidification. For example, TiB2 particles are in use to nucleate aluminum melts (Ref 13), and complex oxysulfide particles nucleate graphite in gray and ductile iron (Ref 14, 15).

Interphase Mass Transport Because high-temperature reactions are rapid, the rate-limiting step in many cases of heterogeneous reaction is the diffusion of reactants to and from the reaction interface. This is particularly the case when the reactants are dilute. Because the reaction rate in these cases is rapid relative to rates of diffusion, the concentration of a reacting species at the reaction interface is either above or below the bulk concentration. The condition that exists depends on whether the activity of the reactive species in the second phase is higher or lower than the activity in the bulk. The model, frequently used in quantifying mass transfer limited kinetics, assumes that there is a thin stagnant film at the interface

across which a concentration gradient exists. The model is schematically represented in Fig. 5. Transport across the boundary layer is assumed to occur only by diffusion. Applying Fick’s first law of diffusion, the flux of species i across the boundary layer is: J¼

   D  C  CiB ¼ km Ci  CiB d i

(Eq 30)

where D is the diffusion coefficient, d is the boundary layer thickness, and Ci and CiB are concentrations of the diffusion species, respectively, at the interface and in the bulk. The term D/d is seldom used; instead, the mass transfer coefficient km is used to represent D/d. When phase 2 is a fluid, d decreases with increasing fluid velocity along the interface (that is, increased convection). Thus, km is a function of the fluid dynamic condition. To determine the rate of reaction, that is, J, it is necessary to know Ci . This is difficult to determine experimentally. To circumvent the problem, one of two assumptions is commonly made. For extremely fast reactions in which the equilibrium constant is very large, Ci can be assumed to be 0. Thus, Eq 30 reduces to: J ¼ km CiB

  J ¼ km Ciequil  CiB

(Eq 32)

The value of Ciequil is determined from the equilibrium constant expression, knowing the activities of all the reactants and products except Ciequil . A number of examples of this treatment are found in succeeding articles in this Section. There are cases in which the reaction rate (rate constant k) at the interface, unimpeded by diffusion, is of the same order as the mass transfer limited rate. In this case, the observed rate and the rate constant k0 are affected by both processes. This relationship is illustrated as follows. Because diffusion of i through the boundary layer and the reaction of i at the interface occur sequentially, the rates of both processes must be the same at steady state. Thus:     J ¼ km Ci  CiB ¼ k Ci  Ciequil

1 ¼ k0



1 1 þ km k

δ

Stagnant boundary layer

Fig. 5

Boundary layer model used for mass transfer limited kinetics

It is clear from Eq 34 and 35 that for the condition k  km, Eq 34 reduces to Eq 32. Applications in which mixed control is important are nitrogen absorption in iron alloy melts (Ref 17) and coke gasification at intermediate temperature (Ref 16).

Kinetics of Alloy Additions As previously noted, the chemical and physical processes involved in alloy addition do not vary much with base metal and can be treated generically. There are two fundamental mechanisms governing the kinetics of alloy additions, depending on whether the alloy is liquid or solid at the temperature of the metal bath. These processes are schematically illustrated in Fig. 6.

Dissolution of Alloy with Melting Point Lower than Bath Temperature

(Eq 33)

where the last term gives the first-order rate of reaction occurring at the interface. Eliminating Ci in the last two terms in Eq 33 yields:   J ¼ k0 Ciequil  CiB

C Bi

Phase 1

(Eq 31)

This assumption has been used to describe the high-temperature gasification of coke (Ref 16). The second assumption is that equilibrium between the two phases is established at the reaction interface. In this case, Eq 30 is written as:

where:

Phase 2 C *i

(Eq 34)

 (Eq 35)

For the case where the alloy addition is liquid at the melt temperature (Fig. 6, top row sequence), addition of the cold alloy at time t0 is quickly followed by formation of a layer of frozen base metal on the alloy surface (t1) due to heat extraction by the alloy. At longer times (t2), the shell thickness increases, while melting occurs at the alloy surface. As the interior of the alloy heats up, the thermal demand decreases. As a result, the frozen base metal shell begins to redissolve while the thickness of the liquid

36 / Principles of Liquid Metal Processing Superheat, ⬚F

Time t1

t0

t2

t3

t4

tm

840

Alloy liquid at bath temperature

32

48

64

80

96

112

60

70

720

Melting time, s

600

Alloy solid at bath temperature

480

360 C 240

Solid alloy Liquid alloy

B 120

Solid base metal

Fig. 6

A

Two kinetic paths for melting and/or dissolving alloy additions in liquid metal baths. Source: Ref 18

0

20

30

40

50

Superheat, ⬚C

Table 2 Data for melting time calculations for a 100 mm (4 in.) diameter sphere of silicomanganese in a steel melt Condition

Fig.

DH, J/g

r0, g/cm3

V0/A0, cm

h, J/g K

TL , K

ðTBL  TL Þ, K

t m, s

387 307

5.6 5.6

1.67 1.67

0.15 0.15

1803 1395

20 418

1200 45

Frozen shell No frozen shell Source: Ref 18

7

Predicted melting/dissolution times for silicomanganese spheres in quiescent steel baths versus steel superheat temperature. Particle diameters of the spheres are as follows: A, 25 mm (1 in.); B, 50 mm (2 in.); C, 100 mm (4 in.). Source: Ref 18

requirement for melting, divided by the rate at which heat is supplied from the bath (Ref 18), that is:

Table 3 Data for melting time calculations for a 10 mm (0.4 in.) diameter aluminum sphere in different liquid metal melts Base metal

DH, J/g

r0, g/cm3

V0/A0, cm

h, J/g K

TL , K

ðTBL  TL Þ, K

tm, s

Steel Cast iron Copper Aluminum

576 576 576 576

2.7 2.7 2.7 2.7

0.167 0.167 0.167 0.167

0.15 0.15 0.15 0.15

1823 1723 1323 1023

890 790 290 90

1.9 2.2 6.0 19.2

annulus increases (t3). In the final stage, beginning at t4, the redissolution of the base metal shell is complete, and as a result, the liquid alloy held in the annular space is released to the melt. Any alloy that remains solid is exposed directly to the liquid metal bath and is progressively incorporated into the melt by the surface melting and convection. Dissolution is completed at tm (melting time). The length of the incubation period (t0  t4) and the length of the alloy dissolution period (t4  tm) are governed by heat transfer. Conditions favoring rapid alloy dissolution are high superheat temperatures, small alloy particle size (large surface for heat transfer), and low alloy thermal conductivity. The heat flux to the alloy particle is given by:   Q ¼ h TLB  TL

(Eq 36) 2

1

where Q is the heat flux (J  m  s ), h is the convective heat transfer coefficient (J  m2  s1  K1), TLB is the temperature of

the bulk liquid (in degrees Kelvin), TL is the melting point temperature of the alloy at the solid/liquid  interface (in degrees Kelvin), and TLB  TL is the superheat temperature (in degrees Kelvin). During the incubation period, the superheat temperature is less than that in the alloy dissolution period because TL corresponds to the melting point of the base metal and alloy in the former and latter cases, respectively. For steel and aluminum melts, superheat temperatures in the incubation period are generally smaller (20 to 100 K) than for cast iron (200 to 300 K). In the alloy dissolution period, the superheat temperature for a given alloy addition increases with increasing melting point of the base metal. Thus, superheat temperatures, and therefore dissolution rates, increase in the following order: aluminum < copper < cast iron < steel. Melting times, tm, can be estimated by assuming that tm is equal to the total heat

tm ¼

Hr0 V0 QA0

(Eq 37)

where DH is the heat supplied to 1 g of alloy at the time it dissolves, and r0, V0, and A0 are the density, volume, and surface area of the alloy particle. The variable Q is defined by Eq 36. Equation 37 is most accurate when applied at the extremes of possible superheat temperatures. For small superheat temperatures, it can be assumed that by the time the frozen shell redissolves, the alloy is a liquid at the melting point of the base metal. At high superheat temperatures, it can be assumed that no frozen layer forms and that the alloy liquefies at its melting point. These two cases have been examined for the dissolution of a 100 mm (4 in.) particle diameter B(dp) sphere of  silicomanganese in a steel melt TL ¼ 1823 K . Data for the calculations are given in Table 2. The very long melting time of 1200 s obtained by restricting the calculations to a small superheat temperature (20 K) is in good agreement with the 900 s obtained with a more sophisticated model (Fig. 7). The difference in tm for the two extreme cases considered in Table 3 underscores the importance of superheat temperature in alloy dissolution processes. This is clearly illustrated in Fig. 7, in which the role of particle diameter is also demonstrated. It is apparent from Fig. 7 that the large value of tm in Table 2 for the case

Chemical Thermodynamics and Kinetics / 37 a common foundry practice that is necessary for control of cast iron composition. The rate at which carbon dissolves in cast iron is controlled by the rate of diffusion of carbon from carbon-saturated iron at the carbon/iron interface into the bulk liquid (Ref 21–23). Thus, Eq 38 can be written:

Density ratio 40

0

0.2

0.4

0.6

0.8

1.0

Liquid iron-carbon alloy

Carbon

C sat * Carbon concentration

20

10

Total immersion time, s

6 4

d 2 Distance C

1

Fig. 9

0.6

B

0.4

A

0.2

0.1 0

1

2

3

4

5

6

7

Addition density, g • cm-3

Fig. 8

Total immersion times versus density ratio for alloys entering a stagnant steel bath at 7.67 m  s1 (25.2 ft  s1) (drop height: 3.0 m, or 10 ft). Particle diameters are as follows: A, 10 mm (0.4 in.); B, 50 mm (2 in.); C, 250 mm (10 in.). Source: Ref 18



 J ¼ km ðCsat  CLB Þ

C BL

 TLB  TL ¼ 20 and dp = 100 mm can be substantially  B reduced with moderate changes in TL  TL and dp. The range of melting times obtained with the addition of an alloy to different base metals is examined subsequently. Consider the addition of pure aluminum to melts of steel, cast iron, copper, and aluminum under conditions where no frozen shell is formed. In this case, TL is the same (933 K), but TLB varies. Pertinent data for the calculation and the calculated melting times are given in Table 3. It is clear from Table 3 that the duration of alloy melting becomes a more serious problem as the melting point of the base metal decreases. The short melting times predicted for steel and cast iron melts in Table 3 highlight another important consideration in alloy addition, namely, immersion time due to alloy buoyancy. Because iron is denser than common alloy additions such as carbon, silicon, and aluminum, buoyant forces prevent these alloys from being easily submerged for times approaching tm. This is illustrated in Fig. 8, which plots immersion times versus the density ratio, ralloy/ rbase metal. For 10 mm (0.4 in.) aluminum particles (density ratio: 0.39) that are dropped

Carbon concentration profiles for a carbon rod dissolving in an iron-carbon melt

a distance of 3 m (10 ft) to the surface of an iron bath, the immersion time is less than 1 s, while the melting time (from Table 3) is 1.9 s. Immersion times can be increased by entraining particles in strong recirculating flows or by wire and rod feeding (Ref 19, 20). These conditions have higher associated heat transfer coefficients, which further facilitate dissolution.

Dissolution of Alloy That Is Solid at Bath Temperatures The processes involved in the dissolution of this type of addition are schematically illustrated in the bottom row sequence in Fig. 6. As in the previous case, heat transfer controls the duration of the incubation period (t0  t4). However, in this case, mass transfer is the controlling factor in the alloy dissolution period (t4  tm). In general, mass transfer controlled dissolution is slower than heat transfer controlled dissolution, and the number of variables controlling the rate is greater. For mass transfer control, melt temperature and fluid dynamic considerations are important, as in the heat transfer controlled case. However, melt composition and diffusion variables are also important. The flux of alloy element to the melt is given by: J ¼ km ðCL  CLB Þ

(Eq 38)

where J is the mass flux of alloy element (g  cm2  s1), km is the mass transfer coefficient (cm  s1), CLB is the concentration of the alloy element in the bulk melt (g  cm3), and CL is the concentration of the alloy element in the melt at the solid/liquid interface (g  cm3). To illustrate mass transfer limited dissolution processes, two industrially important examples are examined as follows. Carbon Dissolution in Cast Iron. Carbon is the most important alloy element in cast iron. The addition of carbon to iron-carbon melts is

(Eq 39)

 where Csat indicates carbon-saturated conditions at the carbon/melt interface. The conditions in the melt are qualitatively described in Fig. 9. Referring to Fig. 9 and Eq 39, the flux of carbon, and therefore the dissolution rate, increases with decreasing bulk carbon concentration CLB , increasing carbon concentration  at saturation Csat (increasing bath temperature), and decreasing boundary layer thickness, d. The boundary layer thickness is a function of the fluid velocity, and it affects the value of km.   CLB ) on The effect of the difference (Csat the dissolution rate of carbon is illustrated in Fig. 10 with data for the dissolution of graphite granules in induction-stirred cast iron melts. Theoretical curves are drawn for different rate constants k and particle shape factors l, where l is defined as:

¼

Sparticle Ssphere

(Eq 40)

where Sparticle is the surface-to-volume ratio of the graphite particles, and Ssphere is the equivalent surface-to-volume ratio if the particles are spherical. The role of boundary layer thickness (fluid dynamics) on dissolution kinetics is illustrated in Fig. 11, which indicates the variation of the rate constant (dissolution rate) with the peripheral velocity (rotational speed) of a cylindrical graphite rod. The line drawn through the data has a slope of 0.7, which is expected from earlier correlations (Ref 24). Dissolution of Steel in Iron-Carbon Melts. Another well-established case of mass transfer controlled dissolution is the dissolution of steel in iron-carbon melts (Ref 25–28). This process is critical in the electric melting of steel and cast iron. For this case, the rate of steel dissolution increases with melt carbon concentration (Fig. 12), melt temperature (Ref 27), and bath velocity (Ref 25, 26, 28) (affecting the boundary layer thickness). The data can be accounted for on the basis that carbon diffuses from the bulk of the melt to the steel/liquid iron interface, where steel is isothermally converted to a liquid with a liquidus composition, CL‘ . The conditions existing in the bath are illustrated in Fig. 13. Figure 13(a) shows the carbon concentration profiles in the liquid bath in the vicinity of a dissolving steel bar. The meaning of the carbon concentration notation is indicated in Fig. 13(b). As indicated in Fig. 13(a), carbon diffuses across the liquid boundary layer, while equilibrium conditions are maintained at the solid/liquid interface. Thus, the carbon concentrations in the solid and the liquid

38 / Principles of Liquid Metal Processing 1.0

1000 15

800

λ = 2.0

10

600 λ = 2.5

0.8

Time to dissolve, s

200

3 λ = 1.4 2

100

λ = 1.6

80 60

k = 0.007 λ = 1.6

1

k = 0.007

Time to dissolve, min

5

0.6

0.4

0.2

k = 0.02

40

Fractional thickness, r /r0

400

B A

k = 0.02 20

0 k = 0.03

0

10

20

30

40

Immersion time, min 10 0

0.5

1.0

2.0

1.5

2.5

3.0

Fig. 12

Melting curves for 75 mm (3.0 in.) diameter steel rods immersed in iron-carbon baths. A, high (4.6% C) iron-carbon bath; B, low (2.1% C) ironcarbon bath. Plotted is the fractional thickness (the fraction of the rod diameter remaining undissolved) versus immersion time. Source: Ref 27

3.5

2 Weight fraction of solution (C*sat – CB L ) ⫻ 10

Fig. 10

 Time of dissolution of carbon particles (10 + 14 mesh) versus (Csat  CLB ). Source: Ref 23

J ¼ km ðCLB  CL‘ Þ

0.1

(Eq 41)

The carbon flux can also be defined in terms of a carbon balance:  B ‘  r C dr J ¼  S ‘L  CSB dt rL

Mass transfer coefficient (km), cm • s−1

0.05 0.7 k = (const) V i

where rBS and r‘ L are the respective densities of solid steel and liquid iron at the liquidus composition, CSB is the bulk carbon concentration of solid steel, and dr/dt is the rate of change of the radius or thickness of the steel with time. By combining Eq 41 and 42 and integrating between the limits r = r0 at t = 0 and r = 0 at t = tm, the isothermal melting time tm is given as:

0.02

0.01

(C *sat – C BL )

tm ¼

0.8–2.2 wt % 2.2–3.7 wt % 3.7–5.0 wt %

0.005

0.002 2

5

10

20

50

100

200

400

Peripheral velocity (Vi), cm • S−1

Fig. 11

Experimental mass transfer coefficient versus peripheral velocity for the dissolution of a rotating carbon rod in an iron-carbon melt. Source: Ref 22

at the interface are the isothermal solidus Css and the liquidus CL‘ concentrations, respectively. Dissolution takes place by isothermal conversion of the steel into a liquid, that is, by a chemical melting process. It is clear from

(Eq 42)

Fig. 13(b) that as temperature increases, CL‘ decreases significantly. Thus, the driving force for diffusion increases with temperature. Based on Eq 38 and Fig. 9, the flux of carbon entering the steel is:

B ðrBS CL‘ =r‘ L Þ  CS r0 ‘ B km ðCL  CL Þ

(Eq 43)

In laboratory studies, Eq 43 accurately predicted steel dissolution rates in a stagnant bath (Ref 27). Research has suggested that melting times are reduced by approximately a factor of 2 in induction-stirred melts (Ref 26). Other work has indicated that tm can be reduced by an order of magnitude with strong convection (Ref 28). On the basis of Eq 43 and the mass transfer correlation for rapidly dissolving rods (Ref 29), correlations were developed relating tm to bath temperature and to carbon concentration (Ref 27). This is given in Fig. 14 for a steel thickness of 2.5 mm (0.098 in.). Because tm is directly proportional to the thickness of the steel, Fig. 14 can be used to estimate tm for a wide range of steel thickness by making the appropriate thickness correction. Checks of the

Chemical Thermodynamics and Kinetics / 39

Steel

Liquid ironcarbon alloy

4. 5.

C BL

Carbon concentration

6.

7.

C *Ll

8. 9.

C *Ss δ

10. C BS

11. Distance (a)

12. 13.

1600 δ+L

2600

C BL

15. 2400

1300 γ+L

1200

2200

1100

Temperature, ⬚F

C *Ll

C *Ss

C BS

1400 Temperature, ⬚C

14.

2800

Liquid alloy

δ

1500

2000 γ + Fe3C

1000

16. 17.

1800

900 0

0.5

Fig. 13

1

2

3

4

5

18.

Carbon, wt%

(b)

(a) Carbon concentration profiles for a steel rod dissolving in an iron-carbon melt. (b) Iron-carbon phase diagram defining the carbon concentrations noted in (a). Source: Ref 27

predictions from Fig. 14 in production operations have indicated that predicted tm values are realistic (Ref 27). ACKNOWLEDGMENT Revised from the articles “Principles of Physical Chemistry” and “Composition Control” by Seymour Katz in Casting, Vol 15, Metals Handbook, 9th ed., ASM International, 1988, p 50–55 and 71–81

REFERENCES 1. Y.K. Rao, Stoichiometry and Thermodynamics of Metallurgical Processes, Cambridge University Press, 1985, p 285–287, 298, 383–394, 880–891 2. E.T. Turkdogan, Physical Chemistry of High Temperature Technology, Academic Press, 1980, p 5–26, 81, 82 3. G.R. Belton and R.J. Fruehan, The Determination of Activities by Mass Spectrometry. I. The Liquid Metallic Systems Iron-

19. 20.

21.

22.

Nickel and Iron-Cobalt, J. Phys. Chem., Vol 71, 1967, p 1403–1409 G.K. Sigworth and J.F. Elliott, The Thermodynamics of Liquid Dilute Iron Alloys, Met. Sci., Vol 8, 1974, p 298–310 G.K. Sigworth and J.F. Elliott, The Thermodynamics of Dilute Liquid Copper Alloys, Can. Metall. Q., Vol 13, 1974, p 455–461 G.K. Sigworth and T.A. Engh, Refining of Liquid Aluminum—A Review of Important Chemical Factors, Scand. J. Metall., Vol 11, 1982, p 143–149 C. Wagner, Thermodynamics of Alloys, Addison-Wesley, 1962, p 51 L.S. Darken, Thermodynamics of Binary Metallic Solutions, Trans. Metall. Soc. AIME, Vol 239, 1967, p 80–89 L.S. Darken, Thermodynamics of Ternary Metallic Solutions, Trans. Metall. Soc. AIME, Vol 239, 1967, p 90 R. Hultgren et al., Selected Values of the Thermodynamic Properties of Binary Alloys, American Society for Metals, 1973 L.S. Darken, Thermodynamics and Physical Metallurgy, American Society for Metals, 1950 L.S. Darken and R.W. Gurry, Physical Chemistry of Metals, McGraw-Hill, 1953, p 235 Aluminum Casting Technology, American Foundrymen’s Society, 1986, p 21 M.J. Lalich and J.R. Hitchings, Characterization of Inclusions as Nuclei for Spheroidal Graphite in Ductile Cast Irons, Trans. AFS, Vol 84, 1976, p 653–664 B. Francis, Heterogeneous Nuclei and Graphite Chemistry in Flake and Nodular Cast Irons, Metall. Trans. A, Vol 10, 1979, p 21–31 S. Katz, The Properties of Coke Affecting Cupola Performance, Trans. AFS, Vol 90, 1982, p 825–833 R.J. Fruehan, B. Lally, and P.C. Glaws, A Model for Nitrogen Absorption in Iron Alloy Melts, Fifth International Iron and Steel Congress Process Technology Proceedings, Vol 6, The Iron and Steel Society of AIME, 1986, p 339–346 R.I.L. Guthrie, Addition Kinetics in Steelmaking, Proceedings of the 35th Electric Furnace Conference, Iron and Steel Society of AIME, 1978, p 30–41 F.A. Mucciardi and R.I.L. Guthrie, Aluminum Wire Feeding in Steelmaking, Trans. ISS, Vol 3, 1983, p 53–59 J.W. Robison, Jr., “Ladle Treatment with Steel-Clad Metallic Calcium Wire,” Paper 35, presented at Scaninject III, Part II, MEFOS, Lulea, Sweden, 1983 L. Kalvelage, J. Markert, and J. Po¨tschke, Measurement of the Dissolution of Graphite in Liquid Iron by Following the Buoyancy, Arch. Eisenhu¨ttenwes., Vol 50, 1979, p 107–110 R.G. Olsson, V. Koump, and T.F. Perzak, Rate of Dissolution of Carbon in Molten

40 / Principles of Liquid Metal Processing Bath temperature, ⬚F 2400

2300 3.5

2500

3.0

2600

2700

2% C bath 4% C bath

1.5% C bath

3% C bath

Melting time, min

2.5

2.0

1.5

1.0

5% C bath 0.5

0 1260

1315

1370

1425

1480

Bath temperature, ⬚C

Fig. 14

Predicted melting times for vertical steel plates and cylinders, 2.5 mm (0.098 in.) scrap thickness and 0.1% C composition, [(%S)i = 0.03] immersed in stagnant iron-carbon baths at different bath temperatures. Source: Ref 27

23. 24.

25.

26.

Fe-C Alloys, Trans. Met. Soc. AIME, Vol 236, 1966, p 426–429 O. Angeles, G.H. Geiger, and C.R. Loper, Jr., Factors Influencing Carbon Pickup in Cast Iron, Trans. AFS, Vol 74, 1968, p 3–11 M. Eisenberg, C.W. Tobias, and C.R. Wilke, Mass Transfer at Rotation Cylinders, Chem. Eng. Prog. Symp. Series, Vol 51, 1955, p 1–16 R.G. Olsson, V. Koump, and T.F. Perzak, Rate of Dissolution of Carbon Steel in Molten Iron-Carbon Alloys, Trans. Met. Soc. AIME, Vol 233, 1965, p 1654–1657 R.D. Pehlke, P.D. Goodell, and R.W. Dunlap, Kinetics of Steel Dissolution in Molten

Pig Iron, Trans. Met. Soc. AIME, Vol 233, 1965, p 1420–1427 27. R.I.L. Guthrie and P. Stubbs, Kinetics of Scrap Melting in Baths of Molten Pig Iron, Can. Metall. Q., Vol 12, 1973, p 465–473 28. K. Mori and T. Sakuraya, Rate of Dissolution of Solid Iron in Carbon-Saturated Liquid Iron Alloys with Evolution of CO, J. Iron Steel Inst. Jpn., Vol 22, 1982, p 964–990 29. P.T.L. Brian and H.B. Hales, Effects of Transpiration and Changing Diameter on Heat and Mass Transfer to Spheres, AICHE J., Vol 15, 1969, p 419–425

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 41-55 DOI: 10.1361/asmhba0005190

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Thermodynamic Properties of Iron-Base Alloys Doru M. Stefanescu, Ohio State University Seymour Katz, S. Katz Associates

THE STRUCTURE AND PROPERTIES of cast metals are sensitive to numerous impurities. Because purification of melts generally adds considerable cost to castings, the lowest cost and surest defense against contamination is careful selection of scrap. Purification is generally reserved for elements that are so pervasive that avoidance is impossible. This is exemplified by sulfur and oxygen removal from cast iron and steel, respectively, and removal of alkali and alkaline earth elements from aluminum. Because of their immediate importance, aspects of the physical chemistry of these processes are reviewed. The structures and properties of steel and cast iron also depend on the control of the solidification structure in castings. To accomplish this, it is necessary to understand and control the thermodynamics of the liquid and solid phases and the kinetics of solidification (nucleation and growth of various phases). The information linked to the practical interests that thermodynamics can provide when considering iron-base alloys encompasses a rather wide range. Only two issues are addressed in sections on thermodynamics in this article. The first is the calculation of solubility lines, relevant to the construction of phase diagrams, and the second is the calculation of the activity of various components, which allows for the determination of probability of formation and relative stability of various phases. Alloying elements are then discussed in terms of their influence on the activity of carbon, which provides information on the stability of the main carbonrich phases of iron-carbon alloys, that is, graphite and cementite. Information on the iron-oxygen system can be found in the article “InclusionForming Reactions” in this Volume.

the level of inclusions, leading to stronger and more fatigue-resistant steels. For cast iron, desulfurization is practiced in the manufacture of ductile iron castings in order to develop spherical graphite morphology. Ductile iron is used in applications where high fracture toughness is needed. Sulfur is removed from iron and steel when the metals are liquid. Although a variety of reagents are employed to remove sulfur, namely, calcium, magnesium, sodium, and rare earths, the most important is calcium. Common forms of calcium include the metal; alloys such as calcium silicon (CaSi); the oxide, calcium oxide (CaO); and the carbide, calcium carbide (CaC2). Despite the use of various forms of calcium, the governing chemical reaction in all cases appears to be the same (Ref 1, 2): CaO þ S ¼ CaS þ O

(Eq 1)

The equilibrium constant for the reaction (Eq 1) is: aCaS hO K9 ¼ aCaO hS

(Eq 2)

For both cast iron and steel, target sulfur concentrations after desulfurization are in the range of 0.006 to 0.010% S. In electric-melted cast iron and steel, the sulfur levels before desulfurization are 0.02 to 0.03% S, while input sulfur levels to the cupola are generally much higher—0.1 to 0.2% S. Requirements for Desulfurization. For reasons related to reaction kinetics and thermodynamics, the final sulfur (%S)f concentration achieved in desulfurization processes depends on three factors:

Purification of Ferrous Melts

 Initial sulfur concentrations (%S)i  Amount of desulfurizer used, usually expressed

One of the most important processes involved in cast iron and steel production is desulfurization. For steels, desulfurization is necessary to reduce

as the weight fraction of desulfurizer to metal, W  Extraction efficiency of the desulfurizer, which is measured by the desulfurization

ratio (DR), that is, the ratio of sulfur concentrations: desulfurizer to metal The three variables can be related as follows through a mass balance on sulfur: ð%SÞf ¼

ð%SÞi 1 þ W ðDRÞ

(Eq 3)

For liquid-desulfurizing slags that are not saturated with respect to calcium sulfide (CaS), the maximum value of DR, that is, the equilibrium value, can be predicted from thermodynamic considerations: ðDRÞmax ¼

  ð%SÞslag ð%SÞT

¼ max

CS fS K14 K15 hO

(Eq 4)

where CS is the slag sulfide capacity, defined as (Ref 3): CS ¼ ð%SÞslag

 1=2 pO2 pS2

(Eq 5)

and K14 and K15 are the respective equilibrium constants for the reactions: 1 O2 ¼ O 2

(Eq 6)

1 S2 ¼ S 2

(Eq 7)

and fS is the activity coefficient for 1 wt% S in the standard state. Equation 4 can be derived from Eq 2. Using known input and desired output sulfur values, Eq 3 gives the required desulfurization ratios as a function of weight fraction desulfurizer. These data are plotted in Fig. 1 for two cases. In the electric-melting case, initial and final sulfur concentrations were assumed to be 0.03 and 0.008% S, respectively. In the cupola-melting

42 / Principles of Liquid Metal Processing

Weight ratio, Wdesulfurizer/Wiron

1

Table 1 Theoretical weight ratios: desulfurizer to iron to achieve 0.008% S in various systems T, K

CS ( 104)(a)

fS

ho ( 104)(b)

44 CaO-15MgO-5Al2O3-36SiO2 CaOsat-Al2O3

1773 1773

2.7 59

5 5

1.3 5.0

90 417

CaOsat-Al2O3 CaOsat-Al2O3

1773 1873

59 316

5 1

5.0 0.45

417 5280

Type of melt

A. cast iron—cupola B. cast iron—cupola (ladle desulfurized) C. cast iron—electric D. steel—electric

0.1 B A

100

(DR)max(c)

W

0.13 0.027 0.0065 0.00052

(a) CS is obtained from optical basicity correlations (Ref 3). fs is based on iron composition (Ref 4). (b) Case A: ho is governed by Si/SiO2 equilibrium based on respective concentration in iron and slag (Ref 5). Cases B and C: ho is governed by Si/SiO2 equilibrium with aSiO2 ¼ 1 due to ladle exposure to air (Ref 6). Case D: ho is governed by Al/Al2O3 equilibrium with %Al = 0.05 (Ref 1), assumed no contact with air. (c) K14 and K15 data from Ref 7

0.01

10-3 10

Slag composition

103

104

104

Desulfurization ratio (DR), (% S)slag/(%S)iron

Fig. 1

 The best desulfurizing cupola slags have lower

sulfide capacity and desulfurization ratio than slags used in ladle desulfurization; the consequence is the need for much larger quantities of slag*.  Compared to steel, cast iron desulfurization benefits from higher fS because of the presence of relatively high concentrations of carbon and silicon in cast iron.  Ladle desulfurization systems that are exposed to air suffer higher hO and, as a result, poorer desulfurization than may otherwise be anticipated. For the cupola slag case in Table 1, the thermodynamically predicted value of (DR)max = 90 is in good agreement with measured data (Fig. 2). This indicates that the cupola desulfurization process operates close to equilibrium levels. Further evidence for this is given Fig. 3, which plots desulfurization data for a cupola operated with varying amounts of municipal ferrous refuse in the charge. The upper portion of Fig. 3 plots CS and oxygen activity. The latter is expressed as the partial pressure of oxygen. *The actual differences are less than those indicated. Calcium sulfide has only limited solubility in CaO-Al2O3 slags. As a result, a minimum W = 0.01 is needed to maintain the CaS-unsaturated condition.

103 Desulfurization ratio, aCaS/as

case, the equivalent concentrations were 0.10 and 0.008% S. The cupola line applies for both cupola iron and ladle-desulfurized cupola iron. The relationships illustrated in Fig. 1 are useful in defining systems that will provide the necessary desulfurization conditions. Four systems are compared in Table 1. These cover cupola- and electric-melted cast iron and electric-melted steel. Also examined are two liquid slag systems: a basic cupola slag (dicalcium silicate saturated) and a CaO-saturated CaOAl2O3 slag. Table 1 shows that:

100

Desulfurization ratio, (%S)slag /(%S)iron

Weight fraction desulfurizer required to achieve given desulfurization ratios. Plots for electricmelted iron and steel (curve A) assume initial sulfur concentrations [%(S)i = 0.03] different from those of cupola-melted iron [(%S)i = 0.10] (curve B).

100

10

10 0.7

0.9

1.1

1.3

1.7

1.5

1.9

2.1

2.3

Slag basicity

Fig. 2

Desulfurization ratio-basicity comparisons for cupola (closed circles) and laboratory data (open circles). Equilibrium values are indicated by the angled line. The vertical line indicates the slag basicity above which slags are saturated at 1500  C (2730  F), with respect to dicalcium silicate. This is the point at which the observed DR should equal (DR)max. Source: Ref 2

The lower portion of Fig. 3 is a comparison of (DR), measured and calculated. The good agreement found is evidence that near-equilibrium conditions existed. For the cases discussed previously, the oxygen activity in cupola iron has been found to be governed by the reaction (Ref 5, 8):

Mn þ O ¼ MnO

(Eq 8)

Therefore, the overall cupola desulfurization reaction is: CaO þ S þ Mn ¼ CaS þ MnO

(Eq 9)

Thermodynamic Properties of Iron-Base Alloys / 43

106

0.01

%C ⬚C 3.7 1550 4.0 3.7 1500 4.0

10−3

10−5 100 Desulfurization ratio

Inlet S 0.015 – 0.03% Inlet S 0.06 – 0.08%

(6 ⫻ 10−4)

Measured Calculated

50

0

0.5

1.0

1.5

3.7 4.0

1450

3.7 4.0

1400

2.0

2.5

3.0

Silicon concentration in iron, wt %

20 10

Fig. 4

Equilibrium sulfur concentrations for CaO desulfurization calculated with Eq 11 and 12, assuming aCaO ¼ aCaS ¼ aSiO2 ¼ 1. Source: Ref 2

5

Measured sulfur concentration, wt %

10−4

Sulfur concentration in iron, wt %

107

10

(po2)−1/2

Slag sulfide capacity

Cs (p o2)−1/2 Mn −3

0.1

(0.02)

108

0.01

0.01

2 1

0

20

40

60

80

100

Municipal refuse in charge, %

Fig. 3

Slag sulfide capacities, oxygen activities, and desulfurization ratios, measured and calculated, for a cupola operated with municipal ferrous refuse as a charge material. Source: Ref 8

Considerably lower sulfur could be achieved (Ref 2) if hO were governed by: C þ O ¼ CO

(Eq 10)

However, equilibrium for this reaction has not been observed. This discussion has concerned CaS-unsaturated slags. However, a desulfurizing slag, saturated with CaS, can in many cases continue to desulfurize as long CaO is present. In this case, the ultimate sulfur levels are not as low as those for CaS-unsaturated slags. Nevertheless, CaSsaturated slags can produce sulfur levels that are more than sufficient for cast iron and steel applications. In fact, CaS-saturated conditions are often employed in industrial applications because much less desulfurizer is required. In addition to the CaS-saturated liquid slags, totally solid slags such as CaO and CaC2 are in the same category, because CaS does not form solid solutions with these materials. Also in this category are desulfurizers containing small amounts of liquid phase. These materials possess the desirable properties of liquid and solid slags. That is, they combine the fast reaction rates of liquid slags with the large desulfurizing capacities (high CaO concentration) of solid slags. The final sulfur concentrations attainable under CaS-saturated conditions can be obtained from Eq 2 by setting aCaS = 1. For ladle desulfurization, the most important industrial desulfurizers are also CaO-saturated, that is, aCaO = 1.

Applying the condition of double saturation to Eq 2 yields:

1 10−3

ð%SÞr ¼

hO K9 fS

(Eq 11)

In the ladle desulfurization of cast iron, hO is governed by the reaction (Ref 2): 1 1 Si þ O ¼ SiO2 2 2

1

(Eq 12)

where aSiO2 ¼ 1. This is attributed to the exposure of the iron to air (Ref 2). Sulfur concentrations obtained with Eq 11, assuming siliconsilicon dioxide (Si-SiO2) equilibrium and aSiO2 ¼ 1, are given in Fig. 4. In continuous ladle-desulfurization processes, equilibrium sulfur levels are achieved when input sulfur levels are low but are only approached when input levels are high. This is illustrated in Fig. 5 with desulfurization data for CaC2 (Ref 2). Other desulfurizers such as CaO-CaF2 behave similarly. High desulfurization rates are needed for effective desulfurization in the short times required. For CaO-base desulfurizers, the presence of relatively small amounts of liquid phase (<25 vol%) significantly increases the rate of desulfurization. This is illustrated in Fig. 6, in which the rates of desulfurization by CaO with varying amounts of calcium fluoride (CaF2) are compared. The faster rates obtained with CaF2 additions were due to the formation of a CaOCaF2 liquid phase that provided a path for CaS diffusion from the reaction interface into the porous CaO particle. This prevented the normal rapid development of an impervious CaS coating on the CaO surface (Ref 10, 11). A similar explanation, involving liquid-phase formation, was used to account for the higher

(6 ⫻ 10−4) 0.0006

0.001

0.002

0.004

Theoretical sulfur concentration, wt%

Fig. 5

Comparisons of the sulfur concentrations in production-continuous desulfurization using CaC2 with the sulfur concentrations from Eq 11 and 12. Open circles indicate electric-melted iron. Closed circles indicate cupola iron. Source: Ref 2

CaO desulfurization rates of steel when aluminum was concurrently added (Ref 12, 13). Another rate-controlling variable in ladle desulfurization is bath agitation. For gas-stirred melts, the desulfurization rate constant k is _ that is, k / Q_ n . a function of gas flow rate Q, For well-dispersed solid-liquid mixtures, the rate of diffusion-controlled interphase reaction _ that is, n  is a relatively weak function of Q, 0.2 to 0.4 (Ref 14–16). For a poorly dispersed system, such as a ladle of iron with a cover slag of desulfurizer, agitation has a much greater influence on the rate constant, with n typically in the range of 1.0 to 2.5 (Ref 14–16). This is due to the entrainment of increasing amounts of top slag into the liquid metal with increasing _ This effect is illustrated in Fig. 7, which Q. plots the apparent mass transfer coefficient for desulfurization versus the gas flow rate. Deoxidation. In the manufacture of steel, FeO-saturated conditions (0.2%O) are approached as a result of oxygen blowing to remove carbon and silicon as the respective oxides. To make a steel product, the oxygen levels are subsequently lowered to 0.005 to 0.02% O by reactive alloy additions of

44 / Principles of Liquid Metal Processing 1

0% CaF2

0.5

2%

0.2

%S/ %So

8%

0.1

5% CaF2

0.05

20%

10% 0.02

0.01 0

10

20

30

40

50

60

Time, min

Fig. 6

Desulfurization rates of carbon-saturated iron, containing 0.4% Si, with CaO and varying amounts of CaF2 at 1450  C (2640  F). Source: Ref 9

For single-element deoxidation, expressed by:

Desulfurization rate constant (k ), s -1 ⫻ 10-3

10

xM þ yO ¼ Mx Oy

5

(Eq 13)

the equilibrium oxygen concentration in the liquid steel, obtained by rearranging the equilibrium constant expression, is:

2 1

•

k a Q0.25

•

k a Q 2.1

0.5

 ð%OÞ ¼

0.2 0.05 0.1

0.2

0.5

1

2

5

10

Gas flow rate (Q ) , m3 • kg-1 • s-1 ⫻ 10-6

Fig. 7

Dependence of CaO desulfurization rate constant on the rate of gas flow through a reactor. Source: Ref 16

electropositive elements. The low oxygen concentrations are necessary to maintain the number of oxide inclusions at a suitable low level, to prevent the formation of CO blowholes, and to ensure effective desulfurization.

a

Mx Oy x y K22 ð%MÞx fM fO

y1 (Eq 14)

The calculated equilibrium values of oxygen solubility in liquid iron at 1600  C (2910  F), based on Eq 14, are given in Fig. 8 for several elements as a function of the concentration of the element, assuming aMx Oy ¼1 . As seen among the common deoxidizers, aluminum produces the lowest oxygen. Although not shown, rare earth metals produce even lower oxygen (Ref 17). Figure 8 plots data for oxygen concentration and oxygen activity. In all cases, oxygen activity decreases with increasing levels of deoxidizer, but the oxygen concentration can go through a minimum. This occurs in cases where the

interaction coefficient eM O is a large negative number, and as a result, fO decreases significantly even at relatively low concentrations of M. To obtain exact values for [%O] or hO, equations for K22 can be found in Ref 1 and 7. As indicated by Eq 14, the oxygen concentration can be reduced if a complex deoxidation product is formed so that aMx Oy <1. Common deoxidation systems of this type are Si-Mn, Al-Si-Mn, or Al-CaO. An advantage, in addition to lower oxygen, is that less deoxidizer is required in solution to achieve a given level of oxygen. This is illustrated in Fig. 9, which plots the oxygen activity of iron as a function of the concentration of CaO in the calcium aluminate inclusion. Separate lines are given for aluminum concentrations ranging from 0.001 to 0.05% Al. Also indicated is the oxygen activity when pure aluminum oxide (Al2O3) is the reaction product. The data in Fig. 9 show that an oxygen activity of 4 ppm can be produced at three aluminum levels—0.002, 0.005, or 0.01%—depending on whether the respective deoxidation product was CaO-saturated calcium aluminate, CaAl2O4-saturated calcium aluminate, or Al2O3. The levels of total oxygen measured in steel melts considerably exceed equilibrium levels (Ref 7). Two possible explanations are the kinetic limitations in the deoxidation reaction and the ineffective separation of deoxidation products from the melt. Consideration of the first possibility suggests that the most important kinetic limitation would be oxygen and/or deoxidant solute diffusion to inclusions in the melt (Ref 7). Considering the presence of 105 to 107 inclusion particles in a cubic centimeter of melt, deoxidation to equilibrium levels of 10 to 20 ppm would require a few minutes. Appreciably longer times would be needed to reduce oxygen to less than 1 ppm, as with rare earth additions. For most cases, oxide formation appears faster than oxide particle separation from the bath. Thus, for applications that depend on dissolved oxygen, such as desulfurization or CO blowhole formation, equilibrium oxygen values can be assumed to exist after relatively short reaction periods. For considerations of the inclusion content of steel, the kinetics of particle separation need to be considered. The limiting case of particle flotation at velocities obtained from Stokes’ law calculations is shown in Fig. 10. Inclusion size versus time of flotation is plotted. The particles produced during deoxidation are generally small (<5 mm, or 200 min.); accordingly, separation times are long. The times indicated in Fig. 10 are representative of conditions in a quiescent bath. Times are considerably shorter for stirred melts because articles grow by collision and coalescence to sizes of 30 to 100 mm (Ref 7). The efficiency of coalescence of particles is material dependent. The energy of adhesion of particles is determined by the wetting characteristics of the particle, measured by wetting angle. For example, the force of adhesion of Al2O3

Thermodynamic Properties of Iron-Base Alloys / 45 20

1

0.002 % Al 0.001

Oxygen concentration Oxygen activity ao ∫ wt% O for %X Æ 0 Iron oxide saturation

Oxygen activity of iron, ppm 0

V2O4

0.1

MnO Cr2O4

SiO2 Oxygen is solution, wt %

16

B2O3

CO

Cr2O3

B2O3

V 2O 3

0.01

Ti2O3

10-3

0.005

0.01

0 30

0.01

CaO saturated

0.05 40

50

60

70

CaO concentration, mass%

ZrO2

Fig. 8

8

0.05

Al2O3

Fig. 9

Ti2O3 0.01

0.002 0.005

4 CaAl2O4 saturated

Ti3O5

10-4 -3 10

12

0.1 Deoxidizer in solution, wt %

1

10

Concentration of oxygen and aluminum in liquid steel in equilibrium with calcium aluminate at 1600  C (2910  F); arrows indicate oxygen values when the reaction product is pure Al2O3. Source: Ref 18

Deoxidation equilibria in liquid iron alloys at 1600  C (2910  F). Source: Ref 17

107

106

10

15 105

20 30 40 0.1

A

B

1

C

10

Z, number of nuclei • cm−3

Inclusion size for 0.05%O removal, µm

5

100

Flotation time, min

Fig. 10

Calculated time of flotation of inclusions in stagnant melts as a function of inclusion size. Melt depth: A, 50 mm (2 in.); B, 500 mm (20 in.); C, 2000 mm (80 in.). Source: Ref 7

particles with a wetting angle of 140 is twice that for SiO2 with a wetting angle of 115 . On this basis, Al2O3 inclusions are expected to cluster more easily than SiO2 inclusions, a phenomenon that has been observed in practice (Ref 7).

Thermodynamics of Ferrous Systems For binary systems, only the iron-carbon and iron-silicon systems are discussed in this article. Information on the iron-oxygen system is available in the article “Inclusion-Forming Reactions” in this Volume.

Fig. 11

Phase diagram for the binary iron-carbon system. Source: Ref 19

The Iron-Carbon System The phase diagram of the binary iron-carbon system includes the stable (iron-graphite) and

metastable (Fe-Fe3C) equilibria. The most recent diagram is shown in Fig. 11. The solubility of carbon in iron is described in Eq 15–20, Eq 21. (Ref 20, 21).

46 / Principles of Liquid Metal Processing For the stable system, that is, with graphite as the equilibrium high-carbon phase, Eq 15–17 are used. For the liquid:

The general equation used to describe the activity of a component 2 in a component 1 according to the Darken formalism (Ref 22) is:

wt%Cmax ¼ 1:3 þ 2:57  103 t

log

(Eq 15) 

in the interval 1152 to 2000 C (2106 to 3632  F), with t being the temperature in degrees Celsius. Using thermodynamic quantities, Eq 15 can be written as: log XCmax

12:728 ¼ T þ 0:727 log T  3:049

(Eq 16)

where XCmax is the maximum solubility of graphite in liquid iron in mole fraction, and T is the temperature in degrees Kelvin. Equation 16 is valid for the interval 1425 to 2300 K. For the austenite: wt%Cmax ¼  0:435 þ 0:355  103 t þ 1:61  106 t2

wt%Cmax

(Eq 18)

(Eq 19)

For the austenite:

(Eq 23)

1180  0:87 T

(Eq 24)

Using the natural logarithm (base e) instead of the common logarithm (base 10), Eq 22 can be written as follows for the iron-carbon systems: 2714 ln gC ¼ 2 T 
2920 þ þ 4:01 2XC  XC2 T

(Eq 25)

  153 aFe-Si ¼  þ 3:02 T

log goSi ¼ 0:914 

The Iron-Silicon System

6863 T

(Eq 26)

(Eq 27)

As shown in Fig. 13, calculations with Eq 26 and 27 are in good agreement with experimental data (note that the equation in Fig. 13 results from introducing Eq 26 and 27 into Eq 22 and using a natural logarithm rather than a common logarithm).

Ternary Iron-Base Alloys Using a Taylor series expansion for the excess partial molar free energy, one researcher derived an expression for the activity coefficient of component 2 in a multicomponent solution, which for dilute solutions can be written as (Ref 24): ln f2 ¼ ln

g2 X2 e22 þ X3 e23 þ X4 e24 ; . . . go2

(Eq 28)

where g2 is the activity coefficient of component 2 dissolved in the three-component system, go2 is the activity coefficient of component 2 in the range of infinitely dilute solution, and e22, e23, e24, . . . are coefficients defined as: @ ln f2 @X2 limit @ ln f2 ¼ X1 !1 @X3 limit

e22 ¼ X1 !1 e23

(Eq 20)

For the ferrite:   10; 908 wt%Cmax ¼ 1:8  103 exp  T

  1270 aFe-C ¼  þ 1:74 T

Equation 25 is in good agreement with experimental data, as shown in Fig. 12.

wt%Cmax ¼  0:628 þ 1:222  103 t þ 1:045  106 t2

where g2 is the activity coefficient of component 2 = a2/X2 (a2 being the activity of component 2 and X2 the mole fraction of component 2); go2 is the activity coefficient of component 2 in the range of infinitely dilute solutions, that is, when %component 2 ffi 0; it is only a function of temperature (g2 ! go2 as X2 ! 0); and a12 is the function of temperature, usually of the shape A/T + B, where A and B are constants. For the activity of carbon in an iron melt, Eq 22 is valid for component 1 (iron) and component 2 (carbon) with (Ref 23):

log goC ¼

For the metastable system, that is, with Fe3C as the equilibrium high-carbon phase, Eq 19–21 apply. For the liquid: wt%Cmax ¼ 4:34 þ 0:1874ðt  1150Þ t  200 ln 1150

(Eq 22)

(Eq 17)

For the ferrite:   11; 460 ¼ 2:46  103 exp  T

g2 ¼ a12 ðX22  2X2 Þ go2

with the following remarks (Ref 23): XSi  0.2; component 1 is iron; component 2 is silicon:

(Eq 29)

Equation 22 can be used to calculate the activity of silicon in the iron-silicon system,

For infinitely dilute solutions, that is, X2 ! 0, X3 ! 0: e23 ¼ e32

Fig. 12

Fig. 13

(Eq 21)

Equation 15–21 can be used to calculate the maximum solubility of carbon in various phases (liquid, austenite, ferrite) as a function of temperature. For example, to calculate the carbon content of the stable eutectic, Eq 15 is used with t = 1152  C (2106  F) to obtain %Cmax = 4.26%. Prediction of phase stability relies heavily on knowledge of the activity of various elements. In the specific case of iron-carbon alloys, an increase in the activity of carbon in the liquid parallels a greater ability of carbon to separate as graphite; that is, the activity of carbon is a measure of an increased tendency of the alloy to solidify as the stable system. On the other hand, a decreased activity of carbon reflects the carbide-promoting behavior of the system, that is, the metastable solidification tendency.

Dependence of the activity coefficient of carbon on the mole fraction of carbon in liquid iron at 1550  C (2820  F). Source: Ref 23

Dependence of the activity coefficient of silicon on the mole fraction of silicon in liquid iron at 1600  C (2910  F). Source: Ref 23

Thermodynamic Properties of Iron-Base Alloys / 47 Simplification of the Darken formalism for infinite dilutions gives the interrelationship between e and a, as follows (Ref 22): e22 ¼ 2a12 2:303 e33 ¼ 2a13 2:303 e23 e32 ¼ ¼ a23  a12  a13 2:303 2:303

ðEq 30Þ

The solute concentration is often given in weight percent. For a ternary system, Eq 28 is written: log fi ¼ eii ½%i þ eij ½%j

(Eq 31)

where the interaction coefficient e is related to E by: eii ¼ 230ðMi =M1 Þeii þ ½ðM1  Mi Þ=M1 eij ¼ 230ðMj =M1 Þeij þ ½ðM1  Mj Þ=M1

log ðEq 32Þ

Data on interaction coefficients in liquid iron for carbon, hydrogen, nitrogen, oxygen, and sulfur are given in Table 2. For metallurgical reactions involving dilute solutions, it is often necessary to change the standard state from pure component to that of 1 wt% in solution. The free energy change is given by: Gs ¼ RT ln

0:5585goi Mi

available in Ref 25. Only the Fe-C-Si, Fe-C-P, and Fe-C-Mn systems are discussed in this article. The Fe-C-Si System. The addition of silicon to a binary iron-carbon alloy decreases the stability of Fe3C, which is already metastable, and increases the stability of ferrite (the a field is enlarged, and the g field is constricted). The equilibrium diagrams in Fig. 14 show that as the silicon content in the Fe-C-Si system increases, the carbon contents of the eutectic and eutectoid decrease (%Cmax decreases), while the eutectic and eutectoid temperatures increase (Ref 26). The activity coefficient, g2, of component 2, dissolved in a three-component system as a function of the concentration X2 and X3 of dissolved components 2 and 3, is described by Eq 34 (Ref 22):

(Eq 33)

Free energies of solution DGs of selected elements in liquid iron are given in Table 3.

g2 ¼  2a12 X2 þ ða23  a12  a13 ÞX3 go2 þ a12 X22 þ a13 X32 þ ða12 þ a13  a23 ÞX2 X3

ðEq 34Þ

where a12 and a13 are the modified interaction parameters of the two-component systems, a23 is the modified interaction parameter that describes the activity of both dissolved components, and go2 is the activity coefficient of component 2 in the range of infinitely dilute solution. Starting from Eq 34, the activity of carbon and silicon in an Fe-C-Si melt can be calculated as a function of temperature and concentration with the following relationships (Ref 23): log aC ¼ð1180=T Þ  0:87 þ ½ð2540=T Þ þ 3:48 XC

Ternary Fe-C-X Systems

 ½ð1270=T Þ þ 1:74 XC2 þ log XC

Detailed equilibrium diagrams of the Fe-C-B, Fe-C-Cr, Fe-C-Cu, Fe-C-Mn, Fe-C-Mo, Fe-CN, Fe-C-Ni, Fe-C-Si, and Fe-C-W systems are

þ ½ð1423=T Þ þ 4:75 XSi 2  ½ð153=T Þ þ 3:02 XSi

 ½ð1423=T Þ þ 4:76 XC XSi

ðEq 35Þ

log aSi ¼  ð6863=T Þ þ 0:914 þ ½ð306=T Þ þ 6:04 XSi 2  ½ð153=T Þ þ 3:02 XSi þ log XSi

þ ½ð1423=T Þ þ 4:76 XC  ½ð1270=T Þ þ 1:74 XC2  ½ð1423=T Þ þ 4:76 XC XSi

ðEq 36Þ

The Fe-C-P System. Because the partition coefficient of phosphorus between solid and liquid iron is low, phosphorus segregates into the liquid during solidification, and the melt becomes supersaturated in phosphorus, which results in the formation of Fe3P. Therefore, the four possible solid phases of the stable and metastable systems are austenite (g), Fe3P, Fe3C, and graphite. The calculated liquidus surfaces for the four phases and the eutectic lines formed by their interactions are shown in Fig. 15, which is a superimposition of the g-Fe3P-Fe3C and the g-Fe3P-graphite (Gr) diagrams. Two ternary eutectic points are shown in Fig. 15. To the left is L ! g + Fe3P + Gr at 954  C (1749  F) and yC = 0.099, yP = 0.123. To the right is L ! g + Fe3P + Fe3C at 948  C (1738  F) and yC = 0.106, yP = 0.123. Here, yP = XP/(1  XC), and yC = XC/(1  XC), where XP is the weight fraction of phosphorus and XC is the weight fraction of carbon. In general, it is agreed that phosphorus does not have a strong graphiteor carbide-stabilizing influence. The Fe-C-Mn System. The influence of increasing manganese contents in the Fe-C-Mn system over the equilibrium phase diagram is shown in Fig. 16. It is evident that manganese:  Decreases the eutectoid temperature  Increases the eutectoid interval, that is, the

range of temperature and carbon contents over which a, g, and carbide can coexist  Decreases the carbon content in the eutectoid and in the eutectic  Increases the eutectic temperature (by approximately 3  C, or 5  F, for each 1% Mn)

Table 2 Interaction coefficients in ternary iron-base alloys for carbon, hydrogen, nitrogen, oxygen, and sulfur at 1600  C (2910  F) Element j

Aluminum Boron Carbon Cobalt Chromium Copper Manganese Molybdenum Nitrogen Niobium Nickel Oxygen Phosphorus Sulfur Silicon Tin Titanium Vanadium Tungsten Zirconium Source: Ref 4, 7

eCj

eHj

eNj

eOj

eSj

0.043 0.24 0.14 0.008 0.024 0.016 0.012 0.008 0.11 0.06 0.012 0.34 0.051 0.046 0.08 0.041 ... 0.077 0.006 ...

0.013 0.05 0.06 0.002 0.002 <0.001 0.001 0.002 ... 0.002 0 0.19 0.011 0.008 0.027 0.005 0.019 0.007 0.005 ...

0.028 0.094 0.13 0.011 0.047 0.009 0.02 0.011 0 0.06 0.01 0.05 0.045 0.007 0.047 0.007 0.53 0.093 0.002 0.63

3.9 2.6 0.13 0.008 0.04 0.013 0.021 0.004 0.057 0.14 0.006 0.20 0.07 0.133 0.131 0.011 0.31 0.14 0.009 (3.0)

0.035 0.13 0.11 0.003 0.011 0.008 0.026 0.003 0.01 0.013 0 0.27 0.29 0.028 0.063 0.004 0.072 0.016 0.01 0.052

Table 3 Free energies of solution in liquid iron at 1 wt% Mi, liquid iron at 1600  C (2910  F) Element i

Aluminum (l) Carbon (g) Cobalt (l) Chromium (s) Copper (l) ½ H2(g) Manganese (l) ½N2 (g) Nickel (l) ½O2 (g) ½P2(g) ½ S2(g) Silicon (l) Titanium (s) Vanadium (s) Tungsten (s) Zirconium (s)

goi

D Gs (cal g atom1)

0.029 0.57 1.07 1.14 8.6 ... 1.3 ... 0.66 ... ... ... 0.0013 0.038 0.1 1.2 0.043

15,100  6.67T 5400  10.10T 240  9.26T 4600  11.20T 8000  9.41T 8720 + 7.28T 976  9.12T 860 + 5.71T 5000  7.42T 28,000  0.69T 29,200  4.60T 32,280 + 5.60T 31,430  4.12T 7440  10.75T 4950  10.90T 7500  15.20T 8300  11.95T

48 / Principles of Liquid Metal Processing

Fig. 14

Influence of silicon content on the solubility lines and equilibrium temperatures of the iron-carbon system. (a) to (c) Source: Ref 26, (d) to (g) Source: Ref 25

The carbide M3C, which is (FeMn)3C, is stable over a wide range of manganese and iron contents, extending at 1000  C (1832  F) from Mn3C to nearly pure Fe3C. At high manganese contents (for example, above 40% Mn), other manganese carbides such as M5C2, M7C3, M23C6, and M4C1+x(e) can form (Ref 25). Influence of a Third Element on the Equilibrium Temperatures. Various elements, dissolving in liquid and solid iron phases, change the equilibrium temperatures on the iron-carbon diagram, as indicated in Table 4. This results in increased or decreased a and g fields.

Of particular interest for the case of cast iron is the influence of third elements on the stablemetastable eutectic interval (Tst  Tmet). In general, elements that increase the Tst  Tmet interval promote graphite formation, while those that decrease the interval promote carbide formation. Based on their specific influence on the Tst  Tmet interval, third elements can be classified into the following four groups (Fig. 17):  Strong graphitizers that increase Tst and

decrease Tmet, such as silicon, aluminum, nickel, and copper (Fig. 17a)

 Weak graphitizers that decrease both Tst and

Tmet but increase the Tst  Tmet interval overall, such as phosphorus and arsenic (Fig. 17b)  Strong carbide stabilizers that decrease Tst but increase Tmet, such as chromium, vanadium, and manganese (Fig. 17c)  Weak carbide stabilizers that decrease both Tst and Tmet, such as molybdenum and tungsten (Fig. 17d) st ¼ Tst  Tmet interval Real values for the Tmet for a number of elements can be calculated with

Thermodynamic Properties of Iron-Base Alloys / 49 the data given in Table 4, which also includes data on the influence of some elements on the eutectoid temperature and on the maximum solubility of carbon in austenite. Two examples of the use of data in Table 4 are provided as follows.

having 2% Si, 0.5% Mn, and 1% Cu by weight: Tst ¼ 1154  C þ 4% Si  2% Mn

Example 1: Calculation of the value of the Tst  Tmet eutectic interval for a cast iron

TFe-C-Mn-Cu ¼ 1166  1117:2 ¼ 48:8 C

þ 5% Cu ¼ 1166  C Tmet ¼ 1148  C  15% Si þ 3% Mn  2:3% Cu ¼ 117:2 C TFe- C  1154  1148 ¼ 6 C

Fig. 15

Calculated liquidus surfaces in the Fe-C-P phase diagram. Source: Ref 27

Fig. 16

Vertical sections through the Fe-C-Mn ternary phase diagram at (a) 4.92, (b) 12.8, and (c) 19.7 wt% Mn

Example 2: Calculation of the temperatures of the point of maximum carbon solubility in austenite, TE and TE0 , for the aforementioned iron: TE ¼ 1154 C  10%Si þ 3:2% Mn  2%Cu ¼ 1133:5 C TE0 ¼ 1148 C þ 2:5% Si  2% Mn þ 5:2%Cu ¼ 1157:2 C Some typical experimental results for silicon, chromium, and vanadium are shown in Fig. 18. These results are in agreement with theoretical predictions. Partition of a Third Element Between Various Phases in the Fe-C-X System. It has been established from the thermodynamic stability of phases involved in the eutectic solidification of Fe-C-X alloy that the differences in the equilibrium partition coefficients DP (= PM  PS, where PM and PS are the partition coefficients of element X between liquid and eutectic metastable or stable solid), the partition of element X between austenite g=L and liquid, PX , and the partition of element X Fe C=g between cementite and austenite, PX 3 , are related to one another and have similar effects on graphitization during eutectic solidification (Ref 28). Calculated and experimental data for partition coefficients of various elements in phases of iron-carbon alloys are given in Table 5. Figure 19 illustrates the correlation between the graphitizing influence of a third element (expressed by the normalized addition of each element required to increase or decrease the chill depth by a given amount) and the DP value. It is quite obvious that a definite correlation exists, although silicon, aluminum, and titanium deviate from the line. These three elements have a strong affinity for nitrogen; this may explain their higher-than-expected graphitizing influence. Indeed, it is accepted that the nitrides may act as

50 / Principles of Liquid Metal Processing nuclei for the stable eutectic, promoting gray solidification. This is why experiments show that titanium, for example, can act either as a carbide or graphite promoter, depending on the nitrogen content. Influence of a Third Element on Carbon Solubility in the Fe-C-X System. The influence of a third element on the solubility of carbon can be expressed by (Ref 20): XCX

¼

XCXmax

 XCmax ðmole fractionÞ

ðEq 37aÞ

concentrations. In the region of low concentration of the third element (%X = 0 to 5%), the change in the solubility of carbon can be represented by the temperature-independent linear equation: XCX ¼ m XX

ðEq 38aÞ

or

%CEX ¼ m0 g %X 0

%C ¼ m %X X

ðEq 38bÞ

%C ¼

X %Cmax

 %Cmax  %Cmax ðwt%Þ

ðEq 37bÞ

where XCX and D%CX represent the increase or decrease of carbon solubility in mole fraction or weight percent, respectively; XCmax and % Cmax represent the saturation concentration in the iron-carbon system calculated from Eq 15 or 16; and XCXmax and %CX max are the saturation concentration in the Fe-C-X system. The changes in carbon solubility resulting from additions of third elements, X, are shown in Fig. 20. These data are valid in the range from 1200 to 1700  C (2190 to 3090  F), with the exception of silicon and sulfur at high

where m and m are solubility factors. In Eq 38a and 38b, m and m0 as well as XCX and D% CX are positive for an increase in the solubility of carbon and negative in the other case. It is also generally accepted that carbide-promoting elements increase the solubility of carbon (activity coefficient of carbon in solution is decreased), while graphite-promoting elements decrease it (activity coefficient of carbon in solution is increased). Therefore, third elements that have a negative solubility factor promote graphitization, while elements that have a positive solubility factor promote carbide formation. The value of the solubility factor is proportional to their effect. It must be noted, however, that the accuracy of the prediction of the behavior of elements based on solubility factor is questioned in

Table 4 Influence of a third element on the temperature change of critical points on the iron-carbon diagram +, increase; , decrease. Maximum solubility of carbon in austenite,  C/wt%X Element

Metastable (E)

Silicon Copper Aluminum Nickel Cobalt Chromium Manganese Molybdenum Tungsten Boron Nickel Titanium Vanadium Phosphorus Sulfur

10 to 15 2 14 4.8 ... +7.3 +3.2   ... ... ... +6–8 180 ...

0

Stable (E )

+2.5 +5.2 +8 +4 ...  2   ... ... ... ... 180 ...

ðEq 39Þ

Data for m0 g are also given in Table 6.

0

X

Ref 28 for elements that have a strong influence on the austenite liquidus. As indicated in Table 6, experimental values are close to the theoretical values for many elements. Equations similar to Eq 38a and 38b can be used to calculate the change in the maximum solubility of carbon in austenite upon the addition of a third element:

Multicomponent Iron-Carbon Systems Carbon Solubility in Multicomponent Systems. Assuming that the values in Table 6 for m and m0 , which were determined for the ternary Fe-C-X system, can be extended to multicomponent systems (in this case, the effect of alloying elements on the solubility of carbon can be considered to be additive, as shown in Eq 41, at least in the region of low concentrations), the saturation concentration of carbon in multicomponent systems can be calculated as (Ref 20): %CSi;Mn;P;S;... ¼ %Cmax þ %CSi;Mn;P;S;... max

ðEq 40Þ

which is Eq 37b rewritten for multicomponent systems. In turn: %CSi;Mn;P;S;... ¼ %CSi þ %CMn þ %CP max





Eutectoid, C/wt%X

Eutectic, C/wt%X 0

Metastable (S) Stable (S ) Metastable C Stable C

+8 ... +10 20 ... +15 9.5 + + ... ... ... +15 + ...

0–30 10 +10 30 ... +8 3.5 + + ... ... ... + +6 ...

10 to 20 2.3 15 6 ... +7 +3   ... ... ... 6–8 37 ...

+4 +5 +8 +4 ...  2   ... ... ...  30 ...

þ %CS þ . . .



Eutectic, C/at. %X(a) 0

Metastable C Stable C

3.25 3.32 7.74 1.33 2.23 4.97 2.33 8.90 12.13 15.47 19.67 16.91 ... 16.05 .. . .

ðEq 41Þ

0

10.91 8.41 0.91 6.50 1.48 10.23 7.15 12.36 15.55 18.53 16.87 18.91 ... 16.98 18.53

(a )Calculated results. Source: Ref 21, 28

Then, using Eq 15 and data from Table 6, Eq 40 becomes: %CSi;Mn;P;S;... ¼ 1:3 þ 2:57  103 t  0:31% Si max þ 0:027% Mn  0:33% P  0:4% S  . . .

ðEq 42Þ

Saturation Degree and Carbon Equivalent. The iron-carbon diagram with letter notations for critical points (Fig. 21) is used in this section. It is well known that mechanical properties of cast iron strongly depend on the amount and shape of graphite. In turn, for hypoeutectic gray irons, the amount of graphite depends on the amount of eutectic. The amount of eutectic can be calculated with the lever rule. For hypoeutectic irons, one can write: Amount of eutectic Amount of eutectic þ amount of austenite %Canal  %C0E ¼ %Canal  %C0E þ %C0C  %Canal %Canal  %C0E ¼ ðEq 43Þ %C0C  %C0E

Sr ¼

Fig. 17

Classification of the influence of a third element in the Fe-C-X system on the graphite- or carbide-promoting tendency in cast iron, based on their influence on the eutectic stable and metastable temperatures. (a) Strong graphitizers. (b) Weak graphitizers. (c) Strong carbide stabilizers. (d) Weak carbide stabilizers

where Sr is the rectified saturation degree, % Canal is the analyzed carbon content of cast iron, and %C0 C and %C0 E represent the carbon content

Thermodynamic Properties of Iron-Base Alloys / 51

Fig. 18

Influence of (a) chromium, (b) silicon, and (c)

Table 5

Equilibrium partition coefficients of a third element X in the Fe-C-X system g=L

Fe C=L

PX (a)

PX 3

(a)

Element X

calc

exp

calc

exp

PS

PM

DP = PM  PS

Silicon Copper Aluminum Nickel Cobalt Chromium Manganese Molybdenum Tungsten Boron Nitrogen Titanium Phosphorus Sulfur

1.71 1.57 ... 1.46 1.18 0.53 0.70 0.41 0.23 0.06 2.04 0.04 0.15 0.06

1.72 1.62 1.15 1.61 1.13 0.55 0.75 0.38 0.42 ... 2.04 ... ... ...

0 0.12 ... 0.43 0.59 1.96 1.03 0.60 0.42 0.22 2.12 0.09 0.08 ...

0.05 0.08 0.03 0.32 0.60 1.95 1.21 0.84 0.88 ... ... 0.27 0.09 ...

1.55 1.43 1.05 1.33 1.07 0.48 0.64 0.37 0.21 0.06 1.86 0.04 0.14 0.06

0.78 0.78 0.55 0.90 0.85 1.32 0.90 0.52 0.33 0.15 2.09 0.07 0.11 ...

0.77 0.65 0.50 0.43 0.21 0.84 0.26 0.15 0.12 0.09 0.23 0.03 0.03 ...

(a) calc, calculated; exp, experimental. Source: Ref 28

temperature of the multicomponent system. Further simplification gives: SC ¼

Fig. 19

Correlation between the partition coefficient, DP, and the graphitizing influence of a third element in the Fe-C-X system. Source: Ref 28

of the eutectic and of the austenite, respectively, at the eutectic temperature in the multicomponent system (Fig. 21). Assuming the solubility factors can be extrapolated to the hypoeutectic region, %C0 C can be calculated with Eq. 42. A similar approach can be used to calculate %C0 E using data in Table 6 for m0 g: %C0E ¼ 2:11  0:11% Si  0:35%P þ 0:006% Mn  0:08% S

ðEq 44Þ

In foundry practice, the following simplified form of Eq 43 is used: SC ¼ ¼

%Canal %C0C %Canal 4:26  0:31% Si  0:33% P þ . . .

ðEq 45Þ

For more accuracy, %Cmax as a function of temperature rather than %Cmax = 4.26 can be used, where the temperature is the eutectic

%Canal 4:25  0:3ðSi þ PÞ

ðEq 46Þ

where SC < 1 for hypoeutectic irons, SC = 1 for eutectic irons, and SC > 1 for hypereutectic irons. Saturation degree is used in most of the European literature. In the Anglo-Saxon literature and foundry practice, carbon equivalent rather than saturation degree is used. Carbon equivalent (CE) can be calculated as: CE ¼ %Canal  %CSi  %CMn  %CP  %CS  . . . ¼ %Canal þ 0:31%Si þ 0:33% P  0:027% Mn þ 0:4% S

ðEq 47Þ

A eutectic iron has CE = 4.26%. It must be noted that although Sr gives directly the amount of eutectic in the structure (for example, Sr = 0.9 means 90% eutectic), SC and CE, although easier to calculate, do not allow for exact estimation of the amount of eutectic.

Structural Diagrams For the practicing metallurgist involved in cast iron production, one of the main applications of the thermodynamics of the iron-carbon system is the calculation of structure-composition

correlations. The so-called structural diagrams are used for this purpose. The first and simplest structural diagram is the Maurer diagram (Fig. 22a), in which the as-cast structure is considered to be solely the product of the carbon and silicon contents of cast iron. Because it was recognized long ago that solidification rate, as well as chemical composition, plays a significant role in the formation of the as-cast structure of casting, especially for cast iron, Laplanche (Ref 29) further developed the Maurer diagram to include the influence of cooling rate through the section size (or bar diameter) of the casting (Fig. 22b). The k lines on the diagram are lines of same structure (same degree of graphitization) at a given cooling rate. The value k is an empirical function of composition, as follows:  k ¼ 4=3 Si 1 

 5 3C þ Si

ðEq 48Þ

The correlation between k and cooling rate is given in Table 7. Thus, if one wants to produce a pearlitic iron in casting with equivalent section sizes of 10 to 30 mm (0.4 to 1.2 in.) bar diameters, Table 7 would suggest a k of 0.85 to 2.35. For example, k = 1.8 is then chosen. A saturation degree, SC, is also selected depending on the required mechanical properties (for example, SC = 0.8), and the intersection of the k = 1.8 with SC = 0.8 will then give the carbon and silicon required for the iron.

52 / Principles of Liquid Metal Processing A further contribution to structural diagrams was made by Patterson and Doepp (Ref 30). This diagram (Fig. 23) encompasses not only structure, composition, and cooling rate but also a broad range of mechanical properties. Part of this diagram is based on Laplanche’s k lines. As an example of the way this diagram is used, assume it is desired to pour a 50 mm (2 in.) diameter cast iron bar (or equivalent section size) with a pearlitic-ferritic structure. A heavy horizontal line is drawn from the y-axis at the 50 mm (2 in.) diameter bar location to the middle of zone IIb. This corresponds to a determination of k = 2.2. This k line can be followed to the lower diagram until it intersects a preselected saturation degree—in this case, SC = 0.6. The composition is now read as C = 2.6% and Si = 3%, together with the equilibrium tem! perature for the ðSiO2 Þ þ 2½C ½Si þ fCOg reaction, which is 1460  C (2660  F). An additional explanation of the significance of this reaction can be found in the article “Solidification of Eutectic Alloys: Cast Iron” in this Volume. It must be pointed out that foundry variables such as superheating and holding during melting, charge composition (raw materials), and inoculation can cause significant deviation from these calculations.

ACKNOWLEDGMENTS Reviewed and adapted from Seymour Katz, Composition Control, Casting, Vol 15, Metals Handbook, 9th Edition, ASM International, 1988, p71–82; and D.M. Stefanescu, Thermodynamic Properties of Iron-Base Alloys, Casting, Vol 15, Metals Handbook, 9th Edition, ASM International, 1988, p 61–70 REFERENCES 1. R.J. Fruehan, Ladle Metallurgy: Principles and Practices, Iron and Steel Society of AIME, 1985, p 7–8 2. S. Katz and C.F. Landefeld, Cupola Desulfurization, Cupola Handbook, American Foundrymen’s Society, 1984, p 351–363 3. I.D. Sommerville, The Capacities and Refining Capabilities of Metallurgical Slags, Foundry Processes: Their Chemistry and Physics, S. Katz and C.F. Landefeld, Ed., Plenum Press, 1988, p 101–133 4. G.K. Sigworth and J.F. Elliott, The Thermodynamics of Liquid Dilute Iron Alloys, Met. Sci., Vol 8, 1974, p 298–310 5. S. Katz and H.C. Rezeau, The Cupola Desulfurization Process, Trans. AFS, Vol 87, 1979, p 367–376 6. S. Katz, D.E. McInnis, D.L. Brink, and G.A. Wilkinson, The Determination of Aluminum in Malleable Iron from Measured Oxygen, Trans. AFS, Vol 88, 1980, p 835–844 7. E.T. Turkdogan, Physical Chemistry of High Temperature Technology, Academic Press, 1980 Table 6

Solubility factors of various third elements for carbon-saturated Fe-C-X melts

Influence of third elements, X, on the solubility of carbon in molten iron. (a) In mole fraction. (b) In weight percent. Source: Ref 20

Fig. 20

5 6 13 14 15 16 22 23 24 25 26 27 28 29 32 33 41 42 44 50 51 52 73 74 76 78 79 92

Symbol

B C Al Si P S Ti V Cr Mn Fe Co Ni Cu Ge As Nb Mo Ru Sn Sb Te Ta W Os Pt Au U

D %CX = m0 %X

DXX C ¼ m XX (a)

Third element X Atomic No.

8. S. Katz and V.R. Spironello, Effect of Charged Aluminum on Iron Temperature, Silicon Recovery and Desulfurization in an Iron-Producing Cupola, Trans. AFS, Vol 92, 1984, p 161–172 9. C.F. Landefeld and S. Katz, Kinetics of Iron Desulfurization by CaO-CaF2, Proceedings of the Fifth International Iron and Steel Congress, Vol 6, Iron and Steel Society of AIME, 1986, p 429–439 10. S. Katz and C.F. Landefeld, Plant Studies of Continuous Desulfurization with CaOCaF2-C, Trans. AFS, Vol 93, 1985, p 215–228 11. S. Katz and B.L. Tiwari, A Critical Overview of Liquid Metal Processing in the Foundry, Foundry Processes: Their Chemistry and Physics, S. Katz and C.F. Landefeld, Ed., Plenum Press, 1988, p 1–52 12. J. Niederinghaus and R.J. Fruehan, Desulfurization Mechanisms for CaO-Al and CaO-CaS in Carbon Saturated Iron, Metall. Trans. B, to be published 13. E.T. Turkdogan, Physiochemical Phenomena of Mechanisms and Rates of Reaction in Melting, Refining and Casting of Foundry Irons, Foundry Processes: Their Chemistry and Physics, S. Katz and C.F. Landefeld, Ed., Plenum Press, 1988, p 53–100 14. S. Asai and I. Muchi, Fluid Flow and Mass Transfer in Gas Stirred Ladles, Foundry Processes: Their Chemistry and Physics, S. Katz and C.F. Landefeld, Ed., Plenum Press, 1988, p 261–292 15. S.-H. Kim and R.J. Fruehan, Physical Modelling of Liquid/Liquid Mass Transfer in Gas Stirred Ladles, Metall. Trans. B, Vol 18, 1987, p 381–390

0 D%CX C ¼ mg %X

m exp

m calc

Validity

m exp

m calc

Validity

m0 g

Validity

0.575 0.67 0.52 0.71 0.81 1.00 +0.508 +0.359 +0.215 +0.105 0.0 0.10 0.20 0.313 0.677 0.734 +0.51 +0.24 0.0 0.70 0.79 0.85 +0.49 +0.256 0.0 0.224 0.333 +0.53

0.51 0.68 0.51 0.68 0.85 1.02 +0.44 +0.33 +0.22 +0.11 0.0 0.11 0.22 0.33 0.66 0.77 +0.33 +0.22 0.0 0.66 0.77 0.88 +0.33 +0.22 0.0 0.22 0.33 +0.55

<0.04 ... <0.05 <0.007 <0.048 <0.004 <0.013 <0.1 <0.13 <0.30 ... <0.20 <0.10 <0.02 <0.1 <0.07 <0.045 <0.14 <0.008 <0.02 <0.05 <0.005 <0.03 <0.05 <0.013 <0.015 <0.024 <0.01

0.539 0.610 0.220 0.310 0.331 0.405 +0.159 +0.105 +0.064 +0.028 0.0 0.027 0.051 0.076 0.144 0.151 +0.058 +0.014 0.020 0.110 0.119 0.122 +0.004 0.015 0.035 0.052 0.060 0.006

0.462 0.622 0.215 0.294 0.349 0.414 +0.138 +0.097 +0.062 +0.029 0.0 0.029 0.055 0.080 0.141 0.159 +0.030 +0.012 0.023 0.104 0.117 0.126 0.009 0.01 0.035 0.052 0.060 0.005

<1.0 ... <2.0 <5.5 <3.0 <0.4 <1.6 <11.0 <10.0 <25.0 ... <40.0 <8.0 <3.8 <14.0 <10.0 <9.0 <2.0 <1.5 <4.5 <15.0 <1.5 <11.0 <18.0 <4.0 <6.0 <10.0 <5.0

... ... +0.04 0.11 0.35 0.08 2.08 +0.18 0.07 +0.006 ... 0.017 0.07 +0.014 ... ... ... 0.3 ... ... ... ... ... 0.12 ... ... ... ...

... ... <2 <6 <0.4 <5 <0.8 <1 <20 <60 ... <20 <6 <5 ... ... ... <4 ... ... ... ... ... <12 ... ... ... ...

(a) exp, experimental; calc, calculated. Source: Ref 20

0

Thermodynamic Properties of Iron-Base Alloys / 53 16. J. Ishida, K. Yamaguchi, S. Sugiura, N. Demukai, and A. Notoh, Denki Seiko, Vol 52, 1981, p 2–8 17. E.T. Turkdogan, Ladle Deoxidation, Desulfurization and Inclusions in Steel—Part I: Fundamentals, Arch. Eisenhu¨ttenwes., Vol 54, 1983, p 1–10 18. E.T. Turkdogan, Slags and Fluxes for Ferrous Ladle Metallurgy, Ironmaking Steelmaking, Vol 12, 1985, p 64–78 19. T.B. Massalski et al., Ed., Binary Alloy Phase Diagrams, Vol 1, American Society for Metals, 1986 20. F. Neumann, The Influence of Additional Elements on the Physico-Chemical Behavior of Carbon in Carbon Saturated Molten Iron, Recent Research on Cast Iron, H.D. Merchant, Ed., Gordon and Breach, 1968, p 659

Fig. 21

21. N.G. Girsovitch, Ed., Spravotchnik po tchugunomu litja (Cast Iron Handbook), Mashinostrojenie, 1978 22. L.S. Darken, The Thermodynamics of Ternary Metallic Solutions, Trans. TMS, Vol 239, 1967, p 80, 90 23. F. Neumann and E. Do¨tsch, Thermodynamics of Fe-C-Si Melts with Particular Emphasis on the Oxidation Behavior of Carbon and Silicon, The Metallurgy of Cast Iron, B. Lux et al., Ed., Georgi Publishing, 1975, p 31 24. C. Wagner, Thermodynamics of Alloys, Addison-Wesley, 1952 25. Metallography, Structures and Phase Diagrams, Vol 8, Metals Handbook, 8th ed., American Society for Metals, 1973, p 400–416 26. E. Piwowarsky, Hochwertiger Gusseisen, 2nd ed., Springer-Verlag, 1958

The iron-carbon diagram with letter notations. S.S., soild solution

27. M. Hillert and P.O. So¨derholm, White and Gray Solidification of the Fe-C-P Eutectic, The Metallurgy of Cast Iron, B. Lux et al., Ed., Georgi Publishing, 1975, p 197 28. A. Kagawa and T. Okamoto, Partition of Alloying Elements on Eutectic Solidification of Cast Iron, The Physical Metallurgy of Cast Iron, H. Fredriksson and M. Hillert, Ed., North-Holland, 1985, p 201 29. H. Laplanche, The Maurer Diagram and Its Evolution and a New Structural Diagram for Cast Iron, Foundry Trade J., No. 1669–1671, 1948, p 191, 225, 249 30. W. Patterson and R. Doepp, Giessereiforschung, Vol 21 (No. 2), 1969, p 91

54 / Principles of Liquid Metal Processing

Fig. 22

Structural diagrams for cast iron. (a) Maurer diagram. (b) Laplanche diagram. See also Fig. 23. Source: Ref 29

Table 7

k lines for various structural ranges

Bar diameter mm

in.

k for mottled structure

k for pearlitic structure

k for pearlitic-ferritic structure

30 20 10

1.2 0.8 0.4

0.65–0.85 0.75–1.10 1.05–1.50

0.85–2.05 1.10–2.25 1.50–2.35

2.05–3.10 2.25–3.40 2.235–3.50

Thermodynamic Properties of Iron-Base Alloys / 55

Fig. 23

Patterson and Doepp structural diagram for cast iron. See the corresponding text for details. Source: Ref 30

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 56-63 DOI: 10.1361/asmhba0005191

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Thermodynamic Properties of Aluminum-Base and Copper-Base Alloys* THIS ARTICLE provides accessible information on the thermodynamic properties of liquid aluminum-base and copper-base alloys. Three alternative means have been used to report the thermodynamic data:
Activities and activity coefficients
Partial and integral molar thermal properties
Interaction coefficients

symbol
Gxsi . It represents the contribution to
Gi resulting from the departure of solution from ideal behavior. Thus, one may write: xs


Gi ¼ RT ln gi xs
Gi

can also be


Gi ¼
H i  T
S i ¼ RT ln gi

(Eq 4)

Corresponding to Eq 1, expressed as: xs

In addition, the utility of phase diagrams is briefly discussed. Knowledge of activities and activity coefficients is necessary in describing solution behavior and in solving problems that involve chemical equilibria. The thermal properties are useful in understanding the liquid state and in correlating data on solution behavior. The interaction coefficients provide a simple means of calculating activity coefficients in dilute solutions and are also used to correlate experimental data on dilute solutions.

Activities and Thermal Properties The activity a of constituent i in a solution is given by the relationship:
Gi ¼
H i  T
S i ¼ RT ln ai

(Eq 1)

where
Gi ,
H i , and
Si are the partial molar free energy, enthalpy, and entropy, respectively, of constituent i in solution relative to the pure constituent i at the solution temperature as the standard state; T is the temperature (in degrees Kelvin); and R is the universal gas constant. Thus, from the knowledge of
H i and
Si as functions of temperature and solution composition, one may calculate
Gi , ai, and gi (activity coefficient of component i relative to pure material standard state) for all conditions of solution. It is important to realize, however, that thermal property values vary with solution composition, unlike the properties of pure substances. Replacing ai in Eq 1 with Xi gi yields:
Gi ¼ RT ln Xi þ RT ln gi

(Eq 2)

where Xi is the mole fraction of component i. In Eq 2, the term RT ln gi is called partial molar excess free energy and is expressed by the

(Eq 3)

xs

xs

However, because
H i for the ideal solution is 0 by definition,
H xsi and are
H i are identical. Therefore, Eq 4 can be rewritten as: xs

xs


Gi ¼
H i  T
S i ¼ RT ln gi

(Eq 5)

Equations 1 and 5 are commonly used to calculate activities and activity coefficients from the thermal properties of the solution. Although the partial molar thermal properties of individual constituents in a solution are most useful in solving chemical process problems, the integral properties are valuable in the discussion of the nature of solutions and also serve in some cases as the source of data on the partial properties. The integral molar free energy is related to the partial molar free energy by the following relationship:
Gi ¼ Xi
Gi þ X j
Gj þ . . . Xn
Gn

(Eq 6)

Similar equations can also be written to express other integral thermal properties. The following tables compile, thermodynamic, data in the form of activities, activity coefficients, partial molar thermal properties, and integral molar properties for selected aluminum-base and copperbase alloys. The alloy systems have been chosen based on their commercial importance and on the concentration of the major alloying elements. In general, the concentrations of the alloying elements in the commercial alloys are below 10 at.%, and under this condition the interaction coefficient method described as follows provides a convenient means of calculating the activities of the constituents in solution. When the concentrations are close to or greater than 10 at.%, the interaction coefficient method does not yield accurate values. With this factor in mind, thermodynamic data have

been selected for the following alloy systems: aluminum-magnesium, aluminum-silicon, copperaluminum, copper-nickel, copper-tin, and copper-zinc.
Aluminum-magnesium thermodynamic data:

Tables 1 and 2


Aluminum-silicon





thermodynamic data: Tables 3 and 4 Copper-aluminum thermodynamic data: Tables 5 and 6 Copper-nickel thermodynamic data: Tables 7 and 8 Copper-tin thermodynamic data: Tables 9 to 17 Copper-zinc thermodynamic data: Tables 18 and 19

Interaction Coefficients The use of interaction coefficients, first suggested by Wagner (Ref 4), provides a convenient means of organizing thermodynamic data on dilute solutions. Wagner proposed a Taylor series expansion for the excess partial molar free energy in order to express the logarithm of the activity coefficient of a dilute constituent in a multicomponent solution. The values of the activity coefficients calculated from this method are as accurate as the original data from which the interaction coefficients are determined, provided it is applied to solutions that are quite dilute (Xi < 0.1). Consider a dilute solution of i, j, and k dissolved in a common solvent, s. If all but the first-order terms of the Taylor series expansion are neglected, the activity coefficient of solute i is expressed by the relationship:   @ ln gi @Xi     @ ln gi @ ln gi þ Xk þ Xj @Xj @Xk

ln gi ¼ ln goi ¼ Xi

(Eq 7)

In Eq 7, the partial derivatives are called interaction coefficients and are expressed by the symbol eji , where the superscript denotes the constituent that affects the activity coefficient of the subscript constituent. Thus:

*Updated from B.L. Tiwari, Thermodynamic Properties of Aluminum-Base and Copper-Base Alloys, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 55–60.

Thermodynamic Properties of Aluminum-Base and Copper-Base Alloys / 57 Table 2 Integral thermal properties for liquid aluminum-magnesium alloys at 1073 K

Table 1 Activities, activity coefficients, and partial molar thermal properties for liquid aluminum-magnesium alloys at 1073 K Al component Alð‘Þ ¼ Alðin alloyÞð‘Þ XAl

aAl

0.9997 0.9992 0.9989 0.9945 0.9886 0.9785 0.9553 0.8870 0.8095 0.6600 0.4175 0.2510 0.1792 0.0450

gAl

DGAl , cal/g-atom

DSAl , cal/K-g-atom

0.9 1.7 2.3 12.8 26.6 45.0 93.5 251.6 473.7 934.7 1893.4 2874.9 3839.5 7988.8

0.0091 0.0016 0.0105 0.0203 0.0417 0.0420 0.0958 0.2157 0.3790 0.7806 1.1548 1.2240 0.2920 3.8840

DGAl , cal/g-atom

0.2 0.0 0.0 0.9 1.9 1.3 4.3 4.3 22.7 48.5 30.3 74.0 173.4 1373.6

0.9996 0.9992 0.9989 0.9940 0.9876 0.9791 0.9571 0.8887 0.8008 0.6451 0.4115 0.2597 0.1652 0.236

0.9999 1.0000 1.0000 0.9996 0.9991 1.0006 1.0020 1.0020 0.9894 0.9775 0.9859 1.0353 0.9219 0.5251

9 0 9 9 18 0 9 20 67 97 654 1561 4153 12157

aMg

gMg

DGMg, cal/g-atom

DSMg, cal/K-g-atom

DHMg , cal/g-atom

DGxs Mg, cal/g-atom

0.00022 0.00063 0.00081 0.0048 0.0107 0.0183 0.0366 0.0929 0.1679 0.3097 0.5176 0.6367 0.7184 0.9198

0.8060 0.7974 0.7814 0.8780 0.9353 0.8535 0.8188 0.8221 0.8813 0.9109 0.8885 0.8500 0.8752 0.9631

18,017 15,696 15,171 11,358 9683 8530 7051 5069 3806 2500 1405 963 705 178

15.4180 13.9463 12.6176 9.6420 7.4045 6.9708 5.4069 3.6123 2.7957 1.6331 1.2733 1.3102 1.7577 1.1672

1470.8 729.8 1630.2 1010.9 1736.7 1049.0 1248.7 1192.1 805.9 747.6 38.2 443.2 1180.8 1074.3

459.73 482.71 525.87 227.42 142.53 337.68 424.12 418.20 269.73 198.80 251.88 346.49 284.20 79.98

Mg component Mgð‘Þ ¼ Mg ðin alloyÞð‘Þ XMg 0.00027 0.00079 0.00104 0.0055 0.0114 0.0215 0.0447 0.1130 0.1905 0.3400 0.5825 0.7490 0.8208 0.9550

XMg

DG, cal/g-atom

DS, cal/K-g-atom

DH, cal/g-atom

DGxs, cal/g-atom

0.00027 0.00079 0.00104 0.0055 0.0114 0.0215 0.0447 0.1130 0.1905 0.3400 0.5825 0.7490 0.8208 0.9550

5.7 14.2 18.1 75.7 136.9 227.3 404.8 796.1 1108.7 1466.9 1608.9 1442.6 1266.9 529.7

0.01320 0.01274 0.02365 0.07366 0.12577 0.19086 0.33346 0.59962 0.83951 1.07043 1.22383 1.28856 1.39046 0.93986

8.61 0.58 7.29 3.34 2.04 22.54 47.28 152.48 207.80 318.19 295.28 59.77 225.25 478.87

0.34 0.39 0.55 2.39 3.53 6.00 14.91 43.49 69.79 99.61 159.37 240.99 264.35 138.20

Source: Ref 1

Table 4 Integral thermal properties for liquid aluminum-silicon alloys at 1100 K ð1  xÞAlð‘Þ þ xSið‘Þ ¼ Alð1xÞ Sixð‘Þ XSi

0.1 0.2 0.3 0.31(a)

DG, cal/g-atom

DGxs, cal/g-atom

1210 1931 2403 2430 (þ100) 

499 837 1067 1076 (þ100) 

(a) Phase boundary. Source: Ref 2

Source: Ref 1

If the solution of i in s obeys Henry’s law over a small range of concentration, then gi is a constant over this range, and eii ¼ 0. It can be shown by using the Gibbs-Duhem equation that:

Table 3 Activities, activity coefficients, and partial molar thermal properties for liquid aluminum-silicon alloys at 1100 K Al component Alð‘Þ ¼ Alðin alloyÞð‘Þ XAl

aAl

1.00 0.90 0.80 0.70 0.69(a)

ð1  xÞAlð‘Þ þ xMgð‘Þ ¼ Alð1xÞ Mgxð‘Þ

xs

DHAl , cal/g-atom

gAl

1.000 0.859 0.698 0.503 0.473 (þ0.01) 

xs

DGAl , cal/g-atom

DGAl , cal/g-atom

0 333 786 1500 1638 (þ50) 

0 102 298 720 826 (þ50) 

1.000 0.954 0.872 0.719 0.685 (þ0.02) 

Si component Sið‘Þ ¼ Siðin alloyÞð‘Þ DGSi, cal/g-atom

XSi

aSi

gSi

0.00 0.10 0.20 0.30 0.31(a)

0.000 0.016 0.051 0.127 0.147 (þ 0.02)

0.040 0.155 0.255 0.424 0.473 (þ 0.60)

1 9106 6510 4509 4194 (þ 300)

xs DGSi , cal/g-atom

7060 4073 2991 1877 1634 (þ 300)

DHSi, cal/g-atom 2502 ... ... ... ... (þ 50)

DSSi, cal/ K-g-atom 1 ... ... ... ...

xs DSSi , cal/ K-g-atom

4.144 ... ... ... ...

(a) Phase boundary. Source: Ref 2

  @ ln gi Xk ¼ 0; Xi ! 0 @Xi Xj   @ ln gi ¼ 0; Xi ; Xj ! 0 eji ¼ @Xj Xk

In Eq 8, goi is the limiting value of gi at infinite dilution:

eji ¼

ln gi ¼ ln goi þ Xj eii þ

Xj eji

þ Xk eki

goi ¼ lim gi (Eq 8)

Xi ! 0

(Eq 9)

eji ¼ eij

(Eq 10)

and this reciprocal relation is most useful in determining values of e from activity data. Values of eii are determined from experimental data on the binary system i-s, and values of eji are determined from experimental data on the ternary system i-j-s. Values of go and e for aluminum-base and copper-base alloys are given in Tables 20 to 24. The standard state for and e and go described previously is the pure material at the temperature of the solution, and the concentration is expressed in mole fraction. In dealing with dilute solutions, however, it is common to use a hypothetical 1 wt% solution as the standard state. Under this condition, the Taylor series expansion corresponding to Eq 8 is: log fi ¼ eii ð%iÞ þ eji ð%jÞ  eki ð%kÞ

(Eq 11)

where fi, which is the activity coefficient for 1 wt% standard state, equals ai/%i and: eji ¼

  @ log fi @%j %k¼0; %i; %j!0

(Eq 12)

58 / Principles of Liquid Metal Processing Table 5 Activities, activity coefficients, and partial molar thermal properties for liquid copper-aluminum alloys at 1373 K

Table 6 Integral thermal properties for liquid copper-aluminum alloys at 1373 K

Al component Alð‘Þ ¼ Alðin alloyÞð‘Þ XAl

aAl

1.0 0.9 0.8 0.7 0.6 0.5

gAl

1.000 0.889 0.759 0.609 0.441 0.266 (þ0.02)  0.116 0.028 0.006 0.001 0.000

0.4 0.3 0.2 0.1 0.0

1.000 0.988 0.949 0.870 0.735 0.532 (þ0.04)  0.290 0.095 0.029 0.008 0.002

(1x)Al‘ þ xCu‘ ¼ Al(1

DGAl, cal/g-atom

xs DGAl , cal/g-atom

DHAl, cal/g-atom

0 320 753 1354 2235 3611 (þ150)  5873 9709 14,062 19,307 1

0 32 144 381 842 1720 (þ150)  3373 6424 9670 13,025 16,640

0 29 44 60 391 1055 (þ300)  2878 5012 5864 7415 8625

xs DSAl , cal/K-g-atom

DSAl, cal/K-g-atom

0.000 0.254 0.580 0.942 1.343 1.862 (þ0.25)  2.181 3.421 5.971 8.661 1

0.000 0.044 0.137 0.234 0.328 0.484 (þ0.25)  0.361 1.028 2.772 4.086 5.838

XCu

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

 x)Cux(‘)

DG, cal/ g-atom

DH, cal/ g-atom

DS, cal/ K-g-atom

DGxs, cal/ g-atom

DSxs, cal/ K-g-atom

1723 2969 3955 4700 5170 (þ 150) 5289 4906 3927 2361

459 960 1445 1863 2163 (þ 150) 2254 1979 1380 800

0.921 1.463 1.828 2.066 2.190 (þ 0.1) 2.210 2.132 1.855 1.137

836 1604 2289 2863 3278 (þ 150) 3453 3240 2562 1474

0.275 0.469 0.615 0.728 0.812 (þ 0.1) 0.873 0.918 0.861 0.491

Source: Ref 2

Cu component Cuð‘Þ ¼ Cuðin alloyÞð‘Þ XCu

aCu

gCu

0.0 0.1 0.2 0.3 0.4 0.5

0.000 0.005 0.013 0.025 0.046 0.085 (þ0.005)  0.166 0.352 0.600 0.839 1.000

0.042 0.052 0.065 0.084 0.115 0.170 (þ0.01)  0.277 0.503 0.750 0.932 1.000

0.6 0.7 0.8 0.9 1.0

DGCu, cal/g-atom

xs DGCu , cal/g-atom

DHCu, cal/g-atom

1 14349 11833 10025 8396 6728 (þ150)  4900 2848 1394 478 0

8671 8067 7442 6741 5896 4837 (þ150)  3506 1875 785 191 0

4225 4849 4975 4675 4071 3271 (þ300)  1838 679 259 65 0

DSCu, cal/K-g-atom

xs DSCu , cal/K-g-atom

1 6.919 4.995 3.897 3.150 2.518 (þ0.25)  2.230 1.580 0.827 0.301 0.000

3.238 2.344 1.797 1.505 1.329 1.141 (þ0.25)  1.215 0.871 0.383 0.092 0.000

Source: Ref 2

Table 7 Activities, activity coefficients, and partial molar thermal properties for liquid copper-nickel alloys at 1823 K Cu component Cuð‘Þ ¼ Cuðin alloyÞð‘Þ XNi

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

aCu

1.000 0.902 0.814 0.740 0.677 0.611 (þ 0.02) 0.534 0.444 0.334 0.191 0.000

gCu

1.000 1.002 1.017 1.058 1.128 1.222 (þ 0.03) 1.334 1.480 1.669 1.912 2.227

DGCu, cal/g-atom

xs DGCu ,

0 374 747 1088 1414 1786 (þ 100) 2275 2941 3974 5992 1

cal/g-atom

0 8 61 204 436 725 (þ 100) 1044 1421 1856 2349 2900

Table 8 Integral thermal properties for liquid copper-nickel alloys at 1823 K (1  x)Cu(‘) þ xNi(‘) ¼ Cu(1 XNi

0.1 0.15 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

DG, cal/ g-atom

948 1195 1377 1617 1745 1786 (þ100)  1742 1604 1349 917

 x)Nix(‘)

DH, cal/ g-atom

DS, cal/ K-g-atom

DGxs, cal/ g-atom

DSxs, cal/ K-g-atom

240 341

0.652 0.842

... ... ...

... ... ...

0.006 0.003 ... ... ... ...

... ... ... ...

... ... ... ...

229 336 435 596 693 725 (þ100)  696 609 464 261

... ... ... ...

Source: Ref 2

Ni component Ni(‘) ¼ Ni (in alloy)(‘) aNi

0.000 0.185 0.341 0.455 0.539 0.611 (þ 0.02) 0.682 0.752 0.826 0.907 1.000 Source: Ref 2

gNi

1.906 1.864 1.704 1.517 1.347 1.222 (þ 0.03) 1.136 1.075 1.032 1.008 1.000

DGNi, cal/g-atom

1 6120 3899 2851 2241 1786 (þ 100) 1337 1031 692 353 0

xs DGNi , cal/g-atom

2336 2222 1931 1510 1079 725 (þ 100) 464 261 116 29 0

In Eq 11, fi is the activity coefficient, and the zeroeth-order term, log fio , disappears because the activity coefficient at infinite dilution fio is equal to 1. The reciprocal relationship, corresponding to Eq 10, is: eji ¼

Mi i e Mj j

(Eq 13)

where Mi is the atomic weight of i. The relationship between the two types of the interaction coefficients is: eji ¼

Ms ej ð2:303Þð100ÞMj i

(Eq 14)

where Ms is the atomic weight of the solvent (aluminum, copper).

Thermal Properties for Hypothetical Standard State It is a common practice in solving problems on chemical equilibria to use the hypothetical pure component (Xi) or hypothetical weight percent (%i) as the standard state. For this purpose, thermodynamic properties for the hypothetical standard states are needed. The values of go can be used to determine Gibbs free energies of mixing for elements in liquid base metal, using Eq 15 and 16:
Goi ðXi Þ ¼ RT ln goi

(Eq 15)

and
Goi ð%iÞ ¼ RT ln

goi Ms 100Mi

(Eq 16)

The values of
Goi ðXi Þ and
Goi ð%iÞ thus calculated for aluminum-base and copper-base alloys are also given in Tables 20 and 22. The principal advantage of using the hypothetical standard states is that at low concentrations, the activity of i can be set equal to its atom fraction or weight percent, depending on the composition scale used to express the standard state.

Phase Diagrams A phase diagram is essentially a map that shows the relative stabilities of various phases present at equilibrium under the varying conditions of temperature and composition. Therefore, for a given composition of an alloy, a phase diagram can be used to determine the phases, which will be present at equilibrium as the melt

Thermodynamic Properties of Aluminum-Base and Copper-Base Alloys / 59 Table 9 Activities, activity coefficients, and partial molar thermal properties for liquid copper-tin alloys at 1400 K

Table 10 Partial molar thermal properties for liquid copper-tin alloys at 633 K

Cu component Cuð‘Þ ¼ Cuðin alloyÞð‘Þ XCu

1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0

aCu

gCu

1.000 0.802 0.539 0.389 0.284 0.220 (þ0.025)  0.169 0.125 0.082 0.038 0.000

Sn(‘) ¼ Sn (in alloy)(‘)

DGCu, cal/g-atom

xs DGCu , cal/g-atom

DHCu, cal/g-atom

0 613 1717 2626 3497 4215 (þ300)  4949 5781 6971 9104 1

0 320 1096 1633 2076 2287 (þ300)  2400 2432 2493 2698 3197

0 159 749 1443 1662 1631 (þ250)  1418 1049 640 81 1050

1.000 0.891 0.674 0.556 0.474 0.440 (þ0.05)  0.422 0.417 0.408 0.379 0.317

DSCu, cal/K-g-atom

xs DSCu , cal/K-g-atom

0.000 0.325 0.691 0.845 1.311 1.845 (þ0.3)  2.522 3.380 4.522 6.561 1

0.000 0.115 0.248 0.136 0.295 0.468 (þ0.3)  0.701 0.988 1.324 1.985 3.034

XSn

0.98 0.99

aSn

gSn

0.965 0.985 (þ0.005) 

0.985 0.995 (þ0.005) 

DGSn, cal/g-atom

47 13 (þ5) 

DGxs Sn, cal/g-atom

20 7 (þ5) 

Source: Ref 2

Sn component Sn(‘) ¼ Sn (in alloy)(‘) XSn

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

aSn

0.000 0.007 0.072 0.197 0.340 0.467 (þ 0.05) 0.580 0.681 0.784 0.892 1.000

gSn

DGSn, cal/g-atom

DGxs Sn, cal/g-atom

DHSn, cal/g-atom

DSSn, cal/K-g-atom

DSxs Sn, cal/K-g-atom

1 13,706 7304 4523 3003 2119 (þ 300) 1516 1069 678 317 0

13,609 7301 2827 1173 454 190 (þ 300) 95 77 57 24 0

8000 5233 1901 252 706 681 (þ 250) 506 311 176 49 0

1 6.053 3.860 3.411 2.649 2.000 (þ 0.3) 1.444 0.986 0.610 0.261 0.000

4.006 1.477 0.661 1.018 0.828 0.622 (þ 0.3) 0.429 0.277 0.167 0.052 0.000

0.007 0.072 0.362 0.656 0.849 0.934 (þ 0.1) 0.966 0.973 0.979 0.991 1.000

Source: Ref 2

Table 11

Heats of formation of solid and liquid copper-tin alloys at 723 K ð1  xÞCuðsÞ þ xSnðsÞ ¼ Cuð1xÞ SnxðsÞ ð1  xÞCuð‘Þ þ xSnð‘Þ ¼ Cuð1xÞ Snxð‘Þ Cuð‘Þ ¼ Cuðin alloyÞð‘Þ

XSn

Phase

0.091(b)

(Cu)

DH, cal/g-atom

xSn

Phase

DCp(a), cal/K-g-atom

DH Cu , cal/g-atom

280

0.825(b)

‘

1105

1.7

...

...

... ... ...

990 760 530 (þ 50) 0

... 1.2 ...

... ... ...

... ... ...

260

... ...

... ...

1.9 (þ2)  ... ...

0.204(b) 0.209

d d

1300 1260

0.850 0.900 0.950

0.244(b)

e

1686

1.000

e

1800 1920 (þ100) 

... ...

0.250 0.255(b)

‘

DH, cal/g-atom

... ...

(a) Cp ¼ heat capacity. (b) Phase boundary. Source: Ref 2

Table 12

Integral thermal properties for liquid copper-tin alloys at 1400 K ð1  xÞCuð‘Þ þ xSnð‘Þ ¼ Cuð1xÞ Snxð‘Þ

XSn

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Source: Ref 2

DG, cal/g-atom

1922 2834 3195 3300 3167 (þ 300) 2889 2483 1937 1196

DH, cal/g-atom

666 979 934 715 475 (þ 150) 264 97 13 52

DS, cal/K-g-atom

0.897 1.325 1.614 1.846 1.923 (þ 0.24) 1.875 1.705 1.392 0.891

DGxs, cal/g-atom

1018 1442 1495 1427 1238 (þ 300) 1017 784 545 291

DSxs, cal/K-g-atom

0.251 0.331 0.400 0.509 0.545 (þ 0.24) 0.538 0.491 0.398 0.245

... ...

DC pCuð‘Þ , cal/K-g-atom

60 / Principles of Liquid Metal Processing Table 13 Measured activities of tin in liquid copper-tin alloys Temperature, K

Table 15 Recommended partial molar thermodynamic quantities for liquid alloys at 1400 K Cu component Cu‘ ¼ Cu(in alloy)‘

Activity

Composition, mole fraction of Sn

Sn

Cu

XCu

aCu

gCu

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.1 0.2 0.3 0.4 0.5 0.65 0.8 0.9 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

0.02 0.082 0.22 0.373 0.499 0.614 0.712 0.811 0.905 0.1 0.055 0.198 0.307 0.466 0.652 0.808 0.919 0.007 0.055 0.214 0.353 0.474 0.575 0.683 0.805 0.903

0.856 0.678 0.498 0.373 0.297 0.227 0.17 0.113 0.064 ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...

1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0

1.000 0.822 0.611 0.448 0.336 0.258 0.197 0.143 0.092 0.045 0.000

XSn

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

1573

1373

1073

DGCu, J/mol

Gex Cu , J/mol

1.000 0.913 0.763 0.640 0.560 0.516 0.492 0.475 0.460 0.450 0.468

0 2282 5744 9356 12,693 15,777 18,929 22,674 27,780 36,100 1

0 1055 3146 5204 6747 7708 8262 8658 9044 9294 8842

aSn

gSn

DGSn, J/mol

Gex Sn , J/mol

0.000 0.013 0.076 0.196 0.336 0.465 0.581 0.690 0.798 0.901 1.000

0.021 0.132 0.378 0.652 0.839 0.930 0.968 0.986 0.997 1.001 1.0000

1 50,370 30,058 19,002 12,716 8913 6326 4316 2633 1213 0

45,239 23,564 11,322 4986 2049 844 380 164 35 13 0

Source: Ref 3

Table 14 Recommended integral thermodynamic quantities for liquid alloys at 1400 K

Table 16 Recommended enthalpies of formation of solid and liquid alloys at 723 K

DG

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

7091 10,607 12,250 12,702 12,345 11,368 9824 7662 4702

DH

3360 4374 4010 2996 1838 857 210 79 94

Gex

DS

2.655 4.452 5.886 6.933 7.505 7.508 6.867 5.529 3.426

3307 4782 5139 4868 4277 3534 2713 1837 918

Note: Values in J/mol or J/mol  K. Source: Ref 3

solidifies. In addition, a phase diagram can also be used to estimate the activity of the components in liquid solution and to understand the behavior of the liquid solution (Ref 9). For example, if the diagram is for a simple eutectic system, it is likely that no strong intermetallic compound exists. If, however, a strong compound in the solid state appears in the diagram, then association of dissimilar atoms is probably occurring in the liquid. Therefore, one can predict whether a liquid solution will exhibit strong negative or positive deviations from the ideal behavior from the shapes of the liquids and azeotropes. The phase diagrams of the aluminum-silicon and copper-zinc systems, which are chosen based on their commercial importance, are shown in Fig. 1 and 2. The aluminum-silicon diagram (Fig. 1) represents a simple eutectic system, and the alloys containing 8.5 to 13% Si are widely used

Sex Cu , J/mol  K

0.000 0.661 1.243 2.123 3.628 5.999 9.360 13.736 19.152 26.317 1

0.000 0.215 0.612 0.843 0.619 0.235 1.742 3.725 5.770 7.172 6.891

Phase

0.090(a)

DH

XSn

(Cu)

1165

0.825(a)

0.205

d

5350

0.85

0.240

e

7650

0.90

0.435

Z

7030(b)

0.95

Phase ‘ ‘ ‘ ‘

49,228 21,394 5854 1526 3930 3704 2455 1176 341 22 0.000

DSSn, J/mol  K

Sex Sn , J/mol  K

1 20.697 17.289 14.663 11.890 9.012 6.272 3.923 2.124 0.882 0.000

2.849 1.552 3.907 4.652 4.272 3.249 2.025 0.957 0.269 0.006

XSn

Phase

GSn, J/mol

671

0.008 0.012 0.021 0.043 0.069

(Cu) (Cu) (Cu) (Cu) (Cu)

54,350 50,354 45,606 38,618 33,600

515

Data given relative to pure liquid Sn. Source: Ref 3

(1  X) Cu‘ þ XSn‘ ¼ Cu(1x)Snx‘ XSn

DHSn, J/mol

Table 17 Recommended partial Gibbs energy of tin in solid alloys at 1000 K

(1  X) Cu(s) þ XSn(s) ¼ Cu(1x)Snx(s)

Sex

0.038 0.291 0.807 1.337 1.742 1.912 1.788 1.368 0.723

0 1356 4004 6384 7614 7379 5825 3444 967 744 806

DSCu, J/mol  K

Sn component Sn‘ ¼ Sn(in alloy)‘

Source: Ref 3

XSn

DHCu, J/mol

DH

263 104

(a) Phase boundary. (b) DH at 237 K. Source: Ref 3

to produce automotive parts. Generally, the higher the silicon content, up to the eutectic composition (12.6% Si), the greater the fluidity and, consequently, the easier an alloy is to cast. The copperzinc diagram (Fig. 2) however, shows a high solubility of zinc in solid copper-zinc solution and the formation of several other phases over a wide range of zinc concentrations. Copper alloys containing up to 35% Zn and significant amounts of other alloying elements are quite common. The copper-zinc phase diagram is used to adjust the compositions in order to produce alloys with the desired properties. REFERENCES 1. B.L. Tiwari, Metall. Trans. A, Vol 18, 1987, p 1645–1651

2. R. Hultgren et al., Selected Values of the Thermodynamic Properties of Binary Alloys, American Society for Metals, 1973 3. P.R. Subramanian, D.J. Chakrabarti, and D. E. Laughlin, Ed., Phase Diagrams of Binary Copper Alloys, ASM International, 1994 4. C. Wagner, Thermodynamics of Alloys, Addison-Wesley, 1952 5. G.K. Sigworth and T.A. Engh, Scand. J. Metall., Vol 11, 1982, p 143–149 6. J.M. Dealy and R.D. Pehlke, Trans. Met. Soc. AIME, Vol 227, Feb 1963, p 88–94 7. G.K. Sigworth and J.F. Elliott, Can. Met. Q., Vol 13, 1974, p 455–461 8. J.M. Dealy and R.D. Pehlke, Trans. Met. Soc. AIME, Vol 227, Aug 1963, p 1030–1032 9. L.S. Darken and R.W. Gurry, Physical Chemistry of Metals, McGraw-Hill, 1953 10. T.B. Massalski et al., Binary Alloy Phase Diagrams, Vol 1 and 2, American Society for Metals, 1986

Thermodynamic Properties of Aluminum-Base and Copper-Base Alloys / 61 Table 18 Activities, activity coefficients, and partial molar thermal properties for liquid copper-zinc alloys at 1200 K Cu component Cu(‘) ¼ Cu (in alloy)(‘) XZn

aCu

0.334(a) 0.4 0.5

gCu

0.432 0.334 0.219 (þ0.05)  0.139 0.085 0.049 0.023 0.000

0.6 0.7 0.8 0.9 1.0

xs DGCu ,

DGCu , cal/g-atom

2003 2614 3621 (þ400)  4705 5869 7187 9010 1

0.648 0.557 0.438 (þ0.1)  0.348 0.284 0.245 0.229 0.235

cal/g-atom

1034 1396 1968 (þ400)  2520 2998 3349 3519 3454

Zn component Zn(‘) ¼ Zn (in alloy)(‘)

aZn 0.132 0.207 0.347 (þ 10.05) 0.505 0.657 0.789 0.900 1.000

xs

gZn

DGZn , cal/g-atom

DGZn , cal/g-atom

0.398 0.517 0.695 (þ 0.1) 0.841 0.939 0.986 1.000 1.000

4814 3757 2521 (þ 400) 1630 1002 564 250 0

2199 1572 868 (þ 400) 412 151 32 1 0

(a) Phase boundary. Source: Ref 2

Table 19

Integral thermal properties for liquid copper-zinc alloys at 1200 K ð1  xÞCuð‘Þ þ xZnð‘Þ ¼ Cuð1xÞ Znxð‘Þ

XZn

DG

0.334(a) 0.4 0.5

2942 3071 3071 (þ 400)

DGxs, cal/g-atom

1423 1466 1418 (þ 400)

xZn

0.6 0.7 0.8 0.9

DG, cal/g-atom

2860 2462 1889 1126

DGxs, cal/g-atom

1255 1005 695 351

(a) Phase boundary. Source: Ref 2

Table 20

Standard Gibbs free energies for solution of elements in liquid aluminum

Solution reaction

Be(s)¼Be Bi(l)¼Bi Ca(l)¼Ca Cd(l)¼Cd Cu(s)¼Cu Fe(s)¼Fe Ga(l)¼Ga Ge(l)¼Ge ½H2 ¼ H In(l)¼In Li(l)¼Li Mg(l)¼Mg Na(l)¼Na Ni(s)¼Ni Pb(l)¼Pb Sb(l)¼Sb Si(s)¼Si Sn(l)¼Sn Zn(l)¼Zn Source: Ref 5

g0, 1100 K

19.6 24.5 0.0086 19.9 0.037 1.6  104 1.1 0.16 ... 12.3 0.40 0.18 293 0.7  106 115 3.4 0.27 4.68 1.92

DGoi for i (X), cal/g-atom

9404  2.639T 5309 þ 1.524T 10,400 7100  0.518T 1135  5.538T 27,000 þ 7.18T 832  0.52T 2761  1.176T ... 6800  1.201T 5800 þ 3.435T 3478  0.30T 8230 þ 3.798T 28,280  2.442T 9970 þ 0.363T 13,100  9.45T 9598  11.291T 5845  2.245T 2538  1.007T

DGoi for i(%), cal/g-atom

9404  9.607T 5309  11.688T 10,400  9.93T 7100  12.498T 1135  16.385T 27,000  3.411T 832  11.551T 2761  12.288T 11,664 þ 6.523T 6800  13.223T 5800  3.014T 3478  9.24T 8230  5.03T 28,280  13.132T 9970  12.832T 13,100  21.589T 9598  20.517T 5845  14.333T 2538  11.911T

Table 21 Interaction coefficients for elements in liquid aluminum i

j

«ji

Temperature, K

Ag Cu H H H H H H H H H Si Cd Cu Ga Ge In Mg Si Sn Zn Zn Li Na Mg Sn H H H H H H H H H H H

Ag H Cu Cu Cu Cu H Si Si Si Si H Cd Cu Ga Ge In Mg Si Sn Zn Si Sn Si Si Pb Ce Cu Cr Fe Mg Mn Ni Th Ti Si Sn

3.1 see eCu H 39.0 16.6 20.1 4.3 0 11.5 6.2 4.2 1.8 see eSiH 5.0 2.2 0.3 þ3.0 4.5 3.0 16.0 6.0 0.9 2.2 16.0 12 9 1.5 100 15 1 1 2 27 19 20 42 7.1 0.6

1273 973–1273 973 1073 1173 1273 973–1273 973 1073 1173 1273 973–1273 1373 1373 1023 1200 1173 1073 1100 1100 1000 963–1053 949 973 973–1073 973–1073 800 700–800 800–900 800–900 700–800 800 700–800 800–900 800–900 700–800 800–900

Source: Ref 5, 6

62 / Principles of Liquid Metal Processing Table 22

Standard Gibbs free energies for solution of elements in liquid copper

Table 24 Activity coefficients at infinite dilution in liquid metals

Temperature Element(a), i

goi , 1200  C (2190  F)

Ag(l) Al(l) As(v) Au(l) Bi(l) C(graphite) Ca(l) Cd(v) Cd(l) Co(s) Cr(s) Fe(s) Fe(l) Ga(l) Ge(l)

3.23 0.0028 4.8  104 0.14 1.25 1.4  105 5.1  104 15.6 0.53 15.4 43 24.1 19.5 0.034 0.009 ... 0.41 0.08 0.044 0.51 0.53 2.22 2.66 ... 5.27 1.3 0.05 ... 0.014 0.002 0.006 0.01 0.048 0.0328 8.5 130 0.146

1=2 H2 ðgÞ

In(l) Mg(v) Mg(l) Mn(l) Mn(s) Ni(l) Ni(s) 1 2 O2 ðgÞ

Pb(l) Pd(s) Pt(s) 1=2 Ss ðgÞ

Sb(l) Se(v) Si(l) Si(s) Sn(l) Te(v) Ti(l) V(s) Zn(l)

DGoi ðXÞ, cal/g-atom



DGoi (wt%), cal/g-atom

3900  0.32T 8630  5.84T 22,350 4630  0.73T 5960  3.6T 8550 þ 17.8T 22,200 25,700 þ 22.9T 1860 8000 11,000 12,970  2.48T 9300  0.41T 10,800 þ 0.61T 16,000 þ 1.52T 10,400 þ 8.4T 9550 þ 4.71T 40,200 þ 22.3T 8670  0.31T 1950 1550  2.31T 2340 6500  2.5T 20,400 þ 10.8T 8620  2.55T 800 10,200 þ 0.87T 28,600 þ 13.79T 12,500 18,200 15,000 2900  7.18T 8900 10,000 6730  0.31T 28,100  9.4T 5640

3900  10.52T 8630  13.84T 22,350  9.44T 4630  12.09T 5960  15.1T 8550 þ 12.0T 22,200 25,700 þ 12.7T 1860  9.0T 8000  9.0T 11,000  8.72T 12,970  11.34T 9300  9.27T 10,800  8.68T 16,000  7.5T 10,400  7.5T 9550  5.58T 40,200  15.1T 8670  7.53T 1950  8.83T 1550  11.14T 2340  9.0T 6550  11.5T 20,400  4.43T 8620  14.01T 800  10.1T 10,200  10.47T 28,600  6.03T 12,500  10.4T 18,200  9.5T 15,000  7.5T 2900  14.68T 8900  10.4T 10,000  10.53T 6730  11.74T 28,100  18.1T 5640  9.1T



C

1100–1200 1100 1000 1175–1325 1100–1300 1100–1300 800–925 ... ... ... ... 1460–1580 1460–1580 1100–1280 1255–1545 1100–1300 700–1000 650–927 650–927 1244 1244 ... ... 1100–1300 1000–1300 1500–1600 ... 1050–1250 1000–1200 1200 1550 1550 1100–1300 1200 1000–1300 ... 1150

(a) l, liquid; v, vapor; s, solid; g, gas. Source: Ref 7

Table 23 Interaction coefficients for elements in liquid copper alloys Temperature i

j

H H H H H H H H H H H H H H H H H O O O O

Ag Al Au Co Cr Fe Mn Ni P Pb Pt S Sb Si Sn Te Zn Ag Au Co Fe

O O O O O O

Ni P Pb Pt S Si

«ij

0.5 6.2 1.9 3.1 1.6 2.9 1.1 5.5 10.0 21.0 8.0 9.0 13.0 4.8 6.0 6.6 6.8 0.7 8.6 68 4.04  106/T 2183 36,000/T þ 17 700,000/T þ 385 7.4 38 19 6300

Source: Ref 5, 6



C

Temperature 

F

1225 1225 1225 1150 1550 1150–1550 1150 1150–1240 1150 1100 1225 1150 1150 1150 1100–1300 1150 1150 1100–1200 1200–1550 1200 1200–1350

2235 2235 2235 2100 2820 2100–2820 2100 2100–2260 2100 2010 2235 2100 2100 2100 2010–2370 2100 2100 2010–2190 2190–2820 2190 2190–2460

1200–1300 1150–1300 1100 1200 1206 1250

2190–2370 2100–2370 2010 2190 2203 2280

i

O S S S S S S Ag Al Au Bi Ca Fe Ga Ge H Mg Mn O Pb S Sb Sn Tl Zn Zn Zn Zn Zn

j

Sn Au Co Fe Ni Pt Si Ag Al Au Bi Ca Fe Ga Ge H Mg Mn O Pb S Sb Sn Tl An Zn Zn Zn Zn

«ij

4.6 6.7 4.8 25,400/T þ 8.7 29,800/T þ 13 11.5 6.9 2.5 14 3.7 6800/T þ 1.65 20 5.7 7 13.4 1.0 9.8 6 24,000/T þ 7.8 2.7 20,800/T 15 10 4.8 4 0.38 0.72 1.185 1.40



C

1100 1115–1200 1300–1500 1300–1500 1300–1500 1200–1500 1200 1150 1100 1277 1000–1200 877 1550 1280 1255 1123 927 1244 1100–1300 1200 1050–1250 1000–1200 1300–1320 1300 1150 902 727 653 604



Temperature

F

2010–2190 2010 1830 2150–2415 2010–2370 2010–2370 1470–1700 ... ... ... ... 2660–2880 2660–2880 2010–2340 2290–2815 2010–2370 1290–1830 1200–1700 1200–1700 2271 2271 ... ... 2010–2370 1830–2370 2730–2910 ... 1920–2280 1830–2190 2190 2820 2820 2010–2370 2190 1830–2370 ... 2100

F

2010 2040–2190 2370–2730 2370–2730 2370–2730 2190–2730 2190 2100 2010 2331 1830–2190 1611 2820 2335 2290 2053 1701 2271 2010–2370 2190 1920–2280 1830–2190 2370–2410 2370 2100 1656 1341 1207 1119

Solvent

Solute

Aluminum

Silver

Copper

Magnesium Zinc

Source: Ref 2, 8



g

0.38 0.47 0.53 0.88 0.14 0.17 0.21 0.24



C

700 900 1000 800 604 653 727 802



F

1290 1650 1830 1470 1119 1207 1341 1476

Thermodynamic Properties of Aluminum-Base and Copper-Base Alloys / 63

Silicon, at.% 1500

10

0

20

30

40

50

60

70

80

90

100 2700 1414 ⬚C

2400

1300

1100

2000

900

1700

700 660.452 ⬚C

1300 577 ± 1 ⬚C

12.6

300

(Si)

(Al)

500

0

Al

10

20

30

40

50 Silicon, wt%

Fig. 1

The aluminum-silicon phase diagram. Source: Ref 9

Fig. 2

The copper-zinc phase diagram. Source: Ref 10

60

70

80

90

900

600 100 Si

Temperature, ⬚F

Temperature, ⬚C

L

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 64-73 DOI: 10.1361/asmhba0005192

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Gases in Metals Richard J. Fruehan, Carnegie Mellon University Prince N. Anyalebechi, Grand Valley Sate University

(Eq 1)

where the underlining of aluminum (Al) indicates that it is dissolved in the metal. The free energy for reaction in Eq 1 is highly negative even for very low concentrations of aluminum, indicating that the thermodynamics for the reaction are highly favorable. Similar reactions can be written for silicon, carbon, and iron. For carbon, carbon monoxide is also evolved, according to the reaction: H2 O þ C ! H2 þ CO

(Eq 2)

For the casting of aluminum and its alloys, the reaction is even more favorable. Whether or not the reaction will actually occur depends on many complex factors, including liquid-phase mass transfer and surface tension; therefore, its occurrence is difficult to predict. However, if there is excessive moisture in the sand, the reaction is highly likely to occur. The other major cause of gas porosity is the evolution of dissolved gases during casting. For example, liquid cast iron may have dissolved hydrogen and nitrogen. The solubility of these gases in the solid may be less than that in the liquid, and the gases may therefore be evolved during solidification. Whether or not this will occur depends on the amount of hydrogen and nitrogen present, the alloy being cast, chemical kinetics, and the surface tension of the alloy. Similarly, hydrogen can dissolve in liquid aluminum, but its solubility is much lower in solid aluminum

Cast Iron (Ref 1) Solubility of Hydrogen and Nitrogen. The gases in cast iron that can cause porosity are hydrogen and nitrogen. Castings from iron produced in a cupola are nearly saturated with nitrogen; therefore, the presence of nitrogen is a major concern (Ref 2). Hydrogen and nitrogen dissolve in liquid iron alloys and other metals as atomic species according to: 1

/2 H2 ðgasÞ ! H

(Eq 3)

/ N2 ðgasÞ ! N

(Eq 4)

12

The underlined symbols H and N denote hydrogen dissolved in liquid aluminum and its alloys. The solubilities of both elements in liquid and solid iron alloys have been extensively measured, and the results have been adequately summarized (Ref 3). The solubility of nitrogen and hydrogen in liquid iron is decreased by carbon. The nitrogen and hydrogen contents in iron-carbon alloys in equilibrium with 1 atm pressure of the respective gases are shown in Fig. 1 as a function of carbon content. The solubility of nitrogen at 1823 K decreases from 450 ppm for pure iron to approximately 80 ppm for an alloy containing 4.5 wt% C. The solubilities of hydrogen for the same metals are 24 and 10 ppm, respectively. The hydrogen solubility decreases with temperature and at

eutectic temperature is 6.5 ppm; temperature has only a small effect on the solubility of nitrogen. Silicon decreases the solubility of nitrogen even further. For a Fe-4wt%C-1wt% Si melt, the solubility of nitrogen is approximately 75 ppm. When cast iron solidifies, it forms austenite and graphite or cementite. The solubility of nitrogen in austenite (Ref 3) is also shown in Fig. 1. For austenite containing approximately 1.8 wt% C, the solubility is approximately 150 ppm. Therefore, when the liquid cast iron freezes, the solubility in the solid is higher than that in the liquid; this interesting phenomenon is discussed later in detail. There is some solubility of nitrogen in cementite (Ref 4). The exact amount is not known but is approximately the same as that in austenite. The solubility of hydrogen in austenite is approximately 7 ppm; therefore, the solubility of hydrogen in the liquid and solid are about equal, and there is little segregation of hydrogen during solidification (Ref 3). Kinetics of Gas-Liquid Reactions. The absorption of hydrogen and nitrogen is controlled by one of the following steps:

500

50

400

40 Hydrogen in liquid

300

30 Nitrogen in liquid

200

20 Nitrogen in austenite

100

0

Fig. 1

Hydrogen, ppm

2Al þ 3H2 O ! Al2 O3 þ 3H2

and can cause porosity. For copper, the evolution of hydrogen, water vapor, or carbon monoxide could cause gas porosity. In this article, the solubilities of the common gases present in ferrous metals, such as cast irons, and nonferrous metals, such as aluminum, copper, and magnesium, are reviewed. The kinetics of the relevant reactions, reactions during solidification, and possible methods of control or removal of the dissolved gases are discussed. The discussion is primarily focused on cast iron because more is known about the thermodynamics and kinetics of iron than other elements. However, th e same basic principles also apply to aluminum, copper, and magnesium.

Nitrogen, ppm

GAS POROSITY is one of the most serious problems in the casting of both ferrous and nonferrous metals. It is generally caused by the evolution of gases during the casting and solidification process. The gases may be the result of a reaction between the casting sand mold and the metal, or they may result from the evolution of gases dissolved in the liquid metal during solidification. An example of a chemical reaction evolving gas and causing porosity is the reaction of moisture (H2O) in a sand mold with elements in the cast iron, such as carbon, silicon, aluminum, or iron itself. For example, aluminum in cast iron often causes porosity problems, and the reaction responsible for the gas evolution is:

10 0

0

1

2 3 Carbon, %

4

5

Effect of carbon on the solubility of hydrogen and nitrogen in the iron-carbon alloys at 1823 K (i.e., liquid) and 1500 K (i.e., austenite) and 1 atm pressure

Gases in Metals / 65

(Eq 5)

where km is the mass transfer coefficient, pBN2 is the pressure of nitrogen in the bulk gas, and pSN2 is the surface pressure of nitrogen in equilibrium with the melt where: pSN2 ¼ KN2 fN2 ðwt%NÞ2

(Eq 6)

where KN is the equilibrium constant for Eq 4, and fN is the activity coefficient of nitrogen with respect to 1 wt%. However, in general, gas diffusion for these reactions is faster than the chemical reaction or liquid-phase diffusion and can be neglected in most cases. The rate of the chemical reaction is controlled by the dissociation of the nitrogen molecule on the surface, and the rate is given by:   dnN2 ¼ kB Að1  yÞ pN2  peN2 dt

(Eq 7)

where kB is the rate constant for pure iron, A is the surface area, peN2 is the equilibrium nitrogen pressure given by an expression similar to Eq 6, and (1  y) is the fraction of vacant sites not occupied by surface-active elements. Certain elements are surface active on liquid iron in that they lower the surface tension of iron by covering most of the surface. For example, oxygen and sulfur are surface active on iron, and for bulk concentrations of 0.03% O or 0.03% S, over 90% of the surface sites will be covered by oxygen or sulfur. These elements therefore retard the rate of chemical reaction. At high coverage, the surface-active element (1  y) is inversely proportional to the activity, a, of the element. Therefore, the rate is given by:  dnN2 kB kA A  ¼ pN2  peN2 dt a

(Eq 9)

where kmN is the liquid-phase mass transfer coefficient of nitrogen, CNB is the bulk concentration of nitrogen, and CNS is the surface concentration of nitrogen. The integrated form of Eq 7 is:   %N  %Ne ArkmN t ¼ ln %No  %Ne W

(Eq 10)

where r is the density of iron, W is the weight of the metal, %Ne is the equilibrium of the nitrogen contents, and %No is the initial nitrogen contents. The rate is often controlled by two processes in series—usually the chemical reaction and liquidphase mass transfer. The rate in this case can be obtained by equating the fluxes given by Eq 8 and 9 and solving for the surface concentration, as indicated by Pehlke and Elliot (Ref 5). 3.0 2.5





1.0

2.0 1.5

1.0 1450⬚C 1600⬚C

0.5

0.75 0.5 0.25 0

0

100

0 0

10

20

30

40

50

Fig. 2

Fig. 4

Effect of sulfur in the rate of reaction of nitrogen with Fe-CSAT-S alloys

1.0 0.75 0.5 0.25 0

Fig. 3

0

2.5

5 1/aBi

7.5

300

400

500

600

Effect of tellurium on the reaction of nitrogen with Fe-CSAT-Te alloys at 1450  C (2640  F)

1.5

1.0

0.5 as = 0.11 Present study Ref 9 0

10

Effect of bismuth on the reaction of nitrogen with Fe-CSAT-Bi alloys at 1450  C (2640  F)

200

1/[% Te]

60

1/as

(Eq 8)

where kA is related to the adsorption coefficient of the element on the surface, and the quantity kBkA/a is the overall rate constant k.




Rate constant, mol/cm2 • s • atm) ⫻ 106

JN2

 km  S p  pBN2 ¼ RT N2

JN ¼ kmN CNB  CNS

The rate of the nitrogen reaction with iron alloys containing oxygen, sulfur, chromium, and other elements has been measured by many investigators, and there is good agreement in most cases (Ref 5–10). However, the rate for carbon-saturated iron containing other elements was not investigated until recently (Ref 11). In the study, an isotope exchange technique was used to measure the rate of dissociation of the nitrogen molecule (N2). In particular, the effects of carbon, sulfur, phosphorus, lead, tin, bismuth, and tellurium on the rate were investigated. Sulfur, phosphorus, lead, bismuth, and tellurium decreased the rate, while carbon and tin had no significant effect. For example, the effect of sulfur is shown in Fig. 2 as the rate constant versus the reciprocal of the activity of sulfur (Ref 10). The activity of sulfur is in weight percent, and the activity coefficient of sulfur in carbonsaturated iron is 6.3. The rate for carbon-saturated iron with no sulfur is approximately 105 mol/ cm2 s atm at 1450  C (2640  F). For example, for 0.009 wt% S (1/as = 18), the rate is decreased to 8  107 mol/cm2 s atm. Bismuth and tellurium had an even larger effect (Ref 11), as shown in Fig. 3 and 4. The activities are relative to pure bismuth and tellurium, respectively, which have larger deviations from ideal behavior. As little as 50 ppm Te reduced the rate by 90%, and 50 ppm Bi decreased it by 80%. On the other hand, carbon is not surface active and does not affect the rate at moderate sulfur contents (0.017 wt% S) up to 4.5 wt% C (Ref 11), as shown in Fig. 5. Therefore, the

Rate constant, mol/(cm2 · s · atm) ⫻ 106

For the desorption of a gas, the same three steps are important but occur in reverse order. The process can be controlled by two or more of the steps in series. For liquid cast iron, nitrogen is the more important of the gases, and it is considered in detail. Gas diffusion is usually not rate controlling for nitrogen absorption from the atmosphere because of the pressure of N2; consequently, the driving force for diffusion is high. As a result, one of the other steps is slower and rate controlling. However, for nitrogen or hydrogen removal by an inert gas, diffusion on N2 or H2 must be considered. The flux of nitrogen,JN2 , away from the surface is given by:

Rate constant, mol/(cm2 ·s · atm) ⫻ 106

from the surface

For dilute solutions, the activity of the surfaceactive element is proportional to its concentration. Therefore, the overall rate constant is inversely proportional to the weight percent of the surface-active element in iron. Liquid-Phase Mass Transfer. If the rate is controlled by liquid-phase mass transfer, the flux of nitrogen atoms is given by:

Rate constant, mol/cm2 ·s· atm) ⫻ 106

 Diffusion of the gas to the surface  Chemical reaction on the surface  Diffusion of the element in the liquid away

Fig. 5

0

1

2 3 4 Carbon content, wt%

5

6

Effect of carbon on the reaction of nitrogen with Fe-C-S alloys at constant sulfur activity

66 / Principles of Liquid Metal Processing pT ¼ KH2 fH2 ð%HÞ2 þKN2 fN2 ð%NÞ2

(Eq 11)

where Ki is the equilibrium constant for the gas reaction, and fi is the activity coefficient. At the end of solidification, the concentrations will be as indicated in Eq 11. Even a small amount of hydrogen may be important. For example, in the case of an Fe-3.8wt%C alloy containing only 4 ppm H, the hydrogen content in the liquid will increase slightly, because of the enrichment, to approximately 4.5 ppm just prior to the eutectic reaction. However, even for this small amount of hydrogen, its thermodynamic pressure is 0.4 atm. Therefore, if the thermodynamic pressure of nitrogen exceeds 0.6 atm, a gas pinhole may result. Calculations indicate that iron with as little as 80 ppm N and 4 ppm H may have a gas pressure exceeding 1 atm. Figure 6 shows a plot of the total pressure of hydrogen and nitrogen just prior to final solidification equal to 1 atm as a function of bulk nitrogen and hydrogen contents in an Fe-3.8C alloy. If the hydrogen and nitrogen contents are such that they are below the line, the pressure will not exceed 1 atm, and a pinhole due to gas evolution will not occur; if above the line and the pressure exceeds 1 atm, a pinhole from this source is possible. For pinholes to develop, it is also necessary to nucleate a bubble. Bubble nucleation is a complex phenomenon and is influenced by surface tension. For melts with low surface tension, bubble nucleation is favored. On the other hand, elements that reduce the surface tension also reduce the rate of N2 formation. Therefore, it is difficult to predict the net effect on the probability of bubble formation of an alloying element that reduces the surface tension of iron. Nitrogen and Hydrogen Removal by Inert Gas Flushing. The preceding discussion indicates that nitrogen and hydrogen evolution during solidification may be a cause of pinholes in castings. Therefore, it would be desirable to 120

remove a portion of these gases if possible. One method is by argon bubbling in the desulfurization reactor or ladle. The desulfurization reactor is a continuous reactor with metal flowing in and out almost continuously. However, for the present calculation, it is assumed to be a batch reactor. Because this is only an order of magnitude calculation, this assumption is not unreasonable. When argon is bubbled through the melt, nitrogen atoms combine on the bubble surface to form N2. The rate is controlled by the chemical reaction on the surface and the liquid-phase mass transfer of nitrogen to the surface in series. For the purpose of the present calculations, the following were assumed:  10 Mg (11 ton) reactor 1 m (3.3 ft) deep  Argon flow rate of 0.005 m3/s (10 scfm)

through the melt

 Initial nitrogen content of 100 ppm  60 mm (2.4 in.) diameter bubbles

The rate constant for iron containing a relatively small amount of sulfur is approximately 1.5  106 mol/cm2 s atm, and the mass transfer coefficient, km, is approximately 0.1 cm/s. The gas velocity, retention time, and surface area were estimated as done previously (Ref 13). Equations 8 and 9 were solved simultaneously, and the results are given in Fig. 7. The calculations demonstrate that the rate is truly mixed control, as indicated by the surface concentration being between zero and the bulk concentration. The results indicate that it would take approximately 40 min to remove 20 ppm; if the argon flow were doubled, it would still take over 20 min to remove 20 ppm. Although these calculations are rather crude, they indicate that it would be difficult to remove nitrogen by argon bubbling. For hydrogen, the chemical reaction and mass transfer are considerably faster than for nitrogen. If equilibrium between the metal and the gas bubbles leaving the system is assumed, it is possible to calculate the amount of hydrogen that can be removed by gas bubbling from thermodynamic considerations alone. The



100 0.012 80 0.010

p N + pH = 1 2

2

60

Nitrogen, %

Nitrogen, ppm

initial rate of nitrogen formation is primarily controlled by chemical kinetics at the surface, and surface-active elements, such as sulfur and tellurium, can reduce the rate. This information can be helpful in controlling the rate of the reaction. For example, if it is desired to remove nitrogen by argon gas flushing, the concentrations of these elements should be as low as possible. On the other hand, it may be possible to retard nitrogen evolution during solidification by the deliberate addition of these elements. Reactions During Solidification. During the solidification of most simple iron alloys, there is enrichment of the alloying element because there is greater solubility in the liquid than in the solid. The difference in the solubilities is expressed by the partition ratio kS/L, which is determined from the slopes of the solidus to liquidus lines on the phase diagram. For most simple alloys, the partition ratio is less than 1, resulting in liquid enrichment during solidification. For example, for an Fe-0.02wt% N alloy, the last portion of liquid to solidify will be enriched and will have a concentration of 0.045 wt% N, which is the solubility of nitrogen in liquid iron at 1 atm (Ref 12). Therefore, even though the liquid alloy is far from being saturated with nitrogen, during solidification, the concentration will increase to the point where the solubility is exceeded and nitrogen gas will be evolved. Researchers have measured the partition ratio for nitrogen in high-carbon iron alloys (Ref 4). The partition ratio was found to be 1.9 and 2.2 for stable and metastable eutectic solidification, indicating that there is greater solubility in the solid than in the liquid. Therefore, in this case, as the alloy solidifies, nitrogen is actually enriched in the solid, and the nitrogen content decreases in the liquid. It may appear that, because there is no enrichment during solidification, the thermodynamic pressure of nitrogen cannot increase. The thermodynamic pressure is defined by an expression similar to Eq 6. Carbon greatly increases the activity coefficient of nitrogen in the liquid and decreases its solubility; therefore, the thermodynamic pressure may increase because the carbon content of the liquid is increasing during solidification. For example, for an alloy containing 3.8 wt% C that is saturated at 1500  C (2730  F) with 110 ppm N just prior to eutectic solidification, the concentration in the liquid will decrease to 97.5 ppm because of enrichment in the austenite. However, the solubility of nitrogen in the liquid has decreased to 90 ppm primarily because of the increase in the carbon content. Consequently, the thermodynamic pressure of nitrogen exceeds 1 atm, and the evolution of nitrogen may cause a pinhole. When determining the possibility of pinhole formation, one must consider the total thermodynamic pressure of all the gases, and if the total pressure, pT, exceeds 1 atm, a gas pinhole may result. For example, for hydrogen and nitrogen, the total pressure is given by:

40

0.008

0.006

20

0 0

Fig. 6

2

6 4 Hydrogen, ppm

8

10

Thermodynamic pressure of nitrogen and hydrogen at the end of solidification equal to 1 atm for an Fe-3.8wt%C alloy

0.004 0

Fig. 7

20

60 40 Time, min

80

100

Rate of removal of nitrogen from an Fe-3.8wt% C alloy with argon bubbling at 0.005 m3/s (10 scfm) in a 10 Mg (11 ton) reactor

Gases in Metals / 67 equations are developed similarly to those for steel (Ref 12), taking the form: 1 1 ¼ kH V  ½H ½Ho 

(Eq 12)

where [H] is the hydrogen content after bubbling, [Ho] is the initial hydrogen content, and V is the total volume of gas used per ton of metal. The constant kH is related to the solubility of hydrogen and therefore depends on the alloy composition. For example, 10 Mg (11 tons) of an Fe-4.5wt%C melt at 1773 K containing 4 ppm H bubbling argon for 20 min at 0.005 m3/s (10 scfm) will reduce the hydrogen content to 1.6 ppm. The previous calculation assumes equilibrium and therefore the fastest rate possible and may be an overestimation of the amount of hydrogen removed. Generally, the efficiency of the flushing gas would be roughly 50% for the conditions encountered in practice. However, the calculation does indicate that it is possible to remove significant amounts of hydrogen by gas bubbling.

Aluminum and Its Alloys Compared to other nonferrous metals, aluminum is the most susceptible to hydrogeninduced defects. This has been attributed to two primary reasons. The first reason is that molten aluminum and its alloys readily absorb hydrogen by the reaction with water vapor: 2AlðlÞ þ 3H2 OðvÞ ! Al2 O3 ðsÞ þ 3H2 ðin AlÞ (Eq 13)

The standard free energy change, DGo, for this reaction is extremely high: Go ¼ 979; 100  719T log10 T þ 413T

(Eq 14)

where T is the temperature in K. Inserting this value of free energy change into the van’t Hoff reaction isotherm yields an activity coefficient that is exceedingly high and, for practical purposes, implies complete conversion to hydrogen of all traces of moisture contacting the metal. That is: Go ¼ RT ln Kp

(Eq 15)

where DGo is the standard free energy change, T is the temperature in K, R is the gas constant, and Kp is the activity coefficient. The activity coefficient is given by the equation: Kp ¼

pH2 aH2 O

(Eq 16)

where pH2 is the partial pressure of hydrogen, and aH2 O is the activity of the water vapor. The second reason is that, unlike other nonferrous metals, the partition coefficient, k, that is, the ratio of hydrogen solubility in the solid state to that in the liquid state, is relatively small. That is, the solubility of hydrogen in liquid aluminum is more than an order of magnitude greater than

that in the solid. The first reason is not unique since, for a similar reaction for magnesium and lithium and their alloys, the free energy change is even higher. This explains the greater propensity for hydrogen absorption and the high solubility of hydrogen in magnesium, lithium, and their alloys. However, the low partition coefficient is unique to aluminum and its alloys and partly explains why, despite the comparatively lower solubility of hydrogen in aluminum, it is still the most susceptible of the nonferrous metals to hydrogen-induced defects. Hydrogen Solubility and Reactions During Solidification. Hydrogen solubility is often a direct indication of the propensity for porosity formation and its evolution during solidification of a casting. It indicates the nature of portioning of hydrogen in both cast and wrought products and the effects of alloying elements on the susceptibility of aluminum alloys to hydrogen-induced defects. It is generally agreed that during solidification, hydrogen in the melt is rejected at the solid-liquid interface (at the liquid side) in the same manner in which alloying elements are rejected from the solid in alloys having an equilibrium partition coefficient k less than unity: k¼

Ss <1 Sl

(Eq 17)

where k is the equilibrium partition coefficient for hydrogen in aluminum, and Ss and Sl are the solubility of hydrogen in the solid and liquid, respectively (in the case of alloys, the solubility of hydrogen at the solidus/eutectic and liquidus temperatures, respectively). The solubility of hydrogen in both the liquid and solid states of aluminum has been measured by several researchers. Of the more than 26 sets of results (Ref 14) reported on the liquid state of the aluminum-hydrogen system, only 9 are assessed to be reliable (Ref 15–25). Combined regression analysis of these reliable sets of results yields the following recommended empirical equation for hydrogen solubility as a function of temperature, T (K), at 1 atm hydrogen partial pressure (Ref 26): Liquid aluminum-hydrogen : 2692 log10 S; cm3=100 g ¼ þ 2:7292 T

Of the 14 sets of results reported on the solubility of hydrogen in solid pure aluminum, only 7 are assessed to be reliable (Ref 14, 15, 26–33). Combined linear regression analysis of these results yielded the following empirical equation for the solubility of hydrogen in pure sold aluminum as a function of temperature at 1 atm hydrogen partial pressure: Solid aluminum-hydrogen: 2965:8 þ 1:8032 log10 S; cm3 =100 g ¼ T

where S is the hydrogen solubility in cm3/100 g, and T is the temperature in K. Hydrogen solubility in both liquid and solid pure aluminum follows Sievert’s law. That is, the solubility of hydrogen is directly proportional to the square root of the hydrogen pressure. Equations 18 and 19 yield values for the solubility of hydrogen in liquid and solid pure aluminum of 0.70 and 0.042 cm3/100 g, respectively, at the melting point, 933 K. This confirms the fact that the solubility of hydrogen in liquid pure aluminum is more than 16 times the solubility of hydrogen in solid aluminum, and the equilibrium partition coefficient of hydrogen in aluminum is significantly less than 1. Both recommended sets of results of hydrogen solubility in liquid and solid pure aluminum are presented graphically in Fig. 8. Alloying elements affect the solubility of hydrogen in aluminum alloys. Unfortunately, there are a limited number of published data on the solubility of hydrogen in aluminum alloys. This is especially the case for multicomponent and commercial aluminum alloys. However, alloying elements have a strong and varying influence on the solubility of hydrogen in aluminum. For example, in liquid aluminum, silicon, copper, zinc, and iron decrease hydrogen solubility, whereas lithium, magnesium, and titanium increase it (Fig. 9). The effects of temperature on the solubility of hydrogen in liquid binary aluminum alloys can be represented by the general equation of the same form as Eq 1 and 2: log10 S; cm3=100 g ¼

(Eq 18) 3

where S is the hydrogen solubility in cm /100 g, and T is the temperature in K. Unlike in the liquid state, published values of hydrogen solubility in solid pure aluminum are limited (Ref 14). This is because the solubility of hydrogen in the solid state of pure aluminum is significantly lower than that in the liquid state. Consequently, determination of the solubility of hydrogen in solid aluminum is a more formidable experimental task than the determination of the solubility of hydrogen in the liquid state. This problem is compounded by the fact that diffusion of hydrogen into aluminum in the solid state is several orders of magnitude slower than that into the liquid.

(Eq 19)

A þB T

(Eq 20)

where S is the hydrogen solubility in cm3/100 g, T is the temperature in K, and A and B are the solubility constants or coefficients that depend on alloy composition. The coefficients for the effects of temperature on the solubility of hydrogen in some liquid binary aluminum alloys, obtained from published data assessed to be reliable, are given in Table 1. In solid aluminum, published results show that silicon, magnesium, copper, lithium, and zinc increase the solubility of hydrogen (Fig. 10). The addition of lithium increases the solubility of hydrogen in solid aluminum by 1 to 2 orders of magnitude and that in liquid aluminum by two- to threefolds. Consequently, the partition coefficient of hydrogen in aluminum-lithium

68 / Principles of Liquid Metal Processing alloys is significantly greater than that in pure aluminum. This explains why aluminum-lithium alloys are less susceptible to gas porosity than other aluminum alloys. This is despite the comparatively higher hydrogen concentration in

lithium-containing aluminum alloys. Figure 11 shows the comparatively higher hydrogen solubility in the lithium-containing aluminum alloy 2090 than in other conventional aluminum alloys.

Reactions During Solidification of Aluminum and Its Alloys. Because the solubility of hydrogen is significantly higher in the liquid aluminum as compared with the solid, there will be enrichment of the liquid during solidification. Assuming no solid diffusion and complete liquid diffusion, at the end of solidification there will be a large increase in the hydrogen concentration. Therefore, even when the hydrogen content is far below the solubility for the bulk liquid during solidification, the concentration in the enriched liquid will increase, and the solubility limit may be exceeded. For example, consider pure aluminum containing a hydrogen concentration of 0.45 cm3/100 g at 750  C (1380  F). By the time the alloy is 90% solidified, the concentration in the remaining liquid will be approximately 4.03 cm3/100 g H, which will exceed the hydrogen solubility for 1 atm, resulting in the formation of hydrogen bubbles. Inert Gas Flushing. It is possible to remove hydrogen from liquid aluminum by inert gas (argon or nitrogen) flushing. As discussed pre viously for cast iron, it is possible to calculate the fastest possible rate, or the minimum amount of purge gas needed to remove hydrogen. The general form of the equation is: 1 1 kH Vt ¼  ½H ½Ho  W

Fig. 8

where [H] is the hydrogen content in ppm at time t, [Ho] is the initial hydrogen content, W is the weight (in kilograms), t is the time (in minutes), kH is a constant related to the solubility, and V is the flow rate (in m s1). In practical units, at 750  C (1380  F):

Hydrogen solubility in liquid and solid pure aluminum at 1 atm hydrogen partial pressure



Temperature, ⬚F 1290 3.5

1380

1470

1560

1650

1740

Table 1 Empirical constants for the calculation of hydrogen solubility (in cm3/ 100 g) in the liquid state of some binary aluminum alloys at 1 atm hydrogen partial pressure

3.0

Hydrogen solubility, cm3/100 g

3.2 wt% Mg

4 wt% Cu Pure AI

2.5

(Eq 21)

Solubility constants, log10 S; cm3 =100 g ¼ A T þB A

B

Al-Mg alloys Al-3.2wt%Mg Al-6wt%Mg

2735.6 2695.5

2.9752 3.1452

Al-Zn alloys Al-1.8wt%Zn Al-3.5wt%Zn Al-6wt%Zn

3193.9 3058.1 2957.7

3.1392 3.0382 2.8602

Al-Si alloys Al-2wt%Si Al-4wt%Si Al-7wt%Si

2777.8 2941.3 2731.0

2.7702 2.9032 2.6562

Al-Cu alloys Al-2wt%Cu Al-4wt%Cu

2955.3 3047.7

2.9060 2.9392

Al-Li alloys Al-1wt%Li Al-2wt%Li Al-3wt%Li

2141.4 2850.7 2869.7

2.5952 3.3832 3.4862

Alloy

2.0

1.5 4.4 wt% Zn

1.0 7 wt% Si 0.5 700

Fig. 9

750

800 850 Temperature, ⬚C

900

950

Hydrogen solubility in liquid pure aluminum and binary aluminum alloys at 1 atm hydrogen partial pressure

Source: Ref 14

Temperature, ⬚F 570 0.10

660

750

840

930

1020

1110

Hydrogen solubility, cm3/100 g

0.08

0.06 7 wt% Si 5.3 wt% Mg 0.04

0.00 300

300

200

100

0

0

350

400

450 Temperature, ⬚C

500

550

1380

1470

Temperature, ⬚F 1560

1650

1740

600

1830

14.0 AA2090 12.0

10.0

ln 6.0

Pure Al AA7050

2.0 Duralumin B 0.0 700

750

800

850 Temperature, ⬚C

900

950

Hydrogen solubility in liquid multicomponent aluminum alloys at 1 atm hydrogen partial pressure

0.2

0.3

0.4

Gas removal ratio for equilibrium in aluminum at 760  C (1400  F)

1000



kg m3ðstpÞ



ðppmÞ s

(Eq 22)

For example, for 10 Mg (2.2  104 lb) of aluminum, a gas purge rate of 0.01 m3/s (0.35 ft3/s) for 100 s could reduce the hydrogen from 0.2 to 0.09 ppm. As the hydrogen content decreases, it becomes increasingly difficult to remove hydrogen (Ref 34), as indicated in Fig. 12. As the hydrogen content decreases, the ratio R of the purge gas to the hydrogen gas removed increases from approximately 15 at 0.4 ppm to over 500 at 0.1 ppm (1 cm3/100 g = 0.9 ppm). It should be emphasized that this calculation indicates the theoretical minimum amount of purge gas required. Due to kinetic factors, the amount of purge gas can be significantly higher. There are two limiting cases with respect to hydrogen removal: the thermodynamic limit discussed previously and the diffusion in the liquid film boundary layer surrounding the gas bubble. For the limiting case of diffusion control, the rate equations are similar to those discussed previously for nitrogen in iron, and the rate can be expressed by:

8.0

4.0

0.1

kH ¼ 62; 000

Hydrogen solubility in solid pure aluminum and binary aluminum alloys at 1 atm hydrogen partial pressure

1290

Fig. 11

400

Fig. 12 4 wt% Cu

Fig. 10

500

Hydrogen dissolved in aluminum (H), cm3/100 g

Pure Al 0.02

Hydrogen solubility, cm3/100 g

Gas removal ratio (R ), (standard ft3 N2)/(standard ft3 H2)

Gases in Metals / 69

%H km At ¼ H %Hi W

(Eq 23)

where %Hi is the initial hydrogen content, r is the liquid density, kmH is the mass transfer coefficient for hydrogen, A is the bubble surface area, and W is the weight of the melt. Both A and kmH increase with decreasing bubble size; consequently, the rate is sensitive to bubble size, with hydrogen removal favored by small bubbles. In general, the rate will be controlled by both diffusion and the thermodynamic limit in series. In an analysis of the case of mixed control, it was found that the purging efficiency depends on the bubble size and hydrogen content (Ref 35), as shown in Fig. 13. Purging efficiency is the ratio of the actual amount of gas required to the theoretical minimums given by Eq 23. Both

70 / Principles of Liquid Metal Processing Temperature, ⬚F

100

1110 10.0

1290

1470

1650

1830

2010

2190

2370

2550

0.05 cm3/100 g

Pure Cu

50

8.0

25 0.1 cm3/100 g 0

2 1 Bubble diameter (d B), cm

0

Fig. 13

3

Purging gas efficiency as a function of bubble size for 0.05 and 0.1 cm3/100 g of hydrogen

the interfacial area and the mass transfer coefficient increase with decreasing bubble size; consequently, the rate of hydrogen removal is enhanced by having small bubbles. The use of porous plugs helps improve the rate by providing small bubbles, but the bubbles generally coalesce in the bath, reducing the beneficial effect. The use of an impeller with a porous plug improves the rate further by dispersing the bubbles. It has also been found that the addition of chlorine (Ref 36) or Freon (E.I. Du Pont de Nemours, Inc.) (Ref 37) improves the rate in some cases. These halogen gases increase the mass transfer coefficient km and the rate. However, if the rate is being primarily controlled by the thermodynamic limit, the addition of these gases will have little effect. It is also possible that the halogen is actually taking part in a hydrogen removal reaction: 2H þ Cl2 ¼ HCl

(Eq 24)

Equation 24, under some circumstances, is thermodynamically favorable. This will increase the amount of hydrogen that can be present in the inert gas bubble; that is, hydrogen can be present as H2 or hydrogen chloride gas, thus increasing the rate of removal.

Copper and Its Alloys As in aluminum and its alloys, gas porosity is a problem in the casting of copper and copper alloys. Hydrogen is the most important dissolved gas in copper and its alloys. It forms dilute solutions in oxygen-free copper. However, in impure copper there are complex interactions among the gas-forming elements hydrogen, oxygen, and sulfur. Thus, in addition to hydrogen, gaseous compounds such as water vapor, carbon monoxide, and sulfur dioxide can also evolve during solidification. Hydrogen Solubility and Reactions. The solubility of hydrogen in pure copper has been experimentally determined by various investigators (Ref 38–47). Combined linear regression

Hydrogen solubility, cm3/100 g

Efficiency, %

75

6.0 5.9 wt% Sn

4.0 11.5 wt% Sn 21.7 wt% Sn 54.8 wt% Sn

2.0

40.2 wt% Sn

0.0 600

Fig. 14

700

800

900

1000 1100 Temperature, ⬚C

1200

1300

1400

Effect of temperature on the solubility of hydrogen in pure copper and copper-tin alloys at 1 atm hydrogen partial pressure

analyses of the sets of reported data for hydrogen solubility in liquid and solid pure copper are given by the following equations: Liquid copper-hydrogen: 2327:59 þ 2:4649 log10 S; cm3=100 g ¼ T Solid copper-hydrogen: 2057:76 þ 1:6829 log10 S; cm3=100 g ¼ T

(Eq 25)

(Eq 26)

Figure 14 is a graphical representation of the effect of temperature on the solubility of hydrogen in copper and copper-tin alloys. The difference in solubility between the liquid and solid pure copper is not as marked as for pure aluminum. At the melting point, 1083  C (1981  F), the concentrations of hydrogen in equilibrium with the gas at atmospheric pressure are 5.6 and 1.46 cm3/100 g for the liquid and solid, respectively. Thus, the partition coefficient at 0.26 for hydrogen in pure copper is comparatively higher than that in pure aluminum, where the partition coefficient is extremely low at 0.04. Hydrogen solubility in both liquid and solid pure copper is directly proportional to the square root of the hydrogen partial pressure, in accordance with Sievert’s law (Ref 38, 45). Published results on the effect of alloying elements on the solubility of hydrogen in copper are sparse. Nickel increases the solubility of

hydrogen in liquid pure copper. For example, at 1200  C (2190  F), the solubility of hydrogen in Cu-10wt%Ni is 11.2 cm3/100 g compared to 7.7 cm3/100 g in pure copper. However, large additions of aluminum, zinc, and tin decrease the solubility of hydrogen in pure copper. For example, at 1200  C (2190  F), the solubility of hydrogen in Cu-5.9wt%Sn is 6.2 cm3/100 g compared to 7.7 cm3/100 g in pure copper. The coefficients for the effects of temperature on the solubility of hydrogen in some liquid binary copper-tin alloys, obtained from data reported by Bever and Floe (Ref 38), are given in Table 2. Hydrogen can enter the copper directly from the hydrogen in the atmosphere, but most likely it enters according to: 2Cu þ H2 O ¼ Cu2 O þ 2H

(Eq 27)

H2 O ¼ 2H þ O

(Eq 28)

Reactions given in Eq 27 and 28 are favored by copper with low oxygen contents. If there are high concentrations of alloying elements with more stable oxides than cuprous oxide, such as tin, the alloying element may react to put hydrogen into solution, for example: Sn þ H2 O ¼ SnO þ 2H

(Eq 29)

The main cause of porosity in the casting results from the evolution of hydrogen or water

Gases in Metals / 71 Table 2 Empirical constants for calculating hydrogen solubility (cm3/100 g) in the liquid state of copper-tin alloys at 1 atm partial pressure Solubility constants, log10 S; cm3 =100 g ¼ A T þB Alloy

Cu-5.9wt%Sn Cu-11.5wt%Sn Cu-21.7wt%Sn Cu-40.2wt%Sn Cu-54.8wt%Sn

A

B

2284.8 2304.0 2551.0 4239.0 3410.8

2.3465 2.2962 2.3289 3.0562 2.3737

Temperature range 

C

1100–1300 1000–1300 1000–1300 1000–1300 1000–1300



F

2010–2370 1830–2370 1830–2370 1830–2370 1830–2370

Source: Ref 38

vapor during solidification. The hydrogen and oxygen contents increase because of enrichment during solidification, as described previously for cast iron. As the concentrations increase toward the end of the solidification, hydrogen or water vapor is evolved by the reverse reactions of Eq 27 and 28, which become thermodynamically favorable because the hydrogen and oxygen contents in the last liquid to solidify are high. This mechanism is supported by the observation that the voids in the casting are normally distributed along the grain boundaries, which are the last parts to solidify. Other gases that can cause porosity are sulfur dioxide and carbon monoxide, which also form during solidification according to Eq 30 and 31, respectively: S þ 2O ¼ SO2

(Eq 30)

C þ O ¼ CO

(Eq 31)

The thermodynamic pressure of carbon monoxide gas can be very high, even at moderate oxygen and carbon contents, because the solubility of the carbon in copper is limited, and therefore, its chemical activity is high. Porosity in Copper Alloys. In copper ingots, gas porosity usually results from H2 and water evolution. The evolution of as little as 1.12 cm3/100 g of hydrogen results in a gas evolution in 44% of the volume of the metal. If the sulfur content exceeds 0.05 wt%, the evolution of sulfur dioxide may also contribute to gas porosity. Only if the oxygen content exceeds 0.01% is carbon monoxide considered to be the cause of porosity (Ref 48). Because phosphorus deoxidizes copper significantly, the oxygen content of phosphorized copper is very low, and water, carbon monoxide, and sulfur dioxide do not contribute to porosity; only H2 need be of concern. Copper-zinc alloys rarely have problems associated with gas porosity, primarily because zinc deoxidizes the metal. Consequently, brasses do not dissolve significant quantities of oxygen. Zinc may also remove dissolved gases because of its high vapor pressure, which results in the zinc vapor flushing hydrogen out of the melt. During casting of copper-zinc alloys, degassing is not usually required. This is because the quantities of hydrogen absorbed by industrially produced copper-zinc melts are usually insufficient to cause problems in castings. The lowest hydrogen contents are usually

found in brasses with the highest zinc contents. However, the copper-tin alloys are the copper alloys that are most susceptible to gas porosity. This is because tin significantly increases the solubility of hydrogen in copper, tin mildly deoxidizes copper, and copper-tin alloys have long freezing ranges. Degassing of Copper Alloys. The two primary methods of removing dissolved gases from copper alloys are oxidation-reduction and inert gas flushing. Although vacuum degassing is theoretically possible, it is rarely used because it is not cost-effective. In oxidation-reduction, the first step is to remove the hydrogen by oxidation using an oxidizing slag or oxygen-rich copper. This involves increasing the activity of oxygen in the melt by melting it under an oxidizing flame or covering it when molten with an oxidizing flux such as a mixture of cupric oxide-boric acid-sodium chloride-sodium fluoride (Ref 49). This drives off the hydrogen in the form of steam. Then, just prior to pouring, the melt is deoxidized by adding phosphorus or other deoxidizers. The phosphorus deoxidizer produces cuprous phosphate, a molten deoxidation product that separates fairly easily from the molten copper metal. Generally, an addition of 0.02 to 0.03 wt% P is sufficient to deoxidize bronzes and gun metals. Calcium boride, boron carbide, and lithium have also been used for deoxidation (Ref 50). A charcoal cover is often used for deoxidation and protection of the melt from reoxidation. For copper alloys that contain strong deoxidizers, such as phosphorus, zinc, and tin, it is not possible to remove hydrogen by oxidation, because these elements will form stable oxides. For these alloys, inert gas flushing with nitrogen is commonly used. The theoretical minimum amount of inert gas is determined by using relationships similar to those derived for iron and aluminum. In practice, 150 to 250 L (5.3 to 8.8 ft3) of nitrogen per 1000 kg (2200 lb) of copper alloy is generally recommended (Ref 51).

Magnesium and Its Alloys Magnesium and its alloys are susceptible to gas-induced porosity. Of the nonferrous metals considered in this article, magnesium dissolves more hydrogen and has the potential for forming hydrides. Magnesium hydride, MgH2, is stable only for very high hydrogen activities

that are not normally encountered in metallurgical operations. At comparable temperature and hydrogen partial pressure, the solubility of hydrogen in magnesium and its alloys is an order of magnitude of that in aluminum and copper and their alloys. However, the propensity for gas porosity formation in magnesium is less than that in aluminum and copper. The principal source of hydrogen in magnesium and its alloys is moisture in the surrounding atmosphere or introduced into the melt when corroded materials are charged or damp fluxes are used. Magnesium and its alloys readily absorb hydrogen by the reaction with water vapor: MgðlÞ þ H2 OðvÞ ! MgOðsÞ þ H2 ðin MgÞ

(Eq 32)

The standard free energy change, DGo, for this reaction is extremely high: Go ¼ 368; 600  19:8T log10 T þ 122T J=mol

(Eq 33)

where T is the temperature in K. Application of the van’t Hoff reaction isotherm yields a value for aH2 =aH2 O that is even higher than the corresponding quotient for aluminum. In the absence of hydride formation, hydrogen dissolution in magnesium and its alloys is in the atomic form. Hydrogen Solubility and Reactions During Solidification. The solubility of hydrogen in pure magnesium has been experimentally determined by several investigators (Ref 52–59). However, the measurement of the solubility of hydrogen in magnesium and its alloys is an experimentally difficult task. This is because magnesium is too volatile and reactive toward traces of water vapor for direct measurement with the Sievert’s method. Combined linear regression analyses of the sets of reported data considered to be reliable yielded the following empirical equations for hydrogen solubility in liquid and solid pure magnesium as a function of temperature: Liquid magnesium-hydrogen: log10 S; cm3=100 g 1290:5 þ 3:042 ¼ T =K

(Eq 34)

Solid magnesium-hydrogen: log10 S; cm3=100 g 1062:47 þ 2:6254 ¼ T =K

(Eq 35)

At the melting point, 651  C (1204  F), the concentrations of hydrogen in equilibrium with the gas at atmospheric pressure are 44.2 and 29.9 cm3/100 g for the liquid and solid, respectively. Thus, the partition coefficient at 0.68 compared to 0.27 and 0.04 for hydrogen in pure magnesium is much higher than that in copper and aluminum, respectively. Hydrogen solubility in both the liquid and solid pure magnesium

72 / Principles of Liquid Metal Processing Temperature, ⬚F 750

570 120.0

930

1110

1290

1470

1650

1830

Pure Mg

Table 3 Coefficients for calculating hydrogen solubility (cm3/100 g) in the liquid state of magnesium alloys at 1 atm partial pressure Solubility constants, log10 S; cm3 =100 g ¼ A T þB

5.3 wt% Al

2 wt% Sr 1.3 wt% Zn Hydrogen solubility, cm3/100 g

A

B

Mg-Al alloys Mg-5.3wt%Al Mg-8.3wt%Al

1127 1201.5

2.8219 2.8532

Mg-Zn alloys Mg-1.3wt%Zn Mg-3.2wt%Zn Mg-6.8wt%Zn

1253.6 1074.7 1262

2.9733 2.7958 2.9577

Mg-Sr alloys Mg-0.5wt%Sr Mg-2wt%Sr Mg-3.4wt%Sr

1198.9 1028.3 886.9

2.9473 2.8323 2.7487

Mg-Si alloy Mg-1wt%Si

1126.1

2.875

Alloy

1 wt% Si

100.0

80.0

60.0

Derived from results given in Ref 38

40.0

that in aluminum and copper alloys. This can be attributed to the following reasons: 20.0

 The significantly higher partition coefficient

0.0 300

400

500

600

700

800

900

1000

Temperature, ⬚C

Fig. 15

Solubility of hydrogen in magnesium and magnesium alloys as a function of temperature and at 1 atm hydrogen partial pressure

Hydrogen solubility, cm3/100 g

60

55

50

Ag Sn Cu Zn Sb

45

40 0

2

6 8 4 Alloying element, wt%

10

12

Fig. 16

Effect of alloying elements on the solubility of hydrogen in binary magnesium alloys at 700 C (1290  F) and 1 atm hydrogen partial pressure 

is directly proportional to the square root of the hydrogen partial pressure, in accordance with Sievert’s law (Ref 52, 53). Alloying elements affect the solubility of hydrogen in magnesium (Fig. 15, 16). With the exception of nickel, zirconium, and strontium, the addition of alloying elements to magnesium generally results in a decrease in the solubility of hydrogen in magnesium. For example, the solubility of hydrogen in both liquid and solid

magnesium decreases with an increase in the addition of zinc, silicon, aluminum, lead, silver, tin, manganese, antimony, and copper (Ref 57, 60). However, because it apparently forms hydrides in the molten aluminum, strontium increases the solubility of hydrogen in liquid magnesium (Ref 57). Although zirconium increases the solubility of hydrogen significantly in solid magnesium, at 700  C (1290  F) it does not have any discernible effect on the solubility of hydrogen in liquid magnesium (Ref 60). In liquid magnesium-zirconium alloys, zirconium apparently acts as a scavenger for hydrogen, precipitating as zirconium hydride, ZrH2, and consequently obviating formation of gasinduced porosity in the castings. In the solid state, the absorbed hydrogen is also precipitated within the structure as zirconium hydride. At above 1 wt%, the addition of nickel significantly increases the solubility of hydrogen in liquid magnesium. The coefficients for calculating the solubility of hydrogen as a function of temperature in some binary magnesium alloys are given in Table 3. Porosity in Magnesium and Its Alloys. Castings of magnesium alloys are susceptible to gas porosity. The prevalence of gas porosity apparently increases with an increase in the hydrogen content of the molten metal. However, the susceptibility of magnesium alloys to gas porosity formation is significantly less than

of hydrogen in magnesium and its alloys reduces hydrogen evolution during solidification.  The nominally higher solubility of hydrogen in liquid magnesium alloys makes the nucleation of gas porosity during solidification rather difficult.  Nucleation of the gas bubble requires that the hydrogen concentration in the solidifying melt should exceed the equilibrium solubility of hydrogen in the melt. The comparatively higher partition coefficient of hydrogen also makes the characteristic of gas porosity in magnesium alloys different from that in aluminum and copper alloys. The gas porosity in magnesium alloys is more interdendritic in nature, more like a shrinkageinduced porosity. This is due to the proportionately much lower decrease in hydrogen solubility during solidification, so that a hydrogen content in the residual liquid sufficient to nucleate gas bubbles is not reached until at a later stage in solidification, if at all. Degassing of Magnesium Alloys. Degassing of magnesium and its alloys is usually accomplished by bubbling nitrogen, helium, nitrogen-chlorine mixture, or chlorine through the molten metal (Ref 61). The most effective of the gas-flushing methods is the bubbling of chlorine through the molten metal at 720 to 780  C (1330 to 1440  F). It can be used to reduce hydrogen content to 2 to 3 cm3/100 g. Bubbling of nitrogen reduces hydrogen to approximately 4 cm3/100 g. Other degassing methods that have been shown to be effective in reducing the hydrogen content of magnesium and its alloys include vacuum degassing and treatment with a flux. Treatment with a flux consisting of 55 wt% KCl, 34 wt% MgCl2, 9 wt% BaCl2, and 2 wt% CaCl2 apparently decreases the

Gases in Metals / 73 hydrogen content to 3 to 4 cm3/100 g. Although more environmentally friendly, vacuum degassing is not as effective as the gas-bubbling methods.

Overcoming Gas Porosity Gas in cast iron, aluminum, copper, and magnesium can be a major problem. The porosity results from reactions with the environment or the evolution of gases during solidification. In determining if gas evolution can cause porosity, the thermodynamics for the reactions must be favorable. In this article, the solubilities of the common gases and the reactions during solidification were discussed. Because of enrichment during solidification, the concentration of the elements increases to the extent that the thermodynamic pressure of the dissolved gases may exceed 1 atm. In addition to the thermodynamics, kinetic factors and interfacial energies determine if the gases will evolve. The most common method for removing hydrogen from aluminum, copper, and magnesium, which can also be used for removing hydrogen and nitrogen from cast iron, is inert gas flushing. The theoretical minimum amount of flushing gas required for removing hydrogen and nitrogen can be computed. In practice, the amount of inert gas required is somewhat higher because of kinetic factors. The kinetics of inert gas flushing can be improved by the dispersion of small bubbles throughout the melt. ACKNOWLEDGMENT Updated from Richard J. Fruehan, Gases in Metals, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 82–87. REFERENCES 1. R.J. Fruehan, in Proceedings of the Physical Chemistry of Foundry Processes Symposium (Warren, MI), General Motors Corporation, 1986 2. S. Katz and C. Landefeld, General Motors Research, private communication, 1986 3. Making, Shaping and Treating of Steel, 10th ed., United States Steel Corporation, 1985 4. A. Kagawa and T. Okamoto, Trans. Jpn. Inst. Met., Vol 22 (No. 2), 1981, p 137 5. R.D. Pehlke and J. Elliot, Trans. TMSAIME, Vol 227, 1963, p 894 6. M. Inouye and T. Choh, Trans. JISI, Vol 8, 1968, p 134 7. R.J. Fruehan and L.J. Martonik, Metall. Trans. B, Vol 11, 1980, p 615 8. P.C. Glaws and R.J. Fruehan, Metall. Trans. B, Vol 16, 1985, p 551 9. P.C. Glaws and R.J. Fruehan, Metall. Trans. B, Vol 17, 1986, p 317

10. M. Byrne and G.R. Belton, Metall. Trans. B, Vol 14, 1983, p 441 11. F. Tsukihashi and R.J. Fruehan, Trans. JISI, Vol 27, 1987, p 858 12. R.J. Fruehan, Ladle Metallurgy: Principles and Practices, Iron and Steel Society of AIME, 1985 13. R.J. Fruehan, B. Lally, and P.C. Glaws, in Proceedings of the Fifth International Iron and Steel Congress (Washington, D.C.), Iron and Steel Society of AIME, 1986 14. P.N. Anyalebechi, “A Comprehensive and Critical Review of Reported Hydrogen Limits in Aluminum and Its Alloys,” unpublished research, 2000 15. C.E. Ransley and H. Neufeld, J. Inst. Met., Vol 74, 1947–1948, p 599–620 16. W. Baukloh and F. Oesterlen, Z. Metallkd., Vol 30 (No. 11), 1938, p 386–389 17. W.R. Opie and N.J. Grant, J. Met. (Trans. AIME), Vol 188, 1950, p 1237–1241 18. C.E. Ransley and D.E.J. Talbot, Z. Metallkd., Vol 46 (No. 5), 1955, p 328–337 19. D.E.J. Talbot and P.N. Anyalebechi, Mater. Sci. Technol., Vol 4 (No.1), 1988, p 1–4 20. P.N. Anyalebechi, Ph.D. thesis, Brunel University, Middlesex, England, 1985 21. D. Stephenson, Ph.D. thesis, Brunel University, Middlesex, England, 1978 22. H. Shahani, Ph.D. thesis, The Royal Institute of Technology, Stockholm, Sweden, 1984 23. J. Kocur, K. Tomasek, and L. Rabatin, Hutn. Listy, Vol 44 (No. 4), 1989, p 269– 275 24. H. Liu, L. Zhiang, and B. Bouchard, Recent Developments in Light Metals, M. Gilbert, P. Tremblay, and E. Ozberk, Ed., The Metallurgical Society of Canadian Institute of Metallurgists, 1994, p 257–268 25. H. Liu, B. Bouchard, and L. Zhiang, Light Metals 1995, J. Evans, Ed., The Minerals, Metals and Materials Society, 1995, p 1285–1291 26. P.N. Anyalebechi, Light Metals 1998, B. Welch, Ed., TMS, 1998, p 827–842 27. W. Eichenauer and A. Pebler, Z. Metallkd., Vol 48, (No. 7), 1957, p 373–378 28. W. Eichenauer, K. Hattenbach, and A. Pebler, Z. Metallkd., Vol 52, (No. 10), 1961, p 682–684 29. W. Eichenauer, Z. Metallkd., Vol 59, 1968, p 613–616 30. M. Ichimura, M. Imabayashi, and M. Hayakawa, Nippon Kinzoku Gakkaishi, Vol 43, (No. 9), 1979, p 876–883 31. E. Hashimoto and T. Kino, J. Phys. F., Met. Phys., Vol 13, 1983, p 1157–1165 32. M. Ichimura, Y. Sasajima, and M. Imabayashi, Mater. Trans., JIM, Vol 32 (No. 12), 1991, p 1109–1114 33. M. Ichimura, Y. Sasajima, and M. Imabayashi, Mater. Trans., JIM, Vol 33 (No. 5), 1992, p 449–453 34. G.K. Sigworth, Trans. AFS, Vol 95, 1987, p 73–78

35. G.K. Sigworth and T.A. Engh, Metall. Trans. B, Vol 13, 1982, p 447–460 36. J. Botor, Metal. Odlev, Vol 6, 1980, p 21–27 37. J. Botor, Aluminum, Vol 56 (No. 8), 1980, p 519–522 38. M.B. Bever and C.F. Floe, Trans. AIME, Vol 156, 1944, p 149–159 39. A. Sieverts, Z. Phys. Chem., Vol 77, 1911, p 11 40. P. Rotgen and F. Moller, Metallwirtschaft, Vol 13, 1934, p 81, 97 41. E. Kato, H. Ueno, and T. Orimo, Trans. Jpn. Inst. Met., Vol 11, 1970, p 351 42. R.B. McLellan, J. Phys. Chem. Solids, Vol 34, 1973, p 1137–1141 43. C. Thomas, Trans. Met. Soc. AIME, Vol 239, 1967, p 485–490 44. F.G. Jones and R.D. Pehkle, Met. Trans., Vol 2, 1971, p 2655–2663 45. M. Mokaram, Ph.D. thesis, Brunel University, London, 1974 46. W. Eichenauer and A. Pebler, Z. Metallkd., Vol 48, 1957, p 373–378 47. W. Eichenauer et al., Z. Metallkd., Vol 56, 1965, p 287 48. R.H. Waddington, J. Inst. Met., 1948– 1949, p 311 49. W.T. Pell-Walpole, J. Inst. Met., Vol 70, 1944, p 127 50. E.R. Thews, Metall., June 1956, p 431 51. W.A. Baker and F.C. Child, J. Inst. Met., Vol 70, 1944, p 349 52. J. Koeneman and A.G. Metcalfe, Trans. ASM, Vol 51, 1959, p 1072–1082 53. T. Watanabe, Y.C. Huang, and R. Komatsu, J. Jpn. Inst. Light Met., Vol 46, (No. 2), 1975, p 76–81 54. Y.C. Huang, T. Watanabe, and R. Komatsu, Proc. Fourth Int. Conf. Vacuum Metallurgy, 1973, p 173–176 55. D. F. Chernega, Gotvjanskij and Prisjasznjokm Bestn. Kiev. Politexn. In-Ta, Kiev, Vol 17, 1959, p 55–58 56. V. Shapavalov, N.P. Serdyuk, and O.P. Semik, Dop. Akad. Nauk Ukr. RSR. Ser. A. Fiz.-Mat. Tekh. Nauki, Vol 6, 1981, p 99–101 57. E. Overlid, P. Bakke, and T.A. Engh, Light Metals 1997, Metaux Legers, Canadian Inst. of Metallurgists, 1997, p 141–154 58. Z.D. Popovic and G.R. Piercy, Met. Trans. A, Vol 6, 1975, p 1915–1917 59. M.V. Sharov and V.V. Serebryakov, Nauchnyye Doklady Vvyshe Shkoly, Metallurgiya, (No. 2), 1958, p 37 60. T. Watanabe, Y. Tachihara, Y.C. Huang, and R. Komatsu, Kei Kinzoku, Vol 26 (No. 4), 1976, p 167–174 61. T. Watanabe, Y.C. Huang, R. Komatsu, and T. Ito, J. Jpn. Inst. Light Met., Vol 26 (No. 6), 1976, p 266–272

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 74-83 DOI: 10.1361/asmhba0005193

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Inclusion-Forming Reactions Revised by Paul K. Trojan, University of Michigan—Dearborn; and Richard Fruehan, Carnegie Mellon University

INCLUSIONS can be defined as nonmetallic and sometimes intermetallic phases embedded in a metallic matrix. They are usually simple oxides, sulfides, nitrides, or their complexes in ferrous alloys and can include intermetallic phases in nonferrous alloys. In almost all instances of metal casting, they are considered to be detrimental to the performance of the cast component. For example, mechanical properties can be adversely influenced by inclusions, which act as stress raisers. There is no set pattern for the effect, but some properties are more sensitive to the presence of inclusions than others. Elongation or reduction in area is usually modified more significantly than ultimate tensile strength. As a result, ductility specifications are common quality-control indices in cast products. Similar observations are found with porosity due to gas or occluded shrinkage. The loss in casting properties measured by a tensile test may then reflect a combination of defects; of these defects, inclusions are only a single but very important source of poor performance. Because imperfections become areas of higher stress concentration, the percentage of property loss becomes greater when the strength requirement is higher. An important example is the decrease in fracture toughness when inclusions are present in higher-strength lower-ductility alloys. Similar pronounced property degradation is observed in tests that reflect slow, rapid, or cyclic strain rates, such as creep, impact, and fatigue testing. In all cases, the degree of property loss will depend on the parameters that control the characteristics of all microstructures. These metallurgical phase controls are the nature, amount, size, shape, distribution, and orientation of the phases. If inclusions are to be controlled, these characteristics must be suitably modified to minimize their impact on casting performance. Although inclusions are generally considered to be detrimental, their intentional introduction in larger quantities can lead to unique dispersion-strengthened materials. An example would be the dispersion of insoluble silicon carbide whiskers or alumina (Al2O3) fibers in a molten aluminum alloy that is subsequently cast into a final shape by squeeze casting methods (Ref 1).

Inclusion Types There are essentially two classifications for all inclusions:
Exogenous: those derived from external

causes


Indigenous: those that are native, innate, or

inherent in the molten metal treatment process Exogenous Inclusions. Slag, dross, entrapped mold materials, and refractories are examples of inclusions that would be classified as exogenous. In most cases, these inclusions would be macroscopic or visible to the naked eye at the casting surface (Fig. 1). When the casting is sectioned, they may also appear beneath the external casting surface if they have had insufficient time to float out or settle due to

1

Fig. 1

density differences with respect to the molten metal. Indigenous inclusions include sulfides, nitrides, and oxides derived from the chemical reaction of the molten metal with the local environment. Such inclusions are usually small and require microscopic magnification for identification. They are often uniformly distributed within the microstructure. However, a grainboundary distribution of indigenous inclusions (Fig. 2) can be particularly damaging to mechanical properties. There are cases in which indigenous inclusions such as oxides may react with mold materials to promote an exogenous inclusion variety that otherwise may not appear in the absence of the indigenous oxide. To circumvent the strict definition as indigenous or exogenous, a number of investigators prefer to catalog all inclusions as either macroinclusions of microinclusions.

2

Ductile iron crankshaft segment essentially free of exogenous inclusions (1, left) and with numerous exogenous inclusions (2, right). Low pouring temperature and poor mold-filling practice were the cause of the inclusions in part 2.

Inclusion-Forming Reactions / 75 Oxygen, at.% 49 50

51

52

53

54

55

56

L2

L1 + L 2

(δ-Fe) 1394⬚ 1373⬚

59

60

61

62

63

64

L2 + O2 (g; 1atm) 28.30

3000

1585⬚

28.07 Fe O + O (g; 1atm) 3 4 2 1459⬚

22.60

1500 1400

58

1599⬚

1600 1529⬚

57

25.31

22.84

1426⬚

28.36

25.60 27.64

2800

30.04

2600

22.91 23.16

2400

1300 FeO (Wustite)

Temperature, ⬚C

1200

2200 Fe3O4 (Magnetite)

1100

2000 (γ-Fe)

Fe2O3 + O2 (g; 1atm)

1000 912⬚

23.10

900

1800

Fe2O3 (Hematite)

Temperature, ⬚F

1700

47 48

1600 Curie temperature

800

Fig. 2

Type II indigenous sulfide inclusions in a 0.25% C steel. Etched using 2% nital. Original Magnification: 500

600

Although such definitions describe the anticipated effect on mechanical properties, there can be a loss in source identification of the inclusion. Whatever definition is used, control of inclusions requires an understanding of their origin and the associated physical chemistry.

770⬚

1400

(α-Fe)

1200

700

560⬚ 23.26

1000

500 400 20

800

30.06 21

22

23

24

25

26

27

28

29

30

31

32

33

34

Oxygen, wt%

Physical Chemistry

Fig. 3

Phase diagrams, thermochemical relationships, and reaction rates must be considered for proper understanding of inclusion formation. The general concepts of each, along with their practical significance, are established in the following sections in this article. Phase Equilibria. Phase diagrams map the occurrence of phases to be expected under equilibrium conditions. Equilibrium also requires long periods of time to achieve. This is certainly not the case in castings, in which even moderate cooling rates may inhibit equilibrium phases. For example, in the iron-oxygen system shown in Fig. 3, wustite (FeO) produced at high temperatures should break down to a-iron and magnetite (Fe3O4) at 560  C (1040  F). However, iron scales may show considerable amounts of retained FeO at room temperature. A further limitation is that pure elements seldom exist in the casting process, and more than two elemental species are usually present. In Fig. 3, pure oxygen is one component, while air (20% O2) is more likely to be the oxidizing source during casting. Furthermore, a cast steel will contain carbon and other elements, such as silicon or aluminum, that will have a greater affinity for oxygen than iron. Therefore, phase diagrams become a tool for understanding, but the predicted equilibrium for inclusion occurrence may not always be observed

for the casting cooled under nonequilibrium conditions. Thermochemistry. Any chemical reaction will occur if the products are at a lower energy level than the reactants. The energy measurement is the chemical free energy, which can be treated simply as the energy of chemical position. Free energy values can be obtained directly from numerous sources or can be calculated from tabulated enthalpy, entropy, and heat capacity data. An equilibrium is then established between the products and reactants such that very little remaining reactant indicates a more or less complete reaction. A more thorough discussion of chemical free energy is provided in standard textbooks on chemical thermodynamics. By way of example, assume pure molten iron contains dissolved oxygen at 1600  C (2910  F). A small amount of aluminum is added to the melt, and the chemical reaction is: 2Al þ 3O ! Al2 O3 ðsÞ where the underlined indicates the elements to be dilute and dissolved in iron while the (s) indicates that Al2O3 is solid at 1600  C (2910  F). (The melting point of Al2O3 is 2072  C, or 3762  F.) For the reaction to occur, the standard free energy exchange at 1600  C (2910  F) must be negative. However, the quantity of each component in the reaction is also important, and the chemical activity of each species must

Iron-oxygen phase diagram

be known. An overabundance of aluminum will react stoichiometrically with oxygen, and the remainder may go into liquid solution or react with the surroundings. In the previous example, the Al2O3 product is insoluble in the molten iron and can result as inclusions trapped in the solidified casting. It is then possible to calculate the free energy exchange for the occurrence of various inclusion chemistries. Their relative stability is dictated by the most negative value of free energy. In the previous example, iron oxide would not occur, because Al2O3 has a lower free energy of formation. In fact, iron oxide would be predictably reduced by aluminum to form Al2O3. As with phase equilibria, thermochemical calculations are again based on equilibrium data, and castings may not be at equilibrium throughout the solidification process. Kinetics. There are several reasons why inclusion-forming reactions are usually incomplete and why reaction rates therefore become the most important practical consideration. Much of the understanding of reaction rates is associated with the following observations:
Not all species diffuse at the same rate; when

coupled with the dependence of diffusion rates on temperature, nonuniform inclusion formation and distribution may be anticipated.

76 / Principles of Liquid Metal Processing
Chemical free energy is also temperature-

Experimental kinetic data can be presented in several ways, including mathematical relationships associated with activation energy and first-order kinetic reaction rate theory. Perhaps more familiar are isothermal and continuous cooling transformation graphical interpretations consistent with nucleation and growth-controlled phenomena.

3990

2200

Liquid 3630

2000

1840⬚ 3270

1800 Cristobalite + liquid

Corundum + mullite solid solutiion

Mullite + liquid 1595⬚

1600

2910 Mullite solid solution

Cristobalite + mullite

1400 0

20

Temperature, ⬚F

1850⬚

40

60

SiO2

80 Mullite 3Al2O3.2SiO2

2550 100 Al2O3

Al2O3, wt%

Control of Inclusions The previous discussion suggests several practical means of controlling the occurrence of inclusions in any cast metal. All of the control methods can be cataloged as chemical, mechanical, or a combination of these. Several of the more common techniques are discussed separately as follows. Chemistry Control. Although control of alloy chemistry may be obvious, composition cannot always be modified to minimize inclusions. The metal compositions may be inherently reactive with the local environment. For example, the presence of magnesium is required in nodular cast iron; the propensity for inclusions is greater than for the same nominal chemistry of a gray cast iron. Similarly, the existence of inclusions is accepted because the alternative is worse. Aluminum added to steels may result in Al2O3 inclusions, but the lack of deoxidation if aluminum were not used would result in gas holes with a greater negative effect on casting performance. However, this does not imply that if a small addition is good then more is better. Phosphorus added to a copper-base alloy can be an effective deoxidizer, but excess phosphorus can be very aggressive and can react with mold materials commonly used with this alloy family. Therefore, the importance of chemistry control cannot be overemphasized if inclusions are to be controlled. Mold/Metal Interface and Refractory Control. Another chemical variable that can lead to inclusions is that of the mold materials and refractories. The SiO2-Al2O3 phase diagram is shown in Fig. 4. It is obvious that the solidus temperature increases with Al2O3

Corundum + liquid

Mullite + liquid Temperature, ⬚C

dependent and may not vary in the same fashion for all of the species under consideration.
Available phase equilibrium data may predict the relative solubility of inclusion-forming species in the liquid state but may not treat decreasing solubility in the cooling liquid or the effects of the solidification process on the equilibrium.
The surface energy of species will change with temperature and will influence the ability of an inclusion to float, sink, agglomerate, or react secondarily with its surroundings.
Reaction products at a surface can become a barrier to further reaction and can block completion of the chemical reaction.

Fig. 4

SiO2-Al2O3 phase diagram. Source: Ref 2

content. Even if a refractory is to be used well below the solidus temperature, a high-Al2O3 refractory may still be preferred because Al2O3 plus mullite would be present rather than the silica (SiO2) plus mullite that would be present at compositions less than 72% Al2O3. The SiO2 does not possess the same thermochemical stability as Al2O3; therefore, aggressive liquid metals or oxides may react more readily with SiO2. Higher temperatures and longer holding times at temperature increase the possibility for reaction and the production of inclusions associated with refractory wear and erosion. Therefore, melt overheating, especially for long periods of time, is to be avoided. The scale on scrap surfaces can also lead to refractory erosion and exogenous inclusions. Rust on ferrous scrap can be considered a hydrated iron oxide scale that, when heated and dehydrated, can form low-melting complex oxides with the refractories. Inclusion difficulties have also been known to increase when adhering mold materials remain on charged revert scrap. The severity depends on the nature of the mold bonding agent; bentonite and organically bonded mold materials would display different tendencies toward refractory attack. The aforementioned observations can be applied to reaction between the mold materials and the molten metal. This is not to be confused with metal penetration, in which molten metal merely fills the void spaces between grains of

the mold material. However, the solution to the problem—for example, a mold wash— may be the same. In this case, the wash fills the voids between sand grains and isolates the mold material from contact with a potentially aggressive molten metal or oxide. Longer solidification times merely increase the possibility for mold-metal reaction and the accompanying inclusion-forming tendency. Separation Techniques. The choice for inclusion separation is between removal before the molten metal enters the mold or use of a removal system as a component part of the mold. Gating systems are discussed in the next section in this article. In both cases, inclusion control is basically mechanical. External methods include simple skimming of ladle slag or dross to minimize exogenous inclusion sources before pouring of the castings. A teapot ladle can be very effective for decanting clean metal from beneath a layer of floating slag or dross. Screens or sieves have been used in gating systems to filter inclusions. In general, metallic filters suffer from the tendency to alloy with the poured metal, and they are difficult to separate from the charge returns. Ceramic foam filters with consistent pore size and structural integrity are being increasingly used for those alloys from which definite cost benefits can be derived (Ref 3–5). Gating Methods. Gating and pouring practice has been known to create and to solve inclusion problems in castings. If exogenous

Inclusion-Forming Reactions / 77

500

475 68

64 425 60 400 56

Alternating stress, ksi

Alternating stress, MPa

450

375 Conventional 52 Calcium treated

350

48

325 104

105

106

107

Cycles to failure

Fig. 5

Comparison of axial fatigue data for untreated and calcium-treated rolled ASTM A516 steel. 51 mm (2 in.) thick plates tested with alternating stress ratio of 0.1. Source: Ref 8

Yield strength, ksi 30 320

50

70

90

110

130 240

280

180

200 160

120

120 Calcium treated 80

60

40 Conventional

0 −40 200

Charpy V-notch USE, ft · lb

240 Charpy V-notch USE, J

inclusions leave the ladle because of inadequate upstream control, the gating system should be designed to remove such inclusions in the pouring basin or in the runner system. Too often, a pouring basin is not designed to fill easily or to remain filled, and inclusions that may float are forced into the casting cavity. Whirl gates can be used to collect the inclusions and to ensure a better opportunity to have clean metal enter the casting proper (Ref 6). In addition, runners can have dead-end extensions that collect the first metal to enter the mold, which is the most likely to contain slag or dross (Ref 7). Pressurized gating systems are often recommended for preventing the aspiration of air that can cause the formation of inclusions during the pouring process. This technique is more important when the metal or alloy has a great tendency toward inclusion formation because of so-called secondary oxidation. More detailed information on correct gating is available in the article “Gating Design” in this Volume; many of the gating practices discussed in that article are used to minimize inclusions. Inclusion Shape Control. As pointed out previously, mechanical properties suffer when inclusions are present, with a more pronounced degradation in strain-rate-sensitive applications. For example, an inclusion at the surface of a casting subjected to cyclic strain can become a source for fracture initiation in fatigue. The most conservative approach in design is to determine the smallest inclusion at the surface that may be present without casting rejection. With no variation in inclusion continuity or size, a spherical shape (lower microstress concentration) would be less damaging than more angular shapes. Inclusions beneath the surface would exhibit the same dependence on shape for fracture propagation rather than initiation. As shown in Fig. 5, calcium treatment of a steel provides longer life under cyclic loading due to inclusion shape modification. Although the data in Fig. 5 are for a rolled constructional steel, similar results would be anticipated for a cast steel, where anisotropy also exists due to directional solidification. The advantage of a spherical inclusion shape is possibly more evident in Charpy V-notch impact data for constructional steels (Fig. 6). SEM fractographs show a pronounced difference in deformation mode with an inclusion shape modification (Fig. 7). Local ductility increases adjacent to the inclusions when the shape is spherical. This would then suggest that if inclusions are to occur and cannot be completely removed, a spherical shape is the least detrimental. In these examples, the MnS inclusions assume a planar shape that allows easier fracture propagation between inclusions or lower observed alloy ductility. The addition of calcium converts the inclusion to essentially CaS with a much higher melting point, spherical shape, and hence, higher alloy ductility. Inclusion shape control, therefore, is a useful technique that is often applied to more demanding casting applications in which added

300

400

500

600

700

800

900

0 1000

Yield strength, MPa

Fig. 6

Comparison of Charpy V-notch upper-shelf energies (USEs) for several grades and thicknesses of untreated and calcium-treated steel. Source: Ref 9

processing costs can be justified. The principle is a control of surface energy between the inclusion and the metal—normally through suitable chemistry modification of the metal. By analogy, the technique may not be unlike the addition of magnesium to cast iron to change the graphite shape from flakes to nodules.

available for modifying the viscosity of slag or dross, thus facilitating their ease of removal. On the other hand, treatment of steels and cast irons is likely to be different, and a few examples, along with the application of appropriate physical chemistry, are given as follows.

Steels

Inclusions in Ferrous Alloys Some applications for inclusion minimization or shape control are immediately obvious. For example, proprietary addition agents are

Oxides. The amount of carbon in iron controls the amount of dissolved oxygen (Fig. 8). The curve has been calculated from thermochemical data with variable CO/CO2 ratios. Experimental points would generally lie above

78 / Principles of Liquid Metal Processing

(a)

(b)

Fig. 7

SEM fractographs showing the effect of calcium treatment on the fracture morphology of ASTM A633C steel impact specimens. (a) Untreated steel with type II manganese sulfide inclusions showing evidence of brittle fracture. (b) Calcium-treated steel with spherical inclusions; the fracture is ductile. Courtesy of A.D. Wilson, Lukens Steel

Company

1.0

0.10

0.5 1705 ⬚C 1600 ⬚C 1515 ⬚C

0.2

0.08

2Fe

O+

0.1

Si

SiO

0.06

2 +2 Fe

0.02 FeO, %

Oxygen, wt%

0.05

0.04

0.01

3Fe

O+

0.005

2A

l Al

0.002 0.02

2O 3 +3 F

e

0.001

Equilibrium with CO + CO2 at 1 atm pressure and 1600 ⬚C

0.0005 0.0002

0

0

0.2

0.4

0.6

0.8

0.0001 0.001 0.002

0.005 0.01 0.02

0.05

0.1

0.2

0.5

1.0

2.0

5.0

10

20

50

Carbon, wt% Alloy remaining in steel, %

Fig. 8

Carbon-oxygen equilibrium at 1600  C (2910  F). Source: Ref 10

the theoretical curve, but the relationship between carbon and oxygen would be the same. Low carbon in solution, as in many cast steels, results in higher dissolved oxygen. The solid solubility of oxygen in iron is low, and during solidification, oxygen is rejected to the remaining liquid, where it can react with carbon to form CO and result in gas porosity. Although the porosity could occur in rimmed steel ingots (induced porosity that is “healed” during subsequent processing), the amount of dissolved oxygen in steel castings is reduced by additions of aluminum or silicon. The resultant oxides are insoluble and form inclusions.

Fig. 9

Equilibrium between FeO and dissolved aluminum and silicon. Source: Ref 11

Because Al2O3 has a lower free energy of formation than SiO2, aluminum is a more effective deoxidizer than silicon (Fig. 9). The FeO content is an index of the amount of dissolved oxygen in the steel where the oxygen can also be reduced by carbon, as shown in Fig. 8. At a given carbon content, the oxygen can then be reduced by aluminum or silicon to still lower levels, because the free energies for SiO2 and Al2O3 are lower than those for CO or CO2 at the reaction temperature. Once the oxide forms, the inclusion removal rate depends on the nucleation of particles, their growth, ultimate agglomeration, and rise to the surface. Although there is a significant

difference in density between the solid particles and the liquid metal, a simple application of Stoke’s law is difficult. The problem is dynamic and is influenced by the presence of other indigenous or exogenous inclusion species. Furthermore, gravitational effects are different while holding the treated metal in the ladle, pouring through air with an opportunity for the metal to reoxidize, and during solidification with the associated thermal convective currents. Therefore, it should not be surprising that oxide inclusions can become trapped in the final casting. Sulfides. Sulfur is also often present in steels and is controlled by the nature of the

Inclusion-Forming Reactions / 79 50 Reduction in area 40

54.2 Izod

40.7

30 Elongation 20

27.1 0.06% S

10 60 Reduction in area

50

67.8

Izod

40

54.2

Elongation 40.7

30 0.04% S

20 60 Reduction in area 50

67.8 Izod 54.2

40 Elongation

40.7

30

(b)

Izod impact strength, J

Izod impact strength, ft · lb Reduction of area and elongation, %

(a)

0.03% S

20 70 Izod

81.3

60 Reduction in area

67.8

50 40

54.2 Elongation

30 20

40.7 0.02% S 0

0.04

0.08

0.12

0.16

0.20

0.24

0.28

Aluminum added to heat, %

(c)

Fig. 10

Illustrations of common inclusion shapes and distributions found in steel. (a) Type I: globular silicates and oxides (no aluminum deoxidizer). (b) Type II: continuous eutectic sulfides and Al2O3 at grain boundaries (low aluminum addition). (c) Type III: duplex irregularly shaped sulfides and Al2O3 (large aluminum additions). Source: Ref 12, 13

slag-refining processes in the furnace. In general, as the process becomes more reducing (that is, the oxygen content of the metal is lower) and the slag more basic, lower sulfur will be present. Because the sulfur content is not reduced to zero, manganese is always present in steels to produce the more desirable manganese sulfide (MnS) inclusions rather than FeS inclusions with lower melting point. The resulting inclusions, although usually identified as manganese sulfides, may have several distributions and shapes, depending on the presence of other elements and the deoxidation procedure. Complex Inclusions. The various elements compete to form compounds with oxygen and sulfur. Furthermore, alloying elements in excess of those necessary to form compounds will modify the surface energy of the resultant inclusions.

Fig. 11

Combined effect of aluminum and sulfur on the ductility and impact strength of cast and normalized medium-carbon steel. Source: Ref 11–13

The shape and distribution of the inclusions will then also vary, depending on the variables discussed at the beginning of this article. Historically, inclusions have been classified into three general types, as shown in Fig. 10. Of these, type II has been demonstrated as particularly damaging to mechanical properties. The argument has been that sulfur, oxygen, and deoxidizer play a combined role in the inclusion type. Figure 11 shows this effect and the resulting decrease in structure-sensitive properties such as ductility and impact strength. The minimum in the curves is attributed to the formation of type II inclusions. The role of aluminum in the modification of inclusion shape and distribution also depends on the oxygen-sulfur equilibrium shift when the aluminum is added. Therefore, complex inclusions such as oxy-sulfides may be present in a cast steel but can be controlled by the quantity of oxide-sulfide-forming elements. A more complete discussion of oxide inclusion formation mechanisms and the relationship to mechanical properties is given in Ref 14.

Inclusion shape control becomes a consideration especially if fracture toughness, impact strength, or fatigue resistance must be optimized. Magnesium and calcium additions can be successfully used for shape control, that is, to produce more spherical shapes (Ref 15). It is noteworthy that both magnesium and calcium form stable sulfides and oxides as indexed by the free energy of formation of the compounds. Several production methods have been employed for calcium or magnesium additions, either as the metallic elements or as compounds. Because of the high vapor pressure of the pure metals and, to a lesser extent, the compounds at the addition temperatures, special procedures have evolved, including injection at the bottom of a filled ladle. In this case, the ferrostatic head provides a pressure that suppresses the boiling action and gives a longer reaction dwell time. The most common practice is to deoxidize the steel with aluminum followed by calcium for inclusion shape control. Solid Al2O3 particles are formed upon deoxidation, occasionally

80 / Principles of Liquid Metal Processing with complex aluminates such as spinels (Al2MgO4). These solids prefer their own kind and can therefore agglomerate into larger, more harmful inclusions. The addition of calcium lowers the melting range to form liquid calcium aluminates (40 to 60% CaO) that do not agglomerate but appear as small and more desirable spherical oxide inclusions. Rare earth elements added in conjunction with calcium, magnesium, and aluminum have also been reported to enhance the spheroidization and fineness of the inclusions by breaking up the Al2O3 galaxies (Ref 16, 17). However, difficulty in removal of the rare earth oxide inclusions, especially in scrap remelt, has limited the practice in present-day steel casting processes. Nitrides. Just as there are elements that preferentially form oxides or sulfides, there are differences in the free energy of formation of the nitrides. These elements control the amount of nitrogen that can be dissolved in both the solid 0.020 1260 ⬚C

Nitrogen, %

0.015 980 ⬚C

1150 ⬚C

0.010 1065 ⬚C 900 ⬚C

0.005

0 0

0.02

0.04

815 ⬚C

0.08 0.06 Aluminum, %

0.10

Fig. 12

0.12

Isothermal precipitation of aluminum nitride. Compositions to the right or above each isotherm cause precipitation. Source: Ref 12

and liquid states. For example, vanadium and manganese increase nitrogen solubility, while carbon and phosphorus decrease nitrogen solubility (Ref 12). The effect of aluminum on nitrogen solubility has been extensively studied, and criteria that result in the precipitation of aluminum nitrides have been developed (Fig. 12). Other nitride-forming elements, such as titanium, zirconium, and vanadium, would show similar but numerically different results. It is not uncommon to consider these nitride inclusions to be beneficial; they peg austenite grain boundaries, and this results in grain refinement. On the other hand, brittleness results when the nitrides become more or less continuous. Exogenous Inclusions. Numerous examples of inclusions occur because of the reaction between the molten steel and the adjacent environment. A few of these are considered as follows. When a steel is killed with aluminum, residual aluminum remains in solution. If the metal contacts a refractory that contains free SiO2, the aluminum can reduce the SiO2 because of the lower free energy of formation of the Al2O3. The eroded or partially dissolved refractory becomes a source of inclusions, as does the Al2O3 that is produced. Careful control of refractory chemistry and good pouring practice minimize this problem. Similar reactions can occur at the mold/metal interface, where the first equilibrium of importance is iron-oxygen-silicon. When conditions are oxidizing, FeO is present, and the corresponding form for silicon is SiO2. The FeOSiO2 phase diagram is shown in Fig. 13. The most significant observation here is the formation of

1800

3270 2 liquids ~1700⬚

1600

2910 SiO2 (Cristobalite) + L

1470⬚ 1400

2550

1371⬚ SiO2 (Tridymite) + L

2FeO·SiO2 + L FeO + L

1208⬚

1200

2190 1180⬚

1175⬚

FeO + 2FeO•SiO2 1000 FeO

20

2FeO·SiO2 (Fayalite) + Tridymite 40

60 SiO2, wt%

Fig. 13

FeO-SiO2 phase diagram. Source: Ref 10

80

1830 SiO2

Temperature, ⬚F

Temperature, ⬚C

Liquid

lower-melting liquid phases as the amount of FeO is increased. With minimal oxidation, burned-on sand and metal penetration will occur. However, when the oxidation level increases, local macroinclusions may form, especially if poor pouring practice, segregated bentonite, nonuniform mold ramming, and excessive mold moisture exist. A number of the problems and solutions associated with exogenous inclusions in steel have been summarized in Ref 18.

Cast Irons Gray Cast Iron. The earlier discussion of the carbon-oxygen equilibrium (Fig. 8) suggests that dissolved oxygen should not be a problem in cast irons because of their high carbon contents. Therefore, oxide inclusions should not occur in gray cast iron. However, because the silicon content of this material is also significant, the carbon-silicon-oxygen equilibrium is of interest. The temperature dependencies of the free energies of formation for SiO2 and CO are different for the two species. At temperatures above approximately 1535  C (2795  F), CO is the more stable compound, while below this temperature, SiO2 is more stable. The exact temperature depends on the carbon and silicon content of the metal. The phenomenon can be observed on the surface of a cast iron melt that appears clean at high temperatures and yet develops a slag covering at lower temperatures. The significance of the equilibrium shift is that skimming the slag from a melt surface may be unnecessary at high temperatures, because little may exist. Inclusions are then found in the casting because of pouring at lower temperatures and exposure of the metal to reoxidation during pouring. These inclusions are often silicates derived from the normal equilibrium change with temperature. Once the molten iron is in the mold cavity, there is less opportunity to form the silicates because conditions are usually less oxidizing. Nitride and sulfide inclusions also exist in gray cast iron but are not as severe a problem as in steel because of the inherently lower dissolved oxygen content of the cast iron. Special gray cast iron chemistries containing alloying elements with an affinity for nitrogen, oxygen, or sulfur may result in inclusions by further reaction with the refractories, mold, or atmosphere. The remedies are usually similar to those given for steel. Ductile Cast Iron. A number of foundries erroneously assumed that ductile (nodular) cast iron would be as easy to process as gray cast iron because the major difference was merely a lower sulfur plus a modest magnesium addition. Although many characteristics were found to be different, one of the more perplexing was the formation of dross usually not found in gray cast iron. The result was the so-called cope-side defect that was very prevalent in early ductile iron castings.

Inclusion-Forming Reactions / 81 Magnesium forms a very stable oxide and can therefore reduce oxides of lower stability, as determined by their free energy of formation. Furthermore, the alloys used to produce the nodular graphite found in ductile iron contain considerable amounts of silicon and are low in density, which allows them to float on ladle surfaces rather than reacting completely. These observations help to explain how dross particles form and can become entrapped in castings. In general, there are three types of dross inclusions in ductile cast iron (Ref 19). Type I consists of large dross particles generally derived from oxidation of the silicon and magnesium used in the nodularizing alloys. The dross particles are present on the treated liquid-metal surface and can be removed by skimming. Silica does not exist in the pure form; rather, magnesia (MgO) and forsterite (Mg2SiO4) are the predominant species. This suggests a driving force from the high-MgO end of the phase diagram shown in Fig. 14. Under some circumstances, especially after a late silicon inoculation, the dross also contains fayalite (2FeOSiO2 or Fe2SiO4) (Fig. 13). Fayalite and forsterite form a complete series of solid solutions known as the olivines and are used as molding materials. Type II inclusions are in a more or less stringlike configuration in conjunction with decayed graphite shapes. Chemically, they are predominantly MgO with some iron and silicon atom replacement for the magnesium in the MgO lattice. Reaction with the oxidizing components in the mold material is one source for this type of inclusion, as is an excess of nodularizing agent. Their position is usually just beneath the casting surface. Control of the mold

atmosphere and the amount of nodularizer usually minimizes the defect. Type III inclusions are fine dispersions of MgO and MgS, occasionally mixed with Mg2Si. These components would normally be expected only if the oxygen content in the immediate vicinity is very low. Their normal uniform distribution does not usually detract from the overall performance of the casting. However, undissolved nodularizer may be a source of type III inclusions, because Mg2Si is the active magnesium source in the nodularizing agent. Finally, the examples provided in this section are not intended to represent all of the inclusions to be anticipated in every ferrous alloy. On the other hand, they are sufficiently representative so that analogies for occurrence and remedy can be derived for the alloys not discussed.

Inclusions in Nonferrous Alloys Several important cases are given subsequently for aluminum, copper, and magnesium alloys. The principles discussed in this section can be extended to include other alloy systems. In the alloys discussed, both nonmetallic and intermetallic compounds are considered deleterious.

Aluminum Alloys Oxides. It is nearly impossible to prevent dross formation in the melting of aluminum alloys. The most common form of dross contains large quantities of Al2O3 and, when agglomerated, can be removed from the melt surface by skimming. Pouring operations should employ 3450

1900 1890⬚

MgO + L 1850⬚

Liquid 3270

1800

skimming ladles to prevent larger dross particles from entering the mold. Furthermore, a gating practice must be used that minimizes Al2O3 formation during mold filling. The fact that Al2O3 forms is not surprising, because the free energy of formation of Al2O3 is very low. Furthermore, the specific gravity of Al2O3 is not greatly different from that of the molten aluminum alloys. Although more dense, Al2O3 forms at an outside surface first if the melt is exposed to the atmosphere. The tenacious skin will float unless it is broken up into small particles, and it can become quite thick when melts are held for extended periods in the furnace. When small oxide particles are formed, they may not sink but may become suspended in the melt. In high-quality castings, these particles must be removed; ceramic filters have been used to remove the particles, as mentioned earlier in this article (Ref 3). The degradation of mechanical properties also depends on the size and distribution of oxide inclusions. Therefore, solidification rate is very important for minimizing property loss. When the casting solidifies rapidly, as in thin sections or in permanent molds, the oxide inclusions are smaller and more uniformly dispersed. This observation is consistent with the general constraints of nucleation and growth, where less time during solidification not only creates smaller particles but also prevents their agglomeration and growth. Complex Oxides. Other elements in the aluminum alloys may react with oxygen sources and produce complex oxides, with Al2O3 as one of the components. This is the case in magnesium-containing alloys, in which the spinel MgAl2O4 may form (Fig. 15). Here, MgO has a lower free energy of formation than Al2O3; the spinel results when they form together. Again, precautions in pouring, gating, and minimizing the reoxidation of the metal are important in inclusion control. The use of filtering systems is also advantageous, especially in high-strength castings in which fracture toughness characteristics are important.

1695⬚

3090

SiO2 + L

1600

2910 1557⬚

20

40

Fig. 14

MgO-SiO2 phase diagram. Source: Ref 10

Liquid + MgAl2O4

2600

2200

MgO + liquid

MgAl2O4 1800

1400

MgO·SiO2 + SiO2

4710

Liquid + spinel

Liquid + Al2O3

Spinel

MgO + MgAl2O4

3990

3270

Al2O3 + spinel

2550

(Enstatite)

60 SiO2, wt%

Liquid

1543⬚

2MgO·SiO2 + MgO·SiO2 1500 MgO

5430

3000

80

2730 SiO2

1000 0 MgO

Fig. 15

20

60 40 Al2O3, mol%

80

1830 100 Al2O3

MgO-Al2O3 phase diagram. Source: Ref 2

Temperature, ⬚F

1700

Temperature, ⬚C

2 liquids

Temperature, ⬚F

(Forsterite)

2MgO·SiO2 + L

Temperature, ⬚C

MgO + 2MgO·SiO2

2030

1100

2012

Temperature, ⬚C

1110

1090

L

1976

1080 1070 1060

1050 0.001

Fig. 16

1994

1958 α

α + Cu2O 0.01 0.1 Oxygen, wt%

Temperature, ⬚F

82 / Principles of Liquid Metal Processing

1940 1922 (0.7)

A portion of the copper-oxygen equilibrium diagram. Source: Ref 20

Other Inclusion Types. Inclusions in aluminum alloys can occur from grain-refining operations in which the grain-refining additions may not be adequately controlled. Titanium, vanadium, and boron are used in various combinations to produce grain refinement, and the diborides can become undesirable inclusions. Again, the fact that a small addition is beneficial does not mean that larger additions will provide greater rewards. Many of the commercial aluminum alloy compositions have been developed to exploit their age-hardening characteristics in conjunction with an alloyed matrix. They then become susceptible to the presence of seemingly minor elements such that undesirable intermetallic particles can be formed during cooling. In those cases of precipitation during solidification, the particles are placed in the category of inclusions. The intermetallic inclusions are particularly troublesome in aluminum alloys containing large quantities of zinc or magnesium. Complexes can be formed with the transition elements. These observations suggest that the remedy is close control of the metal chemistry. There are many examples in which control is inadequate and the problem is not identified until mechanical properties are marginally out of specification. Even aluminum carbide and iron-aluminum compounds have degraded properties by the formation of inclusions. Although the solubilities are low, the problem is additive in that scrap/revert procedures can cause a gradual increase in the impurity levels, and inclusion problems become more severe. The hazards associated with recycling then become apparent, especially when the final product is sensitive to the elements present and when strict mechanical property specifications are imposed. The development of new alloys, such as the aluminum-lithium alloys, brings its own set of metal-processing difficulties that begin with making ingots for forming. The question of casting more complex shapes then becomes a natural evolution. Here, the reaction of lithium with air requires melting under liquid salts that themselves are aggressive with the surroundings, such as refractories or mold materials. The opportunity for trapped inclusions is apparent.

Copper Alloys

Magnesium Alloys

A portion of the copper-oxygen system is shown in Fig. 16. The eutectic shows that the last liquid to solidify may contain 0.39% O, even though the initial melt may contain 0.01% O. Therefore, Cu2O inclusions may be more predominant internal to the casting (liquid enrichment in oxygen due to slow diffusion) and may also be present at grain boundaries. The key to control is to add elements that combine with dissolved oxygen (that is, to form oxides with free energies lower than that of Cu2O). Examples are lithium, boron, magnesium, or phosphorus, where control of the excess addition becomes important to modification of the electrical conductivity in the high-copper alloys. The excess deoxidizer goes into solid solution. A more thorough discussion of gas solubility is available in Ref 20 (see also the article “Gases in Metals” in this Volume). A number of the copper alloys are susceptible to dross formation, and the normal precautions in pouring and gating are absolutely necessary to minimize exogenous inclusions. The drosses are usually complex oxides of copper, zinc, tin, lead, or aluminum in the aluminum bronze family. Iron is added as both a dispersion strengthener and as a grain refiner in aluminum and manganese bronzes. However, iron can become an undesirable inclusion in the other copper alloys. When a foundry pours several different alloy chemistries, scrap segregation becomes important to inclusion control. Consideration can also be given to alternative grain-refining agents such as zirconium. In some cases, the intermetallic inclusions are rich only in iron, and complexes are formed. This is especially true when alloys are high in zinc, such as the high-strength yellow brasses. The problem becomes more severe when a melt is held close to its liquidus for extended periods. In this case, the solubility of the complex intermetallics is lower, and their formation is enhanced. High-zinc alloys offer fewer problems with dissolved oxygen because of the vaporization of zinc and the purging action of the melt. Oxides as inclusions are then a lesser problem. However, zinc oxide (ZnO) can react with refractories and mold materials to produce inclusions during pouring. In addition to careful pouring and gating practices, exogenous inclusions can be removed by filtering processes. However, the practice has not been as well received for copper alloys as for aluminum alloys. Although not strictly an inclusion from the previous definition, lead in many of the copper alloys can form oxides that will react with mold bonding agents such as bentonite. These lead complexes can become leachable in standard tests, and waste sand streams that contain lead compounds may be classified as toxic, with the attendant difficulty in disposal (Ref 21).

Of the nonferrous alloys considered in this section, magnesium has the lowest free energy of formation for its oxide. Therefore, the complete prevention of oxide inclusions is impossible, and a number of methods have been developed to minimize oxide inclusion formation. Agents that evolve gas to partially insulate the melt from the atmosphere and fluxes that agglomerate the oxide (plus other impurities that are to be removed) are two examples of control of the melting practice. Again, great care must be exercised in gating to minimize reoxidation. The magnesium alloys are light in weight, and many of the potential exogenous inclusions are collected as sludge in the bottom of the furnace and ladle. Desludging operations are then necessary. The use of filtering systems, whether metallic or ceramic, is common to all of the higher-quality magnesium castings. As with aluminum alloys, excess grain-refining agents can result in undesirable inclusions that can have complex chemistries because of secondary reactions. The fluxes themselves must be reactive in order to perform their intended function. Excess flux can then react with the atmosphere, refractories, or the mold materials to produce inclusions. On the other hand, insufficient flux causes metal reaction with the atmosphere to produce oxides and nitrides that may be difficult to remove if not in an agglomerated form. Temperature, time at temperature, alloy composition, pouring practice, and gating procedures dictate the optimal fluxing process. So-called fluxless melting has been suggested for magnesium alloys in order to overcome the environmental and metallurgical problems associated with fluxes (Ref 22). Fluxless melting uses nonreactive gaseous atmospheres above the melt. The most effective protective atmosphere is CO2 plus the addition of trace inhibitors such as SO2 and SF6. On the other hand, the potential addition of fluorine-containing gases to a gas waste stream must also be of concern. Finally, the general observations and processing techniques in this article have assumed the presence of air to generate the bulk of the inclusions. Alternative melt processing can exclude air for the primary purpose of inclusion prevention. Both vacuum and inert gases are used to process alloys for more demanding applications such as metal bearings, superalloys, and metal alloy prosthetics. Further discussion is included under specific alloy systems and melt methods elsewhere in this Volume.

ACKNOWLEDGMENT From: P.K. Trojan, Inclusion-Forming Reactions, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 88–97

Inclusion-Forming Reactions / 83 REFERENCES 1. L. Ackermann, J. Charbonnier, G. Desplanches, and H. Kaslowski, Properties of Reinforced Foundry Alloys, Trans. AFS, Vol 94, 1986, p 285–290 2. R.A. Flinn and P.K. Trojan, Engineering Materials and Their Applications, 3rd ed., Houghton-Mifflin, 1986 3. F.R. Mollard and N. Davidson, Experience with Ceramic Foam Filtration of Aluminum Castings, Trans. AFS, Vol 88, 1980, p 595–600 4. L.A. Aubrey, J.W. Brockmeyer, and P.F. Wieser, Dross Removal from Ductile Iron with Ceramic Foam Filters, Trans. AFS, Vol 93, 1985, p 171–176 5. L.A. Aubrey, J.W. Brockmeyer, P.F. Wieser, I. Dutta, and A. Ilhan, Cast Steel Improvement by Filtration with Ceramic Foam Filters, Trans. AFS, Vol 93, 1985, p 177–182 6. P.K. Trojan, P.J. Guichelaar, and R.A. Flinn, An Investigation of Entrapment of Dross and Inclusions Using Transparent Whirl Gate Models, Trans. AFS, Vol 74, 1966, p 462–469

7. D.G. Schmidt, Gating of Copper Base Alloys, Trans. AFS, Vol 88, 1980, p 805–816 8. A.D. Wilson, Calcium Treatment of Plate Steels and Its Effect on Fatigue and Toughness Properties, Proc. 11th Annual Offshore Technology Conference, OTC 3465, May 1979, p 939–948 9. A.D. Wilson, Effect of Calcium Treatment on Inclusions in Constructional Steels, Met. Prog., Vol 121 (No. 5), April 1982, p 41–46 10. The Making, Shaping and Treating of Steel, 9th ed., United States Steel Corporation, 1971 11. C.W. Briggs, The Metallurgy of Steel Castings, McGraw-Hill, 1946 12. R.A. Flinn, Fundamentals of Metal Casting, Addison-Wesley, 1963 13. C.E. Sims and F.B. Dahle, The Effect of Aluminum on the Properties of Medium Carbon Cast Steel, Trans. AFS, Vol 46, 1938, p 65–132 14. W.O. Philbrook, M.C. Flemings, L.H. Van Vlack, A. Gittins, and J. Lankford, Oxide Inclusions in Steel, A series of five papers, Int. Met. Rev., Sept 1977, p 187–228 15. Low Sulfur Steel, Symposium proceedings, AMAX Inc., 1986

16. S.K. Paul, A.K. Chakrabarty, and S. Basu, Effect of Rare Earth Additions on the Inclusions and Properties of Ca-Al Deoxidized Steel, Metall. Trans. B, Vol 13 (No. 2), June 1982, p 185–192 17. M. Olette and G. Gatellier, Effect of Calcium, Magnesium or Rare Earth Additions on Cleanness of Steel, Rev. Metall. Cah. Inf. Tech., Vol 78 (No. 12), Dec 1981, p 961–973 18. R.A. Flinn, L.H. Van Vlack, and G.A. Colligan, Mold-Metal Reactions in Ferrous and Nonferrous Alloys, Trans. AFS, Vol 94, 1986, p 29–46 19. D.R. Askeland and P.K. Trojan, The Approach to Equilibrium and Dross Formation in Nodular Cast Iron, Trans. AFS, Vol 77, 1969, p 344–352 20. Casting Copper Base Alloys, American Foundrymen’s Society, 1984 21. T.R. Ostrom, B.P. Winter, and P.K. Trojan, Lead Transfer from Copper Base Alloys into Molding Sand, Trans. AFS, Vol 93, 1985, p 757–762 22. J.W. Fruehling and J.D. Hanawalt, Protective Atmospheres for Melting Magnesium Alloys, Trans. AFS, Vol 77, 1969, p 159–164

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 87-98 DOI: 10.1361/asmhba0005195

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Electric Arc Furnace Melting*

THE MELTING OF STEEL is performed in both arc furnaces and induction furnaces. The arc furnace is the primary melting method and is used by over 80% of steel foundries (Ref 1). However, arc furnaces are not good holders or efficient at superheating metal, so it is not uncommon to use induction furnaces to superheat steel after the arc melters. Induction furnaces used to melt steel require cleaner charge materials to meet metallurgical requirements and therefore require low-carbon charge to produce lower slag levels. The basic concept of an electric arc melting furnace is a refractory-lined steel vessel having a bowl-shaped hearth and a domed-shaped refractory roof. The hearth and sidewalls are typically lined with bricks, monolithic materials, or a combination of both. Acid refractories (silica-based), basic refractories (magnesitebased), and, in some cases, neutral refractories (alumina-based) are used to line the furnace sidewalls, hearth, and roof. All major additions are made to the furnace through the roof by swinging the roof out of the way and charging scrap, fluxes, and carbon from loaded buckets. The furnace has two or three electrodes mounted vertically through the roof of the furnace that float just above the surface of the cold ferrous scrap or molten bath. Over 80% of arc furnaces used for steel castings are alternating current furnaces with three electrodes. These electrodes strike an arc on the cold metal scrap or molten metal to impart energy to the bath and melt or superheat the furnace. Control of the electrode level above the bath is critical to the melting process efficiency. Arc melting furnaces are a widely used method for melting steel scrap. They can melt a wide variety of ferrous scrap materials and can accept certain levels of dirty scrap. They also use chemical energy to assist in the melting process and to lower the carbon levels of the molten metal. Since steel castings require low-carbon-level metallurgy, carbon from scrap metal feedstock and carbon electrodes must be removed from the molten metal. Injecting oxygen into the molten bath oxidizes carbon and uses the exothermic reaction to assist in melting and superheating the metal. This is referred to

as supplying chemical energy to the furnace. In some instances, different forms of oxyfuels are used to provide additional energy to speed up the melting process. Although refractories line the furnace shell and roof, significant energy losses occur through the surfaces of the furnace and from exposed surfaces during charging and slagging operations. Energy losses from the surface of the bath are due to the large surface-area-to-volume ratio of the furnace, resulting in significant energy losses during delays in the operation. In addition, the large surface area of the furnace results in large losses through the refractory walls, hearth, and roof. Nearly all steel foundries blow oxygen into the bath after most of the scrap is melted, which generates chemical energy from exothermic decarburization, assisting in the melting process. Some operations use additional sources of chemical energy to decrease tap-to-tap time. Electric arc furnaces are used as melters and holders in duplex operations and as melting and refining units. Since its first appearance as a production tool at the beginning of the 20th century, the electric arc furnace(EAF) has become one of the primary melting tools used by foundries and steel mills. The early furnaces had capacities of 910 to 14,000 kg (1 to 15 tons), but new technology has vastly increased EAF productivity. There are currently approximately 100 minimills in the United States capable of producing 50 million or more tons per year, with numerous furnaces capable of producing in excess of 100 tons per hour (Ref 2). This article focuses on the construction and operation of these furnaces and their auxiliary equipment in the steel foundry (shape casting) industry. Primary steelmaking practices differ from that of smaller foundry furnaces, and most primary steelmaking EAF operations have adopted separate practices since the 1980s. For example, methods of improving the efficiency of arc melting furnaces in the primary steel industry are not directly transferable to the smaller furnaces used by steel shape casting producers. Oxygenenriched burners and foaming slag methods are not considered effective on the smaller furnaces typical of steel foundry operations. For primary

melting, the reader is referred to The Making, Shaping, and Treating of Steel, 11th ed., published by The AIST Foundation. More information on ladle metallurgy is also contained in the article “Steel Melt Processing” in this Volume.

Power Supply The current applied to the EAF is supplied by a local electric utility company and passes through an electrical substation that is designed specifically for the furnace(s) to be supplied. A simplified diagram is shown in Fig. 1. The power supplied to the furnace during melting is provided by an electrical arc

Fig. 1

Schematic of electrical network for the electric arc furnace

*Revised from Melting Furnaces: Electric Arc Furnaces, Casting, by Nick Wukovich, in Vol 15, ASM Handbook, ASM International, 1988, p 356–368 from Reviews by Peter Glaws and Craig Darragh, The Timken Co.

88 / Melting and Remelting established from three carbon or graphite electrodes. During the meltdown portion of the heat, the three arcs or phases act as a single phase. Two arcs can strike the furnace charge and draw current without a current going to the third arc or electrode. Because of the type of scrap used and the arc lengthening and shortening that take place, there is a great fluctuation in the current during meltdown, which causes a significant variation in the electrical supply system. If the same power source is the supply for other plant power, a flickering of lights and voltage fluctuation will be noticed on machinery and electrical equipment. During the refining cycle or after meltdown, the arcs tend to stabilize, partially because of a slag cover on the liquid metal and the flatness of the bath. In addition, the arcs have been shortened to direct their energy in a smaller area. A great amount of energy is produced during this meltdown and refining of ferrous metals. Controls for harnessing and directing the energy of the arcs are required in order to produce molten iron and steel in the electric furnace without destroying the furnace refractories. The energy requirements for melting various carbon levels in iron or steel are shown in Fig. 2. Power Factor. The efficiency with which power is transferred to the melt is called the power factor. The power factor, PF, is the ratio of watts, W, divided by volt-amperes, VA: PF ¼




W VA

have been chosen. The electrical system for the furnace includes the transformer and the control panel. Very small furnaces (<910 kg, or 1 ton) are not used in current production schedules; heats of these sizes are normally made in induction furnaces. Furnaces are described as supplying so many tons or pounds per heat under normal conditions, but larger heats can be made by modifying

the furnace. This is done by stopping off the tap hole and raising the breast on the working and charge door (if the furnace has one). The furnace is lined with refractories that determine the type of melting practice (acid or basic) that will be used (Fig. 3). Table 1 lists various refractory furnace materials used in acid and basic melting practices. Depending

 100

One method of measuring the average power factor over a period of time requires the use of two separate meters: a watt-hour meter to accumulate the useful power and a var-hour meter to accumulate the reactive power. The var-hours (in a given time) divided by the watt-hours (in the same time) will give the tangent of the power factor angle. The cosine of this angle is the power factor. The local utility company can be contacted for this information and for recommendations on the effect on energy usage.

Fig. 2

Power consumed in melting iron and steel in the electric arc furnace. Values will vary depending on scrap, transformer, lining, and so on. The melting point of pure iron (0.0% C) is 1535  C (2795  F); of iron containing 4.3% C, 1130  C (2066  F).

Arc Furnace Components The list of equipment associated with an EAF can be extensive. This list is assembled as if a proposal for the furnace has been made. The locations for the furnace foundation and pits or elevated platforms are selected. Factors that also must be considered include the location of the furnaces in the plant, flow and access to raw materials, storage of melt materials, ladle construction, ladle preheating, crane runways, water, air, electrical lines, transformers, laboratory, and temperature equipment. This list is not all-inclusive. Fume and dust collection equipment can be added, as well as used-refractory and slag removal equipment. For the melting portion, a supply of oxygen, hoses, pipe for oxygen blowing, gages, valves, thermocouples, and so on will be needed. The furnace itself is the primary concern after the location, cement work, and electrical supply

Fig. 3

Cutaway view of electric arc furnace showing the typical refractories used

Electric Arc Furnace Melting / 89 on whether the process is basic or acid, slag color may be sufficient to determine the iron oxide content of the slag, as shown: Color

FeO, %

CaO/SiO2

Acid slag Gray-black Dark green Streaked dark green Blue-green (jade green)

30–40 20–30 15–20 <15

... ... ... ...

Basic slag Black Dark brown-gray Light brown Gray White White

14.85 11.50 1.05 1.05 0.56 0.51

3.33 2.19 1.91 1.88 2.01 2.53

The silica content of basic slag can profoundly affect the melt. If present in sufficient (quantities, silica can acid constituent) can significantly reduce the slag’s ability to effectively desulfurize the steel. The size of the furnace can vary from 450 to 91,000 kg (0.5 to 100 tons) or larger. Typical furnace data are given in Table 2.

Table 1

delta section of the roof and can cause excessive refractory wear in this area. The use of water-cooling rings around the electrode ports can help to reduce this wear. The electrodes should be vertical from all sightings around the furnace. When the electrodes are not vertical, the distance and arc flare on the sidewalls are not uniform, and a hot spot may cause the refractory wear to be greater in the areas closer to the electrodes. In addition, because the arc works between the electrodes and the bath, the current per phase can fluctuate (Fig. 5). The roof structure is arched to be selfsupporting and to keep the greatest distance between the arc action in the delta section. Refractory roofs start with a roof ring that is designed as a channel or an angle beam that is rolled or cast into a ring. The ring generally has a water-cooling chamber on all or a portion of its height (Fig. 6). Newer designs of watercooled roofs are cast in gray iron except for the delta section and the exhaust and oxygen lance holes on large roofs, and they are cooled by a network of steel pipes running throughout the structure. The delta section and electrode ports are

The electrode can be graphite or carbon. The graphite electrode has greater current capacity than the carbon unit for the same size electrode. Table 3 lists some data on electrode sizes, weights, and current capacities. The electrodes are supported on three separate arms and held to the electrode arm by a spring-clamped, pneumatically released electrode holder. On older furnaces, the electrode holder had a mechanical wedge-type clamp; on newer, larger furnaces, power-operated clamping devices are used. The electrode arms support the electrodes, holders, bus tubes, and transformer power cables. Smaller bus tubes and power cables can be used if they are air or water cooled. The electrodes are raised and lowered by manual or automatic controls. These controls can be operated manually when the bottoms of the electrodes are raised to the roof or when electrodes are shifted. The electrodes are raised and lowered automatically during the melting and refining process (Fig. 4). The furnace designer will make the electrode circle as small as possible to increase the speed of melting. This will intensify the heat in the

Refractories as a source of slag based on typical composition Composition, %

Type

Fireclay Super duty Medium duty Semi silica Alumina type (high) 50% 60% 70% 80% 90% Mullite Corundum Silica type Silica super duty Conventional Basic type Chrome Magnesite Magnesite high periclase Chrome-magnesite

Table 2

Softening point 



SiO2

Al2O3

Cr2O3

TiO2

CaO

MgO

Fe2O3

Other

C

F

49–56 57–70 72–80

40–44 25–38 18–26

... ... ...

1.5–2.5 1.3–2.1 1.0–1.5

... ... ...

... ... ...

... ... ...

2.5–4.0 4.0–7.0 1.0–3.0

1745–1765 1660–1685 1640–1685

3175–3210 3020–3065 2985–3065

41–47 31–37 20–26 11–15 7½–9 18–34 0.2–1.0

47½–52½ 57½–62½ 67½–72½ 77½–82½ 89–91 60–78 98–99.5

... ... ... ... ... ... ...

2.0–2.8 2.0–3.3 3.0–4.0 3.0–4.0 0.4–0.8 0.5–3.1 Trace

... ... ... ... ... ... ...

... ... ... ... ... ... ...

... ... ... ... ... ... ...

3.0–4.0 3.0–4.0 3.0–4.0 3.0–4.0 1.0–2.0 1.0–3.0 0.3–1.0

1785 1805–1820 1820–1850 1865 1930 1850 2000

3245 3280–3310 3310–3360 3390 3505 3360 3630

95–97 94–97

0.15–0.35 0.45–1.20

... ...

... ...

2.5–3.5 1.8–3.5

... ...

0.3–2.2 0.3–0.9

0.02–0.10 0.10–0.30

1680–1700 1635–1665

3060–3090 2975–3025

3.0–6.0 0.7–1.0 0.5–5.0 4–8

15–34 0.3–1.5 0.2–1.0 16–27

28–33 ... ... 18–28

... ... ... ...

... 1.0–3.5 0.5–1.5 0.7–1.5

14–19 85–93 92–98 27–53

11–17 0.3–7.0 0.2–1.0 ...

1.0–2.0 0.5–1.0 0.0–0.6 ...

1290–1425 1480–1675 1595–1705 1595–1675

2350–2600 2700–3050 2900–3100 2900–3050

Typical dimensions and capacities of foundry-sized electric arc furnace components Refractory

Shell dimensions (inside) Diameter m

1.5 1.8 2.20 2.4 2.67 2.74 3.05 3.35 3.81

Dimensions

Depth

Melting rate

ft

m

ft

5.0 6.0 7.25 8.0 8.75 9.0 10.0 11.0 12.5

1.14 1.4 1.7 1.8 2.01 2.06 2.3 2.51 2.84

3.75 4.5 5.5 6.0 6.58 6.75 7.5 8.25 9.33

kg/h

450–640 910–1,270 1,360–1,910 1,810–2,540 2,270–3,180 2,720–3,810 3,630–5,440 5,440–8,160 7,260–10,900

Sidewall

Liquid metal capacity

Bottom

Normal

Maximum

Electrode diameter

lb/h

mm

in.

mm

in.

kg

lb

kg

lb

mm

in.

Transformer capacity, kVA

1,000–1,400 2,000–2,800 3,000–4,200 4,000–5,600 5,000–7,000 6,000–8,400 8,000–12,000 12,000–18,000 16,000–24,000

254 254 254 254 368 368 368 368 368

10 10 10 10 14.5 14.5 14.5 14.5 14.5

305 381 432 457 483 483 508 508 533

12 15 17 18 19 19 20 20 21

1,360 2,810 4,260 6,080 9,070 9,840 14,500 22,700 29,900

3,000 6,200 9,400 13,400 20,000 21,700 32,000 50,000 66,000

1,810 3,750 5,670 8,120 12,100 13,200 19,100 29,900 39,900

4,000 8,260 12,500 17,900 26,700 29,000 42,000 66,000 88,000

75 152 178 203 229 229 254 305 356

3 6 7 8 9 9 10 12 14

1,000–1,500 1,500–2,000 2,000–3,000 3,000–5,000 4,000–6,000 4,000–6,000 5,000–9,000 7,500–12,500 10,000–15,000

90 / Melting and Remelting

Table 3 Electrode sizes, weights, and current capacities Diameter  length mm

Approximate weight of electrode and nipple in.

Graphite electrodes 102  1015 4  40 152  1220 6  48 152  1520 6  60 178  1220 7  48 178  1520 7  60 203  1220 8  48 203  1520 8  60 229  1520 9  60 254  1220 10  48 254  1520 10  60 305  1520 12  60 305  1830 12  72 356  1520 14  60 356  1830 14  72 Carbon electrodes 203  1520 254  1520 305  1520 356  1520 356  1830

8  60 10  60 12  60 14  60 14  72

Approximate current capacity, A

kg

lb

1,800–3,300 3,500–5,800 3,500–5,800 4,400–7,500 4,400–7,500 5,500–9,300 5,500–9,300 6,700–11,300 8,000–13,300 8,000–13,300 11,300–17,000 11,300–17,000 18,000–25,000 18,000–25,000

13.6 37.6 47.1 46.3 59.9 62.1 77.1 96.6 97.1 118 171 201 228 273

30 83 104 102 132 137 170 213 214 261 378 444 502 602

2,500–4,500 3,100–6,300 4,500–7,900 5,400–10,000 5,400–10,000

79.8 125 178 240 293

176 275 392 528 647

usually constructed of a high-alumina refractory brick, rammed material, or precast shapes. The roofs of all modern electric furnaces are designed for top charging. The roof and electrodes are raised several inches above the top of the furnace shell and swung to one side. The charge is dropped into the furnace, and the roof is swung back into place. The raising of the roof is done by electrohydraulic systems. Schematics of roof-lifting devices are shown in Fig. 7. Tilt Control. The electric furnace is tapped out by tilting the furnace into a position in which the metal and slag will run out of a hole above the slag line in the furnace wall down into a spout and into a ladle. Tapping of the furnace requires that the roof and electrodes remain attached to the furnace shell; therefore, safety locks or stops are found welded to the shell and roof ring. The tilting mechanism can be mechanical for very small furnaces or can be motor driven (either by a screw pitman or a rack-and-pinion mechanism) or hydraulic, and it has rockers attached to the furnace and rails attached to the foundation. Emergency equipment is usually provided to pour the heat from the furnace in case of a power failure. The furnace shell is typically made of carbon steel sections that are rolled and welded to the diameters listed in Table 2. The diameter and height define the furnace size. The furnace bottom is formed and welded to complete the pot. The shell will have door openings cut into it and a cooling frame welded around the door. The door-lifting supports will also be welded onto the shell. The furnace rockers and other lifting equipment attachments are welded onto the shell. With the rockers and door aligned, the spout opening is cut out of the shell and the spout frame attachment welded onto

the shell. To ensure that the furnace is properly grounded, straps, rods, or mesh are attached to the bottom on the inside and ground on the outside prior to lining of the furnace. The choice of refractories will depend on the melting practice selected (Fig. 3 and Table 1). Furnaces built in the 1980s and beyond have used water-cooled panels in the furnace sidewalls above the slag line (Fig. 8). Spout and Tap Hole. The length and shape of the furnace spout have historically created design issues. The melter and metallurgist want the shortest possible spout, while the design engineer is confronted with a network of electrodes, crane cables, pit dimensions, and ladle and tap hole dimensions needed for positioning the ladle in order to obtain a fast and smooth tap from the furnace, using an acceptable spout. New designs and changing the locations of tap holes may reduce the physical and mechanical problems encountered with the spout. One of these alternatives is the side slide gate tap hole (Fig. 9), and another is eccentric bottom tapping with the use of a slide gate (Fig. 10). These alternative devices are primarily used to eliminate furnace slag carryover. In order to do so, they must use a hot heel practice. Eccentric bottom tapping does a very good job of preventing slag carryover at the beginning of pouring. Both do a good job when an effective hot heel practice is employed. Tap-Out and Back-Slagging Pits. The design of the furnace pit should include sufficient depth for the slag boxes, and along with this depth, provision must be made for preventing ground water from entering the pit. Sump pumps are needed to minimize the presence of water and to ensure an environment that is safe from steam explosions. The length and width of the pit should be sufficient to allow movement

of the ladle; this is necessary if the furnace slag is to be kept out of the ladle so that ladle metallurgy or secondary refining can be done following tap out. Water-Cooling System. Water-cooled sidewall panels have become common practice. Even before this innovation, water cooling was used on:  Electrode holders, arms, and clamps  Roof rings and electrode rings in the delta

section

 Around the door  Cooling panels positioned on the outside of

the furnace to cool the furnace wall refractories Preheat and Furnace Scrap Burners. Preheating scrap with natural gas or flue gases helps reduce the electrical cost of melting a charge. On the electric furnace, this preheating may also help stagger furnace operation to keep electrical demand peaks to a minimum. Furnace oxygen/natural gas burners can help melt the charge and reduce electrical charges and electrode usage while shortening the melt time from tap to tap. Improper use of these burners can cause excessive oxidation of the heat and the electrodes and can reduce the life of the refractory lining. Fume and dust collection at the electric furnace has been a required practice in melting during the last 25 to 30 years. The complexity of the fume and dust collection equipment varies from a simple fourth-hole connector to an entire separate room for the melting furnace in which the fumes and dust are collected and where the furnace is tapped into a ladle that is brought to the furnace and then removed through a dust protection door. Figure 11 shows typical dust and fume collectors. The charge bucket is sized to fit within the refractory walls of the furnace and is open at the top, with a lifting bale that folds back or is removable from two trunions. The bottom of the charge bucket can be designed to drop the charge in one of several ways:  The clam shell bottom is actuated with a trip

lever.

 The orange-peel bottom, which is set in a

stand to fold the leaves back and is tied with a rope, burns off when the charge bucket is placed into a hot furnace.  The preheat charge bucket uses a trip lever and, after being charged with scrap, is preheated to some temperature above 260  C (500  F) to up to 815  C (1500  F). Figure 12 shows a typical charge loaded in two different types of charge buckets. Ladles may be part of the equipment assigned to the electric furnace, or they may be assigned to the pouring crew or the mold department. Nevertheless, the heat is not finished until the metal is removed from the furnace.

Electric Arc Furnace Melting / 91 slag will attack the refractory, as in Hadfield manganese steel. In foundry operations, the type of ladle used varies. For holding ladles, bottom-pour, teapot, and lip-pour ladles are used. For pouring ladles, bottom-pour, teapot shanks, lip-pour shanks, and drum ladles are used. The ladle linings have changed from refractory lining to the one-heat type with insulating linings made of insulating board or of one-piece designs on shank ladles to eliminate preheating. The repeated use of the refractory reduces cast steel cleanliness and quality, but the board ladles have overcome this problem (Fig. 13).

Weighing and Chemical Analysis Scales. Charge and alloy scales are required for weighing the charge and alloy additions. The charge scale can be the standard beam scale type or the newer load cell type. These scales should be checked daily or weekly to ensure accurate weighing of the furnace melt. The alloy scale can be a small spring or a large balance beam type capable of measuring in pounds and ounces. These should be periodically checked for accuracy. Quality-Control Testing Equipment. The demands for faster and more accurate chemical analysis methods have spawned advancements in the spectrographic, x-ray defraction, and scanning electron microscope analysis of metals. Carbon, sulfur, alloy, and residual analysis is quick and accurate. Gas analysis, oxygen probes, and nitrogen and hydrogen analysis equipment are within the reach of the average melt shop to assist in quality control. Temperature-control immersion thermocouples that are easy to use and recorders that are accurate to within 3 or 6  C (5 or 10  F) are available.

Scrap and Alloy Storage Purchase scrap, shop returns, and alloys should be stored in a covered area. Wet scrap will produce small errors in weight and increase the gas content of the melt. Purchase scrap should be stored by size and the density of its bulk. Scrap should be stored according to specification or alloy content. Alloys and melt additions, that is, slag formers, fluxes, and deoxidizers, should be organized by usage.

Acid Melting Practice Fig. 4

Schematics of apparatuses for raising and lowering electrodes. (a) Arrangement of electrode and mast. (b) Electrode arm arrangement with power-operated electrode holder in which clamp shoe is held in place by a push rod mechanism. (c) Typical arrangement of electrode arm using power-operated electrode holders; pressure on clamp shoe is exerted by spring-operated pull rod.

In foundries and steel mills, the ladles are refractory lined—usually with an alumina lining. More than 72% alumina is used if a clean

steel low in inclusions due to ladle refractory is desired. In one application, magnesite or olivine is used in situations where the metal and

The melting operation from charge to tap out will vary, depending on the type of charge materials and alloy additions used to produce steel in the acid melting practice. The furnace lining will be made from silica or alumina refractories. These refractories determine the type of slag generated during melting, along with the dirt or silica sand included in the charge. The acid practice is a

92 / Melting and Remelting

Fig. 5

Effect of arcs on furnace refractories due to electrodes not being vertical (a) on section left of center and (b) running too high a tap after meltdown

 Lower refractory costs due to the initial cost

and the refractory life

 Lower power consumption as compared to a

two-slag basic practice

 Greater tonnage per unit of time

Acid Steelmaking Steelmaking practices can be classified as partial-oxidation or complete-oxidation operations. These practices are differentiated by the amount of carbon removal during oxidation compared to the final desired carbon content of the melt. The advantages of acid melting practices using partial oxidation are:  Rapid steel production  Good refractory life due to lower iron oxide

content in the slag

 Lower power input  Lower electrode consumption  No recarburization required

The disadvantages include:  Insufficient removal of gases  Occasional poor fluidity due to oxides  Occasional poor mechanical properties

The advantages of acid melting practices using complete oxidation are:

Fig. 6

Typical roof ring with cooling face. (a) Designed for use with special-shaped or cut starter brick. (b) Designed for use with skewback starter brick. (c) Designed for use with special-shaped or cut starter brick

fast oxidizing operation that does not remove phosphorus or sulfur from the melt. Therefore, charge materials with low phosphorus and sulfur levels should be selected in order to meet the final chemical analysis of the steels produced. The acid melting practice is selected for the following reasons:  Flexibility in producing small quantities of

steel to various chemical compositions

   

Low gas content Good fluidity Good mechanical properties Elimination of nearly all silicon and most of the manganese to a uniform base metal to which additions can be calculated more closely

The disadvantages are:  Slower steel production  Shorter refractory life due to higher iron

oxide content in the slag

 Higher production cost

Fig. 7

Roof-lifting devices and swing mechanisms used on top-charging electric furnaces. (a) Roof-lifting device and swing arm mounted separately from furnace. (b) Roof-lifting device and swing mechanism attached to furnace. (c) Roof lifting by cylinder and cable/swing system by rotating roof with a kingpin arrangement

In partial oxidation, the charge carbon content is 0.05 to 0.15% above the finishing carbon content. If the carbon is above this level, a short blow with the oxygen lance is required. No slag-forming materials are added to the bath or the charge, and a fast melting practice is used. The rust and scale in the charge, along with the oxidation of the charge during melting, are sufficient to produce a light boil. When melting is complete, the carbon content will be approximately 0.25 to 0.35%, and the silicon content will be 0.15 to 0.20%; the actual value depends on the amount of oxidation from the charge and the oxidation during meltdown. By the time the laboratory report on the melteddown carbon and manganese is prepared, the carbon content will have dropped another two

Electric Arc Furnace Melting / 93 or three points, and by the time the tap temperature is reached, the carbon will have dropped so that any alloy additions will bring the carbon content to the desired finishing range. Slag additions are seldom required. The slag will be 3 to 5% of the charge weight because of silica dirt and refractory erosion. Foundries using partial oxidation have minimal mechanical test properties to meet. They

use a good grade of scrap, and their primary concern is the high production of molten steel from their furnace capacity. This practice is losing favor because of quality standards and product liability laws affecting the finished product. Complete Oxidation. Most steel foundries using an acid melting method employ the complete-oxidation practice. This is because nearly all steel castings must meet minimum

mechanical test requirements, along with magnetic-particle, visual, and radiographic inspection. In this process, rapid melting is necessary to start a boil during meltdown, followed by oxidation with the use of the oxygen lance. Some melters add iron ore or mill scale to the charge to obtain a partial boil during meltdown. This boiling action or oxidation is accomplished with the following reaction: FeO þ C ! Fe þ CO

Fig. 8

Fig. 9

Furnace design incorporating a water-cooled roof and upper sidewalls. The location of the water-cooling passages depends on manufacturer’s suggestions.

Furnace slide valve used to hold back slag

Fig. 10

Slag-forming materials are not added to the charge or the bath. Slag is formed from silica and dirt in the charge and erosion from the furnace refractories. A fluid black slag should be obtained by prudent additions of mill scale to ensure consistent silicon and manganese contents at meltdown. Meltdown Period. The meltdown carbon content should be 0.20 to 0.40% above the aimed carbon level. The meltdown carbon level will drop 20 to 40 points after oxygen lancing to approximately 0.20% C. The silicon and manganese contents at meltdown will be approximately 0.13 to 0.20% and 0.25 to 0.35%, respectively. After oxygen lancing, the silicon content will be 0.02 to 0.10%, and the manganese content will be 0.07 to 0.15%. A vigorous boil (oxygen lancing) will eliminate carbon, most of the hydrogen, and small amounts of nitrogen from the bath; the elimination of oxygen depends on the carbon-iron oxide reaction during this period. The slag should be green when the boil is completed. After the boil, silica sand can be added to increase the slag volume and to reduce its iron oxide content. Lime is sometimes added to the slag to reduce some of the iron oxide in the slag. Most slag formers are added to the bath during the refining period. Lime is sometimes added to thin the slag. To minimize the loss of oxidizable elements—silicon and manganese—during

Eccentric bottom tap hole furnace (a) with slide gate attached to furnace bottom (b)

94 / Melting and Remelting melting, the percentages of these elements in the charge should be low. Overoxidizing of the heat should be avoided, and alloying elements should be added immediately before tap out. The Refining Period. After decarburization, the boil will slow to a gentle roll, and recarburization and slag additions can be made if required. The recarburizers may be low- silicon, low-phosphorus pig iron, sorel pig iron, highcarbon steel, broken electrodes, petroleum coke, and so on, and occasional dipping of the carbon electrodes. The addition of carbon gives a carbon boil: O ¼ COðgÞ

This helps keep the gas content down. Small additions of ferrosilicon and ferromanganese or silicomanganese are made to the bath at the end of this boil to hold the bath at the desired carbon level. This is known as a block. The heat should be tapped as soon as possible after this procedure and after the addition of any required alloying elements. Deoxidation is required for preventing the reaction of dissolved oxygen and carbon in producing gaseous products such as CO or CO2, into which hydrogen can find a nuclei and into which it can accumulate and form pinholes in the cast product. The first of these additions is silicon in the form of ferrosilicon or silicomanganese and then aluminum, a powerful deoxidizer. Special deoxidizers such as titanium and

Fig. 11

zirconium help tie up nitrogen, while complex deoxidizers containing silicon, aluminum, calcium, and other elements help modify inclusions.

Melting Heats and Raw Materials The preceding discussion is a general description of the acid melting process. This section discusses the heat and the materials used. The charge will consist of 30 to 50% shop returns (heads, gates, and scrap castings) and 50 to 70% purchase scrap (plate, clean punchings, railroad wheels, and so on). The weight of each group of materials is recorded on the furnace log. The charge distribution into the furnace will normally be heavy scrap and casting on the bottom, followed by light scrap and purchased plate. Nonoxidizable materials such as nickel and molybdenum can also be added with the scrap charge. Carbon raiser additions to the furnace, if desired, are normally done before the charge is placed into the furnace. The roof is swung into place after an inspection of the furnace top to ensure that the scrap does not contact the roof ring, and the roof is then locked down. Melting should proceed as soon as the charge is in the furnace. The No. 1 tap will normally be used for meltdown. To protect the roof from excessive erosion, the No. 2 tap can be used to start the electrodes into the charge, followed by a change back to the No. 1 tap. This procedure will reduce roof wear in the delta section.

Four methods of dust and fume control in electric furnaces. (a) Prepollution control ventilation for dust and fume removal. (b) Direct furnace dust and fume collection (both front view and top view are shown). (c) Total furnace hood for fume and dust collection. (d) Canopy hood for fume and dust collection

A pool of metal will form on the bottom of the furnace as the electrodes burrow through the charge. Once the electrodes have reached this pool of metal, the remainder of the charge will be melted by radiation from the pool and the arc splash off of the electrodes. Rust and any iron ore added to the charge will help in forming a watery slag with the sand, eroded refractory, and slag remaining from the previous heat to form an oxidizing slag on the pool of metal produced from the melting portion of the heat. When the charge is completely melted, the tap is changed to No. 2 or lower to reduce arc splash on the furnace walls and roof. A laboratory sample is taken to determine the carbon level of the bath in preparation for the blowdown, or oxygen lancing. The oxidation following meltdown should not begin until a temperature of at least 1540  C (2800  F) is reached. When heats containing chromium are to be made, temperatures of 1595 to 1620  C (2900 to 2950  F) should be reached before lancing to minimize the loss of chromium to the slag, from which it cannot be recovered. Depending on the amount of carbon, the bath temperature, and the amount of oxygen present in the bath, a decrease of 0.08 to 0.10% C will occur for each minute of oxygen injection. Generally, a steel pipe is used as the lance (ceramic-coated steel pipes are also available). The pipe must be clean and free of grease and oil. The blow time is recorded on the furnace log. After the oxygen blow, a lab sample is taken, and the heat is prepared for finishing and tap out. When the lab test is returned to the melter, adjustments to the carbon level will be made by recarburizing or by additional lancing to the required carbon level. A final check on carbon is made, and the heat is blocked with ferrosilicon. This silicon addition will bring the bath silicon content up to approximately 0.10% or slightly higher to stop any remaining carbon boil. Final additions are made to the furnace to adjust the bath to the final chemistry. These additions may include ferrosilicon, ferromanganese, ferrochromium, ferromolybdenum, silicomanganese, nickel, or other alloys. Following these additions, the bath is heated for a few minutes to ensure that the alloys have melted and that the desired tap temperature has been attained. The bath is quickly tapped out into a ladle and the refractory is preheated to minimize temperature loss and skulling from the refractories. Any alloy additions required to meet the final chemistry can be added to the ladle before tap out. The furnace should be tapped so as to hold back the furnace slag and to give a clean full metal stream. The furnace spout should be smooth, clean, and dry in order to obtain a full metal stream without a spray into the ladle. Deoxidation of the steel in the ladle can be accomplished in several ways. One method is to add the aluminum as ingots to the bottom of the ladle just before tap out with other ladle additions. In another method, a steel bar with stars or doughnut ingots attached to it is plunged into the ladle when the ladle is one-half to

Electric Arc Furnace Melting / 95 of rammable material. Silica materials or materials high in silica will cause melting and refining problems and will attack the basic lining materials. The slag additions for basic melting are lime and high-lime-containing minerals. The basic practice can remove phosphorus from the steel. Sulfur removal with the basic practice can be very successful if a very high lime-to-silica ratio (2.5 to 3.5) is maintained in the refining slag or if a synthetic desulfurizer is used at tap out in the ladle. Because of the ability of slag to remove phosphorus (by oxidation) and sulfur (by reduction), lower-quality scrap can be used in the charge. A great portion of the oxidizable elements can also be recovered from the lime slag during refining. For these reasons, the basic melting practice is the preferred method for the production of highly alloyed steels, stainless steels, high-manganese steels, and specialty alloys. However, the basic steelmaking practice will be slower than the acid melting practice when its abilities to reduce the phosphorus and sulfur from the steel and to reduce alloys from the slag are employed to capacity.

Basic Steelmaking

Fig. 12

Schematics of foundry charge buckets. (a) Orange-peel bottom. (b) Clam shell bottom

three-quarters full. Small additions of less than the weight of an ingot are made with aluminum shot or wire clippings weighed into a bag and tossed into the metal stream at tap out. The amount of aluminum used for deoxidation ranges from 1.0 to 1.5 kg/metric ton (2 to 3 lb/ ton); the higher-carbon heats require less aluminum, while the lower-carbon heats require more aluminum because of their higher oxygen levels in the steel. All alloy additions should be recorded on the furnace log with the wattmeter reading, along with any comments on the heat. Other deoxidizers are often used to provide a complex deoxidation of the steel. These include titanium and zirconium to tie up nitrogen plus calcium and barium to modify inclusions. The type and amounts of these supplemental deoxidizers are determined according to the need and choice of each foundry. After tap out, the furnace should be inspected for slag line and bottom erosion and should be patched with silica sand or a mixture of lowgrade alumina and silica. This prepares the furnace for the next heat. Loss of Alloys. In the acid melting process, alloying elements that are oxidizable are not

completely recoverable, and care must be taken to estimate their loss when the additions are made to the bath and the ladle, as well as the loss while the heat is being melted. This loss of alloying element can be compensated for based on the data given in Table 4. Alloy Additions. The typical analysis of alloy additions made to the electric furnace is shown in Table 5. Because this is an incomplete list, alloy manufacturers should be consulted for an expanded list and additional detail with regard to residual elements, material size, and packaging.

Basic Melting Practice The melting operation from charge to tap out will vary, depending on the type of charge materials and the alloy additions used to produce steel in the basic melting practice. The lining of the furnace bottom and sidewalls is made of basic materials, such as dolomite, magnesite, chrome magnesite, bricks, and/or rammable materials. The roof is generally made of highalumina brick, and the delta section is made

The basic steelmaking practice is divided into two parts: the oxidizing period and the reducing (refining) period. Both periods are important because of the reactions that occur in each portion. The charge for the foundry practice will consist of 30 to 50% shop returns, such as heads, gates, and casting scrap; the remainder will be 50 to 70% purchase scrap. Along with these materials, a carbon raiser and lime or limestone are added to the charge. The amount of carbon raiser added should be sufficient to increase the meltdown to 0.20 to 0.45% C over the finish carbon level. In addition to the charge to the furnace, a small amount of limestone or lime is normally added to the furnace bottom along with the recarburizer. (It is sometimes considered preferable to add lime to the middle or top of the melt.) Heavy scrap is first placed on the bottom, followed by light scrap and finally medium scrap on the top of the charge. Every effort is made to get the entire charge into the furnace at one time to reduce the time loss caused by delays in backcharging the required weight into the furnace. When stainless steels are being produced, a greater portion of the chromium addition is held out of the charge to conserve chromium, to help keep carbon levels down, and to cool the bath after the carbon boil when temperatures may reach 1870 to 1925  C (3400 to 3500  F) to achieve concentrations of 0.02 to 0.08% C. Because of the negative effect of silica in the basic practice, a conscious effort should be made to keep sand out of the charge. Oily and wet scrap should also be avoided in the charge materials to minimize gas pickup. The oxidizing period will require the addition of lime to form a slag on the bath during melting until approximately 1.5 to 2.0% CaO is

96 / Melting and Remelting  A high lime ratio to negate the effect of the sil-

ica in the slag; because this slag must also be fluid, fluorspar additions may be required. High concentrations of manganese oxides also hinder the satisfactory removal of phosphorus. The desired reaction for the phosphorus is: 3CaO þ 5FeO þ 2P ! ðCaOÞ3  P2 O5 þ 5Fe

Fig. 13

Typical types of insulating ladles with cold disposable liners used in foundry operations. (a) Teapot. (b) Bottom pour. (c) Teapot Shank. (d) Lip-pour shank

Table 4

Alloy loss in acid melting practice Loss from charge after blow, %

Loss when making addition, %

Manganese Silicon Nickel Chromium

80–85 90–95 0.0–2 35–50

20–30 5 0.0–2 0.0–3

Molybdenum Niobium Copper

0.0–2 0.0–2 0.0–2

0.0–2 0.0–10 0.0–2

Element

Required element weight to be added as furnace or ladle addition to make up desired weight, %

125 95 100 For less than 1% Cr, and 150% to charge; over 1% Cr, add blow for a total of 120% for >4% Cr alloy. 100 100 100

present in the charge. Once the charge has been melted and temperatures of 1540 to 1595  C (2800 to 2900  F) have been reached, the oxygen lancing can begin. If phosphorus is to be removed, another approach requiring a doubleslag practice should be used. The following conditions must be met to remove phosphorus from the melt during the oxidation period:

after

C (2800  F), because the process is reversible with high temperature.  High iron oxide content in the bath and the slag; the use of iron oxide in the early stages of meltdown is required as a source of oxygen to oxidize the phosphorus.  Low silicon and silica in the bath and the slag  Temperature below 1540



1=3

Tricalcium phosphate is formed in the slag. This slag must then be removed from the furnace to minimize the chance for phosphorus reversion which occurs upon de-oxidation of the melt. This is accomplished by raking off the slag. Limited success is achieved in reducing the phosphorus content, as manifested by the film visible around the sidewalls of the bath following slag removal. This film may contain a major portion of the tricalcium phosphate remaining in the bath, which, if left in the furnace, allows for only minor phosphorus reductions and the reversion of phosphorus to the bath with higher temperatures. The use of slag rakes or a slag flush have become the common methods for removing the slag. After the oxidizing slag has been removed, a new lime-based reducing slag must be added to the melt. Any oxidizable element lost to the first slag must also be added back to the bath during the reducing and final tap out. In the single-slag process, oxygen lancing will remove silicon, carbon, manganese, and chromium from the bath, with manganese and chromium partially reduced from the slag in the reducing period. To minimize chromium loss, a high temperature is required before lancing; therefore, a starting oxygen lancing temperature of 1595  C (2900  F) is suggested. The silicon and manganese contents after oxygen lancing will be between 0.01 to 0.05% and 0.03 to 0.20%, respectively, with carbons in the 0.10 to 0.20% range. An addition of ferrosilicon to bring the melt chemistry to 0.10% Si will stop the boil. Reducing or refining of the heat follows the oxygen lancing of the bath. At this time, a deoxidizer is added to the bath to lower the soluble contert of the melt, after which a reduction in the sulfur can be made. With the temperature of the bath increased after lancing, additions of lime or calcium oxide and fluorspar are made to the furnace to prepare for sulfur reduction and alloy recovery. Sulfur reduction depends on:  A high lime-to-silica ratio in the slag, called

the V-ratio

 High temperature  Low oxygen content in the steel and slag

The reaction may be expressed as follows: ðCaOÞ þ S $ ðCaSÞ þ O

This reaction is reversible. An increase in silica will reduce the activity of CaO, thus retarding the reaction, while highly oxidizing conditions in the slag (high FeO in the slag) will limit the amount of sulfur that can be removed.

Electric Arc Furnace Melting / 97 Table 5

Analysis of alloy additions for the electric furnace Composition, %

Alloy(a)

50% ferrosilicon 75% ferrosilicon Standard ferromanganese 67% silicomanganese Pig iron Low-carbon ferrochromium Molybdenum oxide Low-carbon ferromolybdenum High-carbon charge chromium Standard ferromolybdenum Electrolytic nickel Electrolytic manganese Ferroniobium Copper Ferrovanadium Armco iron Manganese calcium-silicon Calcium silicon Calsibar alloy Hypercal alloy

C

Mn

Si

Cr

Ni

Mo

Cu

Nb

Al

Ca

P

S

V

Fe

Ba

... ... 47–51 ... ... ... ... ... 1.25 max 0.20 max 0.04 ... ... ... ... ... ... 74–79 ... ... ... ... ... 1.25 max 0.60 max ... ... ... ... ... 7.0 78 min 1.0 max ... ... ... ... ... ... ... 0.35 max ... ... ... ... 1.50–2.00 65–68 16–18.5 ... ... ... ... ... ... ... 0.20 max ... ... ... ... 4.0–4.5 0.30–0.50 1.00 max ... ... ... ... ... ... ... ... ... ... ... ... 0.025 max ... 2.00 max 67–72 . . . ... ... ... ... ... ... ... ... ... ... ... ... 4–7 ... . . . 55–60 0.70 max ... Fe + Al ... ... ... ... ... ... 0.12 max 0.25 max 1.5 max ... . . . 65 min 0.70 max ... ... ... ... ... ... ... ... 5–7 ... 2–5 65–70 . . . ... ... ... ... ... 0.03 max 0.07 max ... ... ... 2.5 max 0.25 max 1.5 max ... . . . 60 min 0.80 max ... ... ... ... ... ... ... ... ... ... ... ... 99.9 ... ... ... ... ... ... ... ... ... ... 0.005 max 99.9 0.001 ... ... ... ... ... 0.002 max ... ... ... ... ... ... 0.10 ... 1.3 ... ... ... ... 55.0 + 3.0Ta ... ... ... ... ... ... ... ... ... ... ... ... ... 99.7 ... ... ... ... ... ... ... ... 0.20 max ... 1.50 max ... ... ... ... ... ... ... ... ... 55–60 ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... 98.0 ... ... 14–20 49–55 ... ... ... ... ... 1.75 max 14–20 ... ... ... ... ... ... ... 60–65 ... ... ... ... ... ... 28–32 ... ... ... 5.0 max ... 1.00 max ... 57–62 ... ... ... ... ... 1.5 max 14–17 0.035 ... ... 7.0 max 14–18 ... ... 38–40 ... ... ... ... ... 19–21 10–13 ... ... ... ... 9–12

(a) Aluminum is not included in this list, but it is the primary deoxidizer used in steelmaking. The melter needs to understand that there are various grades of aluminum. For example, primary aluminum in the form of ingots and shot has a purity of 99%, while grades I, II, III, and IV can contain 86–96% Al plus other nonferrous alloy elements. Care should be taken to purchase and add this deoxidizer on the weight contained and not to buy on lowestcost basis.

The slag in the basic melting practice is the key to the production of quality steel. The reduction of iron oxide by the use of coal, aluminum, graphite, or low-sulfur coke is often done to deoxidize the slag and to reduce the content of the less oxides. With this deoxidation, much of the manganese and chromium in the slag are reverted (returned to the bath). After a check on the melt or bath chemistry, alloy additions are made to adjust the chemistry with ferrosilicon, silicomanganese, and ferromanganese, along with chromium, nickel, molybdenum, and other alloys if required. During tapping, final alloy additions are made to the ladle, and the heat is deoxidized with aluminum as the primary deoxidizer plus special deoxidizers as desired.

Melting Heats and Raw Materials The following discussion is a brief review of the basic melting practice. A heat and the raw materials used are also described. Historically, the charge was placed into the furnace with the heavy scrap carbon raiser and limestone on the bottom, the light scrap in the middle, and the medium scrap on the top. Today (2008), it is often common practice to use a cushion of busheling scrap on the bottom. Nonoxidizable materials such as nickel and molybdenum can also be charged into the furnace at this time. The roof is swung into place with an inspection of the top of the charge to ensure that the scrap does not contact the roof ring, and it is then locked down. Melting should proceed as soon as the charge is in the furnace on the No. 2 tap to burrow a hole into the scrap, followed by a change to the No. 1 tap until the charge is melted. The initial use of the No. 2 tap will reduce wear on the furnace delta section.

A pool of metal will form on the bottom of the furnace as the electrodes burrow through the charge. Once the electrodes have reached this pool, the charge will be melted by radiation from the pool and the arc splash off the electrodes. Rust on the scrap, along with oxidizing scrap from the melting charge, will react with the lime and eroded refractories to make a fluid slag. When the charge is completely melted, additional limestone or lime is added to the furnace to increase the volume of slag for an estimated 1½ to 3% of the charge weight (approximately 25 kg/metric ton, or 50 lb/per ton, of lime). At this point, the melt can be processed as a single-slag heat without phosphorus removal, or it can be processed as a double- slag heat where phosphorus removal is attempted. This procedure is described earlier in this section. If a double-slag practice is chosen, the oxygen content of the slag and the bath must be increased while holding down the bath temperature. The old slag must then be removed and a new slag built on the bath. Oxygen lancing can be done before or after this stage to lower the carbon content to the required level, or it could have been done to increase the oxygen content of the bath. Reducing the Heat. The slag must be reduced in iron oxide, and because only a small amount of slag has been added to the furnace up to this point, another 1½ to 3% of charge weight of lime is added to this slag, along with a measured amount of fluorspar. This additional slag volume will dilute the iron oxide content in the slag to reduce the oxygen content and to improve the ability of the slag to retain sulfur. A deoxidizer such as coke, coal, or aluminum can be added to the slag if desired to speed the reaction. Sulfur removal and alloy recovery are the primary concerns while the bath is being heated in preparation for finishing. If a specific sulfur level is needed, further laboratory testing and lime additions may be required

for achieving a higher lime-to-silica ratio, along with fluorspar to thin the slag. If the sulfur level is acceptable, the heat, blocked with ferrosilicon and alloy additions, is finished, and a temperature check is made prior to tap out. If desulfurization is required, it should be done after the heat is blocked. The melt is tapped into a ladle that has been preheated if it is refractory lined, or it is tapped into a cold ladle liner of insulating basic material or boards. Any further alloy additions are made to the ladle prior to tap out. The furnace should be tapped to hold back the slag and to provide a clean full stream. The furnace spout should be clean, smooth, and dry to obtain a full metal stream without a spray into the ladle. Additional sulfur removal may be achieved by adding a desulfurizing agent during tapping. Reductions in sulfur content by 20 to 50% are possible. Final deoxidation of the steel should be accomplished in the ladle; this can be done in several ways. One method is to add the aluminum to the bottom of the ladle and tap onto it. Wire feeding is a popular method that has become a common practice in many steelmaking shops. Another method (poling of aluminum) is to attach stars or doughnuts of aluminum to a steel rod and plunge the rod into the ladle when the ladle is onethird to three-quarters full. (Today, 2008, poling is usually done in the ladle. Historically, it was done in the furnace with even greater effectiveness.) Small additions of aluminum should be in the form of shot or wire clippings weighed and placed in a bag and thrown into the stream while tapping. Finally, ferroaluminum additions can also be used during tapping. The amount of aluminum used for deoxidation is approximately 1.0 to 1.5 kg/metric ton (2 to 3 lb/ton), although higher carbon heats require less, and lower carbon heats require more because of the oxygen levels in the steel.

98 / Melting and Remelting After tap out, the furnace should be inspected for slag line and bottom erosion and then patched with dolomite or other materials. Patching is done by shoveling in or gunning on the patch material. This prepares the furnace for the next heat. Loss of Alloys. A wide range of melting techniques are used in the basic melting practice by foundries—for example, single slag, double slag, deoxidation of slags, desulfurization, and alloy production. Therefore, the individual foundry and melter will need to adjust the recovery and additions to meet particular needs and final chemical analysis based on experience.

Newer Technologies Arc Melting of Iron. In addition to steel melting, electric arc melting of iron is practiced. The first continuous submerged direct current electric arc (Contiarc, SMS Demag AG) furnace for melting iron was installed by American Cast Iron Pipe Company in July 2001. The key to the EAF molten iron production is that it can tap, charge, and arc continuously and simultaneously. As a result, the power does not need to be turned off to tap or charge the furnace. The submerged arc design and the continuous nature of the furnace operation make it more energy efficient than the cupola. In addition, the furnace can take either low-grade steel scrap or direct reduced iron and/or hot briquette iron and combine it with coal and silica to melt steel or smelt iron in the same furnace system, producing 73 Mg/h (80 ton/h). EAF Steelmaking. Several technologies have become common practice in most high-volume

EAF steelmaking shops as well as some foundries. For more expansive treatment of these technologies, the reader is referred to either the article “Steel Melt Processing” in this Volume or to The Making, Shaping, and Treating of Steel, 11th ed. published by The AIST Foundation. The direct current arc furnace has been introduced into steel mills and should be changing the arc melting practices of the future from the three-phase arc furnace to the single-arc direct arc practice (Fig. 14). Use of the direct current arc furnace can depend on the stability of the electricity supply in a given local area. Steel mills using the direct current arc furnace have lower electrode consumption and slightly lower operating costs. The direct current arc furnace was developed for two main purposes:  An alternative to the conventional three-

phase alternating current electric arc

 For the final stage in practices for melting

prereduced material for the production of liquid iron Since its development, this furnace has been used in the production of ferroalloys as well as carbon, low-alloy, and stainless steels. An existing three-phase furnace can be modified into a direct current arc furnace by the addition of a thyristor converter to its controls and the incorporation of a water-cooling system. Modification of the existing furnace requires the removal of the two outboard electrodes and the positioning of the middle electrode to the center of the roof. It has also been shown that a larger electrode will require additional

hardware to conduct current and to provide support. The electrode is the cathode in this system. In the three-phase system, current passes from one electrode to another through the charge; in the direct current system, the current passes from the electrode through the charge or bath to the anode located at the bottom of the furnace. The current from the bottom of the furnace passes through graphite- or tar-impregnated refractories to a copper-base plate to the outside cables, or through pins of steel connected to outside current cables (Fig. 14). The heat distribution in the direct arc is 360 , while in the three-phase furnace, the heat pattern is off the edge of the electrode, causing localized refractory heating and erosion. The electrode usage per ton of steel is reported to be 50 to 60% of a conventional three-phase furnace. The shell bottom of the furnace needs to be modified to accept the current-carrying loads that will result and should be insulated to protect it from shorting out. Thermocouples and cooling air will be required for monitoring overheating or loss of current-carrying ability in the bottom. The furnace foundation may also require modification to accept power cables and air cooling ducts. Patching of the refractory is limited to the slag line, sidewalls, and roof. Noncurrent refractories on the bottom will cause poor conduction. Another consideration to be made by the melter is the heat sequence, because direct arc melting requires long start-ups or a heel of metal over the bottom to make electrical contact. Ladle metallurgy has become the processing method of choice in all large EAF steelmaking shops and several foundries. As their name implies, multipurpose ladle refiners have several important capabilities: precise temperature control, vacuum degassing, inert gas bubbling/ stirring for inclusion agglomeration/flotation and compositional uniformity, and alloy addition systems for precise compositional control. Argon stirring of heats of steel in ladles as a method to promote the removal of nonmetallic inclusions via agglomeration and flotation has become a primary tool for making clean steel. The need for dry gas is very important, because argon containing more than 5 ppm of moisture will induce gas pickup. Systems for the accurate, real-time assessment of hydrogen in the liquid metal have become commonplace and are an invaluable aid to the steelmaker. Calcium wire and other wire and powder injection systems have become important tools for specific deoxidation practices, compositional modification, and inclusion engineering. REFERENCES

Fig. 14

Cutaway view showing the basic components of a direct current arc furnace

1. J.F. Schifo and J.T. Radia, Theoretical/Best Practice Energy Use in Metalcasting Operations, U.S. Department of Energy, May 2004 2. Chap. 2, Steel Industry Technology Roadmap, AISI, Dec 2001

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 99-107 DOI: 10.1361/asmhba0005197

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Cupola Furnaces William L. Powell, ThyssenKrupp-Waupaca (Retired) Alan P. Druschitz, University of Alabama at Birmingham Jim Frost, ACIPCO Melting Department

THE CUPOLA is basically a simple cylindrical shaft furnace that burns coke for fuel. The heat generated is intensified by the blowing of air through the heated bed of coke. Alternate layers of scrap metal and coke are charged into the top of the cupola. In its slow descent, the scrap metal is heated to the melting temperature by direct contact with the upward flow of the hot gases from the coke combustion. The moltenmetal droplets collect in the inner portion of the cupola known as the well, where it is discharged into the foundry from the taphole by intermittent tapping or by continuous flow of molten metal and slag. Melting in cupola furnaces dates back several centuries to old cask-type units that were blown with hand-pumped bellows and hand charged with bits of metallic iron and scraps of cast iron. The cupola as we know it today (2008) was patented in 1794 by John Wilkinson. Prior to the 1950s, the cupola was the predominant furnace for iron melting because of its simplicity and the low cost of molten iron produced. Progressive improvements were made in electric induction furnace equipment since the 1950s. At their inception, induction furnaces required little or no pollution-control equipment because mostly clean, new scrap was charged. This mix of new scrap iron and pig iron yielded a low-sulfur iron that was satisfactory for the fast-growing ductile iron industry. Electric induction furnaces also began to replace many cupolas in small-tonnage operations where both gray and ductile iron or several other types of irons were being produced. The ability to make new chemical compositions in each melted batch provided more flexibility to the foundry operator. Electric induction melting was found to have a lower melting cost and better chemical control under many of these conditions. The number of cupolas in use in the United States has decreased from 4000 in the early 1950s to less than 1400 today (2008). However, the cupola remains the predominant tonnagemelting method, with over 60% of today’s iron tonnage still melted in cupolas. Through the years, the cupola has been progressively refined

and moved toward more efficient combustion, better chemical composition control, and more flexibility in the use of widely varying raw materials. However, during the melting process, the cupola emits considerable amounts of smoke and particulate matter. Known as emissions, this waste gas stream is now cleaned to meet the highest environmental standards, but this has required much engineering effort and considerable expense. With environmental laws requiring more elaborate pollution-control equipment, the cupola lost much of its advantage of low cost and simplicity, especially on small cupolas. In high-iron-tonnage operations, the cupola remains the most efficient source of continuous high volumes of iron needed to satisfy highproduction foundries or the multiple casting machines of centrifugal pipe producers. The continuous stream of cupola-melted iron is not interrupted by charging, slag removal, or chemical composition alteration. On a per ton basis, cupolas usually require less labor to operate and can use lower-cost forms of scrap. In such high-tonnage, continuous-demand situations, the cupola has remained the lowest-cost method of melting. Although the cupola is more difficult to control, many consider it to have favorable iron qualities, such as a low tendency toward chilling, good fluidity, better machinability, and improved shrinkage characteristics compared to electrically melted cast irons.

Progressive Refinements in Cupola Equipment Since the early days of the simple cupola, continuing technological refinements have been progressively applied to the cupola melting process to improve efficiency, flexibility, and control. Many of the refinements have stood the test of time and greatly added to the cupola melting technology. Some other ideas have not proved to be technically sound or have lost appeal because of ever-changing economic conditions, industry needs, state and federal law changes, and a host of other reasons to abandon

them by the melting community. Some of the more successful improvement technologies include the following. Preheated air blast was found to improve efficiency of the combustion reaction by allowing the input of British thermal units (Btus) from an external source, thus increasing melting rate, temperature, and carbon pickup with the same amount of coke, or alternatively, the amount of coke used to melt a ton of iron can be reduced for the same melt rate, temperature of the iron exiting the cupola, or greater carbon pickup. Early blast air heating systems could preheat the incoming air only 150 to 260
C (300 to 500
F), but even this limited additional temperature demonstrated the advantages of the preheated blast. The most successful hot blast systems have been externally fired with gas and are capable of up to approximately 650
C (1200
F) preheat. Higher blast temperatures would continue to improve performance, but the availability of heat-exchange tube materials has a practical economic limit of approximately 650
C (1200
F). Recuperative hot blast systems have begun to replace the externally fired hot blast system. The recuperative hot blast system is an integrated system, which combines both the blast air heating and gas cleaning system. Emission gases are usually first combusted to raise the heat content of the offgas, then the heat is extracted via a gasto-gas heat exchanger (the hot blast portion of the system), and finally, the cooler gases are cleaned of dirt particles. In some designs, the gases are cleaned and then burned. Both have distinct advantages and disadvantages but are effectively used to preheat and clean the offgas stream. Oxygen enrichment and injection, which adds 1 to 15% O2 to the blast air, also intensifies the combustion reaction by increasing temperature, which in turn increases the melting rate. The efficiency is derived from blowing in a higher number of oxygen molecules per volume of air and lower amounts of inert nitrogen molecules. Enrichment is limited to oxygen increases of approximately 2% and is accomplished by adding pure oxygen to the blast ductwork. Injection allows much higher additions of

100 / Melting and Remelting oxygen. Increases of 4 to 15% can be achieved. A pipe is inserted into the center of each tuyere for adding the oxygen. Often, a nozzle is added to the tip of the pipe to achieve supersonic flow rates for the oxygen. This increased velocity allows blast air and oxygen to penetrate deeper into the coke bed. Injection of fine coke and other materials, such as silicon carbide, through the tuyeres has proved to be effective in both increasing carbon and silicon content and shortening response time for chemical changes in the melt composition. The effectiveness of this technology is somewhat dependent on the operating parameters of the particular melting operation. Water cooling of the cupola shell has extended the length of the melting period beyond 24 h and has permitted a single cupola to be operated for weeks before requiring a refractory repair shutdown. Water cooling of the shell allowed the removal of the refractory lining and its effect on the slag chemistry. The same cupola then could operate with a choice of basic, neutral, or acid slag chemistry simply by regulating the amount of basic flux added. The other benefit is the reduction of sulfur in the molten iron, allowing those ductile iron treatment methods requiring low-sulfur-base irons to be used with no external desulfurizing step in the molten-metal processing. Duplex electric holders are large holding furnaces that receive the iron after leaving the cupola. They have ensured that a constant supply of iron is available at all times while also reducing variations of composition and temperature of the iron. Large electric channel induction furnaces hold between 30 and 90 min of iron. These electric channel furnaces can provide a modest boost in iron temperature or maintain iron temperature over cupola melting down-periods. Computer applications have been increasingly used to record, monitor, and control more cupola conditions. Today’s (2008) melting operations are as highly computer controlled as any manufacturing operation. Virtually all control of the motors, valves, flow devices, dampers, and so on are computer controlled. All the various inputs of temperature, flow, pressures, status of valving, and so on are fed back to a computer for data storage and analysis. Continuous analyses of top gases have made it possible to monitor combustion efficiency and to signal any water leaks to avoid explosive gas combinations. Basic slags have made possible the production of low-sulfur iron directly from the cupola and have increased the carbon pickup. By regulating slag basicity, a wide range of iron compositions can be obtained from a wide range of charge materials. Technology Failures. Some technologies have not passed the test of time in melting operations. These have failed the test of providing a true increase in efficiency, being technologically sound, or may have simply been a passing fad. Many are still reflected in

the technical literature of the time, and care must be taken by individuals to determine if they are truly an improvement to the cupola melting operation. Some of these technology failures include the following. Divided-blast cupolas divide the air between two rows of tuyeres in controlled proportions. This was reported to increase combustion efficiency and accomplish either coke reduction or increased melt rate and temperature at the same coke usage. Reversed taper shell was a cupola design meant to concentrate the air across the entire cupola cross section for more uniform melting area and lower coke usage. Ground rubber tires were tried as a coke substitute. Rubber tires are largely composed of carbon and have a high Btu content. Considered a waste product, their use in the cupola would be a way to reduce both a national waste item and coke cost. However, the volatile gases produced did not provide a carbon source and also proved difficult to scrub from the emission gases. Natural gas enrichment to either the blast or the offgases was proposed as a method to raise the Btu content of the blast air. However, the water vapor released during combustion proved to have a negative effect on the heat content within the cupola. Humidity control of the blast air has been employed on some cupolas to eliminate the effect of varying humidity when very close control was important. However, the cost of drying the absorption media proved to exceed the benefits of water removal.

melting needs and inputs: type and size of scrap available in the market place, desired melting rate, melting rate flexibility needed, carbon pickup desired, use or nonuse of additional oxygen, length of desired campaign, length of time for relining, chemistry control limits, and so on. Optimum cupola height has been well studied, and 8.5 to 10 m (28 to 34 ft) of effective melting height is universally accepted as the optimum height for a cupola. Cupola shell design (both diameter and height) as well as tuyere placement, offgas take-off location, and so on require detailed forethought and planning during the design and installation of a new cupola. The foundation of the cupola structure must be adequately engineered for the size of the cupola and the loads it will carry. Footing loadings will be for both the cupola itself and the burden contained within the cupola. A large cupola will often contain 136 to 181 Mg (150 to 200 tons) of scrap iron, coke, stone, and molten iron. An often overlooked component is the

Cupola Construction and Operation This section discusses the shell, intermittent or continuous tapping, tuyere and blower systems, refractory lining, water-cooled cupolas, emission-control systems, and storage and handling of the charge materials.

Shell Design The cupola shell is a vertical, cylindrical steel shell and is the container of the combustion and melting operation. Figure 1 shows a typical modern-day cupola. The size, or specifically, the diameter, of the cupola shell basically defines the limits of operation of the cupola. It limits the minimum and maximum melting rate per hour of operation, the carbon pickup ability, the length of the melting campaign, the maximum size of the scrap that can be charged, the size of the coke purchased for melting, and a host of other parameters. In the past, cupolas were purchased as somewhat standard pieces of equipment. For the past 30 years or more, cupolas have been very much a custom-designed piece of melting equipment. Cupola manufacturers can suggest an optimum diameter based on the user’s specific anticipated

Fig. 1

Typical modern-day cupola showing the construction and arrangements of the principal cupola components

Cupola Furnaces / 101 dynamic loading place on the cupola during charging. Charge drops of 4 to 7 Mg (4 to 8 tons) of scrap falling 1.5 to 6 m (5 to 20 ft) are not uncommon. Charge doors today (2008) are almost always not a door but are actually the opening at the top of the cupola. Traditionally, cupolas were side charged near the top of the cupola. Today’s emission systems demand controlled volumes of offgases. The side charge door allows uncontrolled volumes of air to enter the gas streams, thus causing erratic cleaning of the waste gases. The top-charged cupola employs a “plug” of scrap at the top of the cupola to effectively limit the amount of air being pulled into the cupola stack. A second benefit is the ability to soft-drop a charge onto the burden, thereby reducing the crushing of the coke charge during the charge drop. Bottom doors at the bottom of the cupola stack facilitate emptying of the stack of coke ash and residual metal at the end of an operational melting campaign in preparation for refractory repairs and the resumption of melting. The doors are usually constructed of heavy steel or ductile cast iron and are massive enough to support the entire burden of the charge material and coke as well as the dynamic forces encountered during charging. These doors cover the entire bottom cross section so as not to impose any restrictions during the dropping of the bottom. During the melting operation, the doors are held rigidly in place by heavy steel posts or similar mechanisms. There can be no flexing of the doors, either mechanically or thermally, during the melting operation, or cracks in the bottom refractory will occur that may result in an iron runout. A sand bottom is rammed over the closed bottom doors to contain the melted iron in the well. This sand bottom can either be all sand, all refractory, or a combination of both. In smaller cupolas with short-duration melting campaigns, the bottom can be as simple as foundry molding sand. For larger cupolas with longer melting campaigns, special sands and clays are preferred, with the addition of a refractory layer. There is no general agreement among melters on bottom sand materials and arrangements. What can be agreed upon is that once a system is determined, strict controls must be put in place to assure quality of installation every time a new bottom is put in service. A sand bottom is sloped toward the taphole to ensure that iron and slag continuously flow to the taphole and out of the cupola.

Intermittent or Continuous Tapping Cupolas Iron flow from a cupola can be made to be intermittent or continuous. The first cupolas were intermittently tapped by opening and closing the taphole as in a blast furnace operation. For intermittent tapping, a refractory plug is pushed into the open taphole after iron has

stopped flowing, and the plug is left in the taphole until a volume of iron has accumulated in the cupola well, then the plug is pulled out and iron begins to flow. This process is repeated for the duration of the melting campaign. Slag is accumulated within the cupola well area and flows out a back slag hole. Most modern-day cupolas are tapped continuously, with iron flowing under a refractory dam and over a lip in the cupola spout. The lighter slag floats on top of the stream of molten iron. A notch in the side of the cupola spout (also called the well) allows the slag to flow out of the front spout area. This slag flows continuously to a slag disposal area. Cupola slag has traditionally been quenched in a water basin or a trough of flowing water. However, due to wastewater rules prohibiting the discharge of process water, a variety of dry slagging methods have been developed. The various dams and notches in the front spout area are constructed of refractory materials. Due to refractory wear, the levels of metal dam and slag notch must be carefully calculated and maintained throughout the melt campaign. The dry-bottom cupola is a recent European development in which iron and slag are not collected in the well of the cupola but run into a special vessel outside and beside the cupola. In this outside vessel, the accumulated metal and slag are separated by gravity as they flow through the vessel. The advantages claimed for this method include more consistent iron analysis and easier repair of refractories. Cupolas often have two vessels attached to the front, allowing the repair of one while melting iron through the second vessel. This arrangement allows for longer melting campaigns.

Tuyere and Blower Systems Air for the combustion of the coke is introduced through holes in the shell that have water-cooled pipes, called tuyeres, protruding into the cupola. Tuyeres are constructed of either cast copper or fabricated copper coils twisted around a large piece of pipe. They are equally spaced around the cupola, with the number and size proportioned to cupola diameter. Tuyeres are sized to deliver a specific relationship of air volume, pressure, and velocity. On water-cooled hot blast cupolas, the desirable blast air velocity range is 4600 to 7600 m/min (15,000 to 25,000 ft/min). This velocity has been determined to give the blast air sufficient velocity to maximize the penetration of the coke bed. Blast air blowers generate the necessary blast air volume and pressure to melt iron. The air required to melt 900 kg (1 ton) of iron is 620 to 680 m3 (22,000 to 24,000 ft3) at standard temperature and pressure or 770 to 835 kg (1700 to 1840 lb). Air from the blast air blower is transmitted to the tuyeres through a large-diameter duct system to the cupola wind box. Blast pressure, usually called backpressure, normally ranges from 3.3 to 8.5 N (20 to 65 oz/in.2) or more, depending

on charge density. Blast blowers can be of several designs, such as positive displacement blowers, fans, or centrifugal-type blowers. The latest engineering design preference on large cupolas has been to use two or three blowers in parallel with each other and use automatic air weight controls to control the air volume produced by them. Blast blowers typically consume the second-largest amount of electrical energy in a melting department, second only to the emission system fans. The biggest waste of efficiency in the melting department is due to leakage in the blast air system. This is due both to missing insulation allowing heat loss before the heated blast air reaches the cupola, and leaks in the blast air ductwork system caused by thermal and mechanical movement.

Refractory Lining Conventional cupolas are usually lined with refractories to protect the steel shell against abrasion, heat, and oxidation. Figure 2 shows a cross section of a cupola with the various refractory types indicated. Lining thicknesses range from 114 to 305 mm (4.5 to 12 in.). Most linings are either fireclay brick or a sprayed or gunned alumina material. As the melting campaign progresses, the refractory lining in the melt zone is progressively fluxed away by the high temperature and oxidizing atmosphere, reducing the lining thickness. Early cupola melting practice was to stop melting, drop the bottom, and repair the lining at the end of the day’s melt. After cooling the cupola interior, repairs were made to the refractory lining, restoring it to its original diameter. Often, a pair of cupolas were alternated each day to facilitate the repair and preparation of the cupola lining after each day of use. Today’s (2008) improved refractories and shell water cooling have, for the most part, eliminated the need to have two operational cupolas. Many methods and materials are employed for cupola refractory repair. The most widely used method of lining repair uses a pneumatic gun to blow a mixture of clay and ganister through a nozzle containing water that becomes entrained in the stream as it exits the mixing nozzle. New methods such as shot creting have been satisfactorily used for cupola refractory patching and for extending the life of the cupola refractory. Shot creting creates a denser refractory layer and allows for a greater variety of refractory materials to be used. For repair of the taphole area, refractories containing silicon carbide particles are rammed in with pneumatic hammers. For the upper stack portion of the cupola, brick or blocks are layered with a thin mortar joint. Specialty refractories are being developed for every segment of the cupola to meet the exacting needs of the specific cupola operation. The advent and growth of ductile iron in the 1950s made it desirable to find a way to remove sulfur inside of the cupola. This was

102 / Melting and Remelting accomplished by using a sufficiently basic slag, as is done in basic steelmaking furnaces. However, a highly basic slag cannot be maintained in contact with acid refractories. After considerable experimentation, suitable basic refractories were successfully developed for use in cupolas. Originally, linings were either magnetite or dolomite brick. Patch materials composed of dolomite or magnetite aggregate were then

Fig. 2

developed that could be rammed or blown in using the pneumatic gun.

Water-Cooled Cupolas When the melting operations of foundries were extended to a continuous 16 to 20 h of operation, there was little or no refractory lining

remaining in the melting zone at the end of the melting campaign. When three-shift melting operations became a necessity, the water-cooled cupola was developed. The early water-cooled shells were simple tanks or jackets added to the outside of the cupola shell. Figure 3 shows an early water-cooled cupola design. Water circulated through the tank with little velocity, and cooling was basically just through heat conduction into the water. A limiting factor to the tank method of water cooling was that if a hot spot developed on the shell, it could not be observed nor could additional water cooling be added to the area during the melting campaign. Later water-cooling developments led to the cascading water shell, where a thin layer of water flows rapidly down a cupola shell, wiping heat from the shell in the process. At the top of the water-cooled area, a water curtain is applied to the shell by spray rings encircling the cupola. The water curtain cools the shell as it cascades down the shell, and the water is collected in a trough or moat at the well. One advantage of the open external cooling is that the outside of the cupola shell can be sand blasted at intervals to remove excess deposits of lime and other built-up materials. This sand blasting also restores the thermal conductivity of the steel shell. With increasing dependence on water cooling, it was found that the refractory lining could be eliminated from the melting zone and that the cupola could be operated continuously without a refractory lining. However, the advantages of water cooling are achieved at a loss in thermal efficiency. With demands on lower cost for coke and greater operating efficiencies,

Fig. 3 Cross section of a modern-day cupola showing typical refractories used in lining a cupola

Early water-jacketed cupola used to extend operating hours of a melting campaign

Cupola Furnaces / 103 today’s (2008) cupolas are a hybrid of water cooling and refractory linings. When iron is needed for only two shifts, the water-cooled cupola can be banked during the downshift and brought back on stream when needed. These water-cooled cupolas can even be banked over a weekend as well as operated without the need for refractory repairs for 2 to 4 weeks (or even longer under some operating conditions) before requiring the dropping of the cupola bottom for repairs. Some of the largest cupolas are equipped with double tapholes and double rear siphon slag spouts to enable the cupola to be operated continuously for even longer melting campaigns.

Emission-Control Systems Equipment has been successfully engineered to enable the present-day cupolas to meet the most rigid of environmental regulations. The two most common types of collectors currently in use today (2008) are the high-energy wet scrubber and the dry bag house. However, current environmental regulations have placed even higher demands on the cupola operator for cleaner air emissions, and only dry bag house systems are able to meet these new regulations. High-energy wet scrubbers use high-velocity water jets to impinge atomized water droplets on effluent dust particles passing through a venturi scrubber. The result is a dirty water sludge. This dirt and water sludge requires thickening, separation, and dewatering. Due to the various hazardous chemical elements (typically cadmium and lead) present in the cupola gas stream, the sludge and water are deemed toxic and require further chemical treatment to neutralize the hazardous elements. This is a complex science and a materials handling

challenge. Figure 4 illustrates a typical wet scrubber emission-control system connected to a recuperative hot blast system. The dry bag house environmental system is the best cleaning system for meeting today’s (2008) air emission standards. In this unit, the dust particles are collected on the surface of fabric bags, and then the bags are shaken to free them of the dust. Gases leave the cupola at temperatures ranging from 95 to 1100
C (200 to 2000
F), depending on what is happening in the cupola. The cloth bags can only withstand temperatures of approximately 230
C (450
F); thus, a system for cooling the gas stream before reaching the bag house is necessary to protect the bags from excessive heat. The two most popular heat-exchange systems are the air-to-air and air-to-liquid heat-exchange systems. The engineering for either is complex and a challenge for mass flow and heat-removal requirements. To better understand the operation of the recuperative, dry cleaning system, it can be broken into smaller subsystems. Figure 5 shows a modern dry bag house cupola emission system. The system design is basically simple in terms of what is being accomplished in each stage. Cupola offgas exits the cupola at a temperature between 95 and 540
C (200 and 1000
F) under normal operating conditions. This gas is rich in carbon monoxide (CO) but depleted in oxygen and has the potential for combustion. Air is added to the mixture in a controlled amount, and the mixture is ignited via pilot burners in a large combustion chamber. This combustion results in a stable temperature of approximately 820
C (1500
F). The entire gas stream is now passed through either an air-to-air or air-to-liquid heat exchanger to heat the blast air going to the cupola, or a secondary heat-removal step can be added to heat air for

building heat and/or for other processes in the foundry operation. The gas stream exits the cooling tower at a temperature of 135 to 165
C (275 to 325
F) and is now ready to enter the bag house at a temperature that is low enough not to harm the filter bags. While the pores of the cloth are very small, the particles in the gas stream are much smaller, so, in theory, the dust particles would pass through the bags and into the atmosphere and not be trapped. However, the dust layer that builds up on the bags traps the smaller dirt particles in a porous cake of dust. The bags are mechanically shaken or popped with a shot of air to remove some of the dust cake. As in the wet system, the dust may be a health hazard, depending on the concentration of hazardous elements present. Process chemicals may be added to the dust/gas stream before the bag house to neutralize these hazardous elements. Each lot of bag house dust is tested and then properly disposed of in a suitable landfill.

Charge Materials Storage and Handling of Charge Components Essential to efficient cupola operation is a good materials storage and handling system. Metal charge components are received via dump trailers, rail cars, or bulk containers. Unloading into storage bins can be by selfunloading dump trucks or by using an overhead charge crane with a magnet. The size of scrap storage bins varies widely with how much is to be melted each day, distance from the scrap-processing source, reliability of the transportation system, severe weather issues, and so on. Each charge material needs to be transferred to a weigh hopper, where a designated weight is measured out and transferred to a charge bucket for travel to the top of the cupola. The design possibilities for the charging area are endless. Figure 6 shows a typical charge yard design. Coke and limestone are usually stored in covered overhead storage bins and are discharged into a weigh hopper and on into the charge buckets. One factor in charge yard design that is often overlooked is planning for flexibility. Charge yards are expensive to construct and are longterm structures. The needs of the foundry and the availability of charge materials are likely to change over the lifetime of the facility, so it is necessary to build in some flexibility in design to accommodate for an unknown future need.

Coke

Fig. 4

Typical wet scrubber with heat recuperation hot blast used to clean emission gases

As the major fuel, coke is the most important charge material. High-quality, foundry-grade coke is essential for optimum cupola performance. The amount of coke in the charge mix usually ranges within 8 to 16% of the metal charge, depending on the iron temperature

104 / Melting and Remelting

Fig. 5

Fig. 6

Typical dry bag house emission and recuperative blast air heating system used to clean emission gases

Charge yard arrangement showing scrap delivery, storage bins, weigh hoppers, and charging bucket

required, the percentage of steel scrap used, and many other variables. Foundry coke must be strong to support the charge materials in the cupola stack and have sufficient impact strength to maintain its size during the impact of charging. These mechanical properties are important since they affect hot blast penetration and carbon pickup. Experience has shown that performance is optimized when coke size is consistent. Fine particles of coke should be screened out of the coke portion of the charge, because they fill the interstitial spaces between the coke and scrap iron pieces; this limits air distribution and overloads the emission system with excess carbon monoxide gas. On large cupolas, usually the largest available coke size is preferred, but on small cupolas, coke screened to any consistent size range provides good performance. Coke is manufactured to coke industry standards. Table 1 is a typical foundry specification for coke. Typical foundry coke should have a low ash content (<10%) and a chemical structure that provides for optimum carbon pickup and combustion efficiency. Low sulfur content (<0.70% S) is required for preventing excessive sulfur pickup by the molten iron. Volatiles should be less than 1.0%, and fixed carbon should be above 90% for maximum cupola performance. Physical properties of coke also must be controlled. Coke strength must be high to prevent crushing in the cupola stack. A dropshatter test best measures coke strength. Abrasion resistance, as measured by stability or hardness, must be as high as possible. Efforts have been directed toward the use of limited amounts of less-expensive substitutes

Cupola Furnaces / 105 Table 1 Typical specifications for foundry coke Fixed carbon Volatile Ash Sulfur Moisture Shatter strength 50 mm (2 in.) screen 75 mm (3 in.) screen 100 mm (4 in.) screen Stability index Coke size

91.5–94.5% 0.1–1.0% 5.0–7.5% 0.4–0.7% 0.1–5% 96.0% min 92% min 78% min 38.0% min As required by cupola size

for quality coke, such as briquetted coke fines, anthracite coal, injected fine coke, and various substitute fuels. However, all of these substitutes reduce cupola efficiency, and the effectiveness of each must be weighed against total fuel cost-savings.

Metallic Charge Materials The modern cupola can melt a wide range of metallic charge materials, most of which can be classified as scrap iron and steel, various cast irons, and other ferrous materials. Through its use of scrap materials, the casting industry contributes greatly to environmental cleanup and resource reutilization. Steel scrap was first melted in low percentages to lower the carbon content of gray iron in order to produce higher-strength irons. Today’s (2008) cupolas use steel scrap at a rate of 25 to 90% of the charge mix. Obsolete scrap iron is processed in a variety of ways. It can be cut with a cutting torch, sheared with a large blade shear, shredded in a hammer mill, or bailed into cubes. Each method produces a useable material for cupola melting, but each has its own melting characteristics. Cupolas are generally flexible enough in design to melt any of these prepared materials, but an optimized cupola design can make any scrap preparation method more efficient. Each preparation method has a unique cost associated with it and is local market dependent, so there is not one method or material that is preferred. Cast iron scrap is a common cupola charge material. It is broadly divided into four or five classifications of automotive, machinery, heating, and pipe. The grades useable in any particular foundry are largely dependent on the foundry products being produced. Scrap size and chemical composition are the basic source of the different grades of cast iron. Cast iron has a high carbon content and low melting point. It generally gives a consistent chemistry to the melt and lowers the chill tendency of the melt. Due to its good melting properties and limited availability, it is generally higher in price than scrap steels. Foundry returns, such as gates, runners, and internally generated scrap castings, usually constitute 30 to 50% of the pouring weight of a mold, and thus, they are returned to the charge

yard for melting in approximately this ratio. A melting operation will need to melt all of the foundry returns sent to it. As in cast iron scrap, carbon content is high and melting temperature low. Since this is internally generated scrap, it should have the most consistent chemical content of all metallic charge materials. Pig iron is considered a premium charge material because of its uniformity of composition and size, and it is free of detrimental tramp elements. In the early years, pig iron was the predominant charge component for ensuring quality iron with sufficient carbon content and low quantities of detrimental elements. Despite economics that encourage the use of more scrap and less pig iron, pig iron has remained a basic charge material because of its consistent size and quality. A wide range of scrap metals can be melted in the cupola. However, some forms of scrap metal are not practical to use in a cupola charge. Cast iron borings charged directly into the cupola are excessively oxidized if not blown out of the stack. With briquetting, limited proportions can be used. Steel turnings are more impractical because they are so thin that they readily oxidize high up in the cupola stack. Thin sheet steel has a very low density as received from the source and cannot be used without being compressed into bundles. Compact punchings tend to clog interstitial coke passages and hinder combustion efficiency. Very thick steel parts and large-diameter shafts are not completely melted within the melting zone as the charge descends into the cupola, and this results in low iron temperatures. Recently, a manufactured product consisting of cast iron borings, steel turnings, grinding dust, coke breeze (optional), and a binder has been successfully pressed into a brick. This allows the use of materials previously not usable in the cupola. Brick use must be limited, but use rates of up to 30% of the charge are possible.

Ferrous Alloys Any number of alloys can be added to the cupola for melting into the base iron. Most commonly, these alloys are purchased preblended with an iron-base material. Silicon, which is added to yield the desirable final silicon content, is usually added in any of the following forms:  Lump ferrosilicon: 50 or 75 wt% Si content  Silvery pig: 16 to 20 wt% Si content  Silicon briquettes: composed of silicon fines

briquetted with a suitable binder, often portland cement, with the briquette containing approximately 40 wt% Si  Silicon carbide briquettes: made of silicon carbide plus carbon briquetted together with a suitable binder Manganese briquettes of various compositions can be charged into the cupola in several forms to obtain the desired final manganese

content. The briquettes are generally composed of ferromanganese granular material and a cement binder. Contained manganese percentages are usually 50 to 75 wt%. Alloys of chromium, nickel, molybdenum, and other required alloys can be purchased and added to the cupola. Care must be taken in the selection of alloys to be sure that the size of the additive is sufficiently large to minimize oxidation and losses to the emission system. Briquettes must have sufficient strength to survive charging operations and the descent into the cupola melting zone. Binders must be evaluated to have sufficient hot strength to prevent premature breakdown of the binder system before entering the melting zone.

Fluxes Charge materials contain large amounts of nonmetallic materials, such as dirt on scrap, sand adhering to returns, slag attached to pig irons, and so on. Additionally, coke ash, oxidation products from melting, cement binders, and other noniron materials must be fluxed into low-melting-point compounds. These slags must be sufficiently fluid to be able to exit the cupola and float to the top of the iron so they can be skimmed off and not sent to the holding furnace. Limestone or dolomite additions of 2 to 8% of the metallic charge (or 30 to 45% of the coke portion of the charge) are generally needed in an acid cupola to contribute enough basic constituents of calcium oxide and magnesium oxide to impart sufficient fluidity for the slag to flow freely through the coke spaces, into the well, and down the front slagging trough. In the basic slag cupola, more basic fluxes are added to net an excess of basic slag constituents. Limestone is a natural calcium carbonate or dolomite stone. A lump size of 25 to 75 mm (1 by 3 in.) is preferred. Limestone composition varies widely across the country, so selection of stone for a melting operation is critical. Screening of limestone fines is mandatory for a good flux stone. For limestone (calcium carbonate limestone), CaO contents should be above 45%. For dolomite stones (those containing high levels of calcium-magnesium carbonate), CaO levels should be approximately 30% and MgO levels approximately 20%. Silica is an impurity and should be less than 2%. Several special or supplementary fluxes are occasionally used. When used, they are usually added very sparingly in addition rates of 0.2 to 2.0% of the metallic charge weight. Soda ash (Na2CO3) is a strong slag liquefier and is added as briquettes to the bed to supplement the limestone. Fluorspar (CaF2) is a powerful fluidizer that is used to clean up a cupola clogged with excessively viscous acid slag. In highly basic slags, fluorspar is needed to maintain fluidity. However, excessive fluorspar can cause environmental problems.

106 / Melting and Remelting Calcium carbide (CaC2) serves the multiple functions of deoxidizing, raising carbon, raising temperature, and providing a strong calcium oxide (CaO) flux contribution.

Cupola Control Principles Coke Bed. Preparation of the coke bed is vital to the initial start of the melt and to the maintenance of the melt at the specified level in the cupola combustion zones. Before metal is charged, a bed of good-quality, properly sized coke is ignited and blown to a red heat either internally in the cupola or externally and charging it hot. The height of the coke bed is adjusted to a specific height above the tuyeres that is determined by experience and based on blast volume, cupola diameter, and desired melting conditions. Bed flux and metal charges are then added, the blowers are turned on, and the melting begins within a few minutes. With a high coke bed, melting occurs higher in the cupola stack, producing slower melting rates and hotter iron with less oxidation and more carbon pickup. A low bed melts charge materials lower in the cupola, with faster melting rates and colder iron temperatures with more oxidation. Charge coke is the coke intended to replace the coke burned in the melting of each layer of charge materials. It is weighed and added to each charge to maintain the bed level within the cupola during the melting cycle. While the exact height of the bed during melting cannot be observed, a skilled operator knows its relative position by observing melting rates, slag temperatures, iron temperatures, and final iron chemistry. Melt Rate. A cupola is expected to maintain a melting rate sufficient to meet the needs of mold production. The cupola size must be chosen for the midrange of production demand. Several general rules can predict melting rates. One rule is that 4.5 kg (10 lb) of iron can be melted per hour for each square inch of crosssectional area. A second rule is that 3.2 to 3.3 Mg (3.5 to 3.7 tons) of gray iron can be melted with 0.5 m3/s (1000 ft3/min) blast air, and 2.9 to 3.0 Mg (3.2 to 3.3 tons) of ductile iron can be melted with 0.5 m3/s (1000 ft3/min) blast air. By increasing or decreasing the blast air rate, melt rate can be varied by approximately þ 20% with reasonably good efficiency and  control. Beyond these limits, overblowing a cupola increasingly oxidizes the iron, causing lower carbon and silicon levels and decreasing the iron temperature. Underblowing will lower the iron temperature because the blast will not penetrate the cupola with sufficient air distribution for optimum combustion efficiency. Changes in melting rate can also be accomplished by changes in the charge mix ratio of cast iron components to steel and by varying the hot blast temperature. If oxygen injection is used, changes in oxygen volume can widen the melt rate control range.

Iron Temperature. The iron temperature at the cupola spout must be sufficient to overcome temperature losses due to metal handling from the cupola to the electric furnace or holding ladle, the holder to the pouring ladles, alloying additions, the treatment process (for ductile iron), and radiation losses from the pouring ladles. Iron temperature can be measured by optical pyrometers, immersion thermocouples, and/or continuously recording radiamatic instruments. Carbon and silicon concentrations are most important in the control of iron properties. Sufficient concentrations of carbon and silicon must be maintained to avoid chill and to encourage a microstructure suitable for the section size of the casting and the strength level required. In ductile iron, sufficient carbon and silicon concentrations must be maintained within controlled ranges for proper production of the nodular form of graphite and control of the matrix structure. Excessive carbon content can result in carbon flotation in ductile iron castings. Carbon content of the iron is influenced by charge mix, carbon content of the metallic scrap, the coke-to-metal ratio, and slag basicity. Carbon pickup is increased by increasing the blast temperature and by increasing the oxygen injection percentage. The final carbon content can be altered by changing any one or a combination of blast control parameters, which affect the combustion process. Silicon is controlled in the cupola with the addition of ferrosilicon alloy and/or SiC briquettes. Maintaining the desired silicon content in the final iron requires accounting for the oxidation losses of silicon in the charge. Sulfur in gray iron is a naturally occurring element of melting in a cupola. It primarily comes from the sulfur contained in the coke. Sulfur is not detrimental below approximately 0.150 wt% if the manganese content is maintained at a sufficient ratio in relation to the sulfur content. Sulfur inhibits the conversion of carbon to ductile iron. Ductile irons are treated with a magnesium-containing alloy, which preferentially reacts with sulfur, forming inert MgS compounds. The small amount of magnesium remaining in solution in the iron promotes the nodular graphite shape. Thus, sulfur in ductile base iron should be below 0.030 wt% and preferably below 0.015 wt% for efficient conversion costs. Manganese is needed in gray iron at an amount of 1.7 times sulfur plus an additional 0.20 to 0.30 wt% to ensure that the sulfur is precipitated as manganese sulfide rather than iron sulfide. In pearlitic ductile iron, manganese is added as an alloy to produce a pearlitic matrix. In ferritic ductile iron and austempered ductile iron, manganese should be kept as low as possible to minimize pearlite in the as-cast structure and to minimize segregation, respectively. Phosphorus is controlled by using a variety of charge materials that are low in phosphorus. In gray iron, phosphorus contents should be below 0.10 wt% and even below 0.05 wt%, because high phosphorus levels are detrimental

to machinability and casting soundness in complex shapes. In ductile iron, phosphorus content below 0.05 wt% and preferably 0.03 wt% is desirable for maximum impact resistance and ductility. For low phosphorus levels, careful selection of charge material is required. Steel scrap is generally the lowest-phosphorus-content charge material. Residual Alloys and Tramp Elements. The effects of various inadvertent alloys and tramp elements should be understood, and control ranges and maximum levels should be established. Rapid analysis by modern spectrometers makes it possible to monitor any number of trace element levels. Any scrap source contributing excessive concentrations of detrimental elements should be removed from the cupola charge. Certain tramp elements, such as lead and bismuth, are detrimental to graphite shape (particularly the nodular graphite shape) and must be carefully monitored and maintained below the maximum level allowed. Titanium in trace amounts is beneficial to gray iron but is detrimental to the nodular iron, since it inhibits the formation of the nodular graphite shape. Inadvertent alloys, such as chromium, copper, tin, and molybdenum, if maintained within controlled ranges and combinations, can contribute to the pearlitic matrix structure of high-strength gray irons and high-strength pearlitic ductile irons. However, for ferritic ductile iron requiring the highest ductility, acceptable limits for these residual alloys are much lower.

Control Tests and Analyses The chill test was one of the earliest control tests developed and is still a good indicator of the graphitizing balance of an iron. A chill specimen is cast either as an elongated wedge in a sand mold or as a bar cast against a metal chiller on one side. After a few minutes, the test specimen is broken, and the depth of carbidic white iron chill indicates the balance between carbon, silicon, and other graphitizing elements against carbide-forming elements. This chill depth signals the need for more or less silicon, more or less carbon, and/or an adjustment in the postinoculation. Eutectic arrest methods were developed to provide carbon equivalent and carbon content as well as calculated silicon content. These thermal analysis techniques are rapid and make possible immediate corrective action, that is, adjustment of carbon and silicon. Optical spectrometer equipment greatly facilitates the control of iron chemistry, since all of the important elements can be determined within a few minutes. In large plants, samples are conveyed to the laboratory in pneumatic tubes, and analysis reports are conveyed to the melting center by phone or computer within a few minutes of taking the sample. Mechanical tests of several types are conducted to ensure the uniformity of iron properties and compliance with specifications.

Cupola Furnaces / 107 Tension tests of gray iron ensure that the iron meets the strength level required for the specified class of iron. In ductile iron, ultimate tensile strength, yield strength, and percent elongation are usually measured. Brinell hardness testing of castings monitors part-to-part and batch-tobatch uniformity and is an indicator of the general matrix structure. For ductile irons, microscopic evaluation of each treatment batch is essential to ensure graphite nodularity before castings reach the shakeout operation and lose their identity. Nodular graphite structure in a casting can also be verified with ultrasonic velocity test equipment. Ultrasonic velocity testing of all castings is performed by some companies to ensure the nodularity of safety-critical parts.

Specialized Cupolas Cokeless Cupola. In the 1970s, a cokeless cupola fired with gas or oil was developed in Europe and put into operation in several countries. It is a refractory-lined shaft furnace with a water-cooled grate supporting a bed of refractory spheres, as illustrated in Fig. 7. Heat for melting was generated by either gas or oil burners located below the grate and refractory bed. The iron melted in a cokeless cupola has low sulfur content and low carbon content, both of which are due to the absence of coke. The iron is carburized to a specified carbon content by injecting a fine carburizer (either petroleum coke or graphite) into the iron in the well. Pollution from the cokeless cupola is minimal and reportedly would meet environmental standards in some countries without expensive emission-control equipment. In countries where coke must be imported at a high price and oil or gas is readily available, the cokeless cupola is economically attractive. Production experience has been limited to small cupolas in the 4.5 to 9.1 Mg/h (5 to 10 ton/h) range. No practical experience has been reported on large cokeless cupolas. Attractive fuel (oil or gas) cost is available in some countries, but the high cost of refractory sphere replacements and the restrictions on the percentage of steel scrap used limit the attractiveness of this equipment in many markets. Plasma-Fired Cupola. In the 1980s, a pilot plant plasma-fired cupola was built, and a number of experimental heats were run, with some encouraging results. In the 762 mm (30 in.) inside diameter cupola, a single 2 MW plasma torch was projected into the cupola at tuyere level, as shown in Fig. 8. In the water-cooled torch, an electric arc from a high-amperage direct current is maintained across a gap between two copper sleeves, and the arc is rotated by a magnetic field. Air (or gas) is passed through the electric arc, where it is ionized and heated to very high temperatures unattainable with current hot blast systems. The experimental heats conducted indicated several advantages of using a plasma-fired cupola:

 Coke (fuel) could be considerably reduced    

due to the supplementing electric power from the torch. Lower-quality coke could be used; blast furnace coke, fine waste coke, and anthracite coal were melted successfully. Direct melting of borings and turnings proved practical because of lower air velocity and lower oxidation level. Silicon could be produced by injecting silica sand plus coke breeze. Iron could be produced by injecting as much as 20% iron oxide through the tuyeres.

The many possibilities of the plasma-fired cupola merited further exploration. Economically, the most attractive immediate application was the direct melting of iron borings or steel turnings. In March of 1989, the first commercial plasma cupola started production. This cupola was 3963 mm (156 in.) in diameter and had six conventional tuyeres. A 1.5 MW plasma torch was mounted to the end of each tuyere. Compressed air at 0.7 MPa (100 psig) was used as a plasma source. A booster compressor increased the air supply to 0.9 MPa (130 psig). The compressed air was passed through driers and filters and then to a distribution manifold. After being collected in the manifold, the air was dispersed into the six torches. The operating consumption of air was approximately 0.08 standard m3/s (160 standard ft3/min) per torch. Deionized water was necessary for the cooling of the torches. A conventional 700
C (1300
F) gas heat exchanger was used to preheat the blast air. The air at the tip of the plasma torch was approximately 1650
C (3000
F). The iron temperature at the top of the coke bed was approximately 1540
C (2800
F). The air blast volume was 6.0 to 7.0 m3/s (13,000 to 15,000 ft3/min) at the tuyeres. The development of the plasma cupola for melt operations was to offer melters the opportunity to use low-cost charge materials such as borings and steel turnings. This production plasma cupola was able to use an average of 40% loose borings/turnings in the charge makeup. The highest level achieved was 70%. This offered a significant cost-savings. The increased electric energy cost was partially offset by reduced coke usage.

Fig. 7

Cokeless cupola showing scrap charging, watercooled grates, burner location, and so on

ACKNOWLEDGMENT S.F. Carter, Melting Furnaces: Cupolas, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 383–392. SELECTED REFERENCES  Conference on Cupola Operation, American

Foundrymen’s Society, June 1980

 Cupola Design, Operation, and Control,

BCIRA, 1979

 Cupola Handbook, American Foundrymen’s

Society, 1975 and 1984

Fig. 8

Early version of an experimental plasma cupola used to confirm the concept of plasma heating as a melting technology

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 108-115 DOI: 10.1361/asmhba0005196

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Induction Furnaces Revised by Ralph Y. Perkul, Foundry Solutions & Design LLC

INDUCTION FURNACES have become the most widely used means for melting iron and, increasingly, nonferrous alloys as well. The key to the ready acceptance of induction furnaces has been its excellent metallurgical control coupled with its relatively low pollution operation compared to cupola furnaces. Induction furnaces are very efficient and are made in many sizes. Induction furnaces are relatively simple, and small quantities can be melted quickly. The melting time is relatively short, allowing metal to be delivered at small, regular intervals. A wide range of metals can be melted, although little refining of the metal is possible. Currently, induction furnaces are available in a wide variety of sizes. Coreless units range in capacity from a few pounds, favored by the precision cast metal producers, to 68 metric tons (MT) (75 tons) powered at 23,000 kW. Channel-type units have been built with a capacity of over 2200 MT (2400 tons) powered by 6 inductors rated at 6000 kW each for holding applications. Induction furnaces require much cleaner scrap than cupola furnaces and somewhat cleaner scrap than electric arc furnaces. The capital costs are higher than those of electric arc furnaces, but the operating costs are lower due to reduced refractory wear. In the past, large, high-powered induction furnaces operated at 60 Hz, the frequency of the incoming power. Thanks to major breakthroughs in electronic, solid-state frequency conversion techniques, it is now possible to build a highly efficient furnace operating at medium and high frequencies (70 to 10,000 Hz). Coreless units of 20,000 kW in power for ferrous metals and 7000 kW for nonferrous metals have been built in medium frequencies, and even channel-type units have recently been equipped with solid-state, medium-frequency power supplies. The advantage of medium frequency is that the power density of the furnace system can be increased substantially without increasing its size. Much more melting power can be applied, while maintaining the stirring action at desirable levels. Heat losses, which are a function of furnace size, are reduced, and

overall efficiency of the system is improved. The furnace is also easier to operate because a single potentiometer will typically provide efficient and stepless control of the power.

Types of Furnaces The basic principle of induction furnaces is that a high voltage in the primary coil induces a low-voltage, high current across the metal charge, which acts as a secondary coil. Because of electrical resistance in the metal, this electrical energy is converted into heat, which melts the charge. Once the metal is in its molten state, the magnetic field produces a stirring motion. The power and frequency applied determine the stirring rate. This is controlled to ensure complete melting of the charge and adequate mixing of alloy and fluxing materials, and to minimize temperature gradients in the charge. Excessive stirring, on the other hand, can increase lining damage, increase oxidation of the alloys, generate excess slag, and increase inclusions and gas pickup. There are two classifications of induction furnaces: coreless and channel. Cross sections of each are shown in Fig. 1 and 2. In a coreless furnace, the refractory-lined crucible is completely surrounded by a water-cooled copper coil, while in the channel furnace the coil surrounds only a small appendage of the unit, called an inductor. The term channel refers to the channel that the molten metal forms as a loop within the inductor. It is this metal loop that forms the secondary of the electrical circuit, with the surrounding copper coil being the primary. In a coreless furnace, the entire metal content of the crucible is the secondary. Although both channel and coreless induction furnaces employ similar electrical principles, the two types of furnaces are quite different in their capabilities and operation. The coreless furnace is more widely used for melting and superheating, whereas a channel furnace is better suited for superheating, duplexing, and holding. The channel furnace has been used by nonferrous foundries for many years. Recently,

however, this type of furnace has been increasingly replaced by the coreless and resistance furnace in aluminum-melting applications. It is still commonly used for melting copper alloys. Technological innovations in the channel furnace have been slower in developing than advances in the coreless furnace. Today (2008), medium-frequency power supplies are being installed on some new channel furnaces. Many older furnaces still rely on 60 Hz power supplies and use power-regulating autotransformers for control. Inherent in the channel concept is the fact that only the small amount of metal in the inductor channel receives energy through the surrounding coil. This necessitates superheating the metal in the loop and pumping it out of the channel into the main vessel, where it mixes with the cold metal and heats it by conduction. The ratio of heat transfer is limited, as is the stirring action in the main vessel. This is the primary reason this type of furnace has seen only limited service as a melter. The power that can be connected to a vessel is much smaller than for a similarly sized coreless vessel. The mild stirring action also makes it difficult to get alloy and carbon additions dissolved and mixed rapidly, particularly important in the production of synthetic irons. In addition, the channel furnace must be started with a supply of molten metal, or a melt-out form and a metal heel must be continuously maintained in the furnace. The furnace must not be shut down even for holidays or for other extended periods, because the metal will freeze, and considerable refractory damage will be incurred. The channel furnace is more economical to manufacture in large sizes and uses electric power more efficiently than a coreless furnace. If more power is desired, additional inductors can be connected to the main vessel. Several geometric shapes of vessels have been used. These include configurations that are described as vertical (upright cylindrical), drum (horizontal cylindrical), and low profile (spherical). Typically, all of these units are hydraulically tilted to permit the emptying of the contents.

Induction Furnaces / 109

Fig. 1

Components of a coreless-type induction furnace. (a) Operational elements. (b) Cross section showing water-cooled copper induction coil and key structural components. The entire molten metal bath (which serves as the secondary) is surrounded by the coil (the primary) that encircles the working lining.

Because the magnetic field is transmitted in all directions, special vertical laminations of transformer iron, which form the magnetic yokes (shunts), are evenly spaced around the circumference of the coil to provide additional strength and to pick up the stray flux that would otherwise heat up the furnace frame. Coreless furnaces are typically tilted about the spout through a 95 to 100 angle to empty their contents. The medium-frequency units can be completely emptied for alloy changes or factory shutdowns and quickly started up again with a cold charge. It is not necessary to maintain a molten heel or to use precast starter blocks, as is the case with the 60 Hz units.

Electromagnetic Stirring Fig. 2

A cross section of a channel-type induction furnace showing the water-cooled copper induction coil, which is located inside of a 360 loop formed by the throat and channel portion of the molten metal vessel. It is the channel portion of the loop that serves as the secondary of the electrical circuit, in which the copper coil is the primary.

The coreless furnace is completely surrounded by a spiral of copper tubing of a special cross section that features a water-cooling channel in its center. It must also provide good electrical coupling and strength to withstand substantial electromagnetic forces, yet with provision for thermal expansion.

When alternating current is applied to an induction coil, it produces a magnetic field, which in turn generates a current flow through the charge material, heating and finally melting it. The amount of energy absorbed by the charge depends on the magnetic field intensity, electrical resistivity of the charge, and the operating frequency. A second magnetic field is created by the induced current in the charge. Because these two fields are always in opposite directions, they create a mechanical force that is perpendicular to the lines of flux and cause metal movement, or stirring, when the charge is liquified. The mechanical force stays perpendicular to the

field only in the center of the coil; on both ends of the coil it changes direction. The metal is pushed away from the coil, moves upward and downward, and flows back. Figure 3 shows the four-quadrant stirring. It is this stirring that allows excellent alloy and charge absorption and aids in producing a melt that is both chemically and thermally homogeneous. The stirring is directly determined by the amount of power induced and is inversely proportional to the square root of the frequency. Therefore, the higher the power and the lower the frequency, the more intense the stirring for a given size furnace. The reasons for an increased interest in induction melting for aluminum applications are many, but perhaps most significant is the question of metal loss. Because of the inherent electromagnetic stirring action in the coreless furnace, charge materials quickly become immersed in the bath, minimizing oxidation losses. This effect is particularly pronounced when melting light-gage scrap. Thus, the growing emphasis on recycling favors induction melting as a cost-effective alternative to other metal melting processes.

Power Supplies 60 Hz Line Frequency. In order to meet the range of melting needs, the proper power supply must be selected. Because the vast majority of existing furnaces use line frequency, this arrangement is discussed first (Fig. 4).

110 / Melting and Remelting

Fig. 3

A cross-sectional view of a coreless-type induction furnace illustrating four-quadrant stirring action, which aids in producing homogeneous melt

Fig. 4

Modifying 60 Hz line frequency supplied by utility company to serve as power supply for induction furnaces

Protection of the furnace and the power system, phase balancing, and power correction are required in order to operate an effective installation. The typical 60 Hz (line frequency) power supply basically consists of:        

Primary switchgear Furnace transformer Mainline contactor Starting resistor and bypass contactor (for installations with off-load tap changers) Phase balancing system Power factor correction capacitor bank Instrumentation Controls and supervision

A primary transformation from the highvoltage grid of the local utility is usually

required because the voltage applied to the coil is between 500 and 3000 V. Because of its high inductance, an induction furnace coil has a low power factor. In order to bring the power factor to near unity, capacitors are installed and connected in parallel to the furnace coil. When starting a furnace, in order to reduce the inrush, a soft start is provided. When the transformer with an on-load tap changer is installed, the tap changer travels down to a lower tap whenever the furnace is switched off. This provides a start with a low voltage, resulting in low inrush current. When an off-load tap changer is applied, a starting resistor bank with bypass contactors is provided. Because the induction furnace is a singlephase furnace but its feed is commonly a three-phase line, a phase balancing system consisting of capacitors and reactors must be provided. In ferrous applications, refractory lining wear decreases the coupling distance between the coil and charge, resulting in a power increase. Because power input is limited, the voltage must be decreased to keep power from exceeding its maximum. This is typically done by using steps on a transformer tap changer. As the lining wears and the power increases, the voltage must be reduced by going to a lower tap. Additionally, in order to compensate for the amount of metal in the furnace and the condition of the refractory lining, a portion of the capacitor bank that is installed in parallel with the coil needs to be switched by adjusting the coarse and fine settings on the electrical panel. Medium Frequency. In the modern medium-frequency power supplies, (Fig. 5), many of the functions described are now performed automatically or have been eliminated altogether. The typical conversion is in an ac-dc-ac sequence. While frequency conversion entails electrical losses, modern power supplies incorporate solid-state frequency converters, and the increase in thermal efficiency usually outweighs the conversion losses. Modern converters exceed an efficiency of 97%. In the power conversion commonly used, the incoming 60 Hz ac power is rectified to dc and then chopped by an inverter to a higher frequency. Series and Parallel Inverters. Two types of inverters are typically used, namely, series and parallel. In each case, the furnace and compensating capacitor bank (fixed, in this case) form part of a resonance circuit. In the voltage series converter, the rectifier produces dc voltage, and the energy is stored in a filter capacitor. In a current parallel converter, the rectifier produces dc current, and the energy is stored in a smoothing reactor. The melting power for each of the two types of converters is also controlled differently. In the voltage series converter, the output is controlled entirely by variation of the inverter commutation frequency. The voltage is not controlled and is defined only by the line voltage.

In the current converter, the melting power is controlled by the variations of this inverter frequency as well as the magnitude of the dc voltage. The latter is regulated by variation of the conduction angle or phase control in the rectifier section. In both series and parallel circuits, it is possible to get power control that allows full power to be drawn regardless of the level of the charge in the furnace or the condition of the refractory lining. The midfrequency converters operate on a variable frequency. The load-commutated converter increases frequency as the charge contents of a furnace increase, while it simultaneously compensates for lining wear. No capacitor switching is required as in the line frequency units. In addition, a single potentiometer offers stepless and infinite control, eliminating the need for tap changer systems for power control. Modern induction furnaces take advantage of the latest technology, and many new installations are equipped with programmable controllers in lieu of relay logic. An even more recent advance has been the use of industrial computers for controlling the furnace and auxiliaries. With the variety of software that is currently available, the operator can choose screens such as systems data, automatic lining sintering, automatic cold starting of the furnace after a prolonged shutdown, automatic melting programs for various alloys, continuous calculated-temperature monitoring, fault detection, demand limiting, tabulation of energy consumption, melting cost data, metal inventory control, and chemistry adjustment. Also, by tying the computer to support systems such as charge make-up and hot-metal delivery, the utilization of the furnace can be increased. Packaging of Systems. Until the early 1970s, the electrical components of coreless induction furnaces were typically assembled on site in a concrete vault. Field construction costs made this installation method quite expensive. Modern component assembly techniques have allowed for the power supply to be packaged into steel modules at the equipment manufacturing plants. For furnaces up to 3.6 MT (4 tons) in capacity, current practice allows the manufacturers to completely assemble, wire, and pipe all components on a steel frame, to test, and subsequently to ship the unit as a single module, thus minimizing installation and start-up costs. For furnaces with power above approximately 1250 kW, a step-down isolation transformer from the main distribution voltage is normally used. Below 1250 kW, a 480 V distribution line backed by a disconnecting device can be used. The step-down transformers, when required, can be of the oil-filled outdoor type, which is the least expensive, or the indoor dry or silicon-filled type. Some manufacturers use the 480 V directly in their power supply; others choose to provide a step-up transformer and operate at the 575 V level.

Induction Furnaces / 111 In addition, with channel furnaces, standby generators are often employed to maintain cooling to the inductor.

Lining Material

Fig. 5

Typical medium-frequency induction power supply incorporating (a) a parallel inverter and (b) a series inverter

Water Cooling Systems Because the furnace coils and power supplies of the induction furnaces need to be cooled, a closed-loop water system is generally called for. In a medium-frequency power supply, the electronic components of the system contain a dc leg; it is therefore important to have clean, demineralized water circulating, typically with conductivity set at 50 mS (50 mmho) in order to prevent electrolysis from taking place. The heat exchanger to be used depends entirely on local circumstances such as climate, availability of raw water, and so on. The most widely used are the evaporative, the water-towater, and the air-to-water types. The latter has been frequently adapted to heat recovery systems. In those applications, the hot water from the furnace power supply and coil is piped to a location in the plant where heat is required. The water is circulated through a tube bundle, and a fan dispenses the warm air to the

surroundings. Considering that up to 10% of the power from a channel furnace and 20% of the power from a coreless furnace is lost to the water system, heat recovery systems for induction furnaces are often practical. In all coreless- and channel-type furnaces, the water is typically circulated through the power supply and then routed through the coil and on to the heat exchanger. Sometimes the water system is separated into two — one for the power supply and another for the coil. Usually, two pumps are provided, one as a standby with an automatic switchover, because it is extremely important to circulate water through the coil continuously to prevent it from being damaged. To protect induction furnace equipment during momentary power interruptions or prolonged power failures, manufacturers provide a special pressure-operated valve that, when activated, allows emergency deployment of city water to circulate through the coil. Such a system is shown in Fig. 6.

The selection of lining material is determined by the metallurgical requirements, the operating temperatures, and the type of operation. Refractories used in induction furnaces are oxides of minerals. Typically used are silica (SiO2), alumina (Al2O3), or magnesia (MgO) linings. From a chemical point of view, the silica is classified as an acid, alumina as a neutral, and magnesia as a basic material. Silica is the clear choice in iron melting, because it does not readily react with the acid slag typically produced in high-silica iron, has an extremely forgiving expansion curve, and is easy and economical to use. Alumina is the usual choice for aluminummelting furnaces, and steel-melting furnaces usually employ alumina or magnesia lining. Each of the refractory groups mentioned previously has a different thermal expansion characteristic, as shown in Fig. 7. As can be seen, alumina and magnesia have a nearly linear expansion. Silica, however, completes its expansion at approximately 815  C (1500  F) and remains constant at higher temperatures. With the selection of silica lining, the furnace can be shut down, cooled, and restarted without running the risk of metal penetrating thermal cracks. Magnesia is very sensitive to thermal shock and has the greatest expansion. Once cracked, this type of lining is not likely to seal itself, thus resulting in metal penetration. Alumina has approximately one-half the expansion of magnesia and is therefore less crack sensitive. It is important, however, to heat the furnace lining slowly when undergoing a cold start. This programmed start-up allows the cracks to seal before the metal adhering to the side walls melts and penetrates. This applies particularly to aluminum-melting furnaces because the metal melting temperature is relatively low. In the case of aluminum and some copper alloys, the oxide and metal buildup on the crucible not only poses a potential refractory penetration problem but also effectively increases the lining thickness and reduces the power that goes into the furnace. If not removed, this layer, which contains a small amount of metal, absorbs energy, and becomes superheated. The superheated metal can penetrate the lining and cause premature failure. For these reasons, the furnace crucible walls must be cleaned periodically. Crucible Wall Scrapers. The task of scraping the crucible wall has been made easier with the introduction of a mechanical scraper (Fig. 8), which cleans the sides in a full, hot furnace. The device essentially consists of a number of oscillating blades assembled on a steel frame. To start the operation, the scraper is placed on top of the furnace, and the

112 / Melting and Remelting

Fig. 6

A layout of a closed-loop cooling system showing routing of furnace-cooling water under both normal (black arrows) and emergency conditions (white arrows). Standby emergency pump is activated if the primary pump fails.

Temperature, ⬚F 32 2.0

392

752

1112

1472

1832

2192

2552 A

1.8 1.6

Linear thermal expansion, %

B 1.4

C D E F

1.2 1.0

Fig. 8

A cross section of a patented mechanical scraper used to clean the refractory lining of induction furnaces that melt nonferrous metals

G 0.8

H I K

0.6 0.4 0.2 0 0

100

200

300

400

500

600

700

800

900

1000 1100

1200

1300

1400

Temperature, ⬚C

Fig. 7

Thermal expansion curves of various refractory brick oxide materials used for linings in induction furnaces: A, magnesia; B, chrome magnesia; C, chromite; D, silica; E, zirconia; F, corundum 99; G, corundum 90; H, fireclay; I, sillimanite; J, zircon; K, silicon carbide

oscillating tools travel down the depth of the crucible. Radial pressure on the furnace is adjustable. The loose materials float to the top of the bath and are skimmed off. Ramming Mixes. Most induction furnaces operating in iron foundries are, as stated

previously, lined with a silica ramming mix. Alumina, magnesia, or zircon are typically used for melting steel and alloys. Alumina is typically used for aluminum. The current practice is to use granular materials. For example, the silica ramming mix is granular material combined with the sintering agent boron oxide (B2O3). The amount of boric

acid depends on the operating temperature and the wall thickness of the crucible. The flux or sintering agent fuses the silica particles to each other, forming a hard, glazed, ceramic surface. Ideally, the lining, when sintered, consists of one-third hard, sintered ceramic; one-third fritted material, in which fine particles are sintered and larger particles can still be identified; and one-third loose particles. This loose material has its purpose. When starting up a furnace from a cold start, the lining naturally expands. The loose material provides a cushion to handle the pressure. All granular materials are rammed on the furnace bottom using an electric vibrator equipped with a tamping plate. For the side walls, the lining material is spread behind a lining form that is vibrated with a pneumatic vibrator attached to it. For ferrous applications, the lining form is consumed in the first heat. For nonferrous applications, the lining form can be removed and reused. It may be tapered or collapsible.

Induction Furnaces / 113

Fig. 9

The principal components of a patented push-out device used to extract deteriorated refractory lining from coreless induction furnaces

The sintering process itself takes some time because it is necessary to heat the lining at the rate of approximately 110  C/h (200  F/h) until a temperature of about 30  C (50  F) above operating temperature is achieved. This temperature is held for 1 h before the furnace is ready for production. Lining Push-Out Device. When the linings are to be replaced, they can be chipped out with pneumatic hammers, and the lining procedure is repeated. A push-out device is available that eliminates the need to chip out the spent linings manually, saving both labor and time. The key components of this device, a schematic of which is provided in Fig. 9, are a locking plate that is bolted to the bottom of the furnace and a hydraulic ram assembly that is brought into position and latched to the plate. Once activated, the hydraulic ram moves forward, exerting pressure on a specially designed refractory cone installed in the bottom of the furnace. This distributes the force over the cross section of the lining. The push-out is aided by a slip joint between the coil grout and the working lining. The whole operation takes but a few minutes, with additional time being saved because the cool-down time is cut in half.

Melting Operations There are two distinct ways of operating a coreless induction furnace. One method is a batch operation, in which the entire contents of the furnace are emptied and the unit is

recharged with solids, typically using a conveyor or bucket charger. The other method involves a tap-and-charge operation, in which a portion of the furnace contents, typically one-third to one-half, is tapped and the identical weight in solids is recharged using a charge bucket or charge conveyor. Batch operation has now become feasible because medium-frequency power supplies allow the furnace to be started with ordinary scrap, rather than with starting blocks, which are required of 60 Hz (line frequency) furnaces. In terms of electrical consumption, the batchmelting method provides a 7% higher efficiency in iron-melting furnaces and a 4% higher efficiency in nonferrous-melting units over the tap-and-charge method. This is because the cold charge material couples with the magnetic flux more effectively than does molten metal. In addition to having greater electrical efficiency, a medium-frequency batch operation allows the drying of wet or oily charge material in the furnace, minimizing the need for a scrapdrying system. The tap-and-charge operation is still the most popular approach, however, since the chemistry is much easier to control. Less than one-half of the furnace contents are replenished at any one time, making alloying simpler because most of the furnace contents are already at the current specification. Finally, in order to use a batch operation effectively, the contents of the unit must be quickly and completely discharged, usually into a holding furnace or large ladles. This option may not be possible if small amounts of molten metal need

to be taken out at defined times and the capital and operational costs of a holding furnace cannot be economically justified. In either method of operation, it is particularly important to use the proper power supply efficiently and keep the cover closed to minimize potentially high radiation losses. Utilization is calculated in terms of the time that the power is on. It is obvious that power cannot be on constantly, because certain operations, such as charging, slagging, sampling, and tapping, are best and most safely done while the power is off. The objective is to keep the power on at least 75% of the time. Sometimes, a second furnace added to the power supply improves utilization, because the power can be switched from one furnace to the other while the above-mentioned operations take place. Utilization can also be improved by quickly charging and tapping to minimize waiting time and heat losses as well as by increasing the charge size. Today (2008), induction power supplies can operate two or more furnaces at the same time without the use of mechanical switches. These designs allow the system utilization to go to over 90%. Although medium frequency has made the need for heavy charge materials obsolete, it is still good practice to choose materials that have a reasonably high bulk density, are relatively clean and dry, and are not too long. In order to avoid bridging, the length of the pieces should be no more than two-thirds the diameter of the crucible. If longer pieces such as casting gating systems and structural steel sections are available, they should be broken or cut first. A further aid to utilization efficiency is the recent method of conveyor charging. Made possible by furnace covers acting as close-capture hoods, the furnace lid is kept up for much of the charging cycle. The charging is done continuously, adding more solid charge material as the contents of the furnace liquefy and collapse, allowing room for additional material. This adds safety to the often hazardous charging operation by removing the operator from the proximity of the furnace and allows control to take place from the melt shop control room. There are a variety of conveyor charging concepts. They are usually vibrating steel decks that can swivel between furnaces or extend-retract and move laterally between furnaces. Dry charge material is best suited for melting operations. Oils on the charge surface generate fumes from the furnace, which may need to be exhausted. Moisture on the charge also can be a problem, because a steam explosion can result. If dry materials are not readily available, a charge dryer is often employed to drive off any moisture remaining on the charge. The most common dryer consists of a refractorylined hood equipped with gas burners, which heat the solids on a vibrating conveyor by means of forced convection of the hot gases through the material. Figure 10 shows the various major components of the system. It is

114 / Melting and Remelting

Fig. 10

The major components of a dryer used to preheat induction furnace charge material. Integrated scrap preheat process combines (1) weigh hoppers, (2) preheat hood, (3) material transfer mechanism, and (4 and 5) furnace-charging apparatus into a single automated process.

Fig. 11

A mechanical clamshell-type power skimmer used to remove slag before each tap of an induction furnace

important to get the preheated charge into the furnace quickly to take advantage of the thermal energy in the material and to reduce the electrical energy output required for melting. The average temperature used for drying is typically 315  C (600  F); for preheating, it is 540  C (1000  F). When applied properly, the preheating of the charge material can increase the melt rate by 10 to 15%. Direct impingement

Fig. 12

Robotic attending of furnace. This robot can slag the furnace, take the bath temperature, as well as sample the metal.

Induction Furnaces / 115 of the flames on the charge material, particularly if it is thin, should be avoided because oxidation losses can occur. When dirty material is charged into the furnace, slag forms from the dirt and rust in the scrap as well as from the foundry return sand. Slag is undesirable in induction furnaces because it absorbs energy and erodes the lining. Therefore, it is removed each cycle with a skimmer to prevent slag accumulation. Mechanical Skimmers. Furnace operators can increase the power supply utilization by the use of mechanical skimmers (Fig. 11). This is particularly important in large furnaces because mechanical slagging units can minimize the time that power is either turned off or reduced for skimming during each charge-to-tap cycle.

A power skimmer can easily boost the output and productivity of the melting furnace. A skimmer is typically mounted on a jib. This tool has a long handle, which keeps the operator farther away from the hot furnace. The best and most effective type of slagger is the clamshell type. The skimmer is positioned over the furnace and lowered into the slag. With an automatic raking motion, the jaws of the unit close around the slag, which is then lifted clear of the bath and deposited into a container. The blades are typically dipped into a slurry to reduce slag adhesion during the next charge-to-tap cycle. Today (2008), most larger coreless units use back-slagging, where the furnace is back-tilted to allow the slag to be more easily removed.

A further development has been introduced recently that makes use of an industrial six-axis robot for the first time. In this innovation, the operator, from the safety of the melt deck control room, opens the cover and deploys the robot to skim the furnace, using preset programs (Fig. 12). As an added benefit, this system allows the operator to remotely obtain bath temperature, take samples, and add alloys. This further adds safety to the melting operation. ACKNOWLEDGMENT Updated from: R.Y. Perkul, Melting Furnaces: Induction Furnaces, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 368–374

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 116-123 DOI: 10.1361/asmhba0005200

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Vacuum Induction Melting MELTING UNDER VACUUM in an induction-heated crucible is a tried and tested process in the production of liquid metal. It has its origins in the middle of the 19th century, but the actual technical breakthrough occurred in the second half of the 20th century. Commercial vacuum induction melting (VIM) was developed in the early 1950s, having been stimulated by the need to produce superalloys containing reactive elements within an evacuated atmosphere. The process is relatively flexible, featuring the independent control of time, temperature, pressure, and mass transport through melt stirring. As such, VIM offers more control over alloy composition and homogeneity than other vacuum melting processes. Vacuum induction melting can be used to advantage in many applications, particularly in the case of the complex alloys employed in aerospace engineering. The following advantages have a decisive influence on the rapid increase of metal production by VIM:
Flexibility due to small batch sizes
Fast change of program for different types of

the melting process. Accordingly, the vacuummelted superalloys (compared to EAF/AODmelted alloys) are improved in fatigue and stress-rupture properties. Control of alloying elements also may be achieved to much tighter levels than in EAF/ AOD products. However, problems can arise in the case of alloying elements with high vapor pressures, such as manganese. Vacuum melting also is more costly than EAF/AOD melting. The EAF/AOD process allows compositional modification (reduction of carbon, titanium, sulfur, silicon, aluminum, etc.). In vacuum melting, the charge remains very close in composition to the nominal chemistry of the initial charge made to the furnace. Minor reductions in carbon content may occur, and most VIM operations now include a deliberate desulfurization step. However, the composition is substantially fixed by choice of the initial charge materials, and these materials are inevitably higher-priced than those that are used in arc-AOD.

The charge generally consists of three portions:
A virgin portion, which consists of material

that has never been vacuum melted


A refractory portion, which consists of those

virgin elements that are strong oxide formers and have the tendency to increase the Crucible Power supply

Induction coil

To vacuum pumps

steels and alloys


Easy operation
Low losses of alloying elements by oxidation
Achievement of very close compositional

tolerances


Precise temperature control
Low level of environmental pollution from

dust output


Removal of undesired trace elements with

high vapor pressures


Removal of dissolved gases, for example,

hydrogen and nitrogen Vacuum induction melting is indispensable in the manufacture of superalloys. Compared to air-melting processes such electric arc furnaces (EAF) with argon oxygen decarburization (AOD) converters, VIM of superalloys provides a considerable reduction in oxygen and nitrogen contents. Accordingly, with fewer oxides and nitrides formed, the microcleanliness of vacuum-melted superalloys is greatly improved compared to air (EAF/AOD)-melted superalloys. Additionally, high-vapor-pressure elements (specifically lead and bismuth) that may enter the scrap circuit during the manufacture of superalloy components are reduced during

Process Description A VIM furnace is simply a melting crucible inside a steel shell that is connected to a highspeed vacuum system (Fig. 1). The heart of the furnace is the crucible (Fig. 2) with heating and cooling coils and refractory lining. Heating is done by electric current that passes through a set of induction coils. The coils are made from copper tubing that is cooled by water flowing through the tubing. The passage of current through the coils creates a magnetic field that induces a current in the charge inside the refractory. When the heating of the charge material is sufficient that the charge has become all molten, these magnetic fields cause stirring of the liquid charge. The optimal induction coil frequency for heating the charge varies with the charge shape, size, and melt status (liquid or solid). Older equipment used a single frequency, but newer power supplies are able to be operated at variable frequencies and are adjusted throughout the melt to obtain the most rapid heating/melting conditions.

Fig. 1

Basic elements of a vacuum induction melting furnace

Shunt Heating coil

Brick crucible

Cooling coil Ground detection

Fig. 2

Schematic of vacuum induction melting crucible (shell, coil stack, backup lining, and working lining)

Vacuum Induction Melting / 117 Bulk charger Crucible (pouring)

Melting chamber

Cover (removed) Launder Tundish car

Operating platform

Mold chamber

Electric room Power supply

Molds

Shop floor

Mold car

Fig. 3

Vacuum system

Schematic of a top-opening, double-chamber vacuum induction melting furnace

Table 1

Typical refractories used to line vacuum induction melting crucibles Maximum melt temperature

Refractory

MgO Al2O3 MgO-spinel Al2O3-spinel ZrO2 Graphite



C

1600 1900 1900 1900 2300 2300



F

2910 3450 3450 3450 4170 4170

Refractory density g/cm3

lb/in.3

Resistance to thermal shock

2.8 3.7 3.8 3.7 5.4 1.5

0.101 0.134 0.138 0.134 0.195 0.054

Good Good Poor Relatively good Poor Excellent

solubility of oxides and nitrides in the virgin charge
A revert (or scrap) portion, which consists of both internal and external scrap that previously has been vacuum melted Vacuum-melted scrap has already had its gas content reduced to levels consistent with vacuum production. Scrap, however, has the possibility of having become contaminated during the production process, and care (expense) must be taken in the segregation and preparation of scrap materials for vacuum melting. In most VIM furnaces there is a vacuum lock bulk charger located directly over the crucible (Fig. 3). Charge material may be added to the heat through the bulk charger while melting is in process in the crucible. The material to be added is placed in bottom-opening buckets, placed in the bulk charger, and the charger is evacuated. The valve isolating the charger from the melt chamber is opened, and the bucket is lowered to a point close to the crucible top and the bottom opened so as to drop the charge material into the crucible. In constant operation, if a furnace is not to be opened to the atmosphere, all charge material for a heat will be added by this process.

Applications

Superalloys, high-quality Superalloys, high-quality Superalloys, high-quality Superalloys, high-quality Superalloys, high-quality Copper, copper alloys

steels steels steels steels steels

Older VIM furnaces may have been designed as single-chamber systems with the mold put inside the furnace before the beginning of the melt. The molten charge is then poured into the mold inside the furnace. Single-chamber furnaces thus must be opened after each heat to extract the molds and put in the new molds. Most furnaces have some system of large vacuum locks for transferring the prepared molds into the melt chamber. In double-chamber furnaces (Fig. 3), there is a separate chamber for the molds. The molten metal is transferred via launders (refractory-lined steel troughs) to a refractory tub (tundish). Some systems are designed to pour directly from the crucible into the tundish. The tundish contains a considerable volume of metal and allows residence time for entrained slag to float to the top of the tundish and be removed from the pour stream. The pour stream exits the bottom of the tundish. The pour time is regulated by pour temperature and the nozzle diameter of the tundish. The typical tundish is designed to provide a low-velocity flow path from the point of entry of the metal to the bottom nozzle. Refractories. Crucibles with a capacity of approximately 4500 to 22,500 kg (10,000 to 50,000 lb) are generally built up from refractory brick. Smaller furnaces, used for production of

master melt, may use single-piece crucibles. Refractory brick linings are usually two layers. The backup lining protects the induction coil in the event of a failure of the outer or working lining. The working lining is the primary interface with the metal and is replaced when erosion of the lining becomes excessive. Refractory life is also affected by the expansion of the refractory during the repeated melting cycles. Refractory brick is chosen with regard to resistance to erosion and expansion. Commercially available refractory brick may be incompletely sintered and expands during use, causing loss of crucible integrity. The refractory material used for the crucible lining is based on oxides such as Al2O3, MgO, CaO, or ZrO2 (Table 1). The lining is almost always rammed and sintered; prefabricated brick is used in larger furnaces. Dried silicate, combined with small oxide additions, appears to be very suitable for crucible lining because of its thermal characteristics. Because of an irreversible thermal expansion of 8% above 1000  C (1830  F), a high densification of the lining takes place during sintering. For this reason, this active lining is suitable for foundries. The behavior of the lining refractory with regard to stability at high temperature under vacuum must also be considered. The melting crucible material is not inert and is actually another source of oxygen and other impurities, depending on refractory type and condition. Therefore, both melt refining temperature and refining duration are carefully scrutinized. Proper melt stirring is integral to the deoxidation process and must be optimized through proper furnace power frequency and application procedure to prevent refractory lining erosion, a potential problem particularly during the controlled but more vigorous CO boiling portion of the process. Process Sequence. Figure 4 shows a typical process profile for the VIM of nickel- and cobalt-base superalloys. Before operation or at the completion the preceding heat, the melt chamber is isolated from the mold chamber and the vacuum integrity of the furnace is verified. Specific practices with regard to vacuum measurements will differ in detail. The deterioration rate of the vacuum is a measurement of the inherent vacuum integrity of the furnace. This integrity cannot be measured by vacuum alone, because the large pumping capacities of the pumps used in VIM are able to achieve low vacuum pressures even with significant leakage into the furnace. It is considered bad practice to continuously draw air into the vacuum furnace and across the melt. Immediately after ensuring that the furnace is vacuum-tight, the virgin portion of the charge is placed into the VIM furnace first. This may be done by opening the furnace or, more commonly, by charging the furnace through hoppers lowered through a large vacuum lock (bulk charger) located over the crucible. The furnace is capable of quickly pumping down to or maintaining vacuum levels below 100 mm (and often into the

118 / Melting and Remelting 105

1000

Installed power 1200

104

100 Pyrometer temperature

900

1200 1100

10

1.0

1000 900 800

2732 10 Melting power Pressure rise test

800

1.0

700 Casting end Casting start

600 500

All liquid

400

Additions/Sampling Pressure rise test

300 0.1

200 100

0.01

0

0.1

2552

2192 2012 1832 1652

0.01

1472

10−3

392

Additions Sampling

200

212

100 0

2372

Pyrometer temperature, ⬚F

1300 Melting power, kW

100

1400

1000

Pyrometer temperature, ⬚C

Melt chamber pressure, Pa

103

1100

Melt chamber pressure, mbar

1500

1. Charge 0

3. Charge

2. Charge 1

Melting

2

3

4 Refining

5

6 Superheating

10−4

7

32

Casting

Time, h

Fig. 4

Typical vacuum induction melting protocol for nickel- and cobalt-base superalloys

<10 mm) range. The virgin material is melted by application of current to the induction coils surrounding the refractory crucible. When the virgin material has been completely melted (all molten), it outgasses. The outgassing is monitored until it is complete. The outgassing is a response of the gas in solution in the melt to the low partial pressures of gases in the vacuum chamber. Some degassing is accomplished because carbon in the charge will form CO with the oxygen and also be evolved from the melt. The progress of degassing is followed by measuring the leak-up rate at set time intervals. When a constant leak-up rate is obtained, this indicates that degassing (refining) of the melt is complete. After degassing is complete, the reactive and revert charges are added to bring the charge to its planned weight. A ladle sample is taken after all additions are complete. Based on the analysis of this sample, trim additions are made to bring the melt into a very precise compositional range. Because there are no ongoing chemical changes in the melting, as there are in EAF/AOD, the compositional requirements of a melt may be met as closely as allowed by the reproducibility of chemical analysis. After the trim additions have been made, the temperature of the heat is brought precisely to the desired point, and the heat is poured. The heat, although produced in vacuum, will still have generated significant amounts of slag from the products of deoxidation, desulfurization, and the deterioration of the refractory crucible lining.

Process Controls. Process computerization and automation continue to enhance better reproducibility of the melts and computer modeling. Using charge and alloy calculations, it is easy to achieve the required chemical analysis with minimal cost. For this application, in the ideal case, the computer is linked to the analysis system computer so that the additives can be calculated and, if necessary, weighed and added directly after the analysis. If this alloy calculation is laid out as a charge calculation, the charges and the amounts of scrap necessary can be optimized. The complete calculation system enables a calculation of the alloying elements, starting with the amount of scrap needed and proceeding to the necessary alloy addition at the end of the refining period. A computer-controlled mass spectrometer system, specially developed for the VIM process, can make a significant contribution to the optimization of the melting process and its economics. Using a computer evaluation of the gas composition in the furnace chamber, information about the state of the degassing process and of the chemical reactions in the melt can be continuously obtained. The melting process can thus be controlled more exactly. Process parameters that can be closely controlled include:
Determination of the correct time for the

addition of chemically active elements and the sequence of addition of such elements


Determination of the process steps, such as

completion of the refining period and the suitable time for tapping
Advance detection of leaks in the furnace chamber, cooling water system, or hydraulic system
Monitoring of the state of degassing of the crucible refractory lining Remelting by VIM for Shape Casting. Vacuum furnaces for precision casting are used to remelt alloys that have already been treated in vacuum. The remelted alloy is then precision cast into preheated ceramic molds. Heating and melting in such furnaces almost always takes place by induction, although some electron beam melters and vacuum arc furnaces are used. Electron beam furnaces offer the advantage of ceramic-free melting, but they do not permit a sufficiently accurate and reproducible temperature control for charges of more than approximately 5 to 7 kg (11 to 15 lb). Vacuum arc furnaces are used for titanium precision casting. Remelting with VIM furnaces for precision casting is done with two-chamber furnaces (Fig. 5a), where the melting crucible and mold are in separate chambers. The mold chamber is preferably placed under the melting chamber so that, in connection with another chamber for the charge or melting material, semicontinuous operation with minimal cycle times is possible. For cobalt- or nickel-base superalloys, the melting chamber pressure is of the order of 0.01 Pa

Vacuum Induction Melting / 119 (104 mbar, or 7.5  105 torr), while the mold lock would be evacuated to approximately 5 Pa (0.05 mbar, or 0.038 torr) within 1 min. In addition to the casting of equiaxed, directional solidification, and single-crystal blades for gas turbine engines, vacuum precision casting can be used for large parts (melt weights up to 1000 kg, or 2200 lb) or for the large-scale production of small parts. Examples of large parts produced using vacuum precision casting include compressor sections of jet engines and structural parts for stationary turbines.

A typical small casting produced by this method is a turbocharger wheel for automotive engines (Fig. 5b). Automotive turbocharger wheels weighing 400 to 500 g (0.9 to 1.1 lb) are produced in automatic precision casting machines. Total cycle time, including 40 s for melting, is typically 90 to 100 s. In the production of such small parts, the crucible is often integrated into the mold. Casting yields of 80 to 90% are obtained. Filters. One of the most critical stages with respect to cleanliness is the pouring of the melt.

Ceramic foam filters are used in some master metal operations to remove relatively large melt inclusions by means of entrapment. Foam filters are most effective where extremely high pour rate conditions and gross cleanliness problems prevail. Filter performance often varies because of the occasional use of filters with poor mechanical strength and/or thermal shock resistance. Foam cell particle breakage often results from handling during shipment or tundish installation and, if undetected, results in filter particulate in the alloy bar stock and subsequently cast components. Optimized VIM technology and practice without filters provide a clean alloy, without the inherent risks associated with filter use when applied to master metal production.

VIM Metallurgy

(a)

Fig. 5

(b)

Shape casting with vacuum induction melting, (a) Computer-controlled vacuum furnace with mold chamber. (b) Precision-cast turbocharger wheels for automotive engines. From left: mold with integrated crucible, bar stick, cast part, machined turbocharger wheel

Fig. 6

Vacuum induction melting is often done as the primary melting operation followed by secondary melting (remelting) operations such as electroslag remelting (ESR) and/or vacuum arc remelting (VAR).Various processing routes following VIM are illustrated in Fig. 6. Some superalloys are produced by a triple-melt sequence (VIM/ESR/VAR), where the VIM ingot is typically referred to as an electrode for subsequent ESR and/or VAR operation. The casting weight can vary from a few kilograms to 30 Mg (33 tons), depending on whether the VIM furnace is being used for

Potential processing routes for products cast from vacuum induction melting (VIM) ingots or electrodes. VAR, vacuum are remelting; ESR, electroslag remelting; EB, electron beam; HIP, hot isostatic pressing. Source: Ref 1

120 / Melting and Remelting precision casting or for the production of ingots or electrodes for further remelting. There are many different metallurgical factors that arise with induction furnace melting. The crucible material has an extraordinary effect on the metal/slag reaction because the ceramic outer wall reacts with the liquid metal and with the slag. In a VIM furnace, slag would be transported to the crucible wall by the characteristic bath movement. The result is that the slag solidifies at the wall and therefore has an insufficient reaction with the melt. It is therefore beneficial to use a slag that is more or less saturated with the oxides of the crucible lining in order to minimize heavy slag attack of the lining. In contrast to the ladle metallurgy processes with EAF, the crucible wall lining is susceptible to significantly higher erosion than the brick wall of a ladle. For this reason, metallurgical operations such as dephosphorization and desulfurization are limited. The metallurgy of VIM is primarily limited to the pressure-dependent reactions, such as carbon, oxygen, nitrogen, and hydrogen, and the evaporation of undesired elements with high vapor pressures, such as copper, lead, bismuth, tellurium, and antimony. Cleanliness can be significantly improved if a reactive liquid slag capable of absorbing oxides and sulfides is in contact with the melt. Vacuum induction furnaces are generally not operated with active slags. Therefore, reaction products can precipitate only on the crucible walls, and the melt may not be as clean as with other processing methods. Because the various alloys produced in vacuum induction furnaces must meet the highest quality requirements and because the vacuum induction furnace is primarily a consolidation unit and only secondarily a refining unit, the following methods are used to produce clean melts:
Selection of a more stable refractory mate-

rial for the crucible lining


Rinsing of the melt with inert gas
Minimizing the contact time of the melt in

the crucible


Exact temperature control to minimize cru-

cible reactions with the melt

Fig. 7

Vacuum induction refining process


Suitable deslagging and filtering techniques

during pouring
Conception of a suitable tundish and launder system for good oxide removal For particular applications, however, the quality of the material produced by VIM will not be sufficient to satisfy the highest quality requirements with respect to cleanliness and primary structure. In this case, the VIM-produced material must undergo secondary melting processes (Fig. 6). VIM Refinement Process. The primary process of melt refinement is the removal of meltcontained oxygen by means of a reaction with carbon to form carbon monoxide (CO). The reaction occurs most readily at or near the melt surface, with the reaction kinetics being affected by crucible geometry and melt stirring. The CO bubbles form along the walls and, sometimes, bottom of the melt/lining-refractory interface (Fig. 7). This occurs preferentially at small crevices existing in the lining, with the bubbles growing during movement toward the molten metal/vacuum interface (Ref 2). Actual bubble formation is dependent on the number of gas molecules present, the pressure in the liquid at the level of the bubble, the temperature of the gas, and, for very small bubbles, the interfacial tension between the gas and the liquid metal. Following formation, bubble growth and mass transport within the liquid toward the liquid/vacuum interface is dependent on:
The quantity of the dissolved gas
The decreased pressure exerted on the bub-

ble as it rises in the melt


The bath temperature
The time it takes for the bubble to rise

through the melt to the surface, which, in turn, is a function of melt stirring
The pressure above the melt
The interfacial tension between the bubble and the liquid metal The relatively vigorous, but controlled, portion of the boiling process results in the greatest CO removal. Concurrently, a slight nitrogen

loss is realized because of scavenging associated with the CO bubbles, and a slight sulfur reduction may occur during the CO supersaturation stage via sulfur dioxide (SO2) evolution. Minor tramp elements such as lead, silver, bismuth, selenium, and tellurium are partially evaporated during this period as well as throughout the entire refining process (Ref 3). Some undesirable elements, however, such as arsenic and tin, must be controlled through raw materials selection because they are not removed by vacuum refining. Bath refining from a somewhat vigorous CO boil is undertaken at a temperature and duration long enough to reach the so-called system equilibrium conditions, the assurance of which is provided by the attainment of consistent furnace leak-up rates. At this point, there is the addition of the refractory charge, which includes elements with reactivity toward oxygen (for example, aluminum, titanium, zirconium, and hafnium) that were withheld from the virgin charge because of their reactivity. Revert charge is also added. Once the boiling subsides, surface desorption of additional CO occurs, and it is during this nonboiling period that nitrogen removal (desorption) is most effective (Ref 4). Trace Element Removal. The removal of undesired volatile trace elements, such as arsenic, antimony, tellurium, selenium, bismuth, and copper, from the vacuum induction furnace is of considerable practical importance. These elements must be held to very low concentrations to avoid the risk of premature part failure, particularly for the production of superalloys for critical applications, such as jet engine parts. Figure 8 shows the influence of some trace elements on the stress-rupture properties of alloy 718. Because of the high vapor pressures of most of the undesirable trace elements, they can be kept to very low levels by distillation during melting under vacuum. Figure 9 shows how the different trace elements behave under vacuum. In a nickel-chromium melt, arsenic, antimony, and tin cannot be reduced over the gas phase, while copper, lead, selenium, tellurium, and bismuth can be reduced to a level far below 10 ppm. Nitrogen Degassing. Nitrogen gas can be decreased because its solubility at constant temperature is directly proportional to its partial pressures. As long as no strong nitride-forming elements are present in the melt, there is no difficulty in reducing the nitrogen content to 20 ppm (Ref 7–9). When nitride-forming elements such as chromium, vanadium, aluminum, and titanium are present, the activity of the nitrogen is very low; therefore, removal of nitrogen under high vacuum is difficult. If low nitrogen contents are desired in alloys containing nitride-forming elements, raw materials low in nitrogen must be used. Figure 10 shows the reduction in nitrogen levels obtained in the VIM of a die steel. From an average level of approximately 400 ppm at the beginning of treatment, the nitrogen has

Vacuum Induction Melting / 121 200

99.9 99.5 99

Silver Concentration, %

Rupture life, h

Selenium 120 Lead 80

Frequency, %

160

97.5 95 90

10

40

Fig. 0

10

20

30

50

40

60

70

80

Effect of trace elements on the stress-rupture properties of alloy 718. Test conditions: 650  C (1200  F), 690 MPa (100 ksi). Source: Ref 5

(0.5)

99.9 Arsenic Tin Frequency, %

0.1

Concentration, %

Antimony

Selenium 0.01

99.5 99 97.5 95

After treatment With argon bubbling

90

Without argon bubbling

80 70 60 50 40 30

Before treatment

20

Copper

10 0

100

200

300

400

500

Nitrogen content, ppm Bismuth

Lead

10−3 Tellurium (3⫻10−3)

0

20

40

60

80

100

120

Time, min

Fig. 9

Evaporation of trace elements from Ni-20Cr melts under vacuum. Source: Ref 6

been reduced to a level of 50 ppm. Additional argon purging during melting of the steel did not improve nitrogen removal. In the case of iron-nickel melts, however, the additional argon purging reduces nitrogen to much lower values. Hydrogen Degassing. Like nitrogen, hydrogen can be decreased via the gas phase because its solubility is directly proportional to its partial pressure. The solubility of hydrogen at atmospheric pressure and at 1600  C (2910  F) in iron or nickel melts is in the range of 30 ppm. During VIM, the hydrogen can be removed to very low concentrations. Figure 11

Before treatment

0

1.0

2.0 3.0 Hydrogen, ppm

4.0

5.0

11

Reduction in hydrogen content of X38CrMoV51 die steel after vacuum induction degassing

90

Impurity content, ppm

Fig. 8

80 70 60 50 40 30 20

Bismuth

0

After treatment

Fig. 10

Reduction in nitrogen content of X38CrMoV51 (Fe-0.38C-5.2Cr-1.3Mo-0.4V-1Si-0.4Mn) die steel after vacuum induction melting processing

shows the hydrogen contents of a die steel before and after vacuum treatment. It is evident from Fig. 11 that final hydrogen contents below 1 ppm can be routinely achieved. The reduction in hydrogen during the degassing treatment amounts to approximately 80%. Argon purging had no perceptible influence. The bath agitation in the case of this furnace, which is operating at normal frequency, is sufficient for hydrogen removal. Similar results were also obtained for the VIM processing of superalloys. Deoxidation of the melt in the vacuum induction furnace can be done via the generation of CO gas (C + O ! CO). The removal of oxygen from the melt as CO is favored by decreased melt chamber pressure, elevated bath temperature, and increased carbon activity (Ref 10). Proper melt stirring is integral to the deoxidation process and must be optimized through proper furnace power frequency and application procedure to

prevent refractory lining erosion, a potential problem particularly during the controlled but more vigorous CO boiling portion of the process. The final attainable oxygen content for a given carbon content is directly proportional to the CO partial pressure. The theoretical equilibrium values for oxygen are below 1 ppm in carbon-containing iron or nickel melts at 1600  C (2910  F) and at a pressure of 0.1 Pa (103 mbar, or 7.5  104 torr). However, the actual values are up to 1 order of magnitude higher. These higher oxygen contents result from impurities in the crucible lining, crucible outgassing and leakage, and, above all, the fact that the reaction does not reach equilibrium with decreasing oxygen content because of the difficulty of CO nucleation. In addition, the hydrostatic pressure of the liquid metal in the melt must be taken into account because only a relatively small bath surface area is in direct contact with the prevailing vacuum pressure. However, the influence of the liquid-metal hydrostatic pressure can be minimized through the use of additional agitation effects. The CO reaction occurs in two stages (Ref 1). The first stage is boiling, that is, the formation of CO bubbles within the melt along with a strong bath agitation as a result of this gas formation. The second stage is desorption, in which no more CO bubbles form inside the melt, and CO formation takes place only at the bath surface. Deoxidation with carbon is also a decarburization reaction, which is used for the production of low-carbon high-chromium steels. By using the CO reaction under vacuum, very low carbon contents can be achieved without a noticeable loss of chromium. The oxygen content in such alloys is higher because of the decrease in oxygen activity due to chromium. In this case, precipitation deoxidation using aluminum, silicon, or titanium must be applied. The CO partial pressures at which the lining oxides are reduced by carbon in the melt are:

122 / Melting and Remelting Carbon monoxide partial pressure Lining material

CaO ZrO2 MgO Al2O3 SiO2

kPa

0.04 0.13 0.53 0.53 81.1

torr

0.3 1.0 4.0 4.0 610.0

Source: Ref 1

It is evident that CaO represents the stable crucible material, while SiO2 is reduced at relatively high pressure. Apart from that, SiO2 can be reduced by alloying elements such as manganese, chromium, aluminum, titanium, or zirconium. This can cause a heavy chemical erosion of the lining, accompanied by an undesired silicon pickup in the melt. Because of these undesired reactions with a silica lining, a spinel-forming basic refractory material is used for the melting of high-grade steels and superalloys. The spinel formation during sintering leads to volume growth; therefore, as with the oxidic material, a densification of the rammed lining takes place. The vacuum induction degassing and pouring (VIDP) furnace (Fig. 12) is a variation of the conventional VIM furnace. The VIDP furnace design employs a modular concept that allows for connecting it to casting units for ingot casting, horizontal and vertical continuous casting, or powder production. Because of the smaller volume of the VIDP furnace compared to the VIM furnace and significantly lower desorption and leakage rates, it is possible to obtain very low pressures with lower pumping capacity. The lower part of the furnace can be decoupled and replaced rapidly; therefore, the VIDP furnace enables faster replacement of different furnace vessels with changes of alloy.

Melting Bath Agitation. As noted, melt agitation is an important process factor. Agitation is caused by the induction coil itself, depending on the power input and the installed frequency. Stirring is directly proportional to the amount of induced power and inversely proportional to the square root of the frequency. Therefore, a more intense stirring for a given size furnace occurs at a higher power and a lower frequency. Bath agitation from the induction coil is basically limited to the meltdown period. It can be done with the induction coil alone only if there is a high-power input and therefore strong bath heating. This high temperature increase is not always metallurgically desirable. For this reason, two agitation systems are offered in stateof-the-art vacuum induction furnaces:

vacuum. This ensures that the admitted purging gas flows through the melt and does not take the path of least resistance through the lining of the crucible. Both agitation processes offer advantages with regard to higher reaction rates, correct temperature adjustment, homogenization with simultaneously lower erosion of the crucible lining, and better cleaning of the melt. Figure 14 shows the three different agitation methods operating in a typical VIM process sequence. The agitation effect at 200 to 500 Hz in the melting phase assists in shortening the melting time. In the refining and superheating periods, it is better to use electromagnetic agitation at 50 to 60 Hz or argon purging through a porous plug.


Electromagnetic agitation with an additional

coil

Production of Nonferrous Materials

tom of the crucible

Apart from melting high-grade steels and superalloys, VIM is being increasingly used for the production of nonferrous metals and alloys. Table 2 shows some examples for possible use in nonferrous metallurgy. Aluminum alloys with additives such as zirconium, titanium, beryllium, cerium, tellurium, and cadmium must be melted under vacuum or under inert gas atmosphere because of their high reactivity with air and, in some cases, their toxicity. Aluminum-lithium alloys are also candidates for VIM processing. Copper Alloys. The production of highpurity copper having less than 2 ppm O can be accomplished only in a vacuum induction furnace. Oxygen content influences the electrical conductivity of copper alloys; the lower the oxygen content, the higher the electrical conductivity (Ref 11). For the production of


Agitation by argon purging through the bot-

The difference between the electromagnetic agitation by a separate coil and the agitation caused by the electromagnetic forces of the main coil lies in the fact that the bath can be more vigorously stirred without a temperature increase. A second transformer would be needed. Another method of bath agitation is derived from the principle of argon purging usually applied in ladle metallurgy. Argon purging has a long history of use in atmospheric melting furnaces. A characteristic of this purging is that the porous plug is not in direct contact with the liquid melt. Instead, the plug is covered with refractory material identical to that used for the crucible lining (Fig. 13). The basic difference is that the porous plug is in direct contact with the melt and is suitable for operating under

Crucible lining

Exchangeable porous plug

Basic set

Fig. 12

Schematic of the vacuum induction degassing and pouring furnace

Fig. 13

Argon purging system for vacuum induction melting furnaces

Vacuum Induction Melting / 123 oxygen-free copper, melting and casting must be carried out under vacuum. Selective Evaporation of Alloying Elements. The use of vacuum metallurgy is primarily linked with degassing and decarburization. A side effect of these treatments is the evaporation of

elements with high vapor pressures. In nonferrous metallurgy, this effect is used for the distillation of metals—for example, for the separation of lead and zinc in lead refining, in zinc production, and for the reduction of magnesium and nonalkali metals. Similarly, copper can be refined

from copper scrap by using the vacuum to evaporate volatile elements such as lead and zinc (Ref 12). ACKNOWLEDGMENTS Portions of this article were adapted from:
A. Choudhury and H. Kemmer, Vacuum

Melting and Remelting Processes, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 393–425
M. Donachie and S. Donachie, Superalloys: A Technical Guide, ASM International, 2002
G.L. Erickson, Polycrystalline Cast Superalloys, Properties and Selection: Irons, Steels, and High-Performance Alloys, Vol 1, ASM Handbook, ASM International, 1990 REFERENCES

Fig. 14

(a) Melting and (b) stirring modes of the vacuum induction melting process

Table 2 Possible applications for vacuum induction melting in the processing of nonferrous metals and alloys Metal or alloy

Metallurgical results

Aluminum, aluminum alloys High-purity metals

Low oxygen content, effective removal of oxides Oxygen-free copper (2 ppm O)

“New” alloys

Al-Mg, Al-Ce, Al-Zr, Al-Hf alloys; copper alloys with beryllium, cobalt, titanium, zirconium, and lithium; avoidance of lithium losses in production of Al-Li alloys Production of calcium, barium, lithium, strontium, and magnesium; recycling of hard metals; purification of alloys or melts; removal of zinc and lead

Metal production and recycling

Achieved by

Vacuum tank degassing without additions of polluting gases or materials Melting and pouring under steady, controlled atmosphere or vacuum; gas stirring; carbon as reducing agent Furnace and casting system under inert gas atmosphere; vacuum-tight, highly automated furnace

Thermic/alumino-thermic vacuum distillation; leaching with zinc and subsequent vacuum distillation melting; vacuum distillation

1. J.W. Pridgeon et al., in Superalloys Source Book, American Society for Metals, 1984, p 201–217 2. D. Winkler, Thermodynamics and Kinetics in Vacuum Metallurgy, Vacuum Metallurgy, O. Winkler and R. Bakish, Ed., Elsevier, 1971, p 42–54 3. Evaporation of Elements from 80/20 Nickel-Chromium During Vacuum Induction Melting, Transactions of the Vacuum Metallurgy Conference, American Vacuum Society, 1963 4. V.M. Antipov, Refining of High-Temperature Nickel Alloy in Vacuum Induction Furnaces, Stal0 , Feb 1968, p 117–120 5. W.B. Kent, Int. Voc. Sci. Technol., Vol 11 (No. 6), 1974, p 1038–1046 6. P.P. Turillon, in Transactions of the Sixth International Vacuum Metallurgy Conference (Boston), American Vacuum Society, 1983, p 88 7. G.A. Simkovich, Int. Met., Vol 253 (No. 4), 1966, p 504–512 8. H. Katayam et al., in Proceedings of the Seventh International Conference on Vacuum Metallurgy (Tokyo), The Iron and Steel Institute of Japan, 1982, p 933–940 9. A. Choudhury et al., World Steel and Metalworking Manual, Vol 9, 1987–1988, p 1–6 10. D.R. Gaskell, Introduction to Metallurgical Thermodynamics, Scripta Publishing, 1973, p 268–273 11. O. Kamado et al., Method of Producing Electrical Conductor, European Patent 0121152, 1986 12. J.G. Kru¨ger, Proceedings of the Fifth International Vacuum Metallurgy Conference (Munich), 1976, p 75–80

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 124-131 DOI: 10.1361/asmhba0005201

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Electroslag Remelting ELECTROSLAG REMELTING (ESR) is a secondary refining process (Fig. 1), where a consumable electrode is melted by immersion in a molten flux (slag) consisting primarily of fluorspar (CaF2). The heating is generated by an electrical current (usually ac) that flows from the base plate through the liquid slag and the electrode. Oxides may be added to the slag to raise the resistivity (so as to make melting more efficient electrically) or to modify the slag chemistry with regard to its reactivity with the metal being melted. Electroslag remelting is similar to vacuum arc remelting (VAR), except that ESR is carried out at normal atmospheric pressure and has a greater melting rate than VAR. When the tip of the electrode is melted by the molten slag, metal droplets fall through the liquid slag and are collected in the water-cooled mold (Fig. 1). The slag not only shields the metal pool from contaminating gases but can also be so chosen that it acts to segregate and absorb impurities from the molten metal in its passage through the slag from the electrode to the molten pool. During the formation of the liquid film, the metal is refined and cleaned of contaminants, such as oxide particles. The high degree of superheat of the slag and of the metal favors the metal/slag reaction. Melting in the form of metal droplets greatly increases the metal/slag interface surface area. The intensive reactions between metal and slag refine the melt. The remaining inclusions are very small and are evenly distributed in the remelted ingot. The absence of macrosegregation or heterogeneity in the distribution of nonmetallic inclusions in ESR ingot improves yield to justify the additional cost of remelting. In addition, the very clean and smooth surface of the ESR ingot helps to reduce production costs, because surface conditioning before hot working is not necessary. Another special feature of the ESR process, as in VAR, is the directional solidification of the ingot from bottom to top. The macrostructure is marked by an extraordinarily high density and homogeneity as well as by the absence of segregations and shrinkage cavities. Electroslag remelting is commonly used to produce the highest levels of quality in plate steels, particularly in thick plates. The homogeneity of the ingot results in uniform mechanical properties in the longitudinal and transverse directions after hot working. For steels, the slag

composition can be adjusted to serve as a desulfurizer or dephosphorizer as well as a reservoir for floating inclusions. Sulfur can be lowered to below 0.002%. The cleanliness and homogeneity of ESR ingots results in steels with excellent mechanical properties and superior toughness (Fig. 2) compared to conventionally manufactured steel. In superalloys, ESR is

Fig. 1

often used as an intermediate remelt in a socalled triple-melt process (VIM-ESR-VAR), where the ESR electrode is produced by vacuum induction melting (VIM) and then remelted a third time by VAR after ESR (Ref 4). The theory of ESR was developed in the 1930s and was the subject of a U.S. patent in

Schematic of an ESR furnace with multiple electrodes for large ingot production

Electroslag Remelting / 125 1945 (Ref 5). However, it took more than 30 years for a general breakthrough on a production scale (Ref 6–12) in primary (mill) products. Currently, ESR metallurgy is a well-established production technology (Ref 13). Global production capacity of ESR steel is over a million (106) tons, which includes low-alloy highstrength steel, bearing steel, tool and die steel, and stainless and heat-resistant steels (Ref 14). Ingots from high-quality heavy ESR are over 200 tons, and production cost of heavy ESR manufacturing is significantly reduced by electroslag hot topping. Related technologies include electroslag casting, electroslag turning casting, Boher electroslag tapping, and electroslag centrifugal casting. Process variations of ESR technology include vacuum ESR and pressurized ESR. Electroslag remelting under vacuum (VACESR) is a process variation that has received some attention, because the traditional ESR process is essentially atmospheric and thus ineffective in reducing dissolved gases such as hydrogen and nitrogen. Vacuum ESR is not a superalloy production technology, given the predominance of triple-melt (VIM-ESR-VAR) for superalloys. Pressure electroslag remelting Temperature, ⬚F −4

Impact strength, J

80 60

32

68 104 140 176 212 73.8 ESR (tangential rim) 59.0

ESR (axial core)

44.3

40

29.5

20 Conventional (tangential rim)

0 −40 −20

20

0

40

Conventional (axial core)

60

Impact strength, ft-Ib

−40 100

is a process variation that increases dissolved nitrogen in the melt of austenitic steels for strengthening. Another ESR process variation is electroslag rapid remelt (ESRR). This technique allows production of smaller cross-sectional shapes (rounds and squares), which can be directly processed into smaller shapes on conventional open-die presses and rolling mills, saving conversion time and costs as compared to starting with typical larger round ingots. The ESRR technology is also being developed for nucleated casting (Ref 15), which is a type of spray casting.

ESR Furnaces Major components of modern singleelectrode ESR plants are shown in Fig. 3 (with fixed mold) and Fig. 4 (with a retractable bottom plate). In the case of a retractable bottom plate furnace, current feedback does not take place through a closed copper pipe; instead, it is accomplished with four symmetrically arranged current feedback tubes. The ESR process, like VAR, is a fully automated operation. Unlike a VAR crucible, the ESR crucible is normally self-contained in that the inside and outside shells through which the cooling water runs are assembled as a single piece. A watercooled stool is assembled to the shell to form the bottom of the crucible. A starter or striker plate is usually enclosed in the junction between the crucible shell and the crucible stool.

For crucibles to be used with hot slag starts, the crucible may contain an opening at the bottom (a “mousehole”) through which the molten flux would be introduced. The top of the crucible does not mate to the rest of the furnace, so it operates exposed to air. The electrode is attached to the ram through a stinger (Fig. 5). The ESR stinger is welded to the electrode outside the furnace. The stinger is often tubular, because only the periphery of the electrode is involved in the welding process. The top of the ESR furnace contains the ram drive and load cells. It is connected electrically to the crucible stool through, typically, four vertical bus bars located at 90
intervals around the crucible. The ESR head commonly is able to be translated in the horizontal direction by an XY drive, which allows centering of the electrode in the crucible prior to the start of the melt. (Note that XY capability is inherently different in ESR compared to VAR.) The power supply is most often ac. Most ESR furnaces have two melting stations (crucible setups). While an electrode is melting in one station, the next station is prepared for melting. When one melt is finished, the ESR head is rotated in the horizontal plane to be located over the

14.8

0 80 100

Temperature, ⬚C (a)

167

212

257

3822

in.

122

3500 ESR (0.002% S)

3185

2800

2548

Fracture toughness, ksi

Fracture toughness, MPa

m

Temperature, ⬚F 77

4200

1911

2100 Conventional (0.001% S)

1400

1274 637

700 25

50

75

100

125

Temperature, ⬚C (b)

Fig. 2

Comparison of properties of steel rotor forgings made from ESR and conventionally melted ingots. (a) Impact strength of grade X22CrMoV121. (b) Fracture toughness of grade 30CrMoNiV511. Specimen orientation and location are indicated next to curves. Source: Ref 1–3

Fig. 4 Fig. 3

Basic design of a modern ESR furnace with a fixed mold. 1, ram drive system; 2, electrode ram; 3, XY adjustment; 4, load cell system; 5, sliding contact; 6, four bus tubes; 7, pivoting drive; 8, electrode; 9, mold assembly; 10, coaxial bus tube; 11, base plate; 12, multicontacts

Schematic of an ESR furnace with retractable bottom plate. 1, electrode drive system; 2, load cell system; 3, stinger crosshead; 4, two ball screws; 5, three bus tubes; 6, mold assembly; 7, ingot; 8, ingot withdrawal table; 9, three bus tubes; 10, three ball screws; 11, three sliding contacts; 12, base plate; 13, furnace head drive; 14, electrode; 15, three sliding contacts; 16, XY adjustment

126 / Melting and Remelting second crucible/electrode stinger setup. The ESR process is also capable of using multiple electrodes, which can be melted into a single mold. Electrode preparation is not as critical a part of quality ESR operations as for VAR. Oxides on an electrode surface are incorporated into the slag. Thus, electrodes for ESR are generally not ground. An exception to this is for production of premium-quality material where the end-user specifications require grinding to ensure the removal of potential iron-mold pullout on the electrode surface. Also, the shrinkage cavity at the top of the electrode is not a problem, if the electrode is melted top down. Thus, superalloy ESR electrodes are not commonly hot topped in VIM or cropped prior to melting. There are two possible start-up scenarios: cold start and hot start. In cold start, the slag and small particles of the alloy to be melted (usually machining chips) are placed on the starter plate. The electrode is touched into the slag-alloy mix and backed out to establish an arc. High melt power is used in this start-up phase. The arc melts down both the slag and the metal particles, at which point the electrode becomes immersed in the slag, and the melt process shifts to melting of the electrode surface by the slag. It is necessary that the start-up power be sufficiently high that actual welding (melt-in) occurs between the embryonic ingot and the starter plate. This is required so that the electrical conduction path will be predominantly through the electrode, then the slag to the ingot, from the ingot through the starter plate, and then to the stool. In hot slag starts, the slag is melted externally by electric arc in graphite crucibles. The molten slag is introduced, generally through a bottom mousehole, into the crucible. The electrode is lowered into the slag, and melting

Fig. 5

ESR electrode with stinger

commences. Although high initial melt power is not required to melt slag and metal starter material, a high-power profile start-up is generally used to ensure the melt-in of the ingot to the starter plate (to ensure good electrical conduction paths). ESR Control Parameters. Control of the ESR process is considered from two different aspects: control of pool depth and shape (control of solidification structure) and development of a satisfactory ingot skin. Pool depth and shape are controlled primarily by melt rate (power input). Electroslag remelting generally is run using melt rate control, with continuous amperage adjustments to maintain a uniform melt rate. Due to the high thermal inertia of the ESR process, the melt rate of ESR is more variable than in VAR. Secondary control of pool depth and shape is achieved by maintaining a uniform, shallow immersion of the electrode in the slag. The passage of molten metal droplets through the slag affects slag resistivity and thus causes variation in the operating voltage of the system. The operating voltage appears as a band of values due to this variation. The width of that band is called the volt swing. The electrode immersion is proportional to the volt swing, and the ram drive of the ESR advances the electrode to maintain the desired swing value. Production of satisfactory ingot skin is not actively controlled during the melt but is due to the choice of variables. The most important of these are melt rate, depth of the slag pool, electrode immersion depth in the slag (analogous to arc gap in VAR), ingot and electrode diameter (as in VAR), and choice of slag composition. Higher melt rates produce better ingot surfaces. Because higher melt rates also produce deeper molten pools and move the process closer to the conditions for freckle formation, selection of the correct balance of parameters is critical. The choice of ingot and electrode diameter defines the clearance (annulus) between the crucible and the electrode. Insufficient clearance is generally regarded as a problem in processes using cold slag starts, in that the slag-chip mix may get hung up in the annulus. However, a tight annulus may result in more uniform heat distribution across the molten pool. There is no published information to support either viewpoint. Thus, annulus becomes, much more than in VAR, a parameter dictated by the individual producer’s philosophy. Electrode immersion is generally thought of as the depth of penetration of the electrode into the slag. There is no definitive method for measuring the immersion. In superalloy melts, observation of interrupted melts suggests that the meniscus of slag reaching up the side of the electrode will generally not exceed 6.4 mm (0.25 in.) (Ref 4). Thus, schematics of the ESR process are misleading when they imply significant immersion into the molten slag. However, the degree of immersion, as affected by the ram drive, can produce drastically

different electrical signals. The quality of the melt has been found to be responsive to these signals. Thus, while there is no realistic understanding of what depth of immersion means physically, the parameter that controls it can be demonstrated to have a significant effect on melt quality. The electrode immersion plus the thickness of the slag cap may be considered as the resistance (ac impedance) in the circuit. Changing the weight of slag in the system or the slag composition changes the impedance of the circuit and thus changes the amperage and voltage required to maintain a given power input (for constant melt rate). The slag cap thickness is the primary driver, while electrode immersion may account for 10 to 20% of the total impedance. Measurement of the changes in this resistance or of the corresponding changes in voltage may be used to control the immersion of the electrode in the slag. This suggests that the voltage and impedance changes associated with electrode immersion are actually the impedance of the circuit associated with the interfacial resistance between the electrode melt surface and the slag surface. It thus should be realized that in ESR of superalloys, the electrode immersion is always extremely shallow, and, in fact, the most common electrode positioning is that of “skittering” on the top of the molten slag cap. The degree of this skittering is commonly controlled by the repetitive change in voltage in the process (volt swing). The electrode melt rate is the other important parameter to be controlled. The melt rate is dependent on the melt current. Although some VAR processes are run in constant current rather than in melt rate control, ESR processes are exclusively melt rate control. Melt rates are generally calculated using rolling averages. Load cells are subject to the generation of false signals and to errors in the absolute value. Thus, load cell design and maintenance are important factors in maintaining the quality of ESR. In ESR, the melt current responds to changes in the melt rate and attempts to hold the melt rate constant at the set point. Because of the presence of the slag, which must be heated or cooled, in the system, the response time of ESR to the melt current change is much slower than that of VAR. A fourth factor controlling melt quality and unique to ESR is the choice of slag and the volume of slag used. The resistance of the slag cap depends on both the resistivity of the slag and the thickness of the slag cap through which the melt current must travel. An additional consideration is the degree of chemical reactivity of the slag with elemental components of the electrode. Figure 6 is a nominal plot of process parameters for the remelting of a nickel-base alloy. Figure 7 shows voltage swings and the major control parameters for a hypothetical ESR melt. The melt begins with a high-amp start-up in amperage control and, at the completion of the start-up, is placed into melt rate control.

Electroslag Remelting / 127 6

Melt rate, kg/min

Ingot diam: 460 mm Ingot weight: 2070 kg

Set rate 3.5 kg/min

4

6.6

3 2

4.4

Start

1 Power control 0 5 Power, kW ⫻ 10−2 Voltage, V ⫻ 10−1 Current, kA ⫻ 10−1

8.8

2.2

Hot topping

Melting rate control

Melt rate, Ib/min

11.0

5

0 Power control

4

Voltage

3 Power 2 Current

1 0

 Ensuring consistent remelting over a period 0

1

2

3

4

5

6

7

8

9

10

Time, h

Fig. 6

Electroslag Remelting of Heavy Ingots. At the end of the 1960s, the concept of using ESR plants to manufacture large ingots weighing up to 350 Mg (385 tons) gained acceptance. Increasing demands for energy and the trend toward larger electrical power-generating units required cast ingots weighing 100 Mg (110 tons) and more for the manufacture of generator and turbine shafts. With ESR, it is possible to achieve higher yields and to avoid such internal defects as porosity, macrosegregation, and the accumulation of nonmetallic inclusions. An early example of a large ESR furnace in the 1970s had a retractable bottom plate and manufactured ingots 2300 mm (90.5 in.) in diameter, 5000 mm (197 in.) in length, and weighing up to 160 Mg (176 tons) (Ref 2) (Fig. 8). Remelting of large ingots of this size in retractable bottom plate furnaces demanded the solution of specific technical and metallurgical problems, such as: of several days without interruption

 Prevention of steel and slag breakout when

remelting with a retractable bottom plate

 Achieving a good ingot surface  Ensuring directional solidification to avoid

ESR process parameters for the melting of a nickel-base alloy

macrosegregation and shrinkage cavities

 Controlling the steel and slag compositions

Volt swing - Ram travel (in.) - Resistance Voltage swing and resistance

over the entire ingot height

Ram

 Adjusting for low hydrogen content  Adjusting for a low aluminum content

iion Posit

Volt Swing Resistance

Voltage

Melt rate - Voltage - Amps 40 35 30 25 20 15 10 5 0

Voltage

Melt Rate Amperage

Elapsed time during melt

Fig. 7

Melt trace for ESR showing variation of melt current, melt rate, ram travel, slag resistance, voltage, and volt swing

Note that the change in voltage is the result of the deliberate change in amperage. Formation of a shrinkage cavity several hundred pounds deep into the ingot can occur from solidification of the large molten pool after the end of the melt, if the electrode is completely consumed or if the power abruptly shuts off. This cavity must be removed during subsequent processing. To minimize the amount of material that must be cropped because of the end-of-

melt shrinkage, it is common to step down the melt current as the end of melt approaches and leave a small amount of electrode. Unlike VAR, this biscuit is generally not used to put electrical power into the melt. Rather, it simply prevents radiation loss from the slag. Because of the presence of the slag as a heat source at the top of the ingot, ESR hot topping practices are not as extended as are those for VAR.

(<0.010%) for rotor steels; this has a decisive influence on the creep-rupture properties of the rotors After these problems had been solved, heavy rotors for electrical generators could be manufactured from ESR ingots. Interior defects such as macrosegregation, shrinkage cavities, and nonuniform distribution of inclusions can be avoided in large ingots only if directional solidification is ensured over the entire ingot section. By maintaining the correct melting rate and temperature of the slag, directional solidification can be achieved for large ingot diameters. Accordingly, the ESR ingot is free of macrosegregation in spite of the large diameter. Several electrode changes are necessary to produce such large ingots. In the case of a multi-electrode operation in a single mold, the formation of the ingot can be concluded by a hot topping procedure performed by withdrawing all but one of the electrodes and continuing the process in single-phase mode with the remaining electrode centralized in the mold, for example, by appropriate sideways displacement of the latter on a wheeled support. In such a case, loss of bottom contact to the ingot during the multiphase operation could prevent the subsequent single-phase hot topping step from proceeding. Moreover, loss of contact is more likely to occur with such a multiphase operation than with a single-phase operation, because in the latter the flow of current between the ingot bottom and the mold base has some tendency to keep a thin slag film

128 / Melting and Remelting

Fig. 8

160 Mg (176 ton) ESR ingot before forging into a generator rotor

Fig. 9

Schematic of solidification relationships in VAR (left) and ESR (right) melting processes. Source: Ref 16

between the ingot and mold base in an electrically conductive state. The refining process for producing steel ingots of large sizes (e.g., 200 to 400 ton class) is possible by controlling the rate of solidification in electroslag remelting or with electroslag hot topping techniques.

Ingot Solidification The solidification structure of an ingot of a given composition is a function of the solidification rate and temperature gradient at the

solid/liquid interface. When remelting a consumable electrode in a water-cooled copper mold, the solidification rate is constant because of the constant heat transfer through the cooling water; therefore, a relatively high temperature gradient at the solidification front must be maintained during the entire remelting period to achieve a directional dendritic structure. The structures in the solidified ingot are governed by the same principles as those governing a solidifying VAR ingot. However, the nature of the heat transfer (through an oxide skin) and the presence of a heat reservoir (the molten

slag) at the top of the ingot/molten pool make ESR structure inherently different from that of VAR. Figure 9 schematically illustrates the relationship between the ESR electrode, the solidifying ingot, and the shape and depth of the molten pool. A major feature of ESR solidification is that a “skin” of oxide is incorporated between the ingot surface and the crucible wall. Similar to VAR, the distance between the side of the electrode and the crucible wall is the annulus. The depth of the slag pool is an important operating parameter that controls the electrical resistance of the system. The depth of immersion of the electrode into the slag is an important parameter in controlling the molten pool shape. As shown in Fig. 9, the heat extraction in ESR is of a much greater magnitude than for VAR. Due to the insulating characteristic of the ingot slag skin, heat extraction in ESR is inherently less efficient than in VAR. Thus, ESR is a process with high thermal inertia and slow responses (change in pool shape) that occur in response to changes in power input. The presence of an ingot slag skin not only improves the ingot surface quality but also, more importantly, eliminates the presence of a solute-lean shelf layer on ESR ingots. (As oxide particles from the electrode are dissolved into the slag, they do not agglomerate on the ingot surface.) Because ESR ingots do not have an ingot shelf, they are inherently free of the formation of drop-in (discrete) white spots from this source, but drop-in electrode may still be a source of white spots in ESR. Solidification white spots have not been reported in ESR product. However, one consequence of the ingot slag skin is that heat extraction in ESR is less efficient than for VAR. This is shown schematically in Fig. 9 by the difference in pool shape between the two processes. ESR pools tend to be both deeper and more V-shaped (compared to a U-shaped VAR pool). Deep pools favor freckle formation, which can only be counteracted by reducing ingot size. Consequently, the maximum size of ESR ingot produced (for any given alloy system) is smaller than one that can be produced by VAR. Figure 10 schematically illustrates the structure from an ESR molten pool (the relative sizes of annulus and immersion are not to scale). Again, the most important feature shown is the depth of the molten pool (where the depth and angle of the liquid-solid zone also controls freckle formation in superalloys). Compared to VAR, few studies have been done on the nature of the molten pool in ESR. The pool differs fundamentally from a VAR pool in that the molten slag cap provides a source of heat to the system, and the slag skin on the solidifying ingot provides insulation, reducing heat extraction. Consequently, for comparable melt rates and crucible diameters, an ESR ingot will have a deeper, steeper-sided pool. Compared to VAR, ESR ingots have traditionally been regarded as being more prone to the formation of positive segregation effects. However, even though the ESR pool may inherently be more

Electroslag Remelting / 129 prone to positive segregation, that shape may be altered by judicious choice of melting parameters. Even in the case of directional dendritic solidification, the microsegregations increase with increasing dendrite arm spacing. For optimal results, the objective should be a solidification structure with dendrites oriented as close to parallel to the ingot axis as possible. However, this is not always possible. To achieve a good ingot surface, there is a minimum energy requirement, and therefore a minimum melting rate, which is a function of ingot diameter. This means that the melting rate for large-diameter ingots cannot be maintained for axis-parallel dendrite growth. Like VAR, the melting rates for different steels and alloys are a function of ingot diameter. Ingot Defects. In spite of directional dendritic solidification, remelted ingot may include various defects, such as the formation of:  Tree ring patterns  Freckles  White spots

These types of defects also occur in VAR ingots and are discussed in more detail in the article “Vacuum Arc Remelting” in this volume. In Fig. 10, the schematic for the edge of a solidifying ingot indicates a processing advantage for ESR that has made it desirable to develop this process for use in critical components. The insulating nature of the slag

creates conditions at the edge of the ingot that do not produce a shelf. Thus, with no arc instabilities of concern and no shelf to be undercut, the formation of discrete white spots in ESR product could only occur by some infrequent occurrence, such as drop-in of pieces from an unsound electrode. Positive Segregation. Alloys that have highdensity interdendritic liquids may form unique structures in ESR. The high-density liquid tends to flow downhill toward the center of the ingot. Although channel defects may not be formed, large pools of high-solute alloy may be formed. These regions may form in widely separated parts of an ingot and be separated by “normal” material. Because of the greater number of variables inherent in ESR, the underlying causes for such segregation have not been determined. The nature of these areas and their effect on properties have not been reported. Electroslag Remelting Ingot Surface. The interaction of the molten metal with the slag cap should be understood with regard to obtaining good surface on ESR ingots. As shown in Fig. 10, the molten slag solidifies against the water-cooled crucible side just as the molten metal would. The molten metal pool, in addition to being V-shaped (containing the mushy zone), also contains a fully molten straightsided component or metal head. The metal head remelts some of the solidified slag on the crucible wall, so a thinner slag layer is then contained between the ingot and the crucible wall. Should this slag skin be locally penetrated by the metal head, a thin stream of metal will be driven through the hole. Such a metal fin or bleedout is illustrated in Fig. 11. Note that the penetration of the slag is through a circular region that is essentially the size of a dime.

Metal head

Liquid metal penetration of slag

Ni balls Metal fin

Fig. 11

Fig. 10

Longitudinal schematic of the structure developed in an ESR ingot during melting

Longitudinal section through a metal fin that was formed by a bleedout on the surface of an ESR nickel superalloy ingot. Note: The circular, darketching features are nickel balls introduced in an experiment to delineate the shape of the molten pool. They are not related to the bleedout mechanism.

The molten metal not only moves down the ingot surface in response to gravity but is forced upward from the penetration due to the combined hydrostatic pressure of the molten metal head and the molten slag cap. Bleedout formation is sensitive to melting conditions and slag composition. While there is no published association of bleedouts with detrimental structure, there is an economic factor involved, in that an ingot with bleedouts will require grinding prior to forging or other hot working.

Process Variations Several variations of ESR have been developed: remelting under high pressure, known as pressure electroslag remelting; remelting under reduced pressure; and electroslag rapid remelting. Pressure Electroslag Remelting. The strength of austenitic steels can be increased by alloying with nitrogen as long as the nitrogen is interstitially dissolved in a stable condition. Nitrogen solubility, however, is limited in the case of melting under atmospheric pressure for a given steel composition. Because the solubility of nitrogen in iron is proportional to the nitrogen partial pressure in the furnace atmosphere, there is the possibility of alloying nitrogen into the melt under higher pressure and allowing it to solidify. The ESR process is suitable for this technique. The first trials in this area were carried out in 1971 (Ref 17). For some years, a high-manganese steel has been successfully remelted under a pressure of 4.2 MPa (610 psi). Up to 0.5% interstitial solubility of nitrogen in steel has been achieved. The nitrogen-alloyed austenitic steels are used for the manufacture of retaining rings in generators. Electroslag remelting under vacuum (VAC-ESR) is another interesting process. Alloy 718 has been remelted in a VAC-ESR pilot plant with a fluoride-free slag based on lime and alumina (Ref 18). Because the remelting is carried out under vacuum, problems of oxidation of the melt do not arise. In addition, dissolved gases, such as hydrogen and nitrogen, can be removed. Thus, the advantages of both ESR and VAR are combined in one process (Ref 18). Nonetheless, VAC-ESR is not a production practice for superalloys. Electroslag rapid remelt is a process technology where electrodes are ESRed into smaller cross-sectional shapes (rounds and squares) that can be directly processed into smaller shapes on conventional open-die presses and rolling mills, saving conversion time and costs as compared to starting with typical larger round ingots.

Steel ESR There are a number of publications that deal primarily with improvements in the physical properties of remelted steel (Ref. 6–12).

130 / Melting and Remelting However, they discuss not only the directional solidification of the metal but also the influence of liquid, superheated, and therefore reactive slags. The extent and direction of the metallurgical reactions in the ESR process are determined by the steel composition, the slag used, and the atmosphere used (inert gas, air, etc.). The principal feature of the ESR process is the slag bath. A continuous transport of liquid metal through the slag takes place. During this transport, the slag and the metal compositions change, according to the kinetic and thermodynamic conditions. To perform its intended function, the slag must have some well-defined properties, for example:  Its melting point must be lower than that of

the metal.

 It must be electrically efficient.  Its composition should be such that the

desired reactions, such as removal of sulfur and oxides, are ensured.  It must have suitable viscosity at remelting temperature.

of easily oxidizable elements, such as aluminum and silicon, during remelting. To counteract this oxidation, the slag should be continuously deoxidized, preferably with aluminum. With proper slag composition and

Table 1 Compositions of slags commonly used in ESR Composition, % Slag designation(a)

CaF2

CaO

MgO

Al2O3

SIO2

Comments

100F 70F/30

100 70

... 30

... ...

... ...

... ...

70 70 50 70

20 15 20 ...

... ... ... ...

10 15 30 30

... ... ... ...

Electrically inefficient; use when oxides are not permissible. Difficult starting; high conductivity; use when aluminum is not allowed; risk of hydrogen pickup

40 60 80 60 ...

30 20 10 10 50

... ... ... 10 ...

30 20 10 10 50

... ... ... 10 ...

70F/20/0/10 70F/15/0/15 50F/20/0/30 70F/0/0/30 40F/30/0/30 60F/20/0/20 80F/0/10/10 60F/10/10/10/10 0F/50/0/50

Good general-purpose slags; medium resistivity Some risk of aluminum pickup; good for avoiding hydrogen pickup; higher resistivity

g

Good general-purpose slags Moderate resistivity; relatively inert Low-melting, “long” slag Difficult starting; electrically efficient

60 ⬉10% Al2O3

50 40 30 20

(Eq 1)

(Eq 2)

It is evident that a saturation of the slag with sulfur does not take place; therefore, the desulfurization capacity of the slag remains intact throughout the entire remelting process. With a highly basic slag (CaO/SiO2 > 3), more than 80% of the sulfur can be removed. Behavior of Oxygen. As mentioned previously, the ESR process is usually carried out under a normal air atmosphere. Oxidation of the metal is unavoidable. Oxygen can be transferred into the metal in several ways:  Oxidation of the electrode surface above the

slag bath  Oxidation on the slag surface of elements with variable valences, such as iron and manganese  Oxides attached to the electrode surface Transfer of oxygen into the slag also takes place, and oxygen is transferred into the metal according to Eq 1. This results in losses

for C16 10 Average oxygen content, ppm

The second reaction is the slag/gas phase reaction. In this case, the sulfur absorbed by the slag is removed by the oxygen of the gas phase in the form of gaseous sulfur dioxide: 3 CaS þ O2 ðgasÞ ! CaO þ SO2 ðgasÞ 2

g

(a) The universal designation system for slags is the percentage of constituents listed in the order CaF2/MgF2/CaO/MgO/Al2O3. Thus, a common base slag such as 60%CaF2/20%CaO/20%Al2O3 would be referred to as 60/0/20/0/20. However, industrial designations often omit the 0s. Thus, the previously mentioned slag is generally referred to as 60/20/20. Source: Ref 7

Slags for ESR are usually based on calcium fluoride (CaF2), lime (CaO), and alumina (Al2O3). Magnesia (MgO), titania (TiO2), and silica (SiO2) are also added, depending on the alloy to be remelted. Table 1 shows the compositions of some common ESR slags. Behavior of Sulfur. One of the primary advantages of the ESR process for steel is the good desulfurization of the metal. The final desulfurization is determined by two reactions. The first is the metal/slag reaction, in which sulfur is transferred from the metal to the slag: FeS þ CaO ! CaS þ FeO

remelting techniques, oxygen contents of less than 20 ppm in unalloyed steel are possible. Figure 12 shows the influence of slag composition on the final oxygen content of remelted ingots.

0 40 15–25% Al2O3 30 20 10 0 40 30–45% Al2O3

30 20 10 0

1

2

3

4

5

6

7

8

9

Slag basicity

Fig. 12

Effect of slag composition on the oxygen content of the remelted ingot. Slag basicity is the ratio of CaO to SiO2.

Electroslag Remelting / 131

Superalloy ESR Generally, superalloy ESR uses a cast VIM electrode as a consumable electrode. Unlike VAR, high-vapor-pressure tramp elements are not removed by ESR. Thus, magnesium may be retained at higher levels than is possible in VAR (with beneficial effects on hot workability). A beneficial chemical reaction related to workability is the removal of sulfur. Detrimental compositional changes may occur in reactive elements (notably titanium and aluminum), because these elements change to attain chemical equilibrium with components of the slag. Oxides incorporated in the electrode being melted are effectively removed by the ESR process, producing an ingot that is generally cleaner than a VAR ingot. All ESR slags contain CaF2 as the primary constituent. Oxides may be added to the slag to raise the resistivity (so as to make melting more efficient electrically) or to modify the slag chemistry with regard to its reactivity with the metal being melted. Additions of CaO, MgO, and Al2O3 are made to modify both the resistance of the slag and its melting point. All commercial-purity CaF2 contains SiO2. During melting of titanium-bearing alloys, an exchange will take place between the titanium in the alloy and the silicon in the slag. The slag becomes enriched with TiO2, and the silicon content of the alloy increases until an equilibrium is established, usually within the first few hundred pounds of melting. To minimize this exchange, some commercial slags are buffered with intentional additions of TiO2. Titanium and zirconium are superalloy elements that often interact with the slag during ESR. Many proprietary slags contain additions of TiO2 and/or ZrO2 to prevent excessive loss of these elements into the slag during melting. The loss is due to exchange with any oxide in the slag that is less stable than the element in the electrode. A common problem with titanium-bearing superalloys is that commercially available CaF2 contains SiO2. During ESR, the SiO2 is converted to silicon, which is picked up by the metal. Thus, the slag is enriched in TiO2, which reduces the titanium content of the metal. This reaction generally results in a titanium gradient in the ingot, because the SiO2 content is consumed early in the melt and no longer plays a part in the reaction after the first few hours of melting. When

ESR slag is recycled, it is already depleted in SiO2, and this reaction no longer occurs. The slag always tries to move to chemical equilibrium with the molten metal at the operating temperature of the slag. Thus, repetitive recycling of slag with a given alloy composition produces a slag that is naturally buffered against causing elemental loss from the electrode during ESR. Unfortunately, slag loss (left as a skin on the electrode during the process) generally keeps recovery levels of the slag below 100%, so some fresh slag is always needed in the ESR process.

ACKNOWLEDGMENTS Reviewed by Scott A. Balliett, Latrobe Specialty Steel Company, with portions adapted from Ref 4 and from: A. Choudhury and F. Knell, Electroslag Remelting (ESR), Metals Handbook, 9th Edition, Volume 15, Casting, 1988, pp 401–405.

REFERENCES 1. R. Jauch, A. Choudhury, F. Tince, and H. Steil, Herstellung und Verarbeitung von ESU-Blo¨cken bis zu 160 t, Stahl Eisen, Vol 101, 1981, p 41–44 2. A. Choudhury, R. Jauch, and F. Tince, Low Frequency Electroslag Remelting of Heavy Ingots with Low Aluminium Content, Proceedings of the Sixth International Conference on Vacuum Metallurgy (San Diego), 1979, p 785–794 3. A. Choudhury, R. Jauch, H. Lo¨wenkamp, and F. Tince, Application of the Electroslag Remelting Process for the Production of Heavy Turbine Rotors from 12 %Cr-Steel, Stahl Eisen, Vol 97, 1977, p 857–868 4. M.J. Donachie and S.J. Donachie, Superalloys: A Technical Guide, ASM International, 2002 5. R.K. Hopkins, Method and Apparatus for Producing Cast Metal Bodies, U.S. Patent 2,380,238, 1945 6. B.E. Paton, B.I. Medovar, and W.E. Raton, Ein neues Verfahren—Das Elektroabgiessen von Blo¨cken, Bull. Tech. Informazil NTO, No. 1, Maschprom., Kiew, 1956

7. G. Hoyle, Electroslag Processes, Applied Science, 1983 8. W. Holzgruber and E. Plo¨ckinger, Metallurgische und verfahrenstechnische Grundlagen des Elektroschlacke-Umschmelzens von Stahl, Stahl Eisen, Vol 88, 1968, p 638–648 9. M. Wahlster, A. Choudhury, and K. Forch, Einfluss des Umschmelzens nach Sonderverfahren auf Gefu¨ge und einige Eigenschaften von Sta¨hlen, Stahl Eisen, Vol 88, 1968, p 1193–1202 10. M. Wahlster and A. Choudhury, Beitrag zum Elektroschlacke-Umschmelzen von Sta¨hlen, Rheinstahl Technik, Vol 5, 1967, p 31–37 11. M. Wahlster, G.H. Klingelho¨fer, and A. Choudhury, Neue metallurgusche und technologische Ergebnisse einer 10 t ESUAnlage, Radex-Rundsch., Vol 2, 1970, p 99–111 12. G.H. Klingelho¨fer, A. Choudhury, and E. Ko¨niger, Etude compare´e des caracte´ristique des aciers elabore´s a` l’air et des aciers refondus selon le proce´de´ sous laiter electroconducteur (ESR) pour les cylindres de laminage a` froid et les cylindres calandreurs, Rev. Metall., Vol 67, 1970, p 512–522 13. B.I. Medovar, Arc-Slag Remelting Steel and Alloys, Cambridge International Science Publishing, England, 1995 14. T. Li, Spec. Steel (China), Vol 25 (No. 5), Sept-Oct 2004, p 1–5 15. W.T. Carter, J.J. Jackson, R.M. ForbesJones, and R. Minisandram, Clean Metal Nucleated Casting of Superalloys, Sixth International Symposium on Superalloys 718, 625, 706, and Derivatives, Oct 2–5 2005 (Pittsburgh, PA) 16. G.E. Maurer, Primary and Secondary Melt Processing—Superalloys, Superalloys, Supercomposites and Superceramics, Academic Press, 1989, p 49–97 17. Ch. Kubisch, Druckstickstoffsta¨hle—eine neue Gruppe von Edelsta¨hlen Berg.—und Hu¨ttenma¨nnische Monatshefte, 1971, p 84–88 18. A. Choudhury, “Vacuum Electroslag Remelting—A New Process for Better Products,” Paper presented at the Vacuum Metallurgy Conference, June 1986 (Pittsburgh, PA)

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 132-138 DOI: 10.1361/asmhba0005202

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Vacuum Arc Remelting THERE ARE TWO COMMONLY USED remelting processes for metal refinement: electroslag remelting (ESR) and vacuum arc remelting (VAR). In both processes, an electrode is melted as it advances into the melting region of the furnace. As the working face of the electrode is heated to the melting point, drops of liquid metal that fall from the electrode face are collected in a lower crucible and rapidly solidified. However, ESR and VAR have very different melting methods, which have implications regarding the magnitude of cooling rates obtained and the nature of defects created by the remelting process itself. The VAR process also is carried out in a vacuum, while the ESR process typically occurs under atmospheric conditions—although electroslag remelting under vacuum (VACESR) or under pressure is a process variation of ESR (see the article “Electroslag Remelting” in this Volume for a brief description of VAC-ESR). The VAR process is widely used to improve the cleanliness and refine the structure of standard airmelted or vacuum induction melted (VIM) ingots (typically referred to as consumable electrodes

Fig. 1

Key components of a vacuum arc remelting furnace. Source: Ref 1

for remelting). The VAR process is also commonly used in the triplex production (VIMESR-VAR) of superalloys. The VAR process was the first commercial remelting process for superalloys. It was used in the late 1950s to manufacture materials for the aircraft industry. Current applications of VAR processing include the production of steels and superalloys as well as the melting of reactive metals such as titanium and zirconium alloys. The melt cleanliness and homogeneity from VAR provides benefits in both primary ingot and foundry shape casting for high-integrity applications where improved fatigue and fracture toughness of the final product are essential. This includes aerospace applications, high-strength steels, ball-bearing steels, die steels, tool steels (cold and hot work steels), power generation, defense, and medical and nuclear industries.

Process Description The basic process of VAR (Fig. 1) is the continuous melting of a consumable electrode by means of a direct current arc (electrode negative, melt pool positive) in a vacuum on the order of 0.1 to 1 Pa (7.5
104 to 0.0075 torr). In some cases, the melting is carried out under inert gas with a pressure up to 1000 Pa (7.5 torr). Evaporation losses of volatile alloying

Fig. 2

elements are minimized at this pressure. Remelting currents are up to 40 kA. Melting rates with VAR furnaces are as high as 1150 kg/h (2500 lb/h), and ingot diameters are as large as 1.5 m (5 ft). Figure 2 shows melting rates for various steels and alloys as a function of ingot diameter. These are empirical values that were obtained from experience in operation. These melting rates gave low microsegregation while achieving acceptable surface quality. The diameter of the electrode and its relationship to the crucible is critical and must be matched for the melting rate. The VAR crucible is immersed in a tank of cooling water. Often, a tube is used to further enclose the crucible and limit the thickness of the water stream moving past the crucible surface. This increases the water velocity and thus increases heat removal from the crucible surface. Such tubes are called water guides. The crucible base (stool) has provision for the inlet of a gas (usually helium). The top of the crucible mates with the VAR head. The head contains the vacuum ports and the ram, which drives the electrode into the crucible as it is consumed. The electrode position is controlled to maintain an arc gap between the melting electrode tip and the molten pool on the top of the solidified ingot in the water-cooled copper mold. When an ingot is solidifying in the crucible, it pulls away from the crucible wall as it

Typical vacuum arc remelting rates for various steels and nickel- and cobalt-base superalloys

Vacuum Arc Remelting / 133 cools. This reduces heat conduction to the crucible wall. To maintain the highest possible heat removal from the solidifying ingot, a high-heat-capacity gas such as helium is introduced into the gap that is formed by the ingot shrinkage. The ram in newer designs is capable of X-Y movement. In VAR, this is not movement in the horizontal plane, which would allow centering of the stinger/electrode assembly in the crucible. It is actually angulation of the ram from the vertical, so that if the stinger is not perfectly parallel with the electrode, then the ram may be angled to make the electrode parallel to the crucible sides. The VAR head mates to the crucible flange through an O-ring seal. Most VAR furnaces have two melting stations (crucible setups). While an electrode is melting in one station, the next station is prepared for melting. When one melt is finished, the VAR head is rotated in the horizontal plane to be located over the second crucible/electrode setup. The electrode is attached to the ram through a stinger, which is welded to the electrode. Although the basic design of a VAR furnace has remained largely unchanged over the years, significant advances have been made in the fields of control and regulation of the process with the object of achieving a fully automatic melting procedure. Modern VAR furnaces (Fig. 3) make use of computer-controlled process automation with continuous weighing of the consumable electrode and closed-loop process control of melting parameters such as remelting rate and arc gap based on arc voltage or drop-short pulse rate. This in turn has had a

decisive positive influence on the metallurgical properties of the products. Close control of all remelting parameters is required for reproducible production of homogeneous ingots with a controlled solidification structure free of macrosegregation. Process Capabilities. Because of the high arc temperature and the small pool of liquid metal, sound ingots with dense crystal structure, low hydrogen and oxygen contents, and minimal chemical and nonmetallic segregation are produced. The primary benefits of remelting a consumable electrode under vacuum are:  Removal of dissolved gases, such as hydro-

gen, nitrogen, oxygen, and CO

 Reduction of undesired trace elements with

high vapor pressure

 Improvement of cleanliness by removal of

oxides

 Achievement of directional solidification of the

ingot from bottom to top, thus avoiding macrosegregation and reducing microsegregation In VAR, center porosity and segregation in the ingot resulting from ingot casting are also eliminated because of progressive melting and solidification. Directional solidification of the ingot from bottom to top avoids macrosegregation and minimizes microsegregation. Oxide inclusion removal is optimized because of the relatively short reaction paths during melting of the hot electrode end and because of good drop dispersion in the plasma arc. Oxide removal is achieved by chemical and physical processes. Less stable oxides or

nitrides are thermally dissociated or are reduced by carbon present in the alloy and are removed via the gas phase. However, in special alloys and in high-alloyed steels, the nonmetallic inclusions (e.g., alumina and titanium-carbonitrides) are very stable. Some removal of these inclusions takes place by flotation during remelting. The remaining inclusions are broken up and evenly distributed in the cross section of the solidified ingot. Because the molten metal drops immediately by gravity into the molten pool, it is inherently impossible to superheat the metal in the VAR process. This, coupled with the very high heat-extractive capability of the process, makes VAR the choice for economic manufacture of the largest-diameter ingots of segregation-prone superalloys. Additionally, the exposure of small volumes of molten metal to high vacuum is capable of removing detrimental high-vaporpressure elements. Unfortunately, beneficial high-vapor-pressure elements such as magnesium are also reduced greatly in concentration. Melt Pool in the VAR Process. Figure 4 illustrates the features of a molten pool in a VAR ingot. The most important feature shown is the depth of the molten pool. It generally is assumed that the depth of the liquid zone is representative of the depth of the liquid-solid zone. For a given crucible diameter, the size (depth) and shape of the liquid-solid zone has a direct relationship with heat input (melt rate) and heat extraction (water cooling). The depth and angle of the liquid-solid zone also controls freckle formation for any given alloy. The melt rate is directly proportional to the applied melt current (Fig. 5). However, as seen in Fig. 6, the relationship between pool depth and melt current is direct but is divided into

Fig. 4

Fig. 3

Modern vacuum arc remelting (VAR) furnace. (a) 30 ton VAR. (b) Operational components: 1, electrode feed drive; 2, furnace chamber; 3, melting power supply; 4, busbars/cables; 5, electrode ram; 6, water jacket with crucible; 7, vacuum suction port; 8, X-Y adjustment; 9, load cell system. Courtesy of ALD Vacuum Technologies GmbH

Molten pool region in a vacuum arc remelted ingot. The relative sizes of annulus and arc gap are not to scale. Competing currents are illustrated for stirring by thermal buoyancy (rising of lower-density hot metal along the ingot centerline) versus the stirring by electromagnetic fields (Lorentz stirring) that drive liquid metal from the edge of the crucible to the center of the crucible and down the centerline. Source: Ref 1

134 / Melting and Remelting

Fig. 5

Melt rate versus vacuum arc remelting current (50 cm, or 20 in., ingot; a composite graph from several independent sources). Source: Ref 1

liquid metal from the edge of the crucible to the center of the crucible and down the centerline. Figure 6 thus has two regimes, where the inflection point (nominally 6600 A for a 50 cm, or 20 in., IN-718 ingot) marks the point when Lorentz stirring becomes stronger than thermal buoyancy stirring. The apparent depth of the molten pool is altered not because there is a change in the volume of molten metal, but because the centerline stirring at higher amperages causes the pool to depart from a nominal U-shape and thus become deeper in the center. Due to Lorentz stirring, high VAR amperages thus produce deeper pools and also alter the angle of the liquid-solid zone, with respect to the ingot axis. In addition to deepening the molten pool and increasing the angle of the liquid-solid zone, there is another ramification when Lorentz stirring is dominant. At low currents, oxides and nitrides in the electrode are melted out, drop onto the molten pool surface, and are swept to the sides of the electrode. When melting at high currents, these particles may not migrate to the edge of the ingot but may be incorporated to some extent in the general structure of the ingot, perhaps with a concentration along the centerline. It is thus considered inadvisable to melt at VAR currents in the Lorentz-dominated region. As the volume of metal on the stool (the ingot) builds up, an equilibrium state is reached in which there is a solid ingot, a mushy zone of both liquid and solid above that, and then a zone that is totally liquid. Because the heat extraction in the steady-state condition of melting is fastest through the sidewalls and slower down the ingot and through the stool, the mushy zones and liquid zones are shallower near the sidewalls and deeper in the ingot center. The thickness and growth angles of the mushy zone determine if the interdendritic liquid regions will coalesce to form positive segregation channel defects. The thickness of the mushy zone at the center of a VAR ingot is controlled by the efficiency of heat extraction, the size (diameter) of the crucible, and the melt rate of the electrode face. The melt rate is controlled by the magnitude of current passed through the electrode. Other factors that are controlled so as to positively affect the solidifying structure are the distance of the melting face above the molten pool (arc gap) and the clearance of the sides of the electrode from the crucible (annulus).

Solidification Fig. 6

Vacuum arc remelting pool depth versus melt current. Source: Ref 1

two separate regions. The reason for this is indicated by the two lines in Fig. 4, which indicate the direction of liquid metal flow in the VAR pool. Competing currents are generated

by the tendency of lower-density hot metal to rise along the ingot centerline (thermal buoyancy stirring) versus the tendency of electromagnetic fields (Lorentz stirring) to drive

The solidification structure of a VAR ingot is a function of the local solidification rate and the temperature gradient at the liquid-solid interface. To achieve a directed dendritic primary structure, a relatively high temperature gradient at the solidification front must be maintained during the entire remelting process. The growth

Vacuum Arc Remelting / 135 direction of the cellular dendrites conforms to the direction of the temperature gradient, that is, the direction of the heat flow at the moment of solidification at the solidification front. The direction of the heat flow (and thus the growth direction of the dendrites) is always perpendicular to the solidification front (Fig. 4). The edge of a VAR ingot also is the first metal to solidify. It is thus alloy-lean, as dictated by the phase rule. It is present as a skin around VAR ingots and is known as shelf. The shelf, in addition to being alloy-lean, contains both the oxides and nitrides collected from the melt pool surface as well as portions of the vapor deposits and splash BBs that have been deposited on the crucible wall above the advancing molten pool. If not adequately removed, this shelf material carries over from the ingot. As pool depth increases with the remelting rate, the growth angle of the dendrites, with respect to the ingot axis, also increases. In extreme cases, the growth of the directed dendrites can come to a stop. The ingot core then solidifies nondirectionally, for example, in equiaxed grains, leading to segregation and microshrinkage. Even in the case of directional solidification, microsegregation increases with dendrite arm spacing. A solidification structure with dendrites parallel to the ingot axis yields optimal results. However, a good ingot surface requires a certain level of energy input, resulting in respective remelting rates. Optimal melt rates and energy inputs depend on ingot diameter and material grade, which means that the necessary low remelting rates for large-diameter ingots cannot always be maintained to achieve axis-parallel crystallization. In spite of directional solidification, defects do occur. This can lead to rejection of the ingot, particularly in the case of superalloys. For the most part, these defects also occur in discrete regions, as opposed to continuous defects. Solute-lean (negative segregation) defects are not inherently as detrimental to the properties of a component as are defects resulting from positive segregation. Types of ingot defects include:  Tree ring patterns  Freckles  White spots

Tree ring patterns can be identified in a macroetched transverse section as light-etching rings. They usually represent a negative crystal segregation. Tree ring patterns seem to have little effect on material properties. Freckles and white spots have a much greater effect on material properties as compared to tree ring patterns. Both defects can represent a significant cause for premature failure of turbine disks in aircraft engines. Tree ring patterns are the result of a wide fluctuation in the remelting rate. In modern remelting plants, however, the remelting rate is maintained at the desired value by means of an exact control of the melting rate during

operation, so that the melting rate exhibits no significant fluctuations unless caused by electrode defects. Freckles are dark-etching circular or nearly circular spots that are generally rich in carbides or carbide-forming elements. The formation of freckles is usually a result of a high pool depth or movement of the rotating liquid pool. The liquid pool can be set into motion (rotation) by stray magnetic fields during remelting. Freckles can be avoided by maintaining a low pool depth and by avoiding disturbing magnetic fields through the use of a coaxial current supply. White spots are typical defects in VAR ingots. They are recognizable as light-etched spots on a macroetched surface. There are several mechanisms that could account for the formation of white spots:  Residues of unmelted dendrites of the con-

sumable electrode

 Pieces of the ingot crown that fall into the

pool and are not dissolved or melted

 Pieces of the shelf region transported into

the solidifying interface All three of the aforementioned mechanisms, individually or combined, can be considered as sources of white spots. This indicates that white spots cannot be completely avoided during VAR. To minimize the frequency of occurrence of these defects, the following conditions should be observed during remelting:  Use the maximum acceptable metal rate

permitted by the ingot macrostructure.

 Use a short arc gap to minimize crown for-

mation and to maximize arc stability.

 Use a homogeneous electrode substantially

free of cavities and cracks.

 Use a proper melting power supply.

Under ideal conditions, the arc is distributed uniformly across the surface of the molten pool. (This is called a diffuse arc.) However, the introduction of conductive ionic species into the melt or increases in the resistance (arc gap) in the system may cause the arc to become locally concentrated (constricted arc). Constricted arcs do not dwell in one place on the molten pool surface but move about the surface. When constricted arcs undercut the crown/ shelf, small pieces of the shelf may fall into the molten pool. These pieces move by gravity down the pool profile. They do not readily remelt into the pool, because they have a higher melting point than the pool in general. Ultimately, they are not dissolved and are incorporated into the ingot as regions of low solute content, containing stringers of oxide and nitride. When detected in the final product, they are seen as light-etching defects containing stringers of oxide and nitride and are known popularly as dirty white spots. More correctly, these regions are called discrete white spots, in keeping with their nature as a distinct but isolated structure within the ingot matrix.

Theoretically, a piece of shelf, free of oxides and nitrides, may be undercut and form a discrete white spot that is not dirty. Often, because of the lack of solute, the grain size in discrete white spots will be larger than that of the matrix. When present, dirt stringers act as stress raisers. Thus, discrete white spots, which have a high probability of being largegrained and dirty, are acknowledged to be detrimental to properties in superalloys. Unfortunately, although best practice for electrode preparation and for VAR melting may minimize the frequency of occurrence of discrete white spots, there is no way to guarantee their elimination in VAR products. The probability of their occurrence in the final component must be considered in the design of the component. Currently, there is no method of detecting a subsurface defect except to the extent that, combined with forging deformation, a crack may be generated that will be detectable by ultrasonic inspection. Macroetching of the surface of critical components is a common method for detection of discrete white spots and any other melt-related structure.

Process Variables The manufacture of homogeneous ingots with minimal segregation requires careful control of remelting parameters. The three major parameters defining the melt process are ingot and electrode diameter, arc gap, and melt rate. The choice of ingot and electrode diameter defines the clearance (annulus) between the crucible and the electrode. Insufficient clearance, either through a poor choice of annulus or through an off-center set-up, will lead to excessive current loss from the electrode to the crucible wall, rather than through the arc to the ingot. The choice of ingot diameter controls the amount of possible heat extraction and dictates the choice of melt rate so as to avoid positive segregation defects in the ingot. These choices, of course, are long-term choices dictated in the purchase of VIM molds and VAR crucibles. Control of the VAR process also includes:  Pool depth and shape (control of solidifica-

tion structure)

 Arc stability (control of drop-in defect

formation) As noted, the depth and angle of the solidification front (pool shape) needs to be maintained so that unacceptable solute-rich solidification structures do not occur. Pool shape and depth are controlled by regulating heat input and heat extraction. Of these, the melting current density has the largest influence on the melting bath geometry and conditions of solidification. Melt Rate. The melt rate is dependent on the melt current. Many producers choose to select a given melt amperage and hold it constant. Other producers use load cells built into the

136 / Melting and Remelting VAR furnace to measure electrode weight and thus the change in weight. The change in weight is used to calculate a melt rate, and the applied current is varied to attempt to maintain uniform melt rate. Individual melt rate measurements will vary widely depending on the time frame chosen for the measurement. Melt rates are generally calculated using rolling averages. A common averaging time is 20 min. Modern VAR furnaces are equipped with a load cell system to measure the weight of the electrode at a particular interval of time. Load cell design and maintenance are important factors in maintaining the quality of material melted in melt rate control, because load cells are subject to the generation of false signals. The actual values of the melt rate are compared by computer with the desired set values. Any difference between the measured melt rate and the desired value is eliminated by the proper accommodation of the power input. Figure 7 shows the melt rate and the melting current at startup, during steady-state melting, and during hot topping. Startup and hot topping are usually controlled based on time. The melting phase is controlled based on weight. Hot topping begins when a preselected residual electrode weight is reached. A computer controls the melting parameters, and melting rates can be controlled with a precision of better than  þ2%. Heat input is determined by the melt rate of the electrode. Melt rate is often monitored and corrected over very short time periods by changes in melt current (melt rate control). Alternatively, control of the median melt rate over the duration of a melt may be maintained by operating at constant melt amperage (amperage control). Heat extraction is a function of the contact surface/ingot volume (ingot size) being melted. Larger ingots have poorer heat extraction (lower surface/volume) and must be melted at proportionately lower melt rates to prevent the formation of large liquid + solid zones and thus the formation of unacceptable positive segregation. While, for a given ingot size, high

Fig. 7

melt rate may promote positive segregation, a melt rate that is too low may cause the formation of visible solute-lean structures. The mechanism of formation of these solidification white spots is not completely understood. They differ from discrete white spots in that they are a solidification effect, not a shelf or electrode drop-in, and therefore do not contain oxide/ nitride stringers. Nevertheless, such solute-lean regions are potentially detrimental because they may make it difficult to control grain growth in subsequent processing. Arc Stability. Active control of arc stability is primarily by control of the arc gap throughout the melting of an electrode. The arc gap is the nominal distance of the melting electrode surface above the molten pool at the top of the solidifying ingot. Commercially useful gaps range from a minimum of approximately 2.5 mm (0.1 in.) to a maximum of 12 mm (0.5 in.). The arc gap may be considered as the resistance in this circuit. Because the melt current is held constant, any increase or decrease in the resistance of the circuit (change in arc gap) is seen as a change in voltage. Early VAR controls actually did measure these changes in voltage as a means for maintaining a uniform arc gap throughout the melt. Modern systems control the arc gap by measurement of drip short frequency (DSF). When a drip (drop) of molten metal is formed on the melting electrode surface, there is a point in time at which it is in contact with both the electrode and the molten pool. This causes a sudden decrease in voltage (an electrical short) across the arc gap. The duration of the short is measured in milliseconds. Control systems measure the number of drips of a given duration. It has been demonstrated that the frequency of drip shorts (usually in drips/min) is related to the width of the arc gap. The relationship is not linear over the whole range of melt conditions that may be experienced but does allow useful measurement and control of the arc gap in the ranges that are currently used

Process control parameters and set point functions in vacuum arc remelting

commercially. The higher the DSF, the smaller the arc gap. Making the gap smaller increases the duration of a short and thus increases the number of shorts counted. The relationship between arc gap and DSF also changes with changes in the melt current. At higher currents (higher melt rate), the drips are larger but less frequent. Thus, as in Fig. 8, raising the current requires a reduction in DSF to maintain equivalent arc gap. Drip short frequency is measured over a period of time, usually fractions of a minute. Individual DSF values vary widely. Thus, control signals for the ram drive (electrode feed speed) fluctuate. Normally, the ram is in continuous motion. The response to changes in DSF is to increase or to decrease the ram speed. Monitoring of the consistency of the arc gap may be accomplished by computing a rolling average or by simple observation of the width of the band generated in recording each individual DSF measurement. For melting of superalloys, arc gaps are generally controlled in the range of 6.5 to 13 mm (¼ to ½ in.). To maintain constant arc gap, the electrode is fed into the melt either faster or slower (ram drive) according to the fluctuation in voltage that occurs when the resistance of the system changes in response to changes in the arc gap. Modern control systems for superalloys generally do not use voltage for arc gap control but rather use a related voltage measurement: the DSF. A drip short is the transient downward voltage spike that occurs when a molten metal droplet is in contact with both the electrode and the molten pool. The number of shorts occurring in a given time period is proportional to the arc gap. The relationship for arc gap versus DSF is not melt rate independent. Passive control of arc stability is maintained by controlling the annulus and controlling the quality of the electrode being remelted. Electrodes for VIM generally contain porosity and low levels of residual refractory, both of which may act to destabilize the arc. Vacuum induction melting and pouring practice

Fig. 8

Drip short frequency as a function of arc gap and melt current. Source: Ref 1

Vacuum Arc Remelting / 137 is thus a significant factor in ensuring VAR arc stability and in minimizing the formation of drop-in-type defects. Vacuum Arc Remelting Control Anomalies. Variations in the traces of VAR parameters may indicate changes in melt conditions that will be detrimental to the quality of the solidification structure. Two of the more easily recognized problems are dead shorts and melt rate excursions (MREs). Dead shorts occur when the electrode is driven into the top of the molten pool, causing direct transfer of the melt current to the solidifying ingot. Such a problem may be caused by an unusually shaped primary shrinkage cavity in the electrode. If the normal ram drive programming cannot accommodate the unusual shape, the electrode may be driven into the pool. This will show on the recorded parameters as a sudden reduction of voltage to zero and an immediate ram back-out from the dead short situation. Drip short frequency, of course, would also go to zero while the electrode face was in contact with the pool. (This would not show as zero DSF, because DSF is measured over a time period.) Dead shorts will cause a major change in solidification characteristics in the molten pool in that region. Another cause of multiple dead shorts is the presence of high volumes of oxides and nitrides in the VAR electrode. If the oxide/nitride volume being melted onto the molten pool surface exceeds the capability of the process to sweep the “dirt” to the side and incorporate it into the ingot surface, then the ability to sustain an arc will be deteriorated. Melt rate will drop, and the ram travel, unable to react rapidly to sudden changes in arc gap, will drive the electrode into the pool, causing a short. This may happen repetitively as the ram backs out of the pool and then advances into it again. Transverse cracks in the electrode cause MREs, the most readily definable VAR chart anomaly. Electrodes may develop transverse internal cracks due to thermal stresses. This may happen in the cooling of the electrode, or, often, due to the thermal stresses generated as the VAR melt front moves up the electrode. When the melt front approaches a transverse crack, heat transfer across the crack is diminished, and the material below the crack becomes hotter than normal. Thus, the melt rate begins to increase. The ram speed, which responds to a decreasing DSF, does not respond fast enough to maintain arc gap, and thus, the arc gap increases.

Superalloy VAR The VAR process typically uses a cast (generally VIM or VIM-ESR) electrode. Elements such as bismuth and lead (which can be present in low concentrations even after VIM) are highly undesirable but are reduced to negligible levels by the VAR process. Magnesium, which is considered desirable for improved workability (and

is an addition element in VIM), is reduced in concentration but not completely removed. With the exception of high-vapor-pressure elements, the VIM chemistry is representative of the VAR chemistry. The flotation of oxides and nitrides to the melt surface in VAR improves the cleanliness and reduces the gas content of material processed through VAR. A common factor affecting consumable remelt quality for both ESR and VAR is the quality of the electrode. Electrode quality is affected by composition, cleanliness (low oxide and nitride content), and soundness (freedom from porosity and cracks). The most important characteristic of an electrode is composition. With the exception of reductions in volatile elements, VAR does not change the composition of the electrode. For high-nitrogen electrodes, some reduction of nitrogen is accomplished by flotation of nitrides. As the solubility levels of TiN in the alloy are approached, this mechanism is no longer effective. Electroslag remelting reduces sulfur content but may also cause minor changes in composition through reaction of titanium, aluminum, zirconium, and silicon with components of the slag. These changes are predictable in a mature process. Thus, there is no practical way to achieve composition modification once the master heat has been cast, but departure from normal practice in ESR may change titanium, aluminum, zirconium, or silicon content. Cleanliness. The cleanliness of the cast electrode with regard to entrained oxide and nitride is an important characteristic for VAR electrodes. The introduction of a high volume of oxide or nitride onto the molten pool in VAR may degrade the efficiency of melting of the arc as well as disturb the electrical characteristics measured to maintain control of the arc gap. This is not a problem in electrodes to be melted by ESR. Porosity. The porosity (secondary shrinkage) of an electrode may be of some concern in VAR. Vacuum arc remelting of electrodes with centerline porosity results in the preferred melting of the face in the porosity area. Thus, the melt face is no longer flat. The effect of this with regard to gap control during melting has not been quantified. A secondary problem connected with centerline porosity is the concern that projecting dendrites in this region may become detached from the electrode and fall, unmelted, into the molten pool. Then, they may be incorporated into the structure without being remelted. The composition of such a region thus will be that of a primary dendrite (alloy-lean) rather than representative of the bulk composition. Theoretically, these concerns may also apply to ESR processes, but the necessity for melted or unmelted material to pass through the slag makes ESR products inherently less sensitive to electrode porosity. Cracking. The final electrode characteristic affecting remelt quality is the freedom of the electrode from transverse cracking. In highly alloyed superalloys, the thermal stresses

generated upon cooling of electrodes or upon heating in the remelt process may be sufficient to form transverse cracks. When the melt front of either VAR or ESR approaches such a crack, the heat transfer up the electrode is diminished, because of the thermal barrier presented by the crack. Thus, the material in front of the crack becomes hotter than would occur under equilibrium conditions and will tend to melt at a faster rate. After the melt front has gone through the crack, it encounters cold material, and the melt rate will tend to drop. In both VAR and ESR, the result is that the controls respond with changes in applied power to try to keep the melt rate constant. This, however, is seldom accomplished without some discernable disruption to the continuous nature of the growth of the solidification front. The only defense against these MREs (also called events) is to produce an electrode free of cracks. In the future, it is hoped that more sophisticated melting controls will be able to detect the melt rate changes earlier and thus compensate for them more efficiently.

Titanium VAR Since the commercial introduction of titanium alloys in the 1950s, VAR has been the principal method for the production of titanium ingots. The process is capable of high melt rates with titanium, but steady-state solidification is not possible, and the molten metal pool is relatively deep. Vacuum arc remelting has many advantages, including high purity, good control, and reproducibility in the product. However, if high-density components of an electrode fall into the melt as solids, they can survive in the melt by rapidly dropping to the bottom of the molten pool and solidifying before they can be melted and homogenized in the solidifying ingot. An alternate to VAR is cold hearth melting. The electrodes for making titanium ingots are compacted aggregates (compact or briquettes) of sponge and alloy elements, including both master melt and elemental materials. For commercially pure titanium, the sponge is compacted, and the electrode is melted with appropriate oxygen control. When alloys are to be produced, granules of titanium sponge are mixed and then blended with appropriate alloy-containing material. The resultant alloy mixture is pressed into a compact or briquette. Most titanium and titanium alloy ingot is melted twice using the VAR process. The procedure is known as the double consumable-electrode VAR process. Double melting is considered necessary for all applications to ensure an acceptable degree of homogeneity in the resulting product. For certain critical applications, a third, or triple, melting step was specified at times, especially after the middle of the 1960s when defects were found in rotor-grade titanium alloys produced by double melting. Triple melting achieves even better

138 / Melting and Remelting uniformity of chemistry and structure. Triple melting also reduces oxygen-rich or nitrogen-rich inclusions in the microstructure to a very low level by providing an additional melting operation to dissolve them. In the two-stage process, titanium sponge, revert, and alloy additions are initially mechanically consolidated and then are melted together to form an ingot. Ingots from the first melt are used as the consumable electrodes for secondstage melting. Ingots from the second melt are used as the consumable electrodes for thirdstage melting when it is done. Processes other than consumable-electrode arc melting may be permitted in some instances for first-stage melting of ingot for noncritical applications.

Usually, all melting is done under vacuum, but in any event, the final stage of melting for noncritical applications must be done by the consumable-electrode vacuum-arc process.

 “Vacuum Arc Remelting,” ALD Vacuum

Technologies, GmbH

REFERENCE ACKNOWLEDGMENT

1. M. Donachie and S. Donachie, Superalloys: A Technical Guide, ASM International, 2002

Portions of this article were adapted from:  A. Choudhury and E. Weinga¨rtner, Vacuum

SELECTED REFERENCES

Arc Remelting (VAR), Casting, Vol 15, Metals Handbook, 9th ed., ASM International, 1988, p 406  M. Donachie and S. Donachie, Superalloys: A Technical Guide, ASM International, 2002

 M. Blum, Vacuum Metallurgy, Vulkan, 2005  M. Donachie and S. Donachie, Superalloys:

A Technical Guide, ASM International, 2002

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 139-141 DOI: 10.1361/asmhba0005203

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Skull Melting Reviewed and revised by G. Keough, Consarc

SKULL MELTING refers to the use of furnaces with water-cooled crucibles that freeze a solid “skull” of material on the crucible wall. This technique is useful when melting highly reactive metals (e.g., titanium and zirconium) that would otherwise attack the linings of a crucible with refractories. The solidified metal creates a corrosion-resistant protective layer on the containment vessel surface. Two methods of skull melt casting have evolved. The first, vacuum arc skull melting, was originally developed in the early 1950s by the U.S. Bureau of Mines in Albany, Oregon (Ref 1, 2). Both methods are used to melt significant quantities of contamination free titanium for ingots or near net shape castings. Because the processes are used to melt reactive metals, they are normally carried out in a vacuum, but, in some cases, may also be done in an inert atmosphere. Vacuum arc skull melting was once the dominant method of melting titanium. Currently, other methods for melting titanium have become prevalent. Vacuum arc refining (VAR) furnaces with static crucibles are used in titanium primary melting. Recently, much of the melting of titanium sponge has been converted to plasma cold hearth (PCH) melting. Also, some of the largest rectangular ingots produced for rolling into sheet are now made on large multimegawatt electron beam (EB) furnaces. The total tonnage of material melted by VAR, PCH, and EB is significantly higher than that melted in vacuum arc skull caster furnaces. Vacuum arc skull melting works reasonably well for medium-to-large pour weights, typically in the range of 110 kg (250 lb) and up, but it suffers from a relatively large skull (typically 20% of metal melted). This lowers the thermal efficiency of the system and produces little superheat, so that precision and fine detail are relatively difficult to reproduce. The consumable reactive metal electrode also is expensive to produce, because the alloy shape and size must be preformed. An alternative technique is induction skull melting. Induction skull melting furnaces can melt a wide range of material forms including sponge (compacted), scrap, revert, or master alloy billet. The induction field also promotes vigorous stirring, which ensures uniform alloy composition throughout the melt

and increases superheat (compared to vacuum arc skull casting). Induction skull melting is also used in atomization of melts for powder production of reactive metals such as titanium.

Vacuum Arc Skull Melting Although the possibility of titanium skull melting was considered as early as 1948 and 1949, the first castings were made in 1953. In the late 1950s, this technology was applied by research institutes, which were looking for a practical means of liquefying and pouring uranium into graphite molds, for example, to produce uranium carbide. An early industrial vacuum arc skull melter was built in 1963 for the continuous production of uranium carbide. This furnace had a crucible volume of approximately 0.01 m3 (0.35 ft3) and used a nonconsumable graphite electrode to liquefy the uranium pellets fed into the crucible. The crucible tilting system was hydraulically driven. The molds were stationary. It was not until 1973 that one of the first skull melters for titanium went into operation; this furnace started production in 1974 in West Germany. State-of-the-art titanium vacuum arc skull melting furnaces are often equipped with turntable systems for centrifugal casting (up to 350 rpm). Casting weights of more than 1000 kg (2200 lb) are possible. Vacuum arc skull melting and casting is used for many titanium investment castings for aircraft, aerospace, medical, and chemical applications. Electron beam skull melting is also used for titanium alloys (see the article “Electron Beam Melting” in this Volume).

Furnaces Vacuum arc skull casting furnaces basically consist of a vacuumtight chamber in which an electrode is driven down into a water-cooled copper crucible. The direct current power supply provides the fusing current needed to strike an electric arc between the consumable electrode and the crucible. Because the crucible is water cooled, a solidified skull forms at the

crucible surface, thus avoiding direct contact between melt and crucible. The mechanism of skull formation is similar to the slag splashing on a refractory lining in the steel industry or the formation of a “frozen ledge” on refractories in the aluminum industry. Once the predetermined amount of liquid titanium is contained in the crucible, the electrode is retracted, and the crucible is tilted to pour the melt into the investment casting mold positioned below. For optimal mold filling, the mold can be preheated and/or rotated on a centrifugal turntable. Figure 1 shows the basic components of a vacuum arc skull melting furnace. Operating pressure is in the range of 1 Pa (0.0075 torr), and the specific working current ranges from approximately 1 kA/kg for small furnaces to approximately 0.2 kA/kg for large pouring weights. This batch-type skull melting furnace allows for cycle times of approximately 1 h for a full 50 kg (110 lb) pumping/melting/casting cycle. In principle, three consecutive pours can be obtained from one electrode. This furnace basically consists of a vacuum chamber, an arc voltage-controlled electrode drive system, a skull crucible, a centrifugal casting system with stepless adjustable turntable speed, an automatically sequenced vacuum pump system, a power supply, and an electrical control system with control desk. The cylindrical vacuum chamber is equipped with two large dished doors that support the crucible with the tilting mechanism, the mold platform with the centrifugal casting system, and the casting tundish with its cover. The crucible support system with an additional detachable device is also used for electrode loading. The chamber is jacketed for water cooling in regions that are subject to heat radiation. A top flange with a throat carries the electrode chamber and the electrode feeding system. Viewing ports allow for video monitoring of the melting and pouring. A vacuum pumping port is located in the cylindrical portion of the chamber. Figure 2 shows a semicontinuously operated vacuum arc skull melter for charge weights of up to 1000 kg (2200 lb). The principal difference between this furnace and the smaller model (Fig. 1)—apart from capacity-related

140 / Melting and Remelting

1

2 3 4

5

layout features—is the rectangular vacuum chamber. Again, the crucible and the tilting mechanism are carried by a dished door. In this furnace design, the centrifugal casting system is introduced from the bottom of the chamber to allow the horizontal connection of a separate cooling chamber, if desired. Molds are loaded from the back side into the chamber and can be discharged through a front door, which also allows the use of a cooling and charging chamber with lock valves for continuous mold transport flow. Modern vacuum arc skull melting furnaces are usually equipped with:  Coaxial power feed directly to the skull 

6 7

8 9 10

Fig. 1

Schematic of a vacuum arc skull melting and casting furnace. 1, fast retraction system; 2, power cables; 3, electrode feeder ram; 4, power supplies; 5, consumable electrode; 6, skull crucible (50 kg, or 110 lb); 7, tundish shield; 8, mold arrangement; 9, centrifugal casting system; 10, chamber lid carriage

  

 

crucible to avoid electromagnetic fields that can disturb the melt bath Programmable control systems for crucible tilting to allow repeatable pouring profiles for consistent parameters Highly accurate electrode weighing system for precise determination of pouring weights X-Y adjustment system for coaxial positioning of the electrode in the skull crucible Compact air-cooled power-supply modules of high capacity for achieving high melt rates, thinner skulls, and correspondingly increased yields above 80% Proven vacuum pumping and measuring systems Forced argon cooling systems for faster mold cooling

Induction Skull Melting 1 2

3 4

5

6

7

8

9 10

11

Fig. 2

Schematic of a semicontinuously operating vacuum arc skull melter for charge weights of up to 1000 kg (2200 lb). 1, fast retraction system; 2, power cables; 3, power supplies; 4, electrode feeder ram; 5, consumable electrode; 6, skull crucible; 7, crucible carriage; 8, tundish shield; 9, mold arrangement; 10, vacuum pumping system; 11, centrifugal casting system

In the 1970s, the Bureau of Mines developed an induction melting process for reactive metals known as the induction slag melting process. However, the drawback of that process was that a slag based on calcium fluoride was used to inhibit electrical shorting between segments of the crucible. Later, the Duriron Company further developed the process and, in particular, the crucible design to allow the slag-free melting of highly reactive metals and alloys in what is now referred to as the induction skull melting (ISM) furnace. Induction skull melting has evolved into a process with unique applications. As noted, ISM furnaces have some important advantages compared to vacuum arc skull furnaces. Induction skull furnaces can be charged with various material forms (scrap, revert, billet), and custom alloys can be alloyed in the ISM crucible because the induction field promotes vigorous stirring of the melt. A combination of stirring and levitation ensures uniform alloy composition throughout the melt and more uniform heating and increased superheat (compared to vacuum arc). As an indication of the flexibility of the ISM process, it is claimed that over 3000 alloys have been melted in this process. Part of this flexibility comes from the fact that the process is able to operate at pressure levels from vacuum through 101 kPa (1 atm). The process

Skull Melting / 141

Fig. 3

A 10 kg (22 lb) induction skull melting/casting unit in a vacuum/inert gas chamber

ac coil ISM crucible

Top dc coil

Water in Water out Bottom dc coil Pole piece (a)

ac coil

ISM crucible

Top dc coil

Bottom dc coil Water in Water out Pole piece (b)

Fig. 4

Cross section of induction skull melting (ISM) crucible with two different configurations of alternating current (ac) and direct current (dc) coils used in trials. Source: Ref 3

is also very rapid. A 10 kg (22 lb) charge, for example, can be loaded, melted, cast, and prepared for the next charge on a 15 min cycle. The key design elements of a reliable ISM crucible include effective water cooling and a segmented crucible that allows the field from the induction coil to couple to the load. Modern ISM crucibles use narrow gap technology to minimize the risk of liquid metal penetrating in between the segments and causing a short circuit. Each of the segments is intensively water cooled. A well-designed crucible will typically last 15,000 to 18,000 heats before extensive repair or replacement is required. Figure 3 shows a 10 kg (22 lb) induction skull furnace mounted in a vacuum/inert gas chamber. The chamber and vacuum or inert gas backfill is essential to avoid excessive oxidation of the reactive alloy elements, particularly titanium. Directly below the furnace is a centrifuge table that the molds are mounted on so that the molds can be centrifuged during filling. Skull weights in the range of 5 to 10% of the charge weight can be achieved in a welldesigned ISM furnace. Subsequent charges of the same alloy can be charged into the skull of a prior melt, thus minimizing the material requirements. At the end of a run for a particular alloy, the skull can be removed, identified, and saved for use on a future run. To optimize the efficiency of the process, a relatively high power together with a corresponding frequency are used to achieve both rapid melting and effective stirring of the liquid metal. The individual segments are intensively cooled to protect the copper from erosion and to minimize the risk of contamination of the melt. The combination of high power and high frequency also achieve higher superheat in the melt. This allows the process to be used for smaller pour weights (down to approximately 1 kg, or 2 lb) and for high-precision castings. Typically, the ISM process is recommended for pour weights in the range of 1 to 100 kg

(2 to 220 lb), although further extension of the range for higher pour weights is being developed. Of particular interest is a relatively new development using a combination of direct current electromagnetic fields (Fig. 4) with the traditional alternating current electromagnetic fields to achieve a significant further increase in available superheat of the molten charge (Ref 3). The ISM process also offers other advantages in casting. It can, of course, be used for conventional tilt-pour casting into both static molds and molds mounted on a centrifuge. However, unlike other methods, it can also be used for rollover casting as well as countergravity casting. Rollover casting, in particular, has been shown to be one of the least turbulent methods of skull casting, with promising results for casting complex and thin-walled near-net shape parts. ACKNOWLEDGMENTS Updated from: F. Mu¨ller and E. Weinga¨rtner, Vacuum Arc Skull Melting and Casting, Volume 15, Casting, Metals Handbook 9th Edition, 1988. REFERENCES 1. R.A. Beahl, F.W. Wood, J.O. Borg, and H.L. Gilbert, “Production of Titanium Castings,” Report 5265, U.S. Bureau of Mines, Aug 1956, p 42 2. W.J. Kroll, C.T. Anderson, and H.L. Gilbert, A New Graphite Resistor Vacuum Furnace and Its Application in Melting Zirconium, Trans. AIME, Vol 175, 1948, p 766–773 3. R.A. Harding, M.Wickins, G. Keough, R.J. Roberts, K. Pericleous, and V. Bojarevics, “The Use of Combined DC and AC Fields to Increase Superheat in an Induction Skull Melting Furnace,” LMPC Conference 2005 (Santa Fe, NM), ASM International

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 142-148 DOI: 10.1361/asmhba0005204

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Electron Beam Melting Revised by David W. Tripp, TIMET, USA

ELECTRON BEAM MELTING AND CASTING TECHNOLOGY is accepted worldwide for the production of niobium and tantalum ingots weighing up to 2500 kg (5500 lb) in furnaces with electron beams of 200 to 1500 kW. Other applications in Germany and former Soviet bloc countries are the production of steel ingots weighing 3.3 to 18 Mg (3.6 to 20 tons) using electron beams of up to 1200 kW. Furnaces of up to 4200 kW in electron beam power have been used since 1982 for recycling titanium scrap to produce 16 Mg (17.5 ton) slabs 1420 mm (56 in.) wide. Furnaces of 200 to 1200 kW are used to refine nickel-base superalloys. Other metals, such as vanadium and hafnium, are melted and refined in furnaces between 60 and 260 kW. Many furnaces with melting powers ranging from 20 to 300 kW are in operation in research facilities. These furnaces are used in the development of new grades and purities of conventional and exotic metals and alloys, for example, uranium, copper, precious metals, rare earth alloys, intermetallic materials, and ceramics. The total power

of industrial installed electron beam melting and casting furnaces worldwide was approximately 35,000 kW at the end of 2007. Electron beam melting and casting includes melting, refining, and conversion processes for metals and alloys. In electron beam melting, the feedstock is melted by impinging highenergy electrons. Electron beam refining takes place in vacuum in the pool of a water-cooled copper crucible, ladle, trough, or hearth. In electron beam refining, the material solidifies in a water-cooled continuous casting copper crucible or in an investment ceramic or graphite mold. This technology can be used for all materials that do not sublimate in vacuum. Competing processes include sintering (for example, for refractory metals), vacuum arc melting and remelting (for reactive metals and superalloys), and electroslag melting and vacuum induction melting (for superalloys, specialty steels, and nonferrous metals). Some advantages and limitations of the competing vacuum processes are given in Table 1. Additional information on some of these processes is available in

the articles “Vacuum Arc Remelting,” “Electroslag Remelting,” and “Vacuum Induction Melting” in this Volume.

Process Description The characteristics of electron beam melting and casting technology are:
The flexibility and controllability of the pro-

cess temperature, speed, and reaction


The use of a wide variety of feedstock materi-

als in terms of material quality, size, and shape


The different methods of material processing

available


Product quality, size, and quantity

Contamination of the product is avoided by melting in a controlled vacuum and in watercooled copper crucibles (Fig. 1). The energy efficiency of electron beam processing exceeds that of competing processes because of the control of the beam spot dwell

Table 1 Comparison of characteristics of electron beam melting and competing processes Sintering Metal

Advantages

Vacuum arc melting Limitations

Tungsten, molybdenum

Small grain size; most often used

Refining limited; small batches; high energy consumption

Tantalum, niobium

Small grain size; good workability

Same as above; rarely applied

Hafnium, vanadium

...

Same as above

Zirconium, titanium

...

Not used

Advantages

Limitations

Refining limited; costly Moderate grain size; electrode preparation; acceptable melting dangerous workability; large ingots; low energy consumption Alloying; moderate grain Refining limited; expensive size; large ingots; low electrode; melting energy consumption dangerous Alloying during Almost no refining; costly remelting electrode preparation; melting dangerous

Electron beam melting Advantages

Limitations

Highest possible purity; economical feedstock preparation; large ingots; low energy consumption

Large grain size; brittle product; very rarely applied

Same as above; most frequently used

Alloying limited

Good refining; economical feedstock preparation and ingot production; most often used Economical feedstock Very low contamination; Limited refining; expensive feedstock preparation; only preparation; refining of wide range of alloying round ingots high-density inclusions; possible; large ingots melting of slabs, ingots, low energy and rods; high production consumption; rate; low energy economical melting consumption

High melting costs

Alloying limited; material losses from splatter; high furnace investment

Electron Beam Melting / 143
Large power range of 0 to 1200 kW
Long free beam path of 250 to 1500 mm (10

to 60 in.) and the adjustable beam power distribution
Beam deflection angle of þ45 and spot  frequency up to 500 Hz
Usable vacuum pressure range between 1 and 0.0001 Pa (102 and 106 mbar, or 7.5  103 and 7.5  107 torr)
Reliability of the gun and cathode system The power control and distribution system allows for very accurate distribution of beam power and energy for achieving the required heating for material melting, superheating, refining, and electrothermal effects.

Drip Melting

Fig. 1

Schematic of the electron beam melting process

Fig. 2

Examples of electron beam melting and casting processes. (a) Button melting with controlled solidification for quantitative determination of low-density inclusions. (b) Consolidation of raw material, chips, and solid scrap to consumable electrodes for vacuum arc or electron beam remelting. (c) Drip melting of horizontally or vertically fed feedstocks. (d) Cold hearth refining/melting. (e) Floating zone melting. (f) Investment casting. (g) Pelletizing (manufacture of pellets from scrap and other materials for scrap recycling). (h) Atomization and granulation of refractory and reactive metals

time and distribution at the areas to be melted or maintained as liquid. In addition, unnecessary heating of the ingot pool, as occurs in vacuum arc remelting, for example, is avoided. Power losses of the electron beam inside the gun and between the gun nozzle and the target are very small, but losses up to 40% of the beam power can be incurred because of beam reflection, radiation of the liquid metal, and heat conductivity of the water-cooled trough and crucible walls. From the large variety of electron beam melting and casting processes shown in Fig. 2, only the processes illustrated in (c) and (d) are related to processes used in foundry technology:


Drip melting process for the preparation of

refractory and reactive metal feedstock material for electron beam and vacuum arc remelting skull melting and casting
Cold hearth melting process for the feedstock refining of superalloys for vaccum induction melting and electron beam casting Electron Beam Heat Source Specifications. For all electron beam melting and casting processes, except for the crucible-free floating zone melting process, Pierce-type electron beam guns with separately evacuated beam generating and prefocusing chambers are used. The essential features of these guns are:

The drip melting processes (Fig. 3c and d) are primarily used for the production of clean, mostly ductile ingots of refractory and reactive metals. The feedstock for the first melt (Fig. 3c) can be compacted sponge, granular, powder, or scrap, which may be presintered in a vacuum-heating furnace. In some cases, loose raw materials can be consolidated in a watercooled copper trough (Fig. 3a). The consolidated ingot can then be fed horizontally for drip melting. Raw material that is continuously consolidated in a water-cooled copper crucible with a retractable bottom plate (Fig. 3b) can be fed horizontally or vertically for drip melting. In both consolidation processes, only 20 to 80% of the material is melted. Refining and losses of material by splattering are negligible. Drip melting of horizontally fed compacts is the most frequently used process for the production of ingots from refractory or reactive metals. The resulting ingot is of sufficient purity but has an area of inhomogeneity caused by the shadow of the horizontally fed bar. Two or more electron guns are used in drip melting to make use of reflected electron beams and to reduce evaporation and splattering. The end of a compact is welded to the front of the following one to avoid dropping semisolid material into the pool. Table 2 lists processing parameters that have been successfully used to electron beam melt various reactive and refractory metals. Vertical Feeding of Ingots. Refractory and reactive metal ingots of high purity, homogeneity, and smooth surface are remelted by vertical feeding (Fig. 3d). The molten metal droplets run down the conical, rotating electrode tip, are refined, and then drop into the pool center. The crucible pool is normally of the same diameter as the electrode, but it is not necessary. The crucible pool is kept in the liquid state to allow final refining and to guarantee ingot homogeneity. Because two or more electron guns are used, the entire pool can be equally bombarded; thus, shadow effects of the electrode can be eliminated. Simultaneous melting of horizontally and vertically fed electrodes (Fig. 3e) can be used

144 / Melting and Remelting

Fig. 3

Schematics of electron beam consolidation and drip melting processes. (a) Consolidation of coarse and solid scrap. (b) Continuous consolidation of raw material, chips, and solid scrap by direct feeding into a continuous casting crucible. (c) Drip melting of horizontally fed compacts, sintered bars, or consolidates for initial melting of reactive and refractory metals. (d) Drip melting of vertically fed vacuum induction melted or conventionally melted electrodes. (e) Drip melting of horizontally and vertically fed materials for the production of alloys from feedstocks with very different melting points

Table 2 Melting and refining data of refractory and reactive metals and alloy steels gained in laboratory and pilot production furnaces

Metal

Feedstock size, mm (in.)

Tungsten

40 (1.6) diam

Tantalum

55 (2.2) diam, 100 (4) long 60 (2.4) diam

60 (2.4) square, 160 (6.3) long Molybdenum 100 (4) diam 100 (4) diam 120 (4.7) square, 180 (7.1) long Niobium 80 (3.2) square, 150 (6) long 120 (4.7) square, 180 (7.1) long Hafnium 60 (2.4) square 80 (3.2) diam 100 (4) diam 100 (4) diam Zirconium 60 (2.4) square, 100 (4) long 60 (2.4) square, 150 (6) long Vanadium 60 (2.4) diam, 80 (3.2) long Titanium 100 (4) diam Ti-6Al-4V ... Ti-8Al100 (4) diam 1Mo-1V 4340 steel 80 (3.2) diam

Ingot diameter, mm (in.)

Ingot weight, kg (lb)

Integral melting rate, kg/h (lb/h)

Electron beam power of second melt, kW

... 60 (2.4) ... 115 (4.5) ... 80 (3.2) ... 160 (6.3) ... 100 (4) ... 180 (7.1) ... 150 (6) ... 180 (7.1) ... 80 (3.2) ... 130 (5.1) ... 100 (4) ... 150 (6) ... 80 (3.2) ... 100 (4) ... 100 (4) ... 150 (6) 150 (6)

... 37 (80) ... 200 (440) ... 65 (145) ... 523 (1150) ... 64 (140) ... 408 (900) ... 227 (500) ... 326 (720) ... 40 (90) ... 173 (380) ... 19.5 (43) ... 179 (395) ... 7.7 (17) ... 28.2 (62) ... 28.2 (62) ... 31 (70) 31 (70)

... 10.5 (23) ... 20 (44) ... 16.7 (37) ... 38.4 (85) ... 12.5 (27.5) ... 50.2 (111) ... 17.6 (39) ... 13.2 (29) ... 1.7 (3.7) ... 7.5 (16.5) ... 14.3 (31.5) ... 73 (161) ... 2.8 (6.2) ... 45.2 (100) ... 22.5 (50) ... 10 (22) 80 (176)

... 119 ... 300 ... 130 ... 371 ... 130 ... 290 ... 240 ... 218 ... 80 ... 110 ... 80 ... 250 ... 80 ... 87 ... 60 ... 52 80

Operating vacuum pressure of last melt, Pa (torr)

... 8  103 (6  105) ... 2  102 (1.5  10 4) ... 8  103 (6  10 5) ... 3  103 (2  10 5) ... 8  104 (6  10 6) ... 103 (7.5  10 6) ... 102 (7.5  10 5) ... 5  103 (3.8  10 5) ... 5  103 (3.8  10 5) ... 4  103 (3  10 5) ... 8  103 (6  10 5) ... 2  102 (1.5  10 4) ... 8  103 (6  10 5) ... 0.8 (6  10 3) ... 0.4 (3  10 3) ... 4  103 (3  10 5) 2.0 (0.015)

Interstitial elements in feedstock and final ingot, ppm

Total specific melting energy, kW 7 h/kg

Material yield, %

C

O

N

... 10.3 ... 9.9 ... 6.0 ... 8.30 ... 10.4 ... 5.2 ... 12.3 ... 15.9 ... 38 ... 14.7 ... 4.6 ... 2.6 ... 3.1 ... 1.91 ... 2.66 ... 1.67 1.0

... 93.1 ... 90 ... 92 ... 91.8 ... 95 ... 96.8 ... 87 ... 96.8 ... 93.1 ... 93.5 ... 88.2 ... ... ... 91 ... 99 ... 98 ... 93 99

70 10 200 10 100 10 ... 6 170 12 200 10 160 12 80 6 ... ... 500 100 ... ... ... ... ... ... ... ... ... ... 3900 4300 3622

4100 8 5000 8 1200 75 650 15 810 10 750 12 5220 106 4500 111 1870 170 900 200 950 545 950 735 1045(a) 235(a) 2180 1850 890 730 63 2.4 10

30 5 50 7 140 30 13 13 50 10 60 11 554 60 330 52 95 25 100 50 95 30 95 48 210(a) 95(a) 100 80 70 50 100 26 78



H

10 1 20 1 10 1 10 2 10 2 10 2 35 5 40 8 2 1 2 1 30 3 30 9 16(a) 13(a) 30 16 ... ... 0.3 0.08 0.10

Hardness, HB

... 200 ... 210 ... 65 ... 69 ... 140 ... 140 ... 77 ... 66 ... 160 ... 170 ... 120 ... 125 ... 30–100 ... ... ... ... ... ... ...

(a) The reproducibility of the refining data could not be confirmed.

for the production of critical alloys. In this case, the feedstock should be of the desired purity. Horizontally Fed Ingots. In this process, the feedstock size is smaller than the pool diameter to minimize the shadow effect of the horizontally fed bar. In production units, feeding can be carried out from two opposite sides. Other Process Considerations. To ensure the production of clean, homogeneous metals and alloys in electron beam drip melting furnaces, various aspects of material processing and handling must be controlled. Key considerations include:


Dimensions and quality of the feedstock,






and the feeding system used Ingot cooling and unloading during melting of another ingot Passivation and removal of condensates from the melt chamber Planning of melt sequences to minimize the number of furnace cleanings required Routine preventive furnace maintenance to ensure reliability Material yield and energy consumption

Equipment for Drip Melting. The essential equipment groups required for drip melting—

melting furnaces, control systems, and power supply units—are all important for achieving optimum productivity. The melting furnace (Fig. 4) includes the electron beam gun as the heat source, material feeding and ingot withdrawal systems, a crucible for material solidification, and a vacuum system to maintain the low pressure. Process observation, both visually and with video systems, is possible through viewports. The melt chamber flanges are equipped with x-ray-absorbing steel boards, and interlocking systems prevent operation failures and accidents.

Electron Beam Melting / 145

Fig. 4

Single-gun furnace for horizontal drip melting

Table 3 Principal applications for vacuum-arc-remelted (VAR), electron-beam-melted (EB), and powder metallurgy (PM) reactive and refractory metal ingots Metal

Reactive metals, VAR and EB melting Hafnium Vanadium

Zirconium Titanium

Refractory metals, EB melting and PM Tungsten

Tantalum

Molybdenum

Niobium

Applications

Glow discharge tubes for the electronics industry; control rods and breakoff elements in submarine nuclear reactors Targets for high-deposition-rate sputtering processes in the electronics industry; breakoff elements, fixtures, and fasteners in nuclear reactors; standards for basic research; alloying element for certain high-purity alloys Getter material in tubes in the electronics industry; fuel claddings, fasteners, and fixtures for nuclear reactors Components for bleaching equipment and desalination plants in the chemical industry; superconductive wires; turbine engine disks, blades and housings, rain erosion boards, landing legs, wing frames, missile cladding, and fuel containers in the aircraft and aerospace industries; shape memory alloys; biomedical fixtures and implants; corrosion-resistant claddings Heating elements, punches and dies, and nonconsummable electrodes for arc melting and gas tungsten arc welding for metal processing equipment; targets for x-ray equipment and high-sputtering-rate devices such as very large-scale integrated circuits, cathodes and anodes for electronic vacuum tubes in the electronics industry; radiation shields in the nuclear industry; cladding and fasteners for missile and reentry vehicles Condensers, autoclaves, heat exchangers, armatures, and fittings for the chemical industry; electrolytic capacitors for the electronics industry; surgical implants; fasteners for aerospace applications Dies for conventional and isothermal forging equipment; electrodes for glass melting; targets for x-ray equipment; cladding and fasteners for missile and reentry vehicles Superconductive wire for energy transmission and large magnets for the electrical and electronics industries; heavy ion accelerators and radio frequency cavities for nuclear applications; components for aircraft and aerospace applications

The control system allows the adjustment and control of such operating process parameters as electron beam power, operating vacuum level, material feed rate, and ingot withdraw speed. The control system also records and logs the process data.

Power Supply Units. One or more high-voltage power supply units are needed to supply the electron beam guns with the required continuous voltage (30 to 50 kV). The beam power of each gun can be adjusted between zero and maximum power with an accuracy of  þ2%.

Other Equipment. Large-production furnaces are equipped with lock-valve systems to allow simultaneous melting and unloading of ingots without breaking the vacuum in the melt chamber. Production is thus limited only when the condensate remaining in the melt chamber requires cleaning or when a different alloy is to be melted. Characteristics of Electron Beam DripMelted Metals. Electron beam melted and refined material is of the highest quality. The amount of interstitials present is very low, and trace elements of specific high vapor pressure can be reduced to very low values (Ref 1, 2). Tantalum and niobium ingots have smooth surfaces and are of sufficient ductility that they can be cold worked, and sheets and wires can be produced. Tungsten and molybdenum ingots are also of the highest possible purity, but the ingots are brittle because of the very large grain size and the concentration of impurities at grain boundaries. Hafnium. Electron-beam-melted hafnium is of higher ductility than the vacuum-arcremelted metal (Ref 3). The main application of electron-beam-melted hafnium is as control elements for submarine nuclear reactors. Vanadium is refined by electron beam drip melting. The alumino-thermically produced feedstock is drip melted in several steps. During this procedure, the ingot diameter is reduced at each step by approximately 30 to 40 mm (1.2 to 1.6 in.) to obtain an ingot 30 to 40 mm (1.2 to 1.6 in.) in diameter, regardless of the initial ingot diameter. The clean vanadium ingots are primarily used in nuclear reactor applications (Ref 4). Applications for electron-beam-melted refractory and reactive metals are listed in Table 3.

Cold Hearth Melting The cold hearth melting process (Fig. 5) was developed approximately 10 years after drip melting (Ref 5). Cold hearth melting is mainly used for refining and recycling reactive metal scrap, especially titanium turnings machined with high-density tungsten carbide tool tips (Ref 6) and for refining specialty steels and superalloys. Principles of Cold Hearth Melting. Cold hearth melting (Fig. 6) is the most flexible vacuum metallurgical melting process. It is a two-stage process. The first step (material feeding, melting, and refining) takes place in a water-cooled copper trough or hearth. In the second step, solidification occurs in one of several round, rectangular, or specially shaped water-cooled continuous copper crucibles. Both process steps are nearly independent from each other; they are linked only by the flow of the liquid metal stream. The major refining actions are carried out in the hearth, but some

146 / Melting and Remelting

Fig. 5

Schematic of the cold hearth melting process

postrefining takes place in the pool of the continuous casting crucible, similar to the drip melting of horizontally fed billets. Refinement in cold hearth melting occurs by vacuum distillation in the hearth pool, superheating, and stirring of the molten metal pool. Removal of Impurities. Most impurities with densities lower than that of the melt (for example, oxides in superalloys) can be segregated by flotation and formed into a slag raft. The raft is then held in place by either mechanical or electrothermal means. Impurities denser than the melt, such as tungsten carbide tool tips in titanium, are removed by sedimentation. Inclusions with densities such that efficient flotation or sedimentation does not occur can be partially removed by adhesion to the slag raft or by dissolution in the liquid pool. Hearth dimensions are based on the type and amount of refining required. For example, hearths for vacuum distillation should be nearly square and relatively deep to allow sufficient melt stirring. For flotation refining, the hearth should be

long and narrow. Hearths for titanium alloy scrap recycling can be relatively short if all the materials can be transported to the pool of the hearth rather than to the ingot pool. Longer hearths provide for longer residence times, improving the removal of impurities by dissolution. Feeding. Material feeding criteria include 100% homogeneous material transportation to avoid uncontrolled evaporation of alloying elements and correct feeding into or above the hearth pool. Horizontal feeding of compacted, premelted, or cast material is most often used. Loose scrap and raw material can be used. Feeding of liquid metal was used in one of the first cold hearth melting furnaces to produce ferritic steel in a vacuum induction furnace (Ref 7). Postrefining was carried out in a cascade of five hearths 1.5 m (60 in.) long and 1 m (40 in.) wide. Casting and Solidification. The criteria for material casting and solidification include the shape of the final product and the solidification rate required to avoid ingot tears or other

defects and to ensure a homogeneous ingot structure. The multiple casting of small ingots is sometimes used, especially when forging is impossible because of the brittleness of the solidified material (for example, MCrAly wearresistant coating alloys). The casting of round and rectangular ingots and slabs is common practice. Cold Hearth versus Drip Melting. Table 4 compares the essential features of drip melting and cold hearth melting. Generally, cold hearth melting is used for titanium and superalloys especially when flotation or sedimentation of inclusions is required. Drip melting is used for other refractory metals because of the high melting points and the resulting high heat losses to the water-cooled copper crucible. Depending on production quantity, double or triple drip melting may require less energy than a single cold hearth melt of some materials, such as niobium. Refining and Production Data. Data on cold hearth electron beam melting and refining in laboratory and pilot production furnaces are given in Table 5. The data demonstrate the effectiveness of the process in reducing impurities and interstitial elements. It can also be seen that the selective evaporation of chromium from superalloys can be controlled by the distribution of beam power at the trough pool and by controlling trough pool area and melt rate. The selective evaporation of aluminum from Ti6Al-4V alloy is much more difficult to control; additional aluminum must be used to compensate for the aluminum evaporated. Equipment for Cold Hearth Electron Beam Melting. The equipment required for cold hearth melting is different from that used in drip melting, mainly because of the hearth and the larger melting chamber. In addition, because of the materials often melted in the cold hearth process (superalloys and titanium alloys), additional instrumentation is often provided. This may include an ingot pool level control system, metal vapor and partial pressure analyzers, a two-color temperature control system, and a data logging system. Accurate beam power distribution is achieved in multiple-gun furnaces by microprocessor control, which allows the splitting of a single beam to many locations and the adjustment of dwell time at each location. The beam spot at each of the locations can be scanned to create spot patterns as required by the application. With such systems, the required heating can be achieved without unnecessary power consumption and evaporation of alloying elements. Process observation is accomplished with a video monitoring system. Samples can be obtained from both the trough pool and the ingot pool for nearly continuous control of material quality. Feeding systems for cold hearth furnaces must maintain homogeneity along the length of the feed material. The trough and crucible should be easily accessible for convenient maintenance, especially when different alloys are to be melted in the same furnace.

Electron Beam Melting / 147 Characteristics of Cold-Hearth-Melted Materials. Titanium ingots and slabs can be produced from titanium scrap contaminated with tungsten carbide tool tips. The electronbeam-melted product will not contain any tungsten carbide particles. Oxygen content can be controlled by mixing titanium sponge with titanium revert material in the appropriate ratio. Cold-hearth-melted titanium ingots can be remelted in a vacuum arc remelting furnace. In some cases, depending on alloy and application, the product of the cold hearth furnace can be used without additional melt processing. Superalloys. Cold-hearth electron-beammelted superalloy ingots 150 to 200 mm (6 to 8 in.) in diameter are often used in vacuum induction melting investment casting furnaces. Such ingots are nearly free of nonmetallic inclusions and trace elements (Ref 9).

ACKNOWLEDGMENT Revised from article by W. Dietrich and H. Stephan, Electron Beam Melting and Casting, Metals Handbook, 9th Edition, Volume 15, 1988 pp 410–418

REFERENCES

Fig. 6

Four-gun 1200 kW combined electron beam drip melting and cold hearth melting furnace

Table 4

Comparison of the characteristics of drip melting and cold hearth melting

Characteristic

Power density Inclusions Ingot shape and structure Mass production Competitive economical processes Preferred method

Refractory metals

High Irrelevant Round; coarse grain Low Vacuum arc remelting Drip melting

Reactive metals, superalloys, and specialty steels

Soft; smoothly distributed Must be removed Round or flat; fine grain, segregation-free High Vacuum arc remelting; electroslag remelting Cold hearth melting

1. R.E. Lu¨ders, Tantalum Melting in a 800 kW EB Furnace, Proceedings of the Bakish Conference on Electron Beam Melting and Refining (Reno, NV), R. Bakish, Ed., 1983, p 230–244 2. J.A. Pierret and J.B. Lambert, Operation of Electron Beam Furnace for Melting Refractory Metals, Proceedings of the Bakish Conference on Electron Beam Melting and Refining (Reno, NV), R. Bakish, Ed., 1984, p 208–218 3. H. Sperner, Hafnium, Metallwis. Technik., Vol 7 (No. 16), 1962, p 679–682 4. R. Ha¨hn and J. Kru¨ger, Refining of Vanadium Aluminium Alloys to Vanadium 99.9% by EB Melting, Proceedings of the

148 / Melting and Remelting Table 5 Refining and production data for the melting of reactive and refractory metals and stainless steels in laboratory and pilot production furnaces

Metal

Feedstock size, mm (in.)

Hafnium

60 (2.4) square

Zirconium

100 (4) square

Zirconium

80 (3.2) square

Vanadium

50 (2) square

Ti-6Al-4V

Swarf

Ti-6Al-4V

Solid scrap

Ti-6Al-4V

125 (5) diam

Commercially 160 (6.3) pure titanium Commercially Sponge pure titanium Stainless steel 150 (6) diam Alloy 718

133 (5.2) diam

AISI type 316 stainless steel

150 (6) diam

Trough size, mm (in.)

Ingot size, mm (in.)

Ingot weight, kg (lb)

Melt rate, kg/h (lb/h)

Electron beam power, kW

120  250 (5  10) 120  300 (5  12) 120  300 (5  12) 120  300 (5  12) 120  300 (5  12) 120  300 (5  12) 150  400 (6  16) 150  250 (6  10)

100 (4) diam

83.0 (183)

40 (88)

180

150 (6)

90.5 (200)

42 (92.5)

185

100 (4)

40.2 (89)

80 (176)

140

Operating pressure, Pa (torr)

Specific melting energy, kW 7 h/kg

Composition of feedstock and product



C, ppm

4  10 2 (3  10 4) 3.5  10 2 (2.6  10 4) 3.5  10 2 (2.6  10 4) 1.5  10 2 (1.1  10 4) 2  10 2 (1.5  10 4) 7  10 2 (5.3  10 4) 6  10 2 (4.5  10 4) 6  10 2 (4.5  10 4)

4.5 4.4 1.75

O, ppm

N, ppm

H, ppm

Al, %

V, %

Cr, %

... ... ... ... ... ... ... ... 400 200 1520 ... ... ... ... ...

900 . . . 600 . . . 950 95 540 30 4000 800 1520 210 1045 210 277 50 2600 110 2700 110 1520 75 1320 76 ... ... ... ... ... ... ... ...

... ... 30 3 10 3 10 3 84 22 15 8 ... ... ... ...

... ... ... ... ... ... ... ... 6.0 4.4 6.0 4.8 6.0 3.6 ... ...

... ... ... ... ... ... 99 99 4.0 4.2 4.0 4.1 4.0 4.3 ... ...

... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...

100 (4)

...

20 (44)

130

150 (6)

62.6 (138)

40 (88)

122

150 (6)

62.6 (138)

70 (154)

140

2  75 (3) diam 100  400 (4  16)

2  32 (70.5)

91 (200)

147

96.4 (213)

86.3 (190)

148

150  500 (6  20)

100  400 (4  16)

103.0 (227)

41.2 (91)

226

8  10 2 (6  104)

5.5

...

...

...

...

...

...

...

150  400 (6  16) 150  400 (6  16) 150  400 (6  16)

2  75 (3) diam 2  75 (3) diam 3  65 (2.6) diam

2  55 (121)

136 (300)

144

1.06

2  57 (126)

136 (300)

156

3  41.5 (91.5)

136 (300)

156

6  10 2 (4.5  10 4) 6  10 2 (4.5  10 4) 6  10 2 (4.5  104)

701 536 417 363 ...

97 33 14 17 ...

155 68 52 34 ...

... ... ... ... ...

... ... ... 0.72 ...

... ... ... ... ...

18.25 18.11 19.11 18.73 ...

6.5 3.0 2.0 1.61 1.71

1.15 1.15

Source: Ref 8

Bakish Conference on Electron Beam Melting and Refining (Reno, NV), R. Bakish, Ed., 1986, p 53–67 5. C.d’A. Hunt and H.R. Smith, Electron Beam Processing of Molten Steel in Cold Hearth Furnace, J. Met., Vol 18, 1966, p 570–577 6. H.R. Harker, Electron Beam Melting of Titanium Scrap, Proceedings of the Bakish Conference on Electron Beam Melting and

Refining (Reno, NV), R. Bakish, Ed., 1983, p 187–190 7. C.d’A. Hunt, H.R. Smith, and B.C. Coad, The Combined Induction and Electron Beam Furnace for Steel Refining and Casting, Proceedings of the Vacuum Metallurgy Conference (Pittsburgh, PA), June 1969, p 1–22 8. H. Stephan, Production of Ingots and Cast Parts from Reactive Metals by Electron

Beam Melting and Casting, Proceedings of the Third Electron Beam Processing Seminar (Stratford, U.K.), 1974, p 1b1–1b69 9. M. Krehl and J.C. Lowe, Electron Beam Cold Hearth Refining for Superalloy Revert for Use in Foundry Production, Proceedings of the Bakish Conference on Electron Beam Melting and Refining (Reno, NV), R. Bakish, Ed., 1986, p 286–296

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 149-154 DOI: 10.1361/asmhba0005205

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Plasma Melting and Heating Reviewed and revised by Robin Lampson and Todd Telfer, Retech Systems

PLASMA MELTING is a material-processing technique in which the heat of a thermal plasma is used to melt a material. Plasma torches produce heat from an electrical current that passes through a partially ionized gas (or plasma). The plasma is generated by the phenomenon known as gaseous discharge, whereby a gas is ionized by electrical breakdown. The electrical breakdown of the gas can be accomplished by an applied voltage between two electrodes (as in the case of fluorescent lights). Plasmas also can be generated by various “electrodeless” methods that involve radio frequency (RF) discharges, microwaves, high-frequency induction coils, shock waves, lasers, or highenergy particle beams. Plasma arc torches have several advantages. Unlike heating by consumable electrodes (such as tungsten arc or graphite electrodes), plasma arc heating involves a gas medium instead of a consumable electrode. The gas environment and flow rates can be controlled, and only a small percentage of ionized gas (less than ½% of the total gas flow) is sufficient to create heating temperatures. Without a consumable electrode, plasma arc can generate higher temperatures than the electric arc methods. Products of combustion are eliminated or reduced. In addition, high inert gas pressure (5 kPa, or 50 mbar, or 37.5 torr, and higher) can be obtained with plasma torches. This is an important prerequisite for the avoidance of selective evaporation of alloying elements from the melt, an aspect that is particularly important in the case of titanium and superalloys. Plasma arc torches also supply electric heat to a ladle of molten metal to either maintain the temperature or raise the temperature of the melt. For numerous steel grades, especially those with extremely low carbon contents, plasma arc ladle heating is an alternative to heating with graphite electrodes and the risk of carbon pickup. Highpower plasma torches can be used with minimum modification to the existing ladle design. Prior to their integration into steel ladle metallurgy stations, plasma torches had been used primarily for scrap melting and heat-loss compensation before and during continuous casting. Only thermal (hot) plasmas are considered in this article. The term thermal differentiates

these plasmas from cold (nonequilibrium) plasmas, which are characterized by high electron temperatures and low sensible temperatures of the heavy particles. Cold plasmas are produced in various types of glow discharges, low-pressure RF discharges, and corona discharges. Cold plasmas are not suitable for melting processes. Thermal plasmas are suitable for melting in the production of ingots, slabs, castings, or powders.

Plasma Torches Breakdown (that is, the creation of ionic charge carriers) establishes a conducting path for resistance heating within a plasma torch. Most plasma generators (plasma torches) for melting processes use an electric arc to produce gaseous discharges. The characteristics of an electric arc include relatively high current densities, low cathode fall, and high luminosity of the column. A typical potential distribution along an arc is shown in Fig. 1. The column cross section at the cathode is smaller than at the anode, with contrary current and power densities. This fact influences the design principle of plasma torches. Most plasma heaters operate on direct current (dc) and are designed as either nontransferred or transferred arcs (Fig. 2). However, threephase alternating current plasma torches are employed in plasma heating and degassing applications. These units do not need a bottom electrode but do require more sophisticated power supplies than dc plasma torches do in order to limit the torch current and to ensure a quick, uninterrupted current pass through zero. In a nontransferred plasma torch, the gas is the only resistive element that generates heat. Nontransferred arcs are desirable for hot gas contact in shaft heaters, melters (cupolas), and reactors such as blast furnaces. This mode is suitable for the melting of conductive and nonconductive materials, but it has the disadvantage of lower efficiency compared to the transferred mode. Examples of nontransferred plasma torches are illustrated in Fig. 3. Magnetic coils are used to direct the flow of the plasma in the torch. For the electrodeless RF

torch, the plasma does not depend on the physical connection of electrodes. Transferred Arc Plasma Torches. In the transferred mode (Fig. 2b), one electrode is inside the torch, and the counterelectrode is the material to be melted. With the workpiece being one of electrodes, it provides the medium for resistive heating of both conductive and nonconductive materials. Most nonconductive materials become conductive when heated. Transferred arc heaters are useful for bulk heating of conducting solids and melts, where the gas phase is not a reactant but can be used as an inert shield. Figure 4 shows two typical design principles for torches in the transferred mode: the tungsten tip design and the hollow copper electrode design. In the tungsten tip design (Fig. 4a), the torch electrode is connected as the cathode because high current densities are required. The cathodes are usually made of tungsten, along with small additions of thoria to lower the thermionic work function of tungsten. Still, electron emission requires high electrode temperatures (3500 to 6000 K, or 5800 to 10,300  F) at the attachment of the arc. Therefore, the cathode material is liquid at the arc attachment point even with water cooling at the rear of the

Fig. 1

Typical potential distribution along a plasma arc. Va, anode voltage; Vc, cathode voltage; da, anode current density; dc, cathode current density

150 / Melting and Remelting

Fig. 2

Configurations of (a) Nontransferred arc plasma torch and (b) transferred arc plasma torch

Fig. 3

Examples of some typical nontransferred plasma torch configurations. (a) Electrode-based torches. (b) Radio frequency “electrodeless” plasma torch

cathode. For this reason, oxidizing gases generally cannot be used in direct contact with the tungsten tip. Argon shielding is one solution. Further, it should be kept in mind that portions of the small liquid pool at the cathode tip are ejected into the material being melted. The high melting point of tungsten can result in inclusions in the product; such defects are not tolerable in high-strength metals such as titanium alloys and superalloys for aircraft applications. With argon as the plasma gas, the service life of the cathode in the tungsten tip design is typically between 30 and 150 h. The inert gas consumption and plasma column stabilization of such torches are low. The hollow copper electrode design is shown in Fig. 4(b). The electrode material, usually copper alloyed with chromium or zirconium, is intensively water cooled. Copper is a field emitter of electrons (as opposed to tungsten, a thermionic emitter), because its boiling point is substantially below that required for thermionic electron emission. Current densities at the attachment points on cold cathodes are double those for thermionic emission on hot cathodes. Cathode spots are of high intensity, rapidly moving on the surface of the cold cathode and causing erosion by locally vaporizing copper. For this reason, the polarity is usually reversed in plasma torches with cold electrodes. The rear electrode acts as the anode, and the workpiece or melt is the cathode. The anodic arc attachment is smoother, and erosion is reduced. The plasma-forming gas is blown in tangentially between the rear electrode and the nozzle to create a swirl or vortex to stabilize the arc and to rotate the arc attachment in the rear electrode. With the movement of the arc attachment, a substantially lower power density can be achieved in the rear electrode. Electrode life of 1000 h has been demonstrated in torches using argon or helium as the plasma gas. All inert and reactive gases, or mixtures of these, can generally be used in such torches, but the influence of the gases on electrode erosion varies. The gas consumption of vortex-stabilized torches is higher than that of other designs. The efficiency of plasma torches is less affected by design principles than by process parameters and by the type of plasma gas used. Efficiency is defined as Pw/Pt, where Pt is the electrical power input of the torch, and Pw is the thermal power applied to the workpiece or melt. The efficiency of a 650 kW torch measured in a cold furnace ranges from approximately 25 to 50%, depending on the plasma gas used, furnace pressure, and the distance between torch and workpiece. The highest efficiency is obtained with helium as the plasma gas and a short distance between material and torch. The power loss in torches with hollow water-cooled electrodes is approximately 25%. Remaining additional 25% losses are due to radiation and convection of the plasma column between torch and workpiece. These losses are primarily a function of

Plasma Melting and Heating / 151

Melting Processes

Fig. 4

Design concepts for plasma arc torches in the transferred mode. (a) Torch with tungsten tip and concentric gas flow. (b) Torch with hollow copper electrode and vortex generator

the distance between torch and workpiece, but radiative losses of the plasma column also increase with pressure.

Furnace Equipment Plasma melting furnaces usually consist of double-wall, vacuum-tight, water-cooled constructions that can withstand the radiative and convective heat transfer from the plasma column. Furnaces also usually have material feeding systems. These can consist of vibratory or rotary feeders with bins (for bulk materials) or bar feeding systems. If the furnace is to operate continuously, the feed material must be “locked in” to prevent disturbance of the furnace atmosphere. The furnace must also have a vacuum pump so that it can be evacuated before backfilling. If the furnace is operated under reduced pressure, an offgas pump with an on-line gas control valve is required.

Atmosphere Control Plasma melting furnaces are usually operated under slightly positive pressure to prevent the potential atmospheric contamination by oxygen and nitrogen. However, state-of-the-art furnaces are vacuum-tight and can be operated at pressures between 5 and 200 kPa (50 and 2000

mbar, or 38 and 1500 torr). Vacuum-tightness is essential because the back diffusion of oxygen, nitrogen, and moisture through small leaks can be easily demonstrated even with positive pressure in the furnace. As noted, plasma torches are normally operated with argon or helium plasma gas. A typical gas purity of 99.999% indicates 10 ppm gaseous impurities and corresponds to an absolute pressure of 1 Pa (0.01 mbar, or 0.0075 torr) in a vacuum process. Additional sources of atmospheric contamination are the moisture and gases in the feed material, desorption of the furnace wall, and leaks in the furnace construction. In the case of metals with high affinities for oxygen or nitrogen, these gases are totally absorbed by the melt and will be found as oxide or nitride inclusions in the product. To obtain a clean initial furnace atmosphere, the vacuum-tight plasma furnace is evacuated with a mechanical pump to final pressure (typically approximately 1 Pa, or 0.01 mbar, or 0.0075 torr). The furnace walls can be degassed with the help of warm water running between the double walls of the furnace to enhance moisture desorption. The furnace is then backfilled with high-purity inert gas up to 1 kPa (10 mbar, or 7.5 torr), evacuated again to 1 Pa (0.01 mbar, or 0.0075 torr), and backfilled to operating pressure. This procedure dilutes residual gas impurities by a factor of 1000.

Plasma furnace equipment is of great interest when selecting a suitable heat source for melting, remelting, casting, and atomizing of reactive metals, titanium, and superalloys. Plasma torches, however, are the only nonconsumable heat sources for melting under high inert gas pressures in skull crucibles. High pressure is an essential requirement for preventing the selective evaporation of alloying elements that are characterized by high activity coefficients and/or vapor pressures, such as chromium and manganese in superalloys and aluminum in titanium alloys. Various plasma melting processes have been developed, including plasma consolidation, plasma arc remelting, plasma cold hearth melting, and plasma casting. Plasma cold hearth melting and plasma cold crucible casting are thought to have the most potential for industrial application. Table 1 lists some typical data on feed materials, operating parameters, and products of furnaces for some of these processes. With plasma furnaces, several companies are producing 3 to 5m (120 to 200 in) slabs with sections 50 to 89 cm (20 to 35 in.) thick and 100 to 150 cm (40 to 60 in.) wide. Plasma Consolidation. Consolidation implies that low-density feed material is converted into a high-density product. Plasma consolidation is used for titanium scrap and sponge to produce an electrode for further remelting in vacuum arc remelting (VAR) furnaces. The material is usually fed directly into a watercooled copper crucible and melted by plasma heat from one or several plasma torches (Fig. 5). The water-cooled copper baseplate of the crucible is continuously withdrawn. The shape and depth of the liquid pool in the crucible depend on feed rate and power input. Under these conditions, 100% melting cannot be ensured; some material can be trapped at the liquid/solid interface before melting. Figure 6 shows ingot density as a function of the raw material feed rate and power consumption in the plasma consolidation of titanium. Plasma consolidation processes can reduce manufacturing costs for the preparation of titanium VAR electrodes, which are usually plasma melted from pressed sponge or chip compacts. With plasma consolidation, metal chlorides can be removed from titanium sponge to sufficiently low levels. Tungsten carbide tool tips in titanium chips cannot be removed, and they eventually form high-density inclusions in the product. Therefore, the conventional x-ray inspection of scrap is still necessary. Low-density inclusions such as titanium nitride particles have not been found after plasma consolidation of contaminated titanium scrap and subsequent VAR remelting. Plasma consolidation has also been successfully applied to the processing of titanium aluminide (TiAl), a low-density alloy with excellent high-temperature service properties.

152 / Melting and Remelting Table 1 Characteristics and operating parameters of furnaces for plasma melting processes Type

Feed material

Plasma consolidation Titanium scrap and furnaces sponge Plasma consolidation Titanium scrap furnaces and sponge Plasma arc remelting Titanium, titanium furnaces alloys Plasma arc remelting Specialty steels, furnaces heat-resistant alloys

Plasma cold hearth melting furnaces Plasma cold hearth melting furnaces

Specialty steels, heat-resistant alloys Titanium scrap and sponge Titanium scrap and sponge, superalloys

Plasma cold crucible Titanium scrap melting and and sponge, casting furnaces superalloys

Fig. 5

Product form and size, mm (in.)

Ingot 355–430 (14–17) diam, 3000 (120) long; density >90% Ingot 700 (27.5) diam, 3800 (150) long; density 98% Ingot 125 (5) diam, 500 (20) long Ingot 150–200 (6–8) diam, 1200 (47) long; slabs 70  300  1200 (2.8  12  47) Ingots 650 (25.5) diam, 1500–2300 (60–90) long Slab 245  1125  2450 (9.6  44.3  96.5)

Specific power consumption, kW
h/kg

Plasma gas

250–260 (550–570)

1.4–1.6

Argon

Atmospheric

Tungsten tip with external ignition bar; 6  90 kW

250–670 (550–1480)

0.89–2.4

Argon

Atmospheric

50–70 (110–155) 90–135 (200–300)

2.0–2.5

...

Atmospheric

1.5–2

...

Atmospheric

Hollow copper electrode with graphite liner; 600 kW Probably tungsten tip; 500 kW Probably tungsten tip; 500–1000 kW

400–900 (880–1985)

1.2–1.8

...

Atmospheric

Probably tungsten tip; 300–3600 kW

1–13 Pa (0.01–0.13 mbar)

Six hollow cathode torches; 700 kW on hearth; 1170 kW on crucible Hollow copper electrode, vortex stabilized; 200 kW on hearth, 100 kW on crucible Hollow copper electrode, vortex stabilized; 650 kW

Typical melt rate, kg/h (lb/h)

Typical furnace pressure

330 (730)

9

...

Ingot 195 (7.7) diam

90–227 (200–500)

1.3–3

Helium; gas recycling

Atmospheric

Casting ingots

380 (840)

1.3

Argon; gas recycling

50 kPa (500 mbar)

Plasma consolidation furnaces. (a) Multiple electrode. (b) Single electrode

Plasma Arc Remelting. An electrode that has already been melted by a primary melting process such as electric arc or induction melting can be plasma arc remelted into a water-cooled withdrawal crucible. The major objectives for plasma arc remelting are:  To obtain directional solidification without

changing the chemical composition of the feed material  To improve cleanliness by removal, size reduction, shape control, and redistribution of inclusions  To lower gas impurity content  To alloy with nitrogen Plasma arc remelting is used most frequently in the former Soviet Union countries to produce high-temperature alloys, bearing steels,

superalloys, and titanium alloys. Plasma arc remelting can be advantageous for the nitrogen alloying, deoxidation, or desulfurization of nonreactive alloys. An example of successful reactive melting is the nitrogen alloying of steel during plasma arc remelting with argon-nitrogen mixtures as the plasma gas. Figure 7 shows nitrogen concentrations in various metals as a function of nitrogen partial pressure in the gaseous phase after plasma arc remelting. The data indicate that excited molecules of nitrogen in the plasma react with the metal. In plasma cold hearth melting, bars, sponge, or scrap can be continuously fed into a water-cooled copper trough, usually called a hearth, and melted by means of plasma heating (Fig. 8). The liquid metal flows over the lip of the hearth into a withdrawal mold. The plasma cold hearth melting process is adapted from

Torch design principle and power

the electron-beam cold hearth refining process, which is a well-developed refining and casting method for superalloys. Plasma heat, instead of an electron beam, is employed to minimize selective evaporative losses that occur during electron-beam melting in the pressure range of 0.01 to 1 Pa (10 4 to 0.01 mbar, or 7.5  10 5 to 0.0075 torr). In the hearth, high-density inclusions from tungsten carbide tool tips will sink down to the skull. Low-density inclusions can be dissociated or dissolved in the base metal during interaction with the superheated melt and the plasma jet. Process optimization is necessary to prevent short-circuiting of the hearth by unmelted particles and to establish sufficient dwell time at high melting rates. The power distribution of one or more plasma torches required for smooth melting and fluid flow can be ensured by automated torch-moving devices (Fig. 9). In unconventional configurations, there is no mechanism provided for torch movement. Figure 8(b) shows a modern plasma cold hearth melting furnace concept with a doublewall, water-cooled, vacuum-tight furnace chamber and three plasma torches for the production of ingots 710 mm (28 in.) in diameter. Two torches serve the hearth, and the third the ingot mold. The torches use hollow water-cooled copper electrodes. The high gas consumption of this torch design means that an effective gas recycling system must be provided, especially if helium is used as the plasma gas. A key technology in the plasma melter is gas recycling systems that include reactive filters, oil-free high-capacity pump sets, and compressors. If necessary, the recycling system can be supplemented with cleaning systems for oxygen, carbon dioxide, nitrogen, hydrogen, and moisture. Helium gas recycling systems are

Plasma Melting and Heating / 153 now operating at 99.5% recycle efficiency (less than 0.5% of full gas flow). Plasma cold crucible casting is an alternative process to vacuum arc skull melting and casting or electron-beam skull melting and casting, and it can overcome some of the disadvantages of these processes. The disadvantages of vacuum arc skull melting in the melting and casting of titanium and superalloys are:  Control of superheating is not possible;

increased power results in increased melting rate instead of superheating.

 Expensive electrodes must be used instead

of the high-quality unconsolidated revert material that can be used for plasma cold crucible casting. For small batches of superalloys (up to 8 kg, or 18 lb), electron-beam skull casting is a well-established technique. Electron-beam melting and casting is rarely used for large castings of either superalloys or titanium alloys. This is because of the shallow melt pool in the crucible due to lack of stirring action and the potential for selective evaporation of alloying elements during melting.

The concept of plasma cold crucible melting is shown in Fig. 10(a). Solid material can be charged continuously into the water-cooled copper crucible, and additional material can be fed into the crucible during melting. The melt is poured into stationary molds or centrifugal casting molds situated in the mold chamber. The pouring stream can be superheated during casting or for production of atomized powder. An example of a cold crucible casting furnace (Fig. 10b) shows a horizontally installed mold chamber with the melt chamber on top of the mold chamber. In this example, a programmable torch-moving device is installed on top of the melt chamber. The pressure in the furnace during the melting operation can be preselected; pressures typically range from 30 to 50 kPa (300 to 500 mbar, or 225 to 375 torr). The melting operation can be started at every pressure in this range. The typical casting weight of this furnace is 27 kg (60 lb) of titanium with a 650 kW plasma torch. To achieve a deep molten pool, a high heat flow from the plasma torch to the melt must be established in combination with high stirring velocity in the melt to enhance the effective heat-transfer coefficient. The stirring action in the melt can be regulated by controlling current, gas flow, and torch-to-melt distance. Continuous feeding of the metal to be melted is required in order to achieve a deep pool. In contrast to vacuum arc skull melting, the crucible need not be inspected after each melt. The power densities in the plasma attachment area are adjusted to avoid damaging the crucible if it is inadvertently exposed to the plasma for short periods of time.

Fig. 6

Relationship among raw material feeding rate, ingot bulk density, and specific power consumption in the plasma consolidation of titanium. Source: Ref 1

Fig. 7

Nitrogen concentrations in various metals as a function of the partial pressure of nitrogen after plasma arc remelting. (a) Iron. (b) 16–25–6 stainless steel. (c) Austenitic stainless steel. Curves A and B, induction heating; minimum and maximum nitrogen concentrations, respectively. Curve C, plasma arc remelting. Source: Ref 2

154 / Melting and Remelting

Fig. 8

Plasma cold hearth melting. (a) Concept. (b) Configuration of system for recycling scrap

Fig. 9

Plasma torch-moving device. Courtesy of Leybold AG

REFERENCES 1. T. Yagima et al., Development of the Plasma Progressive Casting Process and Its Application for Titanium Melting, Titanium 1986: Products and Applications, Vol II, Proceedings of the Technical Program from the 1986 International Conference, Titanium Development Association, 1987, p 985–993 2. G.K. Bhat, Plasma Arc Remelting, Plasma Technology in Metallurgical Processing, J. Feinman, Ed., Iron and Steel Society, 1987, p 163–174

Fig. 10

Cold crucible plasma melting. (a) Melting, pouring, and superheating for powder atomization. (b) Industrialsized plasma cold crucible casting furnace with 650 kW torch power

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 155-159 DOI: 10.1361/asmhba0005198

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Crucible Furnaces George Giles, Morgan Molten Metal Systems

CRUCIBLES, in the form of well-fired clay pots, were present at the dawn of the metals industry, some 6000 years ago, and gave early metal workers the ability to melt metal in a durable container that could then be used to pour the molten material into a mold. Few, if any, metallurgical or industrial processes have a more distinguished history than the crucible melting of metals. Crucibles or earthenware smelting pots that were used to refine copper from King Solomon’s mines can still be found in that area of ancient Israel. Some pots had capacities of up to 0.40 m3 (14 ft3). The Wootz steels of India, the famous Damascus sword blades, and the equally famous steels of Toledo, Spain, were produced using crucible melting processes. At their most basic level of form and function, crucibles today (2008) are little changed from their early prototypes. However, in terms of materials, construction, thermal efficiency, and service life, modern 21st century crucibles (Fig. 1) are superior to those in use just a short time ago. Designed for a wide range of traditional applications and new processes, crucibles are formed using a wide variety of materials and processes that impart specific characteristics, capabilities, and limitations to each of the many types of crucibles available from crucible manufacturers. Crucibles may be as small as teacups or may hold several tons of metal.

Fig. 1

Large electric bale-out crucible furnace. Courtesy of Morgan Crucible

They may be fixed in place within a furnace structure or may be designed to be removed from the furnace for pouring at the end of each melt. Crucibles are used in gas- and oil-fired furnaces, electric resistance furnaces, and induction furnaces. They come with or without pouring spouts and in a wide variety of traditional and specialized shapes. There is no direct impingement of the flame on the metal, and the heat loss to the outside is limited by the refractory walls. However, the energy efficiency of the crucible furnace can be low (7 to 19%), with over 60% of the heat loss attributed to radiation (Ref 1). Without proper care, the life of the crucible furnace may be short. Temperature control may also be difficult.

Applications In the metals industry, most crucibles are used to melt, hold, and/or transfer nonferrous metals—commonly aluminum, zinc, and their alloys and copper and copper-base alloys—as well as precious metals. A crucible furnace is the least expensive melting method preferred for melting small volumes of nonferrous metals. It is also used for higher-temperature alloys, such as nickel bronze and cupronickel, and, to a lesser extent, for melting ferrous metals, such as gray iron. Crucible furnaces are popular in jobbing foundries and die casting shops because they are easy to tap and charge with another alloy. Foundry melting and handling operations with crucible equipment include furnace melting, pouring ladle equipment (either separately heated or as a ladle liner), and liners for electric induction melting furnaces. Crucible melting can also be used as a component of qualityadjustment equipment between melter and casting furnaces to alter the chemistry and temperature of the melt for control of such factors as porosity content prior to casting. Metal-processing operations include in-line filtration, fluxing, and hydrogen gas content control. Crucible Advantages. Crucible melting is a simple and flexible process. Crucible furnaces can generally be started or shut down at a moment’s notice. With their unique ability to

melt, hold, and transfer metal using a single vessel and to allow even incompatible alloy changes to be made simply by switching vessels, crucibles provide significant operational flexibility for various foundry applications. Even when fixed within the furnace structure, crucibles offer important advantages when compared to directly heated fuel-fired furnaces and to electric resistance or induction furnaces with rammed refractory linings. Advantages include:  Small, compact size and relatively lower

acquisition cost are economical.

 Smaller (and round) furnaces provide and

maintain good, even temperature profiles.

 Open crucible furnaces, or even those with

pivoting lid covers (Fig. 2), allow easy access for ladling. Lower Metal Loss/Cleaner Metal. In fuelfired furnaces where the metal is heated directly by a flame produced by gas or oil burners, there are both high levels of metal loss and increased oxidation due to the combustion products generated. In fuel-fired crucible furnaces, however, the flame is isolated from the metal, and the metal is heated indirectly with heat conducted through the crucible. This indirect heating greatly reduces metal loss and metal oxidation, resulting in lower metal purchase costs, lower energy costs per pound of metal poured, and less slag or dross formation. Alloy Flexibility. Even when fixed in place in a furnace, the high-density interior surface of a crucible reduces the likelihood of cross contamination when alloy changes are made. Quick Replacement. Removing a worn or damaged crucible from a furnace and installing a new crucible is a faster, simpler, and less technically demanding process than removing and replacing a worn or damaged rammed refractory furnace lining. Also, because crucibles are preformed to exact material and dimensional specifications, there are few concerns about improperly compacted refractory or incomplete sintering. Matching Crucible and Application. Determining the required properties of an application is the most important step in selecting the proper crucible. Knowing the metal to be

156 / Melting and Remelting  Glass: Glass is used to provide a semiperme-

able barrier to gases, controlling the ingress of oxygen and egress of combustion products. Typically, 10 to 20% of a crucible is glass.  Aluminosilicates: Aluminosilicates are used to modify the refractory properties of the crucible or internal glass composition. Unfortunately, there is not a single crucible material recipe or manufacturing process that offers the highest level of every desirable characteristic for every application. Therefore, the best course is to prioritize the list of crucible properties most important for the application and review those requirements with the crucible provider to find the crucible product best suited for the specific needs.

Furnace Types

Fig. 2

Crucible furnace in die casting foundry. Courtesy of Morgan Crucible

melted, held, and/or transferred, however, is just the start in selecting the right crucible type for the application. The specific crucible properties that are important for a particular application must be considered. These properties include:  Maximum temperature limits: It is important











that the highest temperature reached by the application does not exceed the maximum temperature limit for the crucible; otherwise, the crucible will fail. Tolerance of thermal shock: If the application results in rapid heating or cooling of the crucible, a crucible type must be selected that resists thermal shock. Impact resistance and mechanical strength: If the crucible in the application is used to transfer metal or if it is subject to the impact of dropped charge materials, a crucible with the strength and durability to survive these activities must be selected. Erosion/corrosion resistance: If the application produces slag, dross, or other by-products or uses fluxes that mechanically or chemically attack the inside surface of the crucible, then a crucible type must be selected that resists such attacks. Oxidation resistance: When exposed to air at high temperatures, graphite in a crucible will oxidize. If the application is likely to cause melt oxidation or excessively high temperatures (which exist in many foundry casting applications), a crucible should be selected with best oxidation resistance. Nonwettability: It is important in aluminum foundry application that the inside surface of the crucible resist the intrusion of molten metal into pores in the crucible material. Consequently, a crucible should be selected that possesses specific nonwetting agents.

 Electrical properties: If the application

requires a crucible that heats inductively, select a crucible designed for use in an induction furnace and with a specific electrical resistivity that optimizes the coupling power of the crucible.  Thermal conductivity: Not all crucibles offer the same level of heat transfer through the crucible and into the metal. To achieve the lowest energy costs, select the crucible with the highest level of thermal conductivity that meets all of the other requirements of the application. Crucibles have been called the first composites, and modern crucibles use an eclectic mix of materials to achieve the important characteristics listed here:  Graphite: A foundry crucible contains 30 to

50% by weight of flake graphite, depending on the properties required. The carbon content of graphite imparts high thermal conductivity and nonwettability and, coupled with its platelike layered structure, provides high thermal shock resistance. This is critical to foundry applications where temperatures can change by several hundred degrees in seconds.  Silicon carbide: A foundry crucible typically will contain 10 to 50% of silicon carbide, based on the desired characteristics. Silicon carbide provides excellent resistance to high-temperature erosion and corrosion and imparts a high level of thermal conductivity to the crucible.  Silicon metal: Fine powders of silicon metal or alloys are used to increase strength and erosion resistance, to protect the graphite and carbon bonding matrix from oxygen, and to help ensure an even glaze coating.

There are many types of crucibles and shapes produced by crucible manufacturers for the foundry market. Most crucibles used in the metals industry for melting, holding, and/or transferring operations are preformed carbonbonded or ceramic-bonded vessels. Silicon carbide crucibles are a widely used type of carbonbonded crucibles, and clay-graphite crucibles are a good example of ceramic-bonded vessels. It is important to discern which type is appropriate for a given application. Table 1 lists the several shapes and the application that is generally appropriate for their use. The basic types of foundry crucible furnaces can be classified as stationary, tilting, or movable. Stationary furnaces use a crucible set inside a refractory-lined shell that has heating equipment attached. A lift-out crucible is shown in Fig. 3. Normally, molten metal is dipped out with hand ladles, or the crucible containing the molten metal is removed from the shell for the casting operation. By using tong and shank tools to lift and carry the bilge-shaped crucible to the mold, pouring of the casting is done directly from the crucible without the need for potentially damaging metal transfers. The maximum crucible size normally used in this manner holds 140 kg (300 lb) of aluminum or copper. The weight and the strength of the crucible limit this type of operation. Another variation is a pushout furnace (Fig. 4) with induction coils and mechanical lifts to raise and lower the crucibles within the coils. Some stationary crucible furnaces are equipped for hand-ladle dip-out pours. For most metals, it is imperative that the flame and flue gases do not contact the charge. Two methods of controlling this critical factor are illustrated in Fig. 5. The first is the side flue configuration used in an aluminum melter installation (Fig. 5a), and the second is the cover plate configuration used for a brass foundry furnace (Fig. 5b). The cover plate setup incorporates holes on the outside diameter of the crucible top. Also shown is the extensive use of highly insulating fiber refractory material. This material has a low heat

Crucible Furnaces / 157 Table 1

Application of specific crucible shapes

Shape

Advantages

Bilge

Reduced top diameter provides less metal surface area. This helps to conserve heat and cuts metal loss and oxidation. Provides large metal surface area to facilitate manual or automated baling out of metal. Wider lower profile provides more efficient heat transfer in electric resistance furnaces. This shape is also suitable for low-pressure furnaces where die casting stalks are used to feed a die above the metal bath. Provides large surface area to facilitate manual or automated baling out of metal. Narrow lower profile is advantageous in fuel-fired furnaces to provide space for burner flame to circulate. Large-capacity crucibles for bulk melting of metal. Metal is usually then transferred to a holding bale-out furnace. Parallel profile means crucible is equidistant from the induction coil, allowing even heating.

Basin

Bowl

Spouted basin

Cylinder

Furnace type

Application

Typical user

Lift-out (crucible is lifted out of furnace chamber for casting)

General melting work for small castings

“Jobbing” foundries

Bale-out (crucible remains fixed in furnace)

Holding and melt/holding

Die caster

Bale-out (crucible remains fixed in furnace)

Holding and melt/holding

Die caster

Fig. 3

Typical lift-out version of a stationary crucible furnace specifically adapted to the foundry melting of small quantities (<140 kg, or 300 lb) of copper alloys

Melting Basin tilting furnaces (crucible remains fixed in furnace; furnace is tilted to pour metal)

Die caster, ingot producer

Induction tilting furnaces Melting (crucible remains fixed in furnace)

Master alloy producer

Fig. 5

Stationary crucible furnaces equipped for handladle dip-out pours. (a) Side flue configuration. (b) Configuration with cover plate openings on outside diameter of crucible top

Fig. 4

Cross section of a double pushout stationary crucible furnace

content and serves to conserve fuel and to minimize the temperature override that often occurs with the heavier refractory furnace linings. The added refractory flame face is required to accommodate the higher melt temperature. Steel crucibles can also be used in stationary crucible furnaces, as shown in Fig. 6. The high

heat conductivity of the steel offers fuel and meltdown time savings. Steel and ductile iron crucibles are extensively used for magnesium and zinc melting. In these molten alloys, there is little attack of the crucible walls. On the other hand, aluminum alloys will attack the steel crucible walls by forming sludge

(combination of iron, manganese, and chromium), which eventually causes hot spots in the castings produced from this melt. However, there have been successful aluminum foundry uses of steel crucibles. This requires a thorough refractory coating on the crucible surfaces in contact with molten metal and careful cleaning of the rust that tends to accumulate in the firing chamber (for example, flame contact with the

158 / Melting and Remelting

Fig. 6

Cross section of a stationary fuel-fired furnace used for the open crucible melting of magnesium alloys

crucible causes rust and scaling). This can collect in the firing chamber. A special clean-out door is used to remove this residue. Both aluminum and magnesium can react explosively with iron oxide. The use of electrical resistance furnaces and steel crucibles reduces the seriousness of this problem, and long-term use of steel crucibles and resistance furnaces has been successful in the casting of aluminum alloys. Tilting furnaces incorporate a crucible that is supported in the shell such that both are free to rotate on an axis. This allows the molten metal to be poured into ladles or molds. The axis of the pour can be either at the center of gravity or at the lip of the furnace. This design simplifies the handling of molten metal in many foundries. Tilting crucible furnaces are designed to simplify metal removal and have been extensively used in numerous foundries for many years. Figure 7(a) illustrates a center-axis furnace; Fig. 7(b) shows a cross section of a lip-axis furnace. Tilting crucible furnaces are useful where castings require more metal than can be easily handled by stationary crucible furnaces. These units are more flexible and can easily handle production requirements that are too small for the efficient operation of larger hearth furnaces. Figure 8 illustrates the use of a crucible liner for an induction melting tilt furnace. Use of the crucible liner overcomes the problems that would occur as a result of the strong erosive currents existing at the crucible sidewalls. The low electrical conductivity of clay-graphite crucibles makes them ideal for use in such applications. A gradual buildup of molten metal in the heel and voids in the backup refractory, both caused by repeated tilting of the furnace, are detrimental to crucible life. Crucible liners, which provide a durable wear-resistant surface for molten metal, are often used in the brass, steel, and ductile iron foundry industries. This application for the foundry use of crucibles is similar to that for the induction furnace liners shown in Fig. 8.

Fig. 7

Two variations of a tilting crucible furnace. (a) Center axis. (b) Lip axis

Movable crucible furnaces are used in conjunction with automated pouring system methods. Methods include multiple crucible melters arranged on either a shuttle or a turntable, such that one furnace is moved into position for pouring while others are positioned for recharging, stabilizing, and metal treatment.

Design and Operation Proper Furnace Atmosphere. With few exceptions, a slightly oxidizing flame should be used. With aluminum alloys, this is a pale yellow color. For copper-base alloys, this will be tinged with green (due to the oxidation of the copper). A traditional test for the correct furnace atmosphere is to place a cold piece of zinc in the flame. A yellowish color indicates an oxidizing flame, and a black color indicates a reducing flame. The objective is to obtain a flame having a tinge of yellow, which indicates the presence of the desirable, slightly oxidizing atmosphere. Oxidizing conditions were maintained in old coke-fired crucible furnaces at the crucible walls by burning the coke in a 75 mm (3 in.) space around the crucible sidewall. The burning coke maintained a slightly oxidizing atmosphere at the walls and ensured good heat transfer. Burner Control. For efficient melting and optimal crucible life, it is necessary to have even heating of the crucible and to avoid hot and cold areas around the sides of the crucible bowl. Some furnace designers use two or four burners to minimize this problem. This will contribute to longer crucible life, better fuel economy, faster melting, and lower noise levels. Precise periodic adjustment of the burner(s) is essential to efficient crucible furnace operation.

Fig. 8

Cross section of a tilt furnace for the highfrequency induction melting of brass and bronze alloys. Crucible is of clay-graphite composition. Also shown are the locations of the molten metal buildup and the voids in the backup refractory, which shorten crucible life.

Special care is required to avoid overheating of the bath. Approximately 80 to 95% of the heat required to prepare a heat for the foundry is used to heat the charge to the melting temperature and then to convert the solid to a liquid. Very little heat is required to raise the temperature of the newly melted liquid to the required pouring temperature. This can cause problems. For rapid melting of the crucible contents, high heat inputs are normally used. As soon as the charge becomes entirely liquid, little or no additional heat is required. If the equipment or the operator does not immediately turn the heat down or off, the bath temperature can become far higher than required. For most metals, this causes quality-control problems. Currently, most foundries depend on their furnace operators to control this harmful variable. Burner Design. Because of the wide diversity in crucible applications using many different fuels, a summary of these data is well

Crucible Furnaces / 159 beyond the scope of this discussion. It should be noted that developments in the design and operation of burners have significantly improved this furnace component. Any new installations of significance probably should include recent developments in both burner designs and burner control units. Charging Techniques. The most efficient melting (in terms of heat transfer or efficient use of fuel) is achieved with faster melt times, and the highest metal quality is obtained when a heel of metal is used to start the melting. Here, the heat travels easily from the flame through the crucible wall and into the metal. A crucible that contains only a solid charge provides little contact between the crucible wall and the metal to be melted. As a result, there is a considerable time lapse between charging and the initial melting of the charge. For efficient melting, a good heel size is roughly 25% of the crucible capacity. A second point concerning charging is the need to keep a loose charge. When a cold charge is tightly jammed into the crucible, the expansion of this charge upon heating will often break or crack the crucible. The third point concerns the need for gentle charging of all heavy or sharp-cornered material. Crucibles are fragile and easily broken. Many crucibles have been broken by rough and careless charging with heavy and/or sharp ingot or scrap. Handling of Crucibles During Melting and Pouring. It is important to keep in mind that hot crucibles filled with molten metal are fragile and therefore deserve careful handling. Care should be taken to set them on sand or other nonconductive surfaces. Placing a hot crucible full of molten metal on a steel plate or damp floor can easily break or seriously weaken the bottom of the crucible. Crucible Care. With proper care, crucibles will provide long service life. However, like every system used in the processing of molten metal, crucibles must be handled properly and

must be operated within their design parameters. All crucibles may be damaged by:  Excessive temperatures: Crucibles must not









be heated beyond their specified maximum temperatures. Excessive temperatures will cause crucible failure. Thermal shock: Structural damage may be caused by too rapid heating or cooling of the crucible. Uneven heating or frequent thermal cycling between cold and hot also can result in crucible damage, especially with frequent metal-level changes. Physical shock: Damage may be caused to the crucible by the impact of charge materials and/or, in the case of removable crucibles, by impacts during movement. Damage to crucible glazes: Crucible glazes serve as barriers against oxygen attack of the carbon material making up the crucible. These glazes may be damaged by any of the previously noted hazards, and damage to the glaze will lead to crucible failure. Chemical attack: Use of inappropriate or excessive metallurgical additives may lead to corrosion of crucible components, particularly at higher temperatures. Fluxing salts that contain fluorides are especially harmful.

Crucibles must be thoroughly inspected for shipping damage when received. If cracked, chipped, or otherwise damaged, they must not be installed. Crucibles should be stored inside in a warm and dry environment and must be completely dry before use. All clay-graphite crucibles must be properly annealed before being subjected to the first full heat. This will help the crucibles withstand thermal shock throughout their service life. All crucibles should be preheated gradually whenever entering use from a cold, that is, room-temperature, condition, following the crucible manufacturer’s specific recommendations. During operations, the crucible should be inspected

for damage after each heat, and slag or dross buildup inside the crucible must be carefully removed, avoiding damage to the crucible wall. Damaged or worn crucibles must be replaced immediately. Selecting the right crucible for the application; installing it properly; operating it within its design parameters; protecting it from physical, chemical, and thermal damage; and replacing it immediately when damaged or worn are best practices. Following these rules will enhance service life, operational efficiencies, and significant economies in crucible furnace operations. Cleaning of crucibles is impor tant in preventing premature failure. In many, if not most, metal-melting operations, slag or dross buildups develop at the metal surface line, and heavy, unmelted slush residue collects on the bottom. Both of these residues shorten crucible life by causing hot and weak wall areas. The poor and uneven heat transfer around the crucible walls causes hot spots and uneven stresses within the crucible walls. The metallic residues that will expand faster than the crucible upon heating can easily rupture the wall. At times, these residues expand and fracture the crucibles during the normal heating and cooling cycles. Cleanliness and careful workmanship on the melt room floor should be encouraged. ACKNOWLEDGMENT Portions of this article were from Casting, Volume 15, Metals Handbook, 9th ed., 1988. REFERENCE 1. Advanced Melting Technologies: Energy Saving Concepts and Opportunities for the Metal Casting Industry, BCS, Inc., Nov 2005

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 160-169 DOI: 10.1361/asmhba0005350

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Reverberatory and Stack Furnaces David White, Schaefer Furance Martin Reeves, Striko-Dynarad Paul Campbell and David Neff, Metaullics

REVERBERATORY FURNACES have a shallow hearth for melting metal, which is heated indirectly by burners mounted in the roof or in a sidewall of the furnace (Fig. 1). The flame used for melting the metal does not impinge on the metal surface itself but is reradiated off the walls or the roof of the furnace. The term reverberatory furnace is a throwback to the World War II era, when the most common fuels for hearth aluminum melting were coal or coke (Fig. 2). This necessitated locating the heat source or fire box at one end of the bath and the flue at the other end. The flue (stack) was designed to be strong enough to draw the flame over the bath. The roof was slanted to bounce or reverberate the flame (similar to sound echoes) off the ceiling, down across the metal to be melted in the molten bath, and then onto the flue. A primary objective was to heat the refractory walls and ceiling and then use these surfaces to radiate heat to melt the charge. This same flame pattern and the method of using a flame to heat the bath were implemented when

the fuel used was changed from coke to oil and gas. This older concept of aluminum metal melting is now uncommon but may still be used in some foundries or smelters. The modern method is to use roof burners or side burners directed toward the metal surface. In effect, the objective is to heat the metal, not the furnace, for efficient operation. Gas- or electric-heated burners are used in reverberatory furnaces. Aluminum foundry reverberatory furnaces range from small electric or gas-fired vessels of just a few tons volume to upward of 40 to 50 tons. Secondary smelters that produce ingot or supply liquid metal to foundries operate even larger furnaces. Gas reverberatory furnaces are commonly used in large nonferrous casting foundries because they provide a large reservoir of molten metal, ensuring a steady and reliable supply of molten metal to the shop. Small gas or electric everbretatory furnaces (Fig. 3) are also used for melting. Units are also used for holding furnaces, although typically the term reverbretatory furnace is used for melting

units—not holding furnaces. Practical sizes range from 1 to 45 Mt (2000 to 100,000 lb). The advantages of reverberatory furnaces include high-volume processing rates and low maintenance costs. Many foundries and die casters have used reverberatory furnaces for many years. Large- and medium-sized shops use them as central melters to supply multiple operations located within the shop. By using double-end clean-out and preheat hearths, these furnaces are some of the easiest to clean and are much more efficient than just a few years ago. Advances in insulating back-up linings have made these furnaces cooler to work around and more energy efficient than ever before. Many of the newer furnaces can reach an energy efficiency capability of 1250 Btu/lb of metal melted. Typical capacity of the furnace can range widely, from 1 to 75,000 tons. The energy efficiency of reverberatory furnaces is in the range of approximately 20 to 30% (Ref 2). Energy is mainly lost through the hot flue gases. In addition, as the molten

Fig. 2

Fig. 1

Basic components of a reverberatory furnace. Source: Ref 1

Early version of a coal-fired melter hearth furnace used to produce aluminum castings. (a) Plan view. (b) Vertical cross section

Reverberatory and Stack Furnaces / 161

Fig. 5

Fig. 3

Aluminum reverberotory furnace (a) 180 kg (400 lb) per hour electric reverberotory furnace, (b) small gas reverberotory furnace, Courtesy of Schaefer Furnace

Table 1 Melt loss and themal efficiency of electric and gas reverberatory furnaces Melt

Aluminum Aluminum Zinc Zinc

Type

Melt loss, %

Thermal efficiency, %

Electric Gas Electric Gas

1–2 3–5 2–3 4–7

59–76(a) 30–45 59–76(a) 32–40

Major components of a stack melter

metal comes in contact with the furnace gases, it forms slag/dross. The melt loss rate is 3 to 5% in aluminum gas reverberatory operations (Table 1). Improvements in burner technology, fuel/air ratio control, insulating refractories, and temperature control have contributed to a slow but steady improvement in the efficiency of reverberatory furnaces. Efficiency is also obtained in furnaces with preheating hearths (Fig. 4). Fuel savings of approximately 15% per year are possible by allowing a sow or ingot to sweat on the hearth as the internal temperature is raised to approximately 480
C (900
F). There is no moisture left in the sow or ingot, thus making it safer to charge. The stored thermal energy (Btus) provides further energy savings in maintaining the melt.

(a) The primary energy efficiencies of these furnaces are much lower (approximately one-third) than the listed efficiencies due to the use of electricity, which involves sizeable energy losses during generation and transmission. Source: Ref 2

Furnace Types

Fig. 4

In most large facilities, the conventional means for melting and holding aluminum is a gas-fired reverberatory furnace. In this arrangement, radiation from the burners, roof, and sidewalls heats the bath and solid charge. Forced convection through stirring or pumping is also needed to overcome a number of process limitations. Without forced circulation, the concentration of heat at the bath surface readily produces undesirably high temperatures. The high heat flux generated by direct radiation at the bath surface is typically much greater than conductive and natural convective forces, which distribute the heat through the bath. In turn, surface oxidation and dross formation ensue. There are two types of gas reverberatory furnace, based on whether the sows, ingots, and revert materials are preheated before melting or not. A dry hearth reverberatory charges and

Preheat hearth

preheats sows, ingots, and revert materials on a slope between the charging door and the molten metal bath, removing any contamination or water residue prior to melting, which prevents splashing or explosions. A wet bath reverberatory directly dumps the sows, ingots, and reverts into the molten metal bath. The wet bath system is beneficial for melting scrap types with large surface-to-volume ratios, which may oxidize excessively unless melted quickly. Another type is a side-well reverberatory furnace that consists of a number of burners firing inside the hearth, against the furnace hot wall or door. A charging well and a pump well are attached to the furnace hot wall on the outside of the furnace. Both wells are connected to each other and with the furnace hearth by arches, which permit aluminum circulation between the furnace chambers. During the charging period of the furnace cycle, molten aluminum circulates through the charge well where it melts the scrap, then circulates back into the hearth for reheating. Charging continues until the molten bath reaches the desired level. Subsequent steps in the smelting process may include degassing and alloying. Improvements in energy efficiency have also been made with the use of stack furnaces (Fig. 5), which can be considered as a modified reverberatory furnace with improved efficiency by better sealing of the furnace and the use of the flue gases to preheat the charge materials. Stack furnaces are more common than reverberatory furnaces for melting aluminum in Europe. In the United States, 95% of aluminum is melted in gas reverberatory furnaces, which operate with lower energy efficiency but have lower capital cost and provide easier operation and maintenance than stack melters (Ref 2). Dry Hearth Reverberatory Furnace. In a dry hearth furnace, the charge of solid metal

162 / Melting and Remelting is positioned on a sloping hearth above the level of the molten metal so that the charge is completely enveloped by the hot gases. Heat is rapidly absorbed by the solid charge, which melts and subsequently drains from the sloping hearth into the wet holding basin or chamber. Dry hearth furnaces are typically used where castings and/or returns contain nonaluminum inserts or in sand casting operations where casting scrap containing sand must be reclaimed. These castings can be melted on the hearth, scraping the sand or insert off the hearth and keeping them out of the molten furnace bath. This keeps the furnace cleaner and the aluminum from being contaminated with iron (in the case of cast iron inserts). The advantages of the dry hearth furnace include:

the metal is transferred to a holding furnace(s) that feeds the casting line(s). These furnaces are designed to use radiant heat (a highly efficient method of transferring heat into aluminum), where the heat source (burners) is located 46 to 61 cm (18 to 24 in.) off the bath for best practices. The closer the heat source is to the aluminum, the faster the transfer of Btus. The temperature differential between the thermal head and the metal bath provides the driving force for heat transfer. The advantages of the wet bath furnace include:  The up-front cost is lower than high-headroom  

 Safety: Everything melts on the hearth in the

atmosphere of the furnace and not under the bath. Such preheating avoids the potential for moisture entrapment and subsequent explosive force in and beneath the bath surface.  The dry hearth allows the saving of inserts and avoids quality problems with undesirable impurities being incorporated into the melt. Of course, there are other potential sources of impurities that must be considered.  The dry hearth holding chamber is typically smaller, approximately a 4 to 1 hold-to-melt ratio. The furnace holds four times what it melts or less, so the furnace is smaller.  If the flue is maintained correctly, very good temperature control can be achieved in the hold zone. The disadvantages of the dry hearth furnace include:  Typically, these furnaces have two flues: one



 

reverberatory furnaces and shaft melters. Efficiencies can be achieved on average of 1500 Btu/lb of metal melted. Everything is melted under the surface of the bath for greatly reduced metal melt loss (2 to 5% in gas-fired furnaces and less than 1% with electric furnaces). Many options are available: preheat hearths for sows and ingots, circulation wells for scrap, vortexing for chips, flue pressurization, well space for continuous degassing and filtration, and so on. With circulation and preheating, efficiency can be increased to 1250 Btu/lb of metal melted. By adding radiant roof regenerative burners, energy efficiencies can be further improved to 1000 Btu/lb of metal melted.

The disadvantages of the wet bath furnace include:  These furnaces have a higher hold-to-metal

ratio and therefore are larger (higher capital cost). With a typical hold-to-melt ratio of 8 to 1, a minimum of eight times the solid charge weight must be maintained to help

 



 

dilute the cold metal charge and not cause excessive bath temperature reduction. Larger furnaces require extra floor space. Sows and ingots must be preheated before loading in the furnace, whether on the charge well wear plates (i.e., furnace sill) or on a preheat hearth. Electric wet bath reverberatories must be charged evenly or they will lose temperature and take time to recover. Larger electricpowered melting furnaces are extremely expensive and not as efficient as large gasfired furnaces. Low headroom requires care when cleaning. Sows or ingot bundles cannot be double stacked on hearths because of lower headroom.

Stack Melters. For aluminum metal casting, there are substantial opportunities for energy savings in replacing reverberatory furnaces with stack melters (also referred to as shaft melters). Stack melters improve efficiency by better sealing of the furnace and use of the flue gases to preheat the charge materials. The charge materials slide down the shaft and reach the melting zone, where they are melted by the burners, and the molten metal flows down to the holding area (Fig. 5). The hot exhaust gases from the melting zone flow through the shaft to preheat the incoming charge, improving the energy efficiency of the stack furnace by 40 to 50% (Ref 2). A typical example of a shaft melter furnace in a foundry is shown in Fig. 6. One of the central advantages of the stack melter is the physical separation of the charging, preheating, melting, and holding areas of the furnace. Advanced burner controls can augment the advantages of this kind of furnace. The stack furnace also can be coupled with

over the hearth and one over the hold chamber.  Two separately controlled burner systems are required for optimal operation.  High metal melt loss (possibly 7 to 12%) can result due to melting in the atmosphere, which creates a very high oxidation rate at these temperatures (700 to 760
C, or 1300 to 1400
F).  When melting light-gage scrap and small foundry ingot, the metal melt loss is high and the thermal efficiency is low (1800 Btu/lb or higher). When melting larger sows (i.e., typically 540 kg, or 1200 lb), the efficiency is better and approaches the 1500 Btu/lb range. Metal melt loss is lower as the volume-to-surface area of the charge material increases. Wet Bath Reverberatory Furnaces. In a wet bath furnace, the products of combustion are in direct contact with the top of the molten bath, and heat transfer is achieved by a combination of convection and radiation. Wet bath reverberatory furnaces generally are used where there is just one alloy being melted or minimal alloy changes. Melting can be performed in a central furnace, sometimes called a break-down furnace, and then

Fig. 6

Stack melter employed in foundry practice. Courtesy of Striko-Dynarad Corporation

Reverberatory and Stack Furnaces / 163 other melting systems to improve efficiency. For example, it can be used with a crucible furnace, reducing its melt losses since the crucible melts by indirect heat to the metal through the wall of the furnace. The disadvantage of the stack furnace is that, due to the charging mechanism, the height of the stack furnace must be more than 6 m (20 ft) (Ref 3). In addition, as the furnace is charged from the top of the shaft, the refractories at the bottom of the charging door are subjected to repeated impact and wear. This requires additional maintenance. The shaft construction and melting design are at the heart of the advantages offered. The choked shaft design allows the consistent preheating of the charge material prior to melting. The hot exhaust gases from the holding bath pass through the melt zone and from there into the preheating zone before exhausting from the furnace. The additional energy required for melting comes from burners located in the melt zone and focused into the area at the base of the preheating area. Exhaust gases from these burners also exit through the shaft and therefore ensure a greater degree of preheating as well as greater speed and efficiency of melting. Energy Efficiency. Compared to other furnace designs, stack furnaces can result in substantial savings from lower fuel consumption and consequently lower melting costs. In the practical foundry operation, for example, with the use of 50% ingot and 50% bulky return material and a bath temperature of 720
C (1330
F), a power requirement is only 600 kW per hour per ton. This corresponds to a thermal efficiency of over 50%. Improved Melt Cleanliness, Quality, and Yield. By the separation of melting and preheating zones, the production of clean, high-quality metal can be achieved with stack melters. Moisture and other contaminants are evaporated or burnt off in the shaft by preheating with exhaust gases, ensuring that only dry material reaches the melting chamber. The primary source of contamination is moisture from storage of the scrap or ingots; the result of this is an increase in the hydrogen content of the metal, with its consequent quality problems in the cast parts. A further advantage is the avoidance of excessive contact between the melt and atmospheric oxygen, thereby reducing the tendency to form oxides. In order to minimize this danger, the burners in the melt chamber are arranged to avoid direct metal impingement. This allows melting on the hearth only and allows the metal to flow quickly but without turbulence into the holding bath without superheat. This ensures an optimal value for metal yield; typical values lie between 0.8 and 1.5%. A third measure for the maintenance of metal quality is the avoidance of contamination by impurities entering the metal holding bath. The separation of melting and holding ensures that contaminated charge material is not charged directly into the bath. In a crucible furnace or wet bath melter, what goes in stays in and must be removed by additional

metal-cleaning operations. Cleanliness is also assured by allowing the metal to run laterally into the holding bath. In this way, dross and other impurities remain on the melt hearth, to be removed during cleaning. An additional advantage from the separation of melting and holding is an almost continuous availability of molten metal, on demand, with a very consistent temperature (to +/5
C, or 9
F), which itself represents an important feature of quality. Melting continues while the holding bath level remains below maximum. The holding bath also offers the foundryman a source of molten metal at a consistent and correct temperature without fluctuations due to melting. Automated charging (Fig. 7) also contributes to this factor by ensuring that material is always available to melt when required. This is done using an automated lift-and-tilt system. Stack melters can also be combined with more conventional sow-melting technology to improve efficiencies and provide more options for charge materials. Various levels and forms of automation are available to cater to the different needs of the foundry and material supply. This can be by conveyor feed and automatic bin loading, allowing up to several hours, depending on melting rate and metal demand.

Reverberatory Furnace Practice Radiant Heating Factors. Radiant roof reverberatory furnaces rely on three major factors to make the process more efficient:  Distance from the heat source (the burners)

to the load (the aluminum)

 Hold-to-melt ratio  Even charging of the load

One major design criterion for radiant roof melting is based on the Stephan-Boltzmann law of radiant heat, which states that the rate of heat transfer increases the closer an object is to a heat source and the greater the temperature differential. By using low-velocity, flat-flame, highthermal-release burners that heat the refractory block construction to 1400
C (2600
F), a huge temperature differential is obtained between the aluminum and the heat source. In this type of furnace, efficient heat transfer into the aluminum is principally by radiation. The greatest efficiency is obtained when the temperature differential is the greatest between the radiant source (the burner blocks, not the furnace upper chamber) and the receiving medium (the molten aluminum bath). Heat transfer is also more efficient when the radiant source is closer to its receiving medium. When the radiant heat sources are closer, more burners are needed to obtain coverage of the medium. Most reverberatory, radiantly heated melting furnaces will employ practices in aluminum melting wherein the metal charged is either all solid (ingot, foundry scrap, gates, runners, etc.) or a combination of solids plus liquid

Fig. 7

Charge for stack melter

(metal delivered in over-the-road ladles). Consequently, a portion of the charge will be cold (solid) metal. The hold-to-melt ratio is critical as a dilution factor for the cold metal being charged into the furnace. The higher that ratio, the less temperature differential is obtained when the furnace is evenly charged. Either radiant roof or highheadroom side-fired melters should be charged at ¼ of their hourly rated capacity every 15 min. This will ensure that an even temperature is maintained in the bath at +/10
during the melting process. Charging. Overcharging of these furnaces can create sludge in the furnace (if the metal temperature drops below the sludge point of the alloy), which will result in poor-quality metal for casting. Such overcharging and causing greater temperature swing in the melting and already molten bath can also result in additional dross formation at the melt surface, due to increased exposure time (i.e., the overall melt cycle) and the absolute temperature at the bath surface as it continues to absorb heat during recovery of the bath temperature. Typically, the radiant roof approach of melting everything under the surface of the bath greatly reduces metal melt loss (2 to 3% in gas furnaces and less than 1% in electric furnaces) and increases melt efficiencies for melting small, lightweight scrap. Recuperation. While there is little doubt that there is a great deal of savings possible in recouping heat lost out of a flue, some care is needed on how far this can be taken. For example, keeping the recuperation temperature below 370
C (700
F) can substantially reduce the system cost. Below 370
C (700
F), insulated mild steel manifolds can be employed as opposed to more costly stainless steel. Even at 370
C (700
F) preheated combustion air, a 25% savings in fuel usage is possible. An example of the cost trade-off is based on numbers produced by a major burner manufacturer at $10.00/106 Btus on an 1800 kg/h (4000 lb/h) radiant roof melter. One can expect to pay approximately $85,000.00 more for a burner system using an installed recuperator and save approximately $54,750.00 per year, or have a

164 / Melting and Remelting 17.4 month payback on this investment differential. Actual costs may vary due to the age of the furnace and components or the type of furnace and the number of burners. Other factors to consider in evaluating recuperation are:  The flue gases should be dampered to be

able to bypass the exchanger when the furnaces are fluxed. Otherwise, excessive wear and maintenance on the heat exchanger will result.  The nozzles in the burner should be changed to stainless steel to handle the additional temperature of the preheated combustion air.  Older furnaces should consider changing to an ultralow-NOx burner. As the combustion air temperature increases, the NOx emissions go up. Regenerative burners enhance melting efficiency by providing a steady supply of hightemperature preheated combustion area. These burners contain dual firing ports, reversing firing as the air is drawn over heat-storage media in each port. In most cases, initial first-cost of recuperation provides a faster payback than does regeneration on smaller (2700 kg/h, or 6000 lb/h, or less) furnaces. This assumes a new furnace. For larger furnaces, the regenerative burners really begin to pay off. Payback varies depending on the type and age of the furnace in which regenerative burners are installed. Certainly, if the furnace is melting metal at a high end (1800 to 2000 Btus/lb of metal melted), then the return on investment becomes more lucrative. Regenerative burners become cost-efficient with high natural gas prices. As a further example, buying a 3200 kg/h (7000 lb/h) melter with regenerative burners to replace a less energyefficient high-headroom furnace that melts at 2000 Btu/lb of metal melted would result in a savings of over $452,000.00/year in fuel cost. For a melter of this size, the return on investment payback can be two years or less. (These figures are based on 3200 kg/h, or 7000 lb/h, for 20 h of melting/day, 340 days/year, and $10.00/1000 ft3 of gas.) Reducing Fixed-Heat Losses. Highly insulated furnaces with substantial linings help provide for the lowest fixed-heat loss. Fixed-heat loss is the amount of energy being absorbed into and passing through the lining of the furnace and out the doors and open wells on the furnace. An example of fixed-heat loss is:  If the sidewall casing temperature of a given

furnace is 96
C (205
F), then 300 Btus/ft2/h are lost from that casing.  By simply reducing the temperature to 64
C (148
F), 152 Btus/ft2/h are saved. The older refractory configuration concept taught that the freeze plane (the point where the molten metal would solidify if it leaked through the lining) had to be in the hot face or first layer of refractory material. With the

availability of newer, nonwetting insulating materials, the freeze plane can now be designed to be further back in the thickness of the lining. This gives much more flexibility to highly engineer this lining to not only contain the metal but actually save energy. Melting furnace casing temperatures are being designed to 50 to 60
C (125 to 135
F) instead of the old 75 to 85
C (165 to 185
F) of just a few years ago. New superinsulating materials can be used in the roof and floor of the furnaces to further reduce energy consumption. Some furnaces need to have individual burner controls, just to keep the furnaces from overheating over long weekends, because of this thermos-bottle approach to highly insulated linings. The fixed-heat loss has been reduced to the point that the burners on the furnace at low fire will overcome the heat losses and actually raise the temperature if they are not turned off. Covering up open wells will reduce heat loss substantially. An open well of 700
C (1300
F) aluminum with a 0.5 emissivity of the bath surface in the well will lose 8214 Btus/ft2/h. If the average charge well is 2.4 m long by 1 m wide (8 ft long by 3 ft wide) (2.2 m2, or 24 ft2), then 197,136 Btus/h are being lost off of that well. That corresponds to 5.6 m3 (197 ft3 of gas per hour, or 134 m3 (4731 ft3) a day by 340 days a year is 45,560 m3 (1,608,540 ft3) of gas lost. That is $16,085.40 worth of natural gas at $10.00/therm lost a year. Improving and Maintaining Furnace Efficiencies. Burners are not stoichiometric. The combustion system must be set up properly, or more dross will be generated. Typically, most natural gas-fired burners are set up on a 10 to 1 air-to-gas ratio. At this ratio, 95% of the oxygen in the burner system is consumed. However, if a furnace is set up with ratios in excess of this, dross increases in the furnace. On the other hand, a furnace that is set up fuel-rich will produce a colder flame and make the furnace work harder to get the job done. Keep the burner in ratio. A qualified technician should check the combustion system for the proper set-up. The burner system should be firing stoichiometric, or in ratio, air to gas. Typically in reverberatory furnaces, this is a 10 to 1 air-to-gas mixture. By analyzing the flue gases, a technician can determine if indeed the system is running at peak efficiency. This can also be done with a manometer to test pressure (inches of water column) of gas and air, if the technician knows the burner manufacturer’s recommended set-up pressures. Fuel-rich burners produce a colder flame, and some of the gas will ignite when it reaches the atmosphere at the flue level. This results in a wall of flame coming out of the flue. If the burner is set up too lean or has excess air, a colder flame results, together with an increase in oxide formation and dross in the furnace. This may cause the burners to work harder to produce the same Btus, resulting in lower energy efficiency and therefore higher costs of production.

In addition to burner ratio, improving and maintaining furnace efficiencies can be achieved in several ways, as summarized subsequently. Remove all fans that blow on the furnaces. This will result in less heat loss from the casing where the fan is hitting the furnace. Instead, use individual cooling stations that bring the air in from overhead to a specific spot where the operator can stand to cool off. Install well covers (Fig. 8) whenever the furnace is idle for longer than a half-hour. Install a half-flue cover over the flue during long idle conditions. Warning: The flue cover must be removed before turning the furnace to high fire. At low fire, the flue opening is still sized for 100% output of the burners. Without automatic flue pressurization, the holding efficiencies will drop considerably. The furnace may not need to be on high fire as much. Charging. Make sure the furnace is used at full capacity and charged ¼ of its hourly rated capacity every 15 min. Overcharging the furnaces causes dramatic swings in temperature and increased sludge buildup on the floor of the furnace. The more sludge on the floor, the less likely the furnace will melt at its rated capacity. The furnace may not melt anywhere close to the Btus/lb of metal melted that it was designed to do. Reduce molten metal temperature over weekends by 11
C (20
F) or more if feasible. Clean furnaces during idle shifts or idle times, but clean them daily! This will cut down on the amount of oxides growing in the furnace. Oxide is dense and absorbs additional Btus from the metal. Do not overflux the melters. Overfluxing of the walls can break down the binder in the refractory and cause premature erosion of the belly-band area. Clean or change combustion air filters regularly. Hold a dollar bill 50 mm (2 in.) away from the filter. If the bill is not drawn to the filter rapidly, it is time to clean or change the filter on the blower. Double-check and monitor combustion set-up for maximum burner performance (requires a manometer). If a wall of flame is coming out of the flue, a fuel-rich condition probably exists inside the furnace. The raw gas that does not have enough air in the furnace for combustion

Fig. 8

Pneumatic well cover

Reverberatory and Stack Furnaces / 165 ignites when it reaches the atmosphere just outside the flue, because of all the free air available to it at the flue and hood junction. Overpressurization of the furnace can also cause this condition. Preheat Ingots. In reverberatory (radiant roof low-headroom or high-headroom) furnaces, preheat all ingots on a charge well (wear plates) or on a hearth before putting them into the bath. Preheat all sows on a hearth before pushing them into the bath. When a sow or ingot begins to sweat, the internal temperature is close to 480
C (900
F). There is no moisture left in the aluminum, and it can be safely charged under the bath of aluminum, where the stored Btus in the metal help finish the melting process. Ingots left on a well wear plate to preheat will reach 120
C (250
F) after 20 min. This is sufficient to load into the bath. Some bubbling may occur from moisture left in the very center of the ingot. Metal delivery to the holding furnaces should be every 20 to 30 min; less frequently will mean overcharging the melters and drawing the metal down too low in the holders. The closer the metal is to the heat source, the faster the transfer of Btus. A loss of 10% in efficiencies could take place by drawing the holders that low. Ensuring that the metal tenders put back what they take out in the melters results in a difference in metal temperature stability. Remove dross buildup whenever it becomes thicker than 13 mm (½ in.), or the dross will begin to act like an insulator and make the furnace work harder to do the same job. A completely clean bath surface will reflect the heat, so it should not be cleaned any more frequently than once per shift. If a furnace creates a dross layer over 13 mm (½ in.) in less than 6 to 8 h, then the furnace is creating more dross than normal. Electric Pump Motors. To further conserve energy, the pump should have an electric motor rather than an air motor. Plant air is the most expensive energy source in the plant. A small blower keeps the pump cool. Treat the cause of dross buildup, not the symptom. Investigate why the furnace is generating dross. Typical reasons are:  Air leaking in and around the doors  Drawing the furnace down below

the submerged arches or door blades  Burners that have a flame that touches the bath or scrubs the solid metal on a hearth  Too large of a flue opening for the Btus connected, causing the furnace to draw in more air  Keep the burn in ratio!

Molten Metal Circulation In the melting of aluminum and its alloys, forced convectional heat transfer within the melt often plays an important role. There is a considerable latent heat of fusion (95 cal/g) associated with taking aluminum at its melting

point from a solid to a liquid. Forced convection assists this transition, particularly for larger volumes of metal, by better mixing of the solid metal with already molten metal, thus reducing time and energy input into the melting process. Other nonferrous alloys, such as zinc, magnesium, and copper, could also benefit from molten metal circulation, but pumping systems for circulation are not usually employed. Mechanical stirrers are effective, particularly for zinc, due in part to the lower heat of fusion of zinc and the smaller melt sizes usually used. In reverberatory furnaces, forced convection enables the radiant heat energy absorbed by the bath surface to move, that is, to be transmitted throughout the bath depth, so that a steady heat flux is developed and continuing heat transfer to the surface occurs by radiation. Without such forced convection, the surface layer of the molten metal becomes heat saturated, and the surface temperature rises. Ultimately, excessive surface temperature will result in oxidation and dross buildup, which reduces energy efficiency. Forced convection is produced by the circulation of the molten metal bath. Current methods to achieve circulation or metal movement include manual stirring, mechanical pumps, electromagnetic pumps, induction stirrers, porous plugs, jet pumps, and towmotor/boom agitation (see the section “Molten Aluminum Circulation Methods” in this article). Manual stirring is only practical in smaller furnaces (in small pot furnaces, crucibles, and so on). Forced circulation is usually best achieved in large furnaces (2000 to 110,000 kg, or 5000 to 250,000 lb, capacity) by using a submerged discharge molten metal pump. Several types of molten metals pumps have appeared over the years (such as electromagnetic and pressure pumps), but the centrifugal pump has been a common type. These pumps are usually installed in the pump wells and discharge into the charge well of reverberatory furnaces for remelting in large aluminum foundries, die casting operations, secondary smelters producing foundry ingot, and in mill product casting houses. Typical installations are illustrated in Fig. 9 (a and b). A summary of the benefits of forced circulation in large furnaces employed in aluminum foundry practice, as well as a comparison of the different methods of circulation, is in the following sections.

much greater than conductive and natural convective forces, which distribute the heat through the bath. In turn, surface oxidation and dross formation ensue. Figure 10 (Ref 4) provides the oxidation rate of molten aluminum as a function of temperature. Because aluminum oxide is thermally insulative, the variation between surface and bath temperatures increases with the growth of the oxide layer. Temperature differentials in a 915 mm (3 ft) deep bath with a clean surface and without forced convection can typically reach 50 to 85
C (90 to 150
F). Significant melt loss is incurred due to ever-increasing dross formation under these conditions. Without effective heat transfer to the metal, other quality and furnace performance factors may be less than optimal.

Fig. 9

Molten metal circulation pumps. (a) Typical pump installation in a sidewell furnace. (b) Circulation well and pump

Benefits of Forced Circulation The benefits of circulating molten metal in reverberatory furnaces are well documented, especially for aluminum. Forced convection through stirring or pumping overcomes a number of process limitations of reverberatory furnaces for melting and holding. Without forced circulation, the concentration of heat at the bath surface readily produces undesirably high temperatures. The high heat flux generated by direct radiation at the bath surface is typically

Fig. 10

Oxidation of aluminum versus temperature

166 / Melting and Remelting The result of forced convection is to reduce metal temperature stratification in the bath, improve melting rates, enhance metal quality through alloy homogeneity, reduce energy consumption as a result of improved heat transfer, extend refractory life, reduce sludge (dross) formation and therefore lower furnace-tending requirements, and accelerate alloy dissolution rates. Operational benefits of more homogeneous bath temperature of the entire bath of metal (top to bottom, inside and out in the well) include:  Quicker melting with reduction of melt

cycle times by 10 to 50%  Reduce energy consumption by 15 to 25%  Extend refractory life at least 25%  Improve recovery rates by a minimum of 0.25% With these major factors at stake, it is no wonder that a wide variety of techniques have been developed to provide forced circulation. While many production parameters may be difficult to fully quantify, forced circulation clearly improves many key aspects that affect total quality. These benefits of circulation are influenced by a variety of production/operation factors, as briefly described here. The next section, “Molten Aluminum Circulation Methods,” in this article also summarizes the performance of currently applied circulation technologies in large-furnace foundry practice. The review covers six methods that have evolved to some degree of acceptance: mechanical pumps, electromagnetic pumps, induction stirrers, Alcan jet stirrer, porous plugs, and towmotor/boom. With each technique, the production manager must assess the accompanying benefits, characteristics, and drawbacks. Reduced Temperature Stratification. Bath temperature is a key factor that impacts overall product quality. Forced circulation improves bath product quality by ensuring the bath temperature will be consistent throughout the furnace. Typically, reductions in temperature differentials for a 915 mm (3 ft) bath are from 50 to 85
C (90 to 150
F) in an uncirculated furnace to 3 to 7
C (5 to 13
F) for stirred furnaces. Circulation ensures that the temperature of the metal exiting the furnace is virtually the same as that indicated by the furnace thermocouple. Production parameters can be set and more easily controlled when the metal delivered is at a constant temperature. Additionally, there will be less superheat of the metal required because there will be less variance in the furnace temperature. Increased Melt Rate. Constant circulation of the furnace will increase the melt rate of the furnace as well as improve the thermal profile. The high-velocity stream from a circulation pump will increase the melt rate of large submerged solids in the furnace. Insulating boundary layers that are formed around large solids as they melt are swept away by the hot metal stream. Such a device equipped with gas injection can provide bath fluxing and degassing during the melting cycle, eliminating

or greatly reducing the time required at the end of the cycle for metal treatment. The same holds true for temperature control. Continuous circulation provides excellent temperature control, so temperature adjustments are usually not required prior to transfer. Cunard, using a direct-charge furnace in a billet casting process, reported in a study over a five month time period that furnace cycle times increased between 8.1 and 20.6% when the pump was not in operation (Ref 5). He estimated that using a pump produced a conservative value of 10% cycle-time savings. Charging time reductions of 25 to 50% have often been observed in sidewell furnaces when a pump is used. Savings may be obtained by increasing output with existing equipment or by maintaining production levels with fewer furnaces. Manufacturers of foundry reverberatory furnaces estimate that the melting capacity of a given furnace will increase by 10 to 15% when forced circulation is added to an uncirculated furnace. Alloy Homogeneity. One method for stirring a furnace was developed by Alcan during the 1980s. They attached a large pressure vessel to the side of the furnace. The pressure vessel communicates with the furnace through an opening below the metal surface. When a vacuum is applied to the vessel, metal is drawn in from the furnace through the opening. The vessel is then pressurized, forcing metal back into the furnace through the opening. The vacuum is reapplied to the vessel, and the cycle repeats. The results of an Alcan jet stirrer (AJS) installed in a top-charge 70 ton round melter have been documented (Ref 6). Figure 11 schematically represents the main elements of the AJS, while Fig. 12 shows a typical adaptation to a top-charge round melter. The results were obtained during the preparation of an AA4000series alloy with a high silicon content. The silicon specification for this alloy is 7.15 to 7.45%. The silicon content was monitored in the trough during the cast and compared with samples taken from the furnace at the end of the batch casting process. A significant difference was observed between the two samples for the same batch. For eight runs without stirring, the average silicon composition in the furnace at the end of the cast was 7.28%. In contrast, the average for trough samples during the cast was 7.22%, with a standard deviation of 0.08%. In fact, two ingot batches were outside of specification. In five runs where stirring was used, no significant difference between furnace and trough samples was observed. All ingots were within specification, and the standard deviation for the cast samples dropped to 0.0%. In foundry applications, similar results can be expected when the furnace is circulated. Without forced circulation, alloy additions must be stirred in mechanically to ensure uniformity. Since mechanical stirring is done over a relatively short period of time, it is difficult to ensure that the alloy is mixed uniformly

Fig. 11

Schematic of the main elements of the Alcan jet stirrer

Fig. 12

Typical Alcan jet stirrer installation in a top-charge melter

throughout the furnace. Some additions will also begin to sink to the bottom or float to the surface if not continually stirred. When a stirring device such as a mechanical pump is used, alloy elements are constantly mixed with the rest of the metal in the furnace. In a typical foundry application, the volume of metal in the furnace passes through the pump every 15 min. Since the pump is operated 24 h per day, samples taken from any point in the furnace will show uniform chemistry. One other consideration with foundry alloys is the formation of sludge. This is an Fe-CrMn compound that precipitates out of the metal if the temperature drops below a critical point. This is discussed in detail by Jorstad (Ref 7). With a temperature differential of 50 to 85
C (90 to 150
F) between the top and bottom of the bath, it is easy to see how the metal at the bottom of the bath can cool below the critical point if the furnace temperature is monitored and controlled by a thermocouple located near the top of the bath. As previously discussed, temperature is uniform throughout the bath when the furnace is constantly stirred. This allows for accurate temperature control and the avoidance of sludge formation. Reduced Energy Consumption. The rate at which energy is transferred from the burners

Reverberatory and Stack Furnaces / 167 and refractories to the metal is referred to as the heat flux (q). Heat flux in a reverberatory furnace is determined by the following equation (Ref 8): q ¼ Ah½ðt1 Þ4  ðt2 Þ4  where A is the surface area of the bath inside the furnace, h is the heat-transfer coefficient of the area inside the furnace above the bath, t1 is the temperature of the heat source or refractory walls and ceiling, and t2 is the temperature of the bath surface. From this equation, it can be seen that the rate at which energy is transferred into the metal is dependent on the difference between the surface temperature of the metal and the temperature of the refractories, to the fourth power. Constantly stirring the furnace allows heat to be removed from the surface to the rest of the bath by convection. As heat flows more efficiently away from the surface of the bath, the bath surface temperature is reduced. As the surface temperature drops, the temperature differential between the refractories and the metal increases, which translates into much higher heat-transfer rates. This subject is covered in more detail by Campbell (Ref 9). Since stirring occurs during the melting process and with the doors closed, energy that would have been lost out the door when the bath is stirred manually is saved. Additional savings are possible from the elimination of throughthe-door fluxing operations if the circulation pump is equipped with gas injection capability. Actual furnace efficiency improvements vary from operation to operation. Experience has shown that a minimum of a 15% improvement can be expected. A 25 to 30% improvement is possible, depending on furnace and operationspecific variables. In the study done at VAW in Ellenville, New York (Ref 5), savings ranged from 10.3 to 20.2% over a five-month period. Longer Refractory Life. By reducing the metal surface temperature, more heat is transferred to the metal and less is absorbed by the refractories. Roof and stack temperatures are usually reduced by 80 to 115
C (145 to 205
F). These lower operating temperatures translate into longer refractory life. Lower temperatures will also result in less corundum formation at the metal line. Less buildup means less need for mechanical cleaning and therefore less opportunity for damage from cleaning tools. While specific information on savings from improved refractory life is not widely shared, dramatic improvements have been reported by some companies using circulation pumps (Ref 10). Decreased Dross Formation. Another aspect that should not be ignored is melt loss. Oxidation increases with temperature (Fig. 10); at temperatures above 775
C (1425
F), the rate at which aluminum oxidizes increases quickly. Reducing the bath surface temperature can significantly reduce the rate of dross formation. Minimizing surface agitation with submerged circulation removes another source of dross formation.

Many foundries do not pay enough attention to melt loss that results from dross formation. Every pound of aluminum that is lost to oxidation in the form of dross is lost to the manufacturing process and must be replaced. While not widely documented, the typical dross reduction in the hearth when using a centrifugal pump for forced circulation is approximately 0.5%. Using a conservative value of 0.25%, a foundry that melts and casts 454 MT (1 million 1b) of aluminum a month would save the replacement cost for 1.1 MT (2500 1b) of metal every month. In another study, a large die casting company augmented its production facility with centrifugal pumps (Ref 11). They estimated a 98% recovery of charged material during the first six months of pump use versus a 96% recovery prior to its use. Easier Alloy Adjustments. The benefits associated with alloy additions can manifest themselves in three areas: alloy addition dissolution, treatment during melting, and improved melt quality before downstream treatments. By continually providing forced circulation, alloy additions and adjustments are easier to control, since the bath will remain homogeneous. Silicon and other alloy additions go into solution much faster when exposed to the discharge stream of a pump. When heavy chlorination is required due to scrap charge composition and desired alloy chemistry, a gas injection pump is an extremely effective tool (Ref 12). The dispersion and subsequent reaction of chlorine gas as it exits the pump in the molten metal stream creates a very effective kinetic environment for removal of magnesium from the melt. Figure 13 provides a typical representation of demagging results in the production of foundry ingot in a 68 MT (150,000 lb) furnace with a 34 MT (75,000 lb) heel from the previous heat. The magnesium content was decreased consistently and effectively during the normal charge cycle.

Fig. 13

Demagging results using a gas injection pump

Results from tests conducted at Golden Aluminum in 1995 show that with a mixture of 95% Ar and 5% Cl, a gas injection pump can significantly reduce sodium and calcium levels without affecting magnesium content (Ref 13). The tests also showed an improvement in porous disc filtration analysis results and a reduction in gas levels while the gas injection pump was in operation. By removing inclusions and hydrogen in the melting furnace, the load is reduced on downstream operations such as degassers and filters.

Molten Aluminum Circulation Methods As noted, methods of molten metal circulation include manual stirring, mechanical pumps, electromagnetic pumps, induction stirrers, porous plugs, jet pumps, and towmotor/ boom agitation. These technologies have evolved to facilitate efficient and consistent molten aluminum processing in melting and holding furnaces as the commercial importance of aluminum continues to grow. The features, limitations, and costs of each device must be considered with the requirements of the operation to determine the best solution. While all of these devices impart motion to the molten aluminum bath, the effectiveness of heat and mass transfer, capital investment, running expense, maintenance issues, ease of installation, and capabilities for alloy modification vary significantly. One of the most important factors in today’s (2008) competitive environment is product quality. While the economic benefits of quality can be difficult to quantify for a specific operation, the benefits of improved productivity, reduced cycle time, and better thermodynamics can be readily translated into economic savings. Many papers have been published on the application and performance of mechanical circulation and gas injection pumps, as provided

168 / Melting and Remelting in the references. These are typically centrifugal pumps with a rotating impeller in a volute casing. Materials of construction consist of graphite and/or ceramics so that they can survive in contact with the molten metal. They require an external sidewell (Fig. 9a) on the furnace. Mechanical pumps offer the most flexibility, low capital cost, and moderate maintenance requirements. On the other hand, they are not suitable for every application. The operating principle of an induction stirrer and an electromagnetic pump is the same as that of a linear motor (Ref 14). An arrangement of a bottom-mounted induction stirrer with a furnace is provided in Fig. 14. The melt is stirred by the force created by the inductor attached to the bottom of the furnace. A variation of this mounts the stirrer on the side of the furnace. Electromagnetic stirrers typically have the highest capital and energy costs, but with no metal contact, they tend to be the most reliable of all the systems considered here. In contrast, the electromagnetic pump moves molten metal through a tube within a coil by the generation of an electromagnetic force created by an applied magnetic field and induced eddy currents (Ref 15). Figure 15 depicts the plan view of an electromagnetic pump in a furnace with charge and dross wells. Electromagnetic pumps offer many of the advantages of a mechanical pump, generally require less maintenance, and can be used in applications not suitable for mechanical pumps. Capital costs

Fig. 14

Fig. 15

Schematic side view of an induction stirrer

Plan view of an electromagnetic pump, charging well, drossing well, and furnace

and energy costs are higher than mechanical pumps. The operating principle of the AJS is based on the repetition of two phases: a vacuum/accumulation phase followed by the pressurization/ expulsion phase. Metal is drawn into the unit through an opening with the furnace during the vacuum phase, then forced out through the same opening during the pressurization phase. For several years, Reynolds Metals used porous plugs in melting and holding furnaces at the alloys plant in Muscle Shoals, Alabama (Ref 16). This technology uses porous refractory plugs that are permanently installed in the furnace floor. Gas, usually nitrogen or argon, passes through porous refractory and produces a plume of fine bubbles over each plug. This gas plume creates localized circulation over each plug. Finally, a towmotor/boom system is used in some foundries and is considered to provide circulation within the furnace hearth. With this method, the door is opened, and the towmotor operator works a boom-and-rake arrangement back and forth to stir and skim the molten melt. Maintenance Comparisons. As a general rule, mechanical pumps will require more maintenance than electromagnetic pumps and stirrers. Mechanical pumps have seen significant improvements in designs and materials of construction in recent years, and reliability has improved to a point where at least 1 manufacturer now offers a 1 year warranty on the major components. Electromagnetic pumps and stirrers have no moving parts and therefore require little maintenance (Ref 17). This is somewhat offset by higher capital and energy costs. Porous plugs require no maintenance. Life and reliability are determined by installation and operation techniques. Major refractory repairs are required when they need to be replaced. Flow Rate/Effectiveness Comparison. As described in the first section of this article, there are many benefits of strong, forced circulation. Mechanical pumps, electromagnetic pumps, and induction stirrers receive high marks for delivering these benefits to the production process. Porous plugs, which are probably better categorized as a treatment technique rather than forced circulation, rate as moderate in overall performance. The towmotor/boom system is a “get-what-you-pay-for” arrangement. Not only is the circulation poor and intermittent, there are quite a number of negative factors that are associated with this technique. Since the furnace doors must be opened for access, there is a high heat loss. With this type of thermal cycling and exposure, the refractory life is significantly reduced. While this method is inexpensive from a capital standpoint, the increased dross formation and associated metal losses can be a large operating loss. Finally, safety is an issue. Certainly, the towmotor operator is at risk in this environment; it is often a topic of lively discussion at aluminum

meetings. Additionally, the back-and-forth action that is required presents a hazard in an aisle that is common to both plant personnel and equipment traffic. Alloy Modification. In today’s (2008) demanding production environment, concurrent processes are often added to classical charging, melting, and homogenization schemes to reduce overall production time. These processes include chlorine injection to remove magnesium from the melt, gas injection to remove hydrogen and increase inclusion flotation, and solids addition for alloy modification. A mechanical pump structured as a gas injection model has documented performance for all three key needs. An electromagnetic pump can be used for alloy modifications by the addition of solid materials, but it has not been shown to be effective for gas injection. In contrast, the induction stirrer does not provide any of these benefits as it is currently configured. Porous plugs are excellent for hydrogen and inclusion reduction. They are not employed for demagging, and their effectiveness for alloy additions is not documented. Towmotor/booms are often used to stir in alloy additions, but they cannot effectively improve operations over any of the other stirring methods. Ease of Installation/Retrofit. The application of a mechanical pump requires an open well and a residual molten metal heel during normal furnace cycles. A floor-mounted induction stirrer system is usually only considered for new construction, since the installation requires a basement area to house the inductor, and the furnace floor must be stainless steel. Side-mounted induction stirrers can be added to a furnace fairly easily. Electromagnetic pumps can be adapted to most furnaces. The furnace refractory and steel shell must be modified so that the equipment can be bolted to the side of the furnace. In order to install porous plugs, a major furnace reline is required. The towmotor with boom may be used in all instances. REFERENCES 1. “Refractories for Industrial Processing: Energy Reduction Opportunities,” Oak Ridge National Laboratory, Dec 2004 2. Advanced Melting Technologies: Energy Saving Concepts and Opportunities for the Metal Casting Industry, BCS, Inc., Nov 2005 3. J.F. Schifo and J.T. Radia, Theoretical/Best Practice Energy Use in Metalcasting Operations, U.S. Department of Energy Industrial Technologies Program and Keramida Environmental, Inc., May 2004, p 36 4. D.V. Neff, Molten Aluminum in Motion, Light, Metals, TMS, 1993, p 197–211 5. T. Cunard, Application of Graphite Pump to Direct Charged Aluminum Melting Furnace, Aluminum Industry Energy Conservation Workshop X, Oct 20–21, 1987, Aluminum Association, p 159–185

Reverberatory and Stack Furnaces / 169 6. M.A. Thibault, F. Tremblay, and J.C. Pomerleau, Molten Metal Stirring: The Alcan Jet Stirrer, Light Metals, TMS, 1991, p 1005–1011 7. J. Jorstad, Understanding Sludge, Die Casting Engineer, NADCA, 1986 8. M.C. Mangalick, Reduced Fuel Consumption Through Mechanized Molten Metal Circulation, Soc. Die Cast. Eng., Vol 17 (No. 5), Sept–Oct 1973, p 22–31 9. P.S. Campbell, Use of Molten Metal Pumps to Reduce Energy Consumption in Reverberatory Furnaces, Aluminum Industry Energy Conservation Workshop XI, Nov 1–2, 1990, Aluminum Association, p 537–546

10. T. Thornton, C. Hammond, J. van Linden, P. Campbell, and C. Vild, “Improved UBC Melting Through Advanced Processing,” 136th Annual Meeting and Exhibition, TMS, 2007 11. L-35 gas injection customer, private communication 12. D.V. Neff, Use of Gas Injection Pumps in Secondary Aluminum Metal Refining, Recycle and Secondary Recovery of Metals, Dec 1–4, 1985, Metallurgical Society of AIME, p 73–95 13. R.S. Henderson et al., “Quantification of Molten Metal Quality Improvements Using an L-Series Gas Injection Pump,” Third Int. Symposium on Recycling of Metals

14. 15.

16.

17.

and Engineered Materials, Nov 12–16, 1995, TMS C. Clark et al., A Molten Metal Pump Installation at New Zealand Aluminum Smelters Limited, TMS, 1988 R.M. Jones, A New Electromagnetic Metal Pumping and Stirring Technology, Fourth Australian-Asian-Pacific Course and Conference, TMS, 1995, p 275–289 J.R. Guttery and W. Evans, A New Generation of Fluxing in Aluminum Melting and Holding Furnaces, Light Metals, TMS, 1994, p 921–927 A. Katyal, Metal Stirring Using an Electromagnetic Pump, Optimization ’95 Conference, Proc. BCIRA, Vol 8, 1995, p 1–20

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 173-184 DOI: 10.1361/asmhba0005352

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Transfer and Treatment of Molten Metal—An Introduction MOLTEN METALS are processed and handled in various ways, depending on the alloy and melting practice. The very nature of melting is a way of combining elements for the purpose of alloying based on the laws of physical chemistry (see the article “Chemical Thermodynamics and Kinetics” in this Volume). However, the reactivity of metals and the high temperatures of melting also introduce the deleterious effects of dissolved gases and inclusions in molten metals, as described in the articles “Gases in Metals” and “Inclusion-Forming Reactions” in this Volume. Molten-Metal Treatments. Of primary importance during melting and the handling of molten metal is the mitigation and reduction of dissolved gases and inclusions in the melt. This objective may be accomplished in different ways, depending on the base metal, alloying constituents, and residual impurities. Fluxing of molten metal is often the first step in obtaining a clean melt. Fluxing is a term commonly used in foundries and melting departments that refers to molten-metal treatments from the addition of reactive chemical compounds or inert gases. The various functions of molten-metal fluxing additives, either singly or in combination, include:


To render the high-melting refractory oxide

impurities or gangue fusible and separable from the molten metal
To provide a medium with which the impurity elements or compounds would combine in preference to the metal In steelmaking, the primary basic fluxes are either limestone (CaCO3) or dolomite ((Ca, Mg)CO3), depending on the other slag constituents present and the required ability of the slag to remove sulfur. Limestone is preferred for large sulfur removal. The slag-forming compounds in steel melting can be classified as either basic or acidic (see the article “Electric Arc Furnace Melting” in this Volume). Likewise, secondary refinement of steel melts may include various fluxing techniques in converter and ladle metallurgy (see the article “Steel Melt Processing” in this Volume). Fluxing is also commonly done to some extent with virtually all nonferrous mill and foundry melting operations. In particular, fluxing is used extensively in controlling dross formation and removing oxides from aluminum melts and melting furances. The basic types of aluminum fluxes are:
Cover fluxes designed to barrier “blanket”

on the surface of molten aluminum

the article “Casting of Copper and Copper Alloys” in this Volume)
Fluxing of zinc alloys with chloride-containing fluxes that form fluid slag covers, which can be used to minimize melt loss if they are carefully skimmed from the melt before pouring (see the article “Casting of Zinc Alloys” in this Volume) Degassing. Removal and reduction of dissolved gases in molten metal is typically accomplished with injection of inert gases in the melt (see the article “Degassing” in this Volume). In addition to dissolved gases, other sources of gas porosity include various mold-metal reactions (see the article “Gases in Metals” in this Volume). Molten-Metal Filtration. In addition to fluxing, inclusions in molten metal can also be controlled with various types of filters, such as screens or foam. Filters are commonly used in both aluminum melting furnaces and in-mold systems (see the article “Molten-Metal Filtration” in this Volume). Filters are also used in steel foundries. Molten-metal handling systems are also important components in the effective transfer, holding, and pouring of melts. After the melting process, molten metal may be handled and processed in various systems that may include:


Removing inclusions, impurities, or dissolved


Drossing fluxes with wetting action that


Launders, which are channels for transport-

gases from the melt
Preventing excessive formation of solidphase inclusions
Preventing and/or removing oxide buildup on furnace walls
Injecting or distributing alloying constituents

promotes coalescence of aluminum
Melt-cleaning fluxes designed to remove aluminum oxides from the melt
Wall-cleaning fluxes specifically designed for the softening and removal of excessive aluminum oxide buildup that occurs on melting furnace walls

Thus, fluxes are commonly used to some extent with virtually all molten-metal operations in both the foundry and in the production of mill products. Fluxing can occur throughout the various stages of melting and handling of molten metal—from the melting or holding furnaces to the transfer ladles. Molten-metal fluxing agents may be chemically active solids, inert gases, or a chemically active gas. The type of flux depends, of course, on the composition of the melt and refractories. In iron and steelmaking, for example, the function of fluxes is twofold:

These fluxing treatments of molten aluminum are described in more detail in the article “Aluminum Fluxes and Fluxing Practice” in this Volume. Fluxing of other nonferrous alloys includes:

ing molten metal from melting units to tundishes, ladles, or molds Tundishes, which are intermediate holding vessels that have the ability to regulate molten-metal flow using slide gates, sliding nozzles, or a stopper Holding furnaces or transport crucibles that maintain a melt (or superheat) over a period of time Molten-metal transfer pumps Teeming ladles Dosing furnaces for delivery of controlled quantities (doses) of molten metal to the mold or die


Fluxing of magnesium alloys using salt

fluxes or inert gas as a cover (see the article “Magnesium and Magnesium Alloy Castings” in this Volume).
Fluxing of copper alloys to remove gas or prevent its absorption into the melt, to reduce metal loss, and to remove specific impurities and nonmetallic inclusions (see










This article describes these typical systems for transporting, holding, or delivering molten metal to the mold/die system. Specific moltenmetal treatments and refractories are covered in other articles, as noted in the preceding overview on fluxing.

174 / Molten Metal Processing and Handling

Launders Launders are refractory-lined steel troughs or channels for transporting molten metal. The refractory may be brick or, if the launder size is not excessive, cast and sintered as single pieces of refractory. Launders often incorporate projections (dams and weirs) into the line of flow of the molten metal to help prevent the transfer of slag into the molds during pouring. Dams are the projections rising from the bottom of a launder. Weirs are projections coming down into the top of the pour stream. Launders are used in the tapping of blast furnaces (several hundred in 30 min) and cupolas (1=10 that of blast furnace tapping). Launders are also used in smaller furnaces, where melt is transferred from one chamber to another (see, for example, the article “Vacuum Induction Melting” in this Volume). With systems opertaing under atmospheric conditions, the launder is usually covered so the metal is not

Fig. 1

Fig. 2

Electric low-energy holding furnace with laundered inlet. Courtesy of Schaefer Furnace Inc.

directly exposed to the atmosphere. The covers are heavily insulated and are hinged to permit inspection of any portion of the launder. The trough is lined with heavy insulation having low thermal conductivity. This insulation is also nonwettable by the molten metal. Metal conveyance free from turbulence throughout the launder system is important, so the metal moves smoothly and continuously usually under a protective layer of stationary metal oxides. Most importantly, metal temperature fluctuations are thus significantly reduced, and manual work is lessened. The size of launders varies, of course, depending on the scale of the production operation. Small holding furnaces may have an integrated launder inlet (Fig. 1). A much larger launder unit is shown in Fig. 2 with its connection to a tilting aluminum furnace. In this case, the tilting furnace is connected to the launder with two cast iron parts. These cast iron parts need maintenance after each cast in order to keep them working properly. If they start to leak, it takes quite some time to prepare them or replace the cast iron pieces before the next cast. Periodic skimming of the launder is also another maintenance task. Flexible launder joints (Fig. 3) are also used for the transfer of molten aluminum. They are very lightweight and easy to install. One-layer launder joints are used only once because of possible leakage. Flexible launder joints also include multiple-layer designs that can extend the load-carrying capacity and be used multiple times. The thickness and density of the blanket and the hot-face cloth can be adjusted according to the operating temperature, possible chemical reaction, and whether or not there is any flame contact.

Transfer area from a tilting aluminum furnace to a launder. Courtesy of AGN Nachrodt

Tundishes The tundish is an intermediate holding vessel used in steel continuous casting systems or in the handling of other metals such as superalloys. The metallurgical role of a tundish is to facilitate separation of inclusions and slag, which floats with sufficient residence time of the metals in the tundish. Tundishes also perform the function of a metal reservoir that allows unabated casting during ladle changeovers for sequence casting of heats. In this case, it is imperative that the ladle change be done in the shortest possible time. The tundish is a ceramic bricked or cast tub into which the metal flows after leaving a teeming ladle. Some systems are designed to pour directly from the melting crucible into the tundish. A typical tundish design has significant depth and provides a low-velocity flow path from the point of entry of the metal to the bottom installed nozzle through which the metal exits into the molds or feeding system. Tundishes are preheated prior to metal pouring and are often covered to minimize radiation heat losses. Tundishes may also be heated. For highly deoxidized steels, they are protected with an argon gas cover to reduce reoxidation and inclusion formation. An artificial slag is usually added to absorb inclusions that are floating out. Metal is poured away from the nozzle location directly on a wear resistant pad. A constant metal height above the tundish nozzle is maintained to discharge the metal at a constant rate and, in turn, to maintain constant casting speeds. The height of metal in the tundish is kept constant by regulating the metal pour rate from the ladle. The height of the liquid head in the tundish makes the flow (pour rate) out of the tundish nozzle a fixed rate. For specialquality steels or other melts with stringent surface and cleanliness requirements, shrouding is commonly employed to protect the steel from air, particularly for aluminum-killed grades, which have a potential of forming alumina inclusions. The shrouding systems include inert gas shrouding and refractory-tube shrouding. The concept of a refractory-tube shrouding

Fig. 3

Flexible launder joint in aluminum unit. Courtesy of Alunorf and Ceranex

Transfer and Treatment of Molten Metal—An Introduction / 175

Fig. 4

Schematic of a refractory-tube shrouding system for minimizing oxidation during pouring. Courtesy of Metaullics Systems Division, Pyrotek, Inc.

system is shown in Fig. 4 for ladle-to-tundish and tundish-to-mold pouring. Refractory-tube shrouds for steel are made of fused silica or alumina graphite. Nominal residence time of the metal in the tundish is important, because this controls the extent to which entrained slag will float to the top of the tundish and be removed from the pour. This has led to a trend toward tundishes with larger volumes and thus longer residence times for a given throughput. Flowcontrol devices, such as refractory dams and weirs, are attached to the tundish to distribute the metal flow, to minimize turbulence, and to eliminate dead flow zones (Ref 1). Refractory baffle walls in tundishes are also used to help direct the flow of liquid, so inclusions have an easier chance to float out at a later point. The holes in the baffle walls are typically approximately 50 mm (2 in.) and do not catch or trap inclusions (as in the case of molten-metal filters for inclusions in aluminum melts). Dams and weirs also are used in the tundish to collect slag particles. Bottom-projecting dams direct metal flow to the top of the tundish as a flotation aid for the slag particles. Flow modifiers have no influence on the very small inclusions collected by ternary refining, but they can help clean the steel of midsized inclusions in the 50 to 200 mm (2 to 8 mils) range. Flow controls are not needed for larger inclusions (which have higher Stokes rising velocities). The refractory lining in a tundish includes a service lining and a back-up safety lining. Refractories in the tundish are subject to corrosive wear, slag penetration, and thermal shock. Until recently, refractory wear in a tundish was usually of secondary importance to the ease of installation and tearout, as well as metallurgical considerations of the steel. During deskulling, the hot-face service lining is usually lined with a thin, 2.5 to 7.5 cm (1 to 3 in.), coating layer of olivine (magnesium iron silicate), MgO, or a mixture of these. This coating is

usually a slurry spray material or a dry vibratable powder that is installed behind a steel form that is heated to set a phenolic resin binder. With the drive for longer and longer cast sequences, the emphasis of tundish linings is changing toward more corrosion-resistant coatings that last longer and are cost-effective but that also maintain good deskulling properties. In steel continuous casting, the tundish continues to play a critical role in postfurnace steelmaking practice (along with steel ladle metallurgy treatments, as described in the article “Steel Melt Processing” in this Volume). Properly designed tundish flow modifiers provide substantial benefits in terms of both quality and productivity, and extensive work in physical and mathematical modeling has been done to understand and improve the fluid flow pattern in the tundish (Ref 2–4). Mathematical prediction of flow pattern has become more widely used in analyzing turbulence. Molten-Steel Filters. For critical applications of steel, the presence of inclusions with a diameter greater than approximately 50 mm (2 mils) needs to be prevented. There was a move in the industry in the 1990s to use filters (foams or screens) for such applications to help eliminate inclusions under approximately 50 mm (2 mils) in size, which are not susceptible to flow modifiers. However, this use of filters in tundishes for steel continuous casting did not prove effective, because the large number of inclusions would quickly clog the filter.

Molten-Metal Pumps Molten metals have traditionally been transferred from the melting vessel to a mold or casting unit by tapout or lip pouring (essentially gravity flow systems) or by using an intermediate ladle. Molten-metal transfer/recirculation pumps are now commonly used with aluminum, magnesium, zinc, and lead alloys. The pumps may be a mechanical device or a device operating on the principle of electromagnetic pumping. Mechanical pumps are typically centrifugal pumps with a rotating impeller in a volute casing. They can be used as either transfer pumps or as recirculation pumps in melting or holding units. Mechanical pumps are also used for gas injection. Electromagnetic pumps are also widely used in industry, because they have no moving parts, so routine maintenance is minimized.

Mechanical Pumps Mechanical pumps for molten-metal recirculation and transfer (Fig. 5) traditionally have graphite posts, which must be rebuilt approximately three times per year when they become oxidized or broken. Newer techonolgy includes molten-metal pumps with ceramic posts (Fig. 5) that are held in compression from a tension rod made of high-temperature alloy steel.

By replacing the posts with a stronger material that will not oxidize, the life of the pump has been extended, with fewer rebuilds. In this design, fragile ceramic sleeves are no longer needed to protect graphite posts from oxidation and abrasion. Ceramic sleeves can also hide post oxidation, which could result in unexpected failures. The ceramic posts held in compression are stronger than graphite; if they do break, the base is still supported by the steel rods. Mechanical recirculation pumps are used to improve heat transfer by forced convection. This is particularly important in aluminum reverberatory furnaces (see the article “Reverberatory and Stack Furnaces” in this Volume). In reverberatory furnaces, forced convection keeps the entire melt at a more uniform temperature, so that the burners above the melt transfer more radiant heat to the surface of the melt. If the surface temperature becomes hotter than the lower parts of the bath, then the heat transfer for melting become less efficient in reverberatory furnaces. Other nonferrous alloys, such as zinc, magnesium, and copper, also benefit from molten-metal circulation, but pumping systems for circulation are not usually employed. Mechanical stirrers are effective, particularly for zinc, due in part to the lower heat of fusion of zinc and the smaller melt sizes usually used. Mechanical transfer pumps are now commonly used with aluminum, magnesium, zinc, and lead alloys. A mechanical pumping system has several advantages. Pressurized flows permit larger volumes of metal to be transferred per unit time, with resultant productivity gains. The ability to move greater volumes of metal in shorter times also means that less superheat must be put into the metal to accommodate temperature losses during transfer. Thus, shorter overall furnace melting cycles are possible. Because most metal-transfer pumping systems have submerged intakes, dross entrainment from the surface is avoided, providing better metal quality. Transfer pumping per se does not produce additional oxidation or dross losses, because molten nonferrous metals are always subjected to turbulent flows (high Reynolds numbers), even with gravity flow tapout systems. Transfer pumps can often be used with closed piping systems to provide transfers surmounting physical obstacles (over other equipment, for example) or overcoming head and height limitations when gravity flow is limited or not possible. Melting and casting facilities can therefore be mixed and matched and need not be directly in line with one another. Transfer pumps are now commonly used in many segments of the aluminum casting industry to move molten metal from melting furnace to holding furnace, to feed or empty auxiliary refining vessels (used for alloying, degassing, and filtration), to empty and fill transfer ladles, to directly pour castings and fill molds, or to feed casting machines in the wrought industry. In the magnesium, zinc, and lead metal casting industries, mechanical pumps transfer metal

176 / Molten Metal Processing and Handling

Fig. 5

Fig. 6

Molten-metal pumps with ceramic posts and steel tension rods. (a) Recirculation pump. (b) Transfer pump. (c) Pump elements. Courtesy of Metaullics Inc.

Schematic of metallic centrifugal pump used to transfer magnesium

between furnaces, fill ingot and pig molds, feed die casting machines, cast electrogalvanizing anodes, and empty galvanizing kettles when repairs or desludging is required. Molten copper alloys are also transferred by pressure pump systems. Of the various kinds of mechanical pumping systems, a positive-displacement pressure pump is often quite useful for small furnaces or small flows (<45 kg, or 100 lb). When actuated, external air pressure displaces the volume of air in the cylinder, which forces an equivalent volume of liquid metal through the delivery tube. Another type of pump is the siphon pump. Both of these pumps are quite capable of filling small molds where the accuracy of fill demanded is moderate (several percent variation) and where pressure head requirements are minimal (0.3 to 0.6 m, or 1 to 2 ft, maximum). When larger volumes of metal, more sustained flow, or higher head is required, a centrifugal pump can be used. Flow is directed by an impeller up through a riser tube. Operated either by air or electric drive, these pumps are

typically constructed of graphite and are easily operated in molten aluminum, magnesium, and zinc alloys. It is important to keep the riser tube open, using insulation and careful external heating if necessary, to prevent freezeups, especially if use is intermittent. Another type of centrifugal mechanical pump often used with magnesium alloys is shown in Fig. 6. Pump construction is usually cast iron, low-carbon steel, or 400-series stainless steel. Nickel impurities in metal pump components should be specified to be below 0.20% to avoid the dissolution of nickel in the molten magnesium and the subsequent reduced corrosion resistance of magnesium castings. Centrifugal pumping systems for zinc are usually constructed of graphite, but metallic centrifugal pumps are often used as well. Pumping systems for lead are usually constructed with cast iron because little corrosion is experienced. Cast iron can also be used to pump zinc as long as the temperature is kept below 480  C (900  F), above which molten zinc begins to dissolve iron. Austenitic stainless

Transfer and Treatment of Molten Metal—An Introduction / 177 steels can be used as pump materials with aluminum-containing zinc alloys (0.2 to 3.5% A1). Stainless steel must not be used, however, with pure zinc, because liquid metal embrittlement and intergranular attack occur quite rapidly. With zinc-aluminum galvanizing alloys (40 to 60% Al), the high operating temperature (595 to 650  C, or 1100 to 1200  F) is in the region where austenitic stainless steels become sensitized to greater intergranular carbide precipitation and subsequent accelerated corrosion. Stainless steel pumps should therefore not be used for this application. A useful but expensive alternative for the latter two applications is an Mo-30W or Mo-40W pump shaft, which provides excellent corrosion resistance but is fairly brittle at high temperatures.

Electromagnetic Pumps Electromagnetic pumps have been developed for circulation of molten metal and controlled mold-filling capabilities. This type of pump is constructed from ceramic materials and has a linear motor that induces pumping on the basis of the left-hand rule of electromagnetic field theory, using the molten metal as the secondary field. It is particularly useful for metals with good electrical conductivity. One specific advantage of this type of pump is that there are no moving parts. However, the moltenmetal channel opening must be kept very clean and free of oxide buildup or dross; otherwise, the field strength and resulting pumping capacity can be greatly reduced. Electromagnetic Pumps for Furnace Circulation and Scrap Recycling Technology. Electromagnetic (EM) pumping systems are used for application of forced circulation in both foundry and casthouse aluminum melting. The EM pumps are supplied to suit the furnace size and application, with mass flow capability from 1 metric tonne per minute to greater than 20 metric tonnes per minute. All pumps are operated at main voltage and frequency and are fully controllable for both speed and direction. The consumable pump tube is replaced approximately every 6 months, especially with mixed-charge scrap melting, although with certain applications, up to 18 months of continuous operation are possible. An EM pump is especially suitable for those melting furnaces that may not have sidewells, or furnaces that are not stationary but rather tilt-pour. With the consolidation in the foundry industry, larger, volume-oriented castings producers are increasing their central melting furnace sizes well above 10 tons, making the application of an EM pumping system to circulate the furnace a viable option. In terms of operation, an EM pump is connected via insulated tubes to the furnace charge chamber and an external refractory-lined charge well, where a unique internal bowl shape develops a vortex with, low-turbulence scrap submergence (LOTUSSE Metaullics

Systems) in the molten-metal stream. The vortex creates an ideal entry point for high-surface-area, low-density scrap and alloying materials that need rapid submergence to avoid oxidation with direct flame impingement. These include machining chips, borings, turnings, flashings, used beverage containers, and so on. While many foundries would only melt a small portion of such materials, cast wheel producers as well as producers of castings with inhouse extensive machining produce a sufficient quantity of such light-gage materials (machining chips) to make the EM pump/LOTUSS scrap recycling technology quite feasible and economical. Indeed, the same LOTUSS scrap recycling technology can be employed with mechanical pumps as well. The choice of which system to employ—EM pumping or mechanical pumping—will depend on many specific user’s economical and process factors. Metal flow capability and electrical requirements of typical EM pumping parameters are: Metal flow rate, t/min

Kilovolt amperes (VA) rating

8–10 8–10

110 110

15

130

8–10

110

15

130

20 12–14

150 110

Features

100 mm (4 in.) pump 100 mm (4 in.) pump and redesign tie bars 100 mm (4 in.) pump and laminations 150 mm (6 in.) pump and redesign tie bars 150 mm (6 in.) pump and laminations 150 mm (6 in.) pump, 9 poles 200 mm (8 in.) standard pump

Use of a either a centrifugal or electromagnetic metal pumping system has the following benefits:
Increased metal yield (typically for dry








Alloying materials in foundry melting are prone to problems with dissolution, incomplete melting, and less-than-optimum recovery. The combination of rapid-submergence charging capability, coupled with the forced convectional circulation provided by the pumping action, helps to ensure quick dissolution and higher recovery than would otherwise be possible. Pure copper, silicon, or aluminum master alloys can be processed by this technology. Smaller lump forms, granules, and small ingot bars are preferable to powdered additions. Recent advances (Ref 5) in EM pumping technology have integrated several other functions into the heretofore single application of forced circulation. The EM pumps are now capable of providing transfer of metal out of the closed system into a transfer ladle, launder, or a holding furnace, and with suitable gas injection systems, degassing is also possible. Additionally, such systems can be outfitted with noncontact sensor devices (video camera, level sensor, etc). Mold-Filling EM Pumps. Electromagnetic pumps are easily used for controlled mold filling in such applications as permanent mold casting, large die casting, low-pressure die casting, and perhaps small ingot casting. They offer exceptional control of metal flow throughout mold filling, and turbulence is minimal since there are no moving parts. This puts the highest-quality metal into the mold. Other benefits are:
Metal filtration is easily performed using a

chips, 98%)




compared to a static bath)




yield; silicon, +95% yield)





Increased melt rate (typically 10 to 20%
High-efficiency alloying (magnesium, +95%
Reduced energy use (equivalent to increased

output)


Chemical and thermal homogeneity within

  the furnace (þ /5 C, or 9 F, within the furnace)
Reduced dross formation (depending on scrap types)

With the LOTUSSE scrap recycling system, many different materials are possible to be charged into the special-shaped bowl. However, it is not possible to charge large sows or large scrap castings through the system. The following are typical smaller, lighter-gage aluminum materials that are suitable for processing in this vortex scrap charging system:
Machining chips, turnings, borings (which are

most prevalent in integrated foundry operations)


Shredded sheet scrap
Extrusions

Large fragmented castings Small castings Scalper chips Used beverage containers Aluminum chips





filter over the pump inlet, filtering all metal to the casting. The response to a desired change in flow rate is very fast. Plant safety is increased due to automation— no need for hand ladling. The alloy in the furnace can be tested at any time, and additions made without interfering with production. In low-pressure casting, the furnace is not a closed pressure vessel. The temperature of the metal delivered by a pump is very close to that of the moltenmetal bath, so superheating of the furnace is not required.

Types of EM Pumps. The types of EM molten-metal pumps include:






Single-phase conduction pumps Jumping ring pumps Multiphase linear pumps Pinch-effect pumps Direct current conduction pumps

Single-phase EM pumps (Fig. 7) can be applied to almost any metal casting process. They are particularly suited to low-pressure casting,

178 / Molten Metal Processing and Handling

Fig. 7 Example of two sizes of single-phase conduction pumps with metal filters over the bottom inlet where they fill the mold from below while exhibiting very good control over the rate of metal delivery, along with the ability to almost instantly change the flow rate. Repeatability is generally excellent. It is also very easy to hold the metal just below the mold between castings to minimize oxide generation and to reduce overall cycle time. Other uses include squeeze casting through the side of a vertical shot sleeve, just above the top of the shot tip, to eliminate dross from top-filling; high-pressure die casting; bottom-fill lost foam; and all kinds of gravity casting. Single-Phase Conduction Pump. The singlephase conduction pump uses alternating current applied to a coil of many turns of wire that is wound around a laminated ferrous core, which constitutes the primary winding of a transformer. Around the same laminated core is a ceramic material that is compatible with aluminum, and this ceramic has an internal passage into which molten aluminum flows from the metal bath. This passage forms a complete circuit (loop) around the laminated core; hence, the molten aluminum therein forms a one-turn shorted secondary winding on the transformer. By virtue of this single turn, a high-amperage flow is induced into the molten aluminum. A portion of the loop is flattened to allow for close spacing of the magnet pole pieces and to make the current density in that portion of the loop relatively uniform throughout the cross section. This is the area (pumping vein) in which the pumping force will be produced. Two pole pieces are placed as closely as possible to the broad face of the pumping vein, and additional electrical coils are around the laminated core that forms the magnet pole pieces. This creates an alternating magnetic field across the pumping vein. Since the transformer primary and the magnetic field coils are powered from the same alternating current source, both the current flow in the metal and the applied magnetic field alternate polarities together. If either the current in the metal or

the magnetic field were static while the other one alternated, the metal would experience an alternating force, first in one direction, then in the other. Since the magnetic field and the current in the metal are both alternating together, the force on the metal simply pulsates from zero to maximum at twice the applied line frequency. These pressure pulsations may prove beneficial when pumping thixotropic materials such as metal-matrix composites, because the pulsations introduce shear into the moving metal, thus greatly increasing the fluidity, which helps the metal to flow into mold details. Since the electrical current is induced directly into the molten metal, there is no need for electrodes in contact with the metal, also eliminating the need for heavy cables and very large direct current power supplies. This is the type of pump most commonly used for mold-filling applications in aluminum foundries, because of its ease of use, reliability, and repeatability. Jumping Ring Pumps. If a coil of wire is wound around a straight ferromagnetic bar or rod, and a ring of any conductive metal is placed near it over the same bar, with alternating current applied to the first (primary) coil, the ring is very strongly repelled from the coil. This is called the jumping ring pump. These pumps do not generate much force in the molten metal and are most often used for stirring metal in a furnace. Multiphase Linear Pump. If a multiphase electric motor stator (the stationary part with the electrical windings) were cut open and rolled out so the windings are all in a line, when power is applied, there would be a moving magnetic field. If a conductive material is placed above this stator, electrical current would be induced into that material. This would also produce corresponding magnetic fields, which would interact with the moving magnetic field, and the interaction between these magnetic fields would produce a force on the conductive material, tending to move it in the direction of the moving magnetic field. As described, this would be a very inefficient pump due to poor control over the magnetic field. There are two common ways to improve this. One is to place a plain pole piece opposite the stator while keeping the molten metal between the stator and the coil assembly. The second improvement is that the stator could be rolled to form a long tube, with the moving magnetic field traveling down the length of the tube. This type of pump is typically used for high-volume metal transfer and stirring of metal in large furnaces. Pinch-Effect Pumps. Another form of pump induces current into a cylinder of molten metal that has a tapered ceramic mandrel in the center. Since the metal is trying to collapse on itself, it is forced toward the narrower end of the taper, which is also the metal discharge area. These pinch pumps typically use a highfrequency coil to improve the efficiency of the induction, and they also typically use water cooling to keep the coil from overheating.

Direct current conduction pumps use electrodes to pass a large direct current into the fluid to be pumped, and a permanent magnet supplies the magnetic field. These pumps are not suitable for use with aluminum, since they would require electrodes to be in contact with the molten metal, and there is no suitable material available for this purpose. Additionally, the amperage required is very high, which would need very heavy cables and a direct current power supply capable of supplying as much as 10,000 A.

Pouring and Dosing As noted, molten metals traditionally have been transferred to a mold or casting unit by tapout or lip pouring by using an intermediate ladle. Traditional metal transfer from the molten-metal supply to the mold is done with open-top ladles ranging from hand held to multiton, crane-supported, mechanized tilt ladles. Underpouring from a ladle, either through the incorporation of a teapot spout or bottom stopper rod and nozzle, reduces slag inclusions in castings (Fig. 8). In addition to gravity-feed pouring, a number of other methods are also used to fill molds. For example, molten-metal pumps (as briefly described in the preceding section of this article) are used. Other methods include the use of dosing furnaces in die casting operations (e.g., see the article “Automation in High-Pressure Die Casting” in this Volume). Dosing furnaces are replacing ladle furnaces and are described in the next section of this article. Another method is low-pressure filling systems for sand casting of heavy metals. In lowpressure pouring (Fig. 9), the molten metal is pushed into the open mold from below. Compared with gravity casting, the low-pressure method offers a number of important advantages:
Reliable, simple, and tight mold-to-nozzle

docking action


Adaptable pressure buildup for active control

of mold-filling characteristics


High dosing accuracy based on pressure and

(where applicable) level control


Essentially laminar mold-filling flow
Cleaner melt supply

Finally, automated methods encompass a wide range of mechanized methods and specialized equipment. This includes mechanized ladles and pouring robots that are progarmmed to transfer measured amounts of metal from a storage vessel or a controlled pouring unit to the mold. Pumps and dosing furnaces are also suited to integrated automation. Benefits of automated pouring include:
Increased productivity due to the speed and

consistency of pours from one mold to the next

Transfer and Treatment of Molten Metal—An Introduction / 179

Fig. 8

Four types of pouring ladles. (a) Open lip-pour ladle. (b) Teapot ladle. (c) and (d) Bottom-pour ladles


Improved process control with tracking of

production and inventory information. In some cases, automatic pouring equipment also allows a reduction in pouring temperatures with significant savings in energy and refractory costs.
Improved working conditions in terms of health and safety, environmental, and operational factors

Pouring Methods

Fig. 9

Low-pressure mold filling. (a) Bottom fill. (b) Fill from top of the mold


Increased yield due to fewer overpours,

spills, and allowance for smaller pour basins. Consistent, controllable pour rates also may allow for the design of less complex gating and riser systems as well as for simultaneous multiple-pour streams in the same mold.
Reduced scrap costs due to reduction of inclusions in castings and the reduction of pouring failures such as underpours, interrupted pours, and irregular pour rates
Improved quality control with better temperature control within the pouring vessel, more consistent metallurgy, and automatic feeding of alloy and inoculant additions into the pouring ladle or metal pour stream

The two key variables in pouring are pouring rate and duration. The optimum pour rate is the maximum rate at which a mold will accept metal, and the optimum duration is the time from the start of the pour until the mold will no longer accept metal. Other important considerations in automatic pouring systems include the relative positioning of the mold and pour spout or nozzle and the sequencing of the pour. In manual pouring, the worker compensates for variables in the mold, ladle, and metal while pouring to produce the best results for each pour. Duplication of this effort is the challenge of automatic pouring. Optimum pouring time is based on the weight and shape of the casting, the temperature of the molten metal and its solidification characteristics, and the heat-transfer characteristics and thermal stability of the mold. Most foundries may also pour many different castings from one heat, or even from one ladle. Therefore, these foundries do not attempt to control pouring time, but they do control the speed at which molten steel enters the mold cavity. This control is achieved through the design of the gating system. Pouring Ladles. The three basic types of pouring ladles are:
Bottom pour
Teapot
Lip pour

Ladle capacity normally ranges from 45 kg to 36 Mg (100 lb to 40 tons), although ladles having much larger capacities are available.

Ladle Linings. Ladles with capacities above approximately 360 kg (800 lb) are lined with layers of firebrick covered with a thick refractory coating. Smaller ladles are protected by monolithic linings prepared by ramming or by pouring and vibrating refractory mixtures around forms. For molten steel, the optimum refractory for this purpose is high-purity alumina, but because of the high cost of alumina, alternative materials such as olivine and zircon are more widely used. The service lives of these monolithic linings, which are most commonly used for small ladles, vary considerably, depending on the refractory mixture and the type of service. Thermal expansion and contraction can cause hairline cracks, into which hot metal enters. This results in erosion of the lining. Erosion not only shortens the life of the lining but also contributes to excessive dirt in the steel. Disposable fiber linings that do not require preheating are gaining acceptance. Preheating the ladle helps to prolong the life of the ladle lining. Heating the lining to a temperature close to that of molten steel serves to lessen the thermal shock to the lining when the molten steel contacts it. Common practice is to preheat ladle linings to 815 to 1095  C (1500 to 2000  F) with oil-fired or gas-fired torches. Ladles lined with high-purity alumina can be preheated to temperatures as high as 1370  C (2500  F). Bottom-pour ladles have an opening in the bottom that is fitted with a refractory nozzle. Stopper rods (Fig. 10a) were commonly used to control flow from the unit, but more typically a valve-type mechanism (Fig. 10b) is often used. Bottom pouring is preferred for pouring large castings from large ladles, because it is difficult to tip a large ladle and still control the stream of molten steel. Also, the bottom-pour ladle delivers cleaner metal to the mold. Inclusions, pieces of ladle lining, and slag float to the top of the ladle; thus, bottom pouring greatly reduces the risk of passing nonmetallic particles into the mold cavity. Pouring ladles may also include options for melt treatments or shrouding. For example, Fig. 11 shows a schematic for a bottom-pour

180 / Molten Metal Processing and Handling

Fig. 13 Fig. 10

Bottom-pour mechanisms (a) stopper rod (b) sliding value

Fig. 12

Fig. 11

Bottom-pour teeming with argon insertion ring. The term teeming typically applies to the pouring of molten metal from a ladle into ingot molds.

mold setup with an argon insertion ring that shrouds the pour from oxygen and nitrogen pickup from the atmosphere. Mold powders or fluxes are added to the advancing metal front to improve surface quality by providing thermal insulation from the mold surface. At the end of the pour, exothermic compounds may be added to the top of the mold to insulate the top of the solidifying ingot and provide a degree of hot

Typical teapot ladle used to pour small- to medium-sized castings

topping, where the molten metal will fill the shrinkage occurring upon solidification lower in the ingot. Argon-oxygen decarburized electrodes also may be top poured, but the need to continually close the ladle tap, transfer the ladle, and reinitiate argon shrouding produces variable conditions from ingot to ingot. Should difficulty in closing the nozzle be experienced during top pour, then a safety problem arises. Thus, bottom pouring is the more generally used process. On the other hand, it is impractical to pour molten metal into small molds from a large bottom-pour ladle. The pressure head created by the metal remaining in the ladle delivers the molten metal too fast. Also, the time required to fill a small mold is short, thus requiring that a large bottom-pour ladle be opened and closed many times in order to empty it. This may cause the ladle to leak, although special nozzles have been developed to minimize leakage. Despite the fact that the size of bottom-pour

Schematic of automatic mechanized ladle

pouring

with

ladles could be scaled down for pouring smaller castings, this is unnecessary because of the almost equal ability of the teapot ladle to deliver clean metal. The teapot ladle incorporates a ceramic wall, or baffle, that separates the bowl of the ladle from the spout. The baffle extends almost four-fifths of the distance to the bottom of the ladle (Fig. 12). As the ladle is tipped, hot metal flows from the bottom of the ladle up the spout and over the lip. Because the metal is taken from near the bottom of the ladle, it is free of slag and pieces of eroded refractory. The teapot design is feasible in various sizes, generally covering the entire range of casting sizes that are below the minimum size for which the bottom-pour ladle is used. Lip-pour ladles resemble the teapot type but have no baffles to hold back the slag. Because the hot metal is not taken from the bottom of the ladle, this type of ladle pours a more contaminated metal and is seldom used to pour high-alloy castings. Nevertheless, it is widely used as a tapping ladle (at the melting furnace) and as a transfer ladle to feed smaller ladles of the teapot type. Methods of Automatic Pouring. There are two basic methods of automatic pouring: mechanized ladle pouring and direct pouring. Mechanized ladles generally have capacities matched to single mold requirements and include load cell weighing systems and controlled tilt rates (Fig. 13). In the case of direct pouring, metal is metered directly into the mold from the pouring vessel, with the appropriate controls to ensure the desired results. The use of a mechanized ladle allows for inladle alloying and inoculation before pouring. Dripping of molten metal onto the mold is eliminated when a mechanized ladle is used in combination with a stopper rod and nozzle

Transfer and Treatment of Molten Metal—An Introduction / 181

Fig. 14

Fig. 15

Automatic pouring into molds on continuously moving line. (a) Pouring into flaskless molds on an indexing line from bottom-pour unit with stopper rod. (b) Synchronized pouring on a continuously moving line with mechanized ladles capable of x-axis and y-axis travel and synchronization with the pouring line.

Rotary mechanical pourer with molten metal level detector and control

pouring unit. This is very important when pouring lost foam molds. Ladle tilting about an axis in line with the pouring lip can reduce moltenmetal stream travel during pouring. The capacities of mechanized pouring ladles are based on anticipated metal requirements. The dispensing of a metal stream from a ladle at a specific point requires that the nozzle or pouring lip not move, relative to the mold,

during pouring. In semiautomatic pouring units, bottom-pour ladles may be mounted above indexing molding lines to pour into molds indexed into position beneath the ladles (Fig. 14a). Synchronized pouring on a continuously moving mold line (Fig. 14b) may also require the use of a pouring vessel to measure predetermined weights of metal and either one or two (depending on line speed) mechanized

pouring ladles. Another approach is a mechanized rotary pourer with optical or laser monitoring for automatic control of melt level (Fig. 15). Direct bottom-pour ladles use either slide gates or stopper rods and nozzles to control and stop the flow of metal to the mold. Slide gates and nozzles are most commonly used to control flow rates on bottom-pour ladles for continuous and semicontinuous strand casting operations. Slide gates (Fig. 10b) offer more reliable shutoff than stopper rod valves; however, service life of the slide gate can be shorter than that of the stopper rod. Stopper rods and nozzles offer more control of pour rates than slide gate valves and are less expensive. For these reasons, bottom-mounted stopper rods and nozzles are more common in foundry pouring operations. Controls of Pouring. Satisfactory pouring results can be obtained in some automatic pouring applications without complex electronic control systems. Some requirements can be met by establishing a flow rate and simply timing the pour. A slightly more complex system includes dual pour rates and pour times. Simple infrared photosensors designed to detect the presence of molten metal can be used to end the pour cycle. If the exact weight of metal required for a mold is known and a mechanized ladle is weighed while filling, at least one variable is under control. If the movements of the operator controlling the pouring of a mechanized ladle are measured and then stored in a programmable controller, these movements and the results can be duplicated under automatic control. The programming of a controller by duplication of remote control is referred to as a teach-in system. The program is referred to as the pouring profile. Weighing of a bottom-pour ladle and pouring by weight can also be effective in some cases, although the relatively small capacity of the

182 / Molten Metal Processing and Handling ladle results in substantial variations in metal head pressure above the stopper and subsequent variations in pour rate. Increasing the capacity of the pouring vessel, facilitated by electric heating, reduces the variation in head pressure. The incorporation of a process control system to compensate for reduced head pressure through stopper positioning by feedback of weight, anticipated weight, or actual bath height is effective in some cases. If the rate of metal flow through a nozzle is constant for a constant head pressure, the repeated variations in the position of a stopper rod above the nozzle will yield repeatable and controlled pouring rates. A teach-in system is used to program the pouring profile. If slag buildup on the nozzle and variations in metal head pressure cause flow variations, the system pouring profile program must be corrected by a process control feedback loop. Feedback can be provided by a timing and metal-presence sensing system that checks the rate of mold fill or by a more sophisticated control system for mold-level measurement. Mold-level measurement and control systems fall into two categories: laser systems and vision systems. Both of these electronic process control systems monitor the level of metal in the pour cup and control the position of the stopper rod to maintain a nearly constant level throughout the pour. Optical sensors provide input to help regulate the electromechanical stopper drive in such a way that the liquid level in the filling head is kept at a constant level. This type of feedback system most closely duplicates the efforts of a skilled operator. Camera or laser level sensors are used depending on production factors. Manufacturers who mainly use a sprue cup diameter of 85 mm (3.3 in.) or larger will usually select the camera-based solution, whereas the laserbased solution will be used particularly where the focus is on minimizing the circulation material and a sprue cup diameter smaller than 80 mm (3.1 in.) is employed. The laser reflection angular detection system (Fig. 16a) determines the position of a lowpower laser beam reflected from the surface of the molten metal in the pouring basin. The laser beam is directed at the pouring basin from an angle. The distance to the molten-metal level will affect the incident and reflected angle and therefore the position of the beam on a sensing unit. The comparative value of the beam position on the sensor is used to control the position of the stopper rod; thus, the level of the metal in the pouring basin is controlled. A vision system uses a video camera to observe the pouring process (Fig. 16b). The video signal is displayed on a monitor for the operator and processed electronically in a manner that separates the brightness levels of the molten metal from the other portions of the picture. The molten-metal stream is electronically blanked from the composite picture of the metal in the pouring basin and the area of metal fill compared to the actual area of the pouring

Fig. 16

Optical sensors for pouring control. (a) Laser sensor. (b) Vision sensor

basin. The comparative value is used to control the position of the stopper rod, and the level of the metal in the pouring basin is controlled. The most sophisticated pouring systems may incorporate more than one set of controls. The use of integrated control schemes allows the system backup, improved safety, and increased flexibility that are required for pouring a variety of castings. For example, the benefits of moldlevel measurement and control are enhanced when used in combination with a teach-in duplicated pouring profile in some pouring applications. Pouring Furnaces. Radiant element, electric resistance-heated pouring furnaces are used in both nonferrous and ferrous pouring applications. For nonferrous, low-temperature applications, metal or silicon carbide heating elements are used. For ferrous high-temperature applications, graphite rod heating elements are used. Radiant-heated furnaces are typically bottompour units with stopper rods and nozzles mounted in the furnace bottom or in a detachable pour box. The furnace can be tilted to drain the pour box and to contain the molten metal beneath the radiant heat source. Crucible-type electric resistance-heated and coreless induction-heated furnaces are most commonly used in nonferrous pouring applications. These applications vary from the simple dip-and-pour mechanized ladle systems to the underpour stream discharge, axial-pour, and trunion-pour furnaces. Underpour stream

discharge in aluminum pouring for continuous and semicontinuous casting processes uses the oxide skin on the metal surface to protect the metal stream from further oxidation while transferring metal from the melting furnace to the automatic pouring ladle. Channel-type electric induction-heated pouring furnaces are sealed and designed to use gas pressure to discharge metal from the furnace vessel. Pouring can be accomplished over the lip of a pour spout or through a nozzle, with or without a stopper rod, in a pour box mounted on the furnace vessel. Vessel gas pressure can be varied to control the rate of pouring or to control the level of metal over the pouring nozzle (Fig. 9b). Control of the level of metal in the pour box establishes a constant static head pressure above the nozzle and ensures consistent pour rates. Venting of the pressure in the vessel causes the metal in the pour box to drain back into the furnace. Inert gas can be used as the pressure medium in applications where the metal is particularly susceptible to oxidation.

Dosing Furnaces Dosing furnaces are replacing ladle furnaces in many applications, especially in aluminum high-pressure die casting (see the article “Automation in High-Pressure Die Casting” in this Volume). Aluminum castings traditionally have been produced by either direct pouring from the melting furnace into the mold, or from a

Transfer and Treatment of Molten Metal—An Introduction / 183 transfer ladle in which further treatment has been performed just prior to pouring the mold, or from a holding furnace dip-well from which the casting is poured either by automatic robotic ladle or by manual means. Increasingly, automated and controlled direct pouring is becoming popular, especially in high-pressure die casting operations where the dosing furnace technology can be meshed with the automated die casting machine controls. A number of different units exist commercially that are comprised of holding furnaces with discharge capabilities directly into the mold cavity. These devices provide a very precise dosage of molten metal and are thus known as dosing furnaces. The dosing operation (Fig. 17) begins with the introduction of compressed air into the pressure-sealed furnace, thereby causing the ascension of the melt in the riser tube. With further increase in the pressure, the melt is discharged from the riser tube and flows via the dosing channel into the shot sleeve of the pressure die casting machine. Flow of the melt is stopped by rapid discharge of the compressed air, the molten metal returns to the level of that in the dosing furnace, and the die casting machine can now execute the casting operation. There has been constant further development in dosing with the dosing furnace to fulfill casting demands. The following describes several of the key technologies that represent the current state of the art. The following demands are placed on the overall process:
High and reproducible dosing accuracy
Short main time and minimum waiting time
No loss of time with fluctuations in the cycle

cycle times without further adjustments, thereby excluding human influence. It is only necessary to enter the melt temperature and dosing weight. The dosing furnace automatically synchronizes itself with the die casting machine, thus always resulting in an optimized dosing cycle. Detection of Melt Level. It has been and still is necessary to be able to detect the level of the melt in front of the discharge edge at the rising tube. This was previously achieved by an electrode that generated an electric signal on contact with the surface of the melt (Fig. 18a). In order to avoid deposits of aluminum, this electrode had to be rapidly swung out after contact by means of a pneumatically operated device. Heat, oil, spraying agents, and sometimes open fire often led to failures in this component. A pneumatic scanning system (Fig. 18b) today (2008) replaces the electrode for detection of the melt level. Pneumatic scanning employs no moving components, requires no adjustment, and is especially advantageous with regard to high operational reliability and the absence of maintenance in a particularly harsh environment. The scanning accuracy is a precondition for exact dosing and is very well fulfilled by this system. The high degree of reliability in a pneumatic scanning system allows a level-scanning process known as the “top-stop-position” that has been used in the dosing furnace technology since approximately 2005. In this process, the height of the melt in the riser after dosing is not reduced to the level of the melt in the furnace but is held a little below the discharge

edge (Fig. 19). The furnace is thus always ready for dosing, so that there is no delay in discharge of the melt on receipt of the signal from the die casting machine. This compensates for fluctuations in the cycle of the die casting machine, and there is no loss of time through nonsynchronized process sequences. As a rule, this system achieves up to an 8% savings in the cycle time. Because the dosing furnace is a closed vessel that cannot be looked into, the filling level previously had to be determined by a weight load cell. This method was inaccurate since external influences on the furnace were also detected and could falsify the reading. With the development of new technology, the level in the furnace is precisely detected during the dosing process by means of the dosing contact and the system, so that the result is free from external influences. Level scanning is used to ensure that the furnace also remains ready for dosing while it is being refilled with molten metal (Fig. 20). Level scanning is started after opening of the filling cover, the level and the required amount of refill being semicontinuously detected in short cycles with indication of the current level. Furnace Recharging and Melt Holding. The quality of the melt in the transport ladle is only conditionally decisive for the casting quality. The decisive factor is the quality of the melt in the holding or dosing furnace and, finally, in the mold itself or the shot sleeve of the pressure die casting machine. If the quality of the melt in the shot sleeve is to be maintained or increased, closer consideration must

Fig. 18

Fig. 19

time Controlling and Monitoring the Dosing Operation. A digital control unit provides for real-time pressure control in the dosing furnace, and process information is monitored on a computer display with a large-surface keyboard for the operator. Precision hardware ensures adherence to dosage accuracy. The dosing program for the cold chamber pressure die casting machine optimizes the dosing to the shortest

Fig. 17

Dosing furnace

Level detection in dosing furnace. (a) Contact electrode. (b) Pneumatic scanner

Dosing sequence. (a) Melt hold, ready. (b) Melt flow (casting). (c) Melt again held, ready

184 / Molten Metal Processing and Handling

Fig. 20

Detection of melt level by level scanning

Fig. 21

Fig. 22

Recharging a dosing furnace

Dosing furnace with porous plugs

be given to the recharging operations and melt holding in the dosing furnace. Each filling or refilling operation is associated with a reduction in the quality of the melt. A correctly implemented filling of the dosing furnace can thus also substantially restrict the loss in quality. Figure 21 shows a correctly implemented filling operation. The filling funnel is kept full, comparable with the filling of a die in gravity die casting. This results in a lower formation of oxides, which remain on the surface, do not enter the dosing furnace, and can be skimmed off at the end of the filling operation. Holding of an aluminum melt in the dosing furnace allows for settling of inclusions to a major extent. Measurements confirm that the density index or other quality parameters usually improve during quiescent holding periods. The quality of the aluminum melt thus continuously improves with the holding time, and the melt quality is not negatively influenced by the dosing process itself with dry compressed

air (protection by the oxide skin). One option to provide further improvement in melt quality within the dosing furnace is through the installation of porous plugs (Fig. 22). Introduction of inert gas, preferably argon, at the lowest point in the furnace enables reliable maintenance of a hydrogen content in the furnace to acceptable low levels. ACKNOWLEDGMENT The editors extend thanks to the following for contributions to this article:
Edward S. Szekeres, Casting Consultants

Incorporated, and Robert D. Pehlke, University of Michigan, on tundishes in continuous casting of steel
Martin Reeves, Striko-Dynarad, on dosing furnaces
Lee Gouwens, CMI Novacast, on electromagnetic mold-filling pumps
Dave Neff and Paul Campbell, Metaullics Systems Division, Pyrotek, Inc.

REFERENCES 1. I.V. Samarasekera and J.K. Brimacombe, The Continuous Casting Mould, Continuous Casting, Vol 2, Heat Flow, Solidification and Crack Formation, ISS of AIME, 1984, p 33–44 2. Y. Sahai and Toshihiko, Melt Flow Characterization in Continuous Casting Tundish, ISIJ Int., Vol 36 (No. 6), 1996 3. M.M. Collur, D.B. Love, and B.V. Patil, “Use of Flow Modifiers to Improve Performance of a Tundish,” ISS Steelmaking Conference, April 1997 4. M.L. Lowry and Y. Sahai, Investigation of Steel Flow in a Continuous Casting Tundish with Multiple Baffles Using Mathematical Models and Tracer Studies, 1989 Steelmaking Conference Proceedings, p 71–79 5. A.M. Peel, Electromagnetic Pumps and Stirring Solutions for the Aluminum Industry, Alum. Times, Vol 7, 2005, p 32–33

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 185-193 DOI: 10.1361/asmhba0005353

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Degassing David V. Neff, Pyrotek, Inc.

GAS POROSITY is a major factor in the quality and reliability of castings, and there are various sources of gas porosity in castings. One source is the reaction between the casting mold and the molten metal. For example, the moisture in sand molds can react with the elements in the cast iron (carbon, silicon, aluminum, or iron itself) to form dissolved gases such as hydrogen or carbon monoxides (see the article “Gases in Metals” in this Volume). In addition to reactions with the mold wall, porosity also occurs from entrapped air during filling, centerline shrinkage that occurs during the final solidification, and blow holes from unvented cores (Ref 1). The other major cause of gas porosity in castings is the evolution of dissolved gases from melting, and dross or slag containing gas porosity. For example, liquid cast iron may have dissolved hydrogen and nitrogen. The solubility of these gases in the solid may be less than that in the liquid, and the gases may therefore be evolved during solidification. Whether or not this will occur depends on the amount of hydrogen and nitrogen present, the alloy being cast, chemical kinetics, and the surface tension of the alloy. Similarly, hydrogen can dissolve in liquid aluminum, but its solubility is much lower in solid aluminum and can cause porosity. For copper, the evolution of hydrogen, water vapor, or carbon monoxide could cause gas porosity. Gases can be a major problem in the casting of iron, aluminum, magnesium, and copper. Dissolved hydrogen is a particular concern because of its solubility in these metals (see the article “Gases in Metals”). The most common method for removing hydrogen from aluminum, copper, and magnesium is inert gas flushing, as described in this article. Inert gas flushing can also be used for removing hydrogen and nitrogen from cast iron (see the article “Gases in Metals”). Degassing of steels is described in the article “Steel Melt Processing” in this Volume.

Aluminum Foundry Degassing Degassing of aluminum alloy melts is a critical melt treatment technology necessary to provide readiness of the melt for casting into many

applications (Ref 1–8). Hydrogen is the only gas that is soluble in aluminum, as monatomic hydrogen. Figure 1 exhibits the hydrogen solubility of typical aluminum-silicon foundry alloys. Clearly, a substantial amount of hydrogen can be present in the liquid state, but as the melt solidifies, there is a substantial reduction in the amount of hydrogen that can exist in the solid state. How this change takes place will determine whether porosity will form. If not removed or reduced to an acceptable level determined by prior process and quality studies, the gas dissolved in molten aluminum, if not treated, can result in detrimental properties to castings. Figure 2 is a classic example of gas porosity developed during solidification in an aluminum alloy casting. Loss in mechanical properties, such as tensile strength, elongation, fatigue strength, and fracture toughness, are common results. In addition, interconnected porosity can result in “leakers” during pressure-tightness testing and blistering upon subsequent heat treatment as the pores grow and open up; this is especially a problem on the surface when surface finish/cosmetic properties are required. However, gas can arise not only from entrapped hydrogen dissolved during the melting process and evolving as gas during the liquid-solid transition, but also gas can result from aspirated air or gas evolution from molding materials (i.e., cores, binders, etc.). Thus, it is not totally clear in Fig. 2 that the porosity is due to evolution of hydrogen gas. Observable porosity may also very well be caused by lack of feeding, that is, shrinkage, during solidification. In fact, Fig. 2(a) depicts porosity that follows very well the interdendritic structure and hence is very likely shrinkage, not gas. In Fig. 2(b), the rounded porosity would suggest gas evolution during final stages of solidification and hence could very well be caused by hydrogen, However, further microscopic investigation may be necessary, since this porosity may also be caused by entrapped air or lubricant decomposition. In actual practice, it is therefore best to degas the melt to an acceptable level before casting. A typical target value, below which hydrogen-caused porosity is not normally a problem in the final casting, is that the degassing process should achieve a

level of 0.15 mL H2/100 g Al, the usual standard of measurement in a quantified test such as Alscan (Alcan Aluminum Ltd.). If one is instead using the reduced-pressure test as a measure of minimal porosity potential, it is not possible to accurately predict lack of hydrogen-caused porosity by this means alone. However, reaching a high percentage value in the reduced-pressure test related to the specific alloy theoretical density is a good measure, for instance, 2.62 in A356, whose theoretical density is 2.68 g/cm3.

Hydrogen Sources Given the propensity of aluminum to readily absorb hydrogen, there are many ways that hydrogen can and will always be present in a melt to a certain degree. Fuel-fired furnaces (natural gas, oil) have hydrogen available from fossil fuel decomposition. Water vapor is almost always present in the atmosphere, even at high temperatures directly in contact with the melting furnace. Indeed, high humidity conditions exacerbate not only hydrogen dissolution in the melt but also can retard the ability of any melt

Fig. 1

Hydrogen solubility in aluminum casting alloys. Source: Ref 2

186 / Molten Metal Processing and Handling

Fig. 3

Fig. 2

Typical porosity configurations in aluminum-silicon casting alloys. (a) Shrinkage pore found within a casting. (b) Gas pore in an Al-8% Si alloy. (c) Microporosity (gas plus shrinkage). (d) Microporosity (gas plus shrinkage). Source: Ref 2

treatment to remove hydrogen, because a degassed melt can then readily redissolve additional hydrogen. Ingot, additives, and fluxes are other common sources of hydrogen. In the case of fluxes, fluxing compounds are usually mixtures of various salts, including halides, and become very hygroscopic when exposed to air. In addition, the tools used to add and work in fluxes—rakes, spoons, and so on—will often retain flux and oxide debris after use, which will then adsorb moisture from the air as they cool down. Reuse of these tools, without intermediate cleaning, not only poses a potential safety hazard but also can reintroduce hydrogen. Scrap castings, machine turnings and borings, die cast trim press scrap, gates and risers, and sand and other nonmetallic molding material debris are additional sources of hydrogen.

Aluminum Degassing There are many means available to minimize hydrogen pickup or to degas a melt. Noting the increased solubility with temperature, as depicted in Fig. 1, merely minimizing the melting or holding times and temperatures will lessen the hydrogen absorption. Reducing the temperature will result in natural outgassing, given time, which is not always available. Therefore, additional degassing measures are often needed.

The basic principle of practical foundry degassing is that any gaseous component has the ability of collecting hydrogen gas, since all gases are insoluble in aluminum, with the exception of hydrogen, which exists as a single atom. Thus, a purge or process gas can function as in Fig. 3, depicting gas fluxing, which not only collects the hydrogen but also removes particulate matter, that is, inclusions, by flotation. The first application of this principle was mainly through the decomposition of a solid product that reacted in molten aluminum. Historically, the degassing material commonly employed was a hexachloroethane tablet (C2Cl6). When submerged and plunged into the melt, the hexachloroethane decomposed into its respective components. The chlorine component immediately forms a metastable aluminum chloride gaseous compound, which is insoluble in the melt and served as the purge gas. With any purge gas, the monatomic hydrogen diffuses into the purge gas bubble, combines with another hydrogen atom to form molecular hydrogen, and the bubble rises to the surface, where both gaseous species are released. The fluxing components in the hexachloroethane tablet, which are often halide salts, served not only to bind the tablet, but these salt components do aid in wetting of the gas bubble in the aluminum, reducing the interfacial energy

Schematic of gas fluxing in molten aluminum

and allowing easier transmittal of the hydrogen atoms into the collector aluminum chloride gas bubble. This use of hexachloroethane tablets has given way to other degas methodologies, in most instances because of environmental concerns and more efficient technologies. However, some foundries still employ this technique, especially on small crucible furnaces. The second most popular degassing technique in historical use—and still used today (2008)—is a simple flux wand or static lance. This can be a steel pipe or, more commonly, a graphite lance. A purge gas such as nitrogen or argon is injected under low pressure (less than 207 kPa, or 30 psi) into the melt. The insoluble nitrogen or argon gas bubble then rises within the melt, and the hydrogen atoms diffuse to this process gas bubble and form a hydrogen molecule. When the bubble reaches the surface of the melt, the hydrogen is released. If the relative humidity in the atmosphere is particularly high, this will be a dynamic process, because new hydrogen may become absorbed into the melt. Fundamentally, the lance degassing technique is relatively inefficient, as is seen later in this section, in relation to newer technologies. The reason for this is the relatively large gas bubble that emanates from the bottom of the lance and then agglomerates and rises quickly to the surface. The rate of rise increases with bubble size, and the bubble accelerates as it climbs toward the surface. Fig. 4(a) depicts a water model showing that a static lance only influences a small volume of fluid surrounding the lance, unless it is manually moved about. While this is possible in relatively small-diameter melting and holding vessels, it is impractical in larger ones. In Fig. 4(b), the use of either a disperser on a static lance or a rotor on a rotating shaft achieves a much finer bubble size, more dispersion, and better mixing of the gas bubbles and the liquid throughout a greater vessel volume. Visually, when performing this evaluation, one would see that the rate of rise of these smaller bubbles is measurably slower than with the larger gas bubbles in the static lance. This greater residence time increases the probability of reaction between the purge gas bubble and the hydrogen, thus increasing the efficiency. The efficiency of a degassing process is directly

Degassing / 187

Fig. 4

Water model with (a) static lance and (b) disperser or rotating shaft/rotor

Fig. 5

Fig. 6

to manipulate the gas injection into the system, and the shaft/rotor assembly itself. Equipment can be as simple as a well-mounted drive unit or more complex, with a fixed-mast drive-up ladle treatment station or a mobile unit that can be moved from furnace to furnace. Many manufacturers offer these various equipment designs, together with varying degrees of automated controls, and options to inject flux (see the article “Aluminum Fluxes and Fluxing Practice” in this Volume). For the operating foundry person, several important variables must be considered in developing the degassing process in the foundry to meet specific casting requirements. The parameters that must be integrated include:

Relative degassing efficiencies. Source: Ref 4

influenced by the bubble size, as shown in Fig. 5. A classic depiction of the fundamental difference in degassing efficiency between a static lance, a rotating injector, and a porous plug (i.e., a fine pore disperser device) is shown in Fig. 6 (Ref 4). It is clear that high hydrogenremoval efficiencies are possible with a disperser, but a rotating injector with an impeller achieves a much faster rate of reduction. Rotary Impeller Degassing. This technique has been borrowed from the chemical process industry and greatly improves mixing capability between the aluminum and the process gas. This technique was introduced to the aluminum

Degassing efficiency as a function of purge gas bubble size. Source: Ref 5

foundry industry in the mid-1980s and has become the preferred technology of choice for use in degassing wells in melting furnaces, transfer ladles, and in continuous-flow launder systems. In this technique, the purge gas is injected through a bored shaft, usually made of graphite. The incipient gas volume is then fractured into many small bubbles, that is, bubble shear, by the rotor attached to the bottom of the shaft, as the gas leaves the shaft. Many different rotor designs, sizes, and configurations are now employed throughout the industry. The basic components of a rotor degassing system include a mechanical drive unit to rotate the shaft/rotor assembly, a gas source, controls


Initial hydrogen level versus desired final

hydrogen level as determined by evaluation. This is a dynamic, ever-changing process where the initial hydrogen level is influenced by charge materials, melting practice, and relative humidity.
Vessel (well or transfer ladle) size and volume. In a continuous-flow launder process, the metal volume residence time must be considered.
Available time in a transfer ladle for melt treatment. The metal must then be delivered to the casting station(s) without excessive temperature loss, considering the relationship between rotor type and revolutions per

188 / Molten Metal Processing and Handling minute (rpm), gas volume being injected, avoidance of surface effects (vortexing, splashing), and time necessary and available to achieve the desired degassing results. The interplay among these variables must be determined on a case-by-case basis by the individual foundry and often on the specific casting(s) to achieve the optimum combination of process and equipment parameters. In general, the optimum result desired is to achieve the necessary degassing specification in as short a time as possible, at the lowest cost, and without excessive turbulence. As a general rule, most well-developed foundry degassing practices should achieve the desired degassing level in 5 min when treating a vessel with 455 kg (1000 lb) volume and when optimizing the foregoing parameters of rotor type, rpm, and purge gas flow rate. Selecting a Process Gas. Since all gases except hydrogen are insoluble in molten aluminum, there are theoretically many choices of which gas to use. Nitrogen is the most commonly employed and is the least expensive. Nitrogen creates a wet dross; that is, the dross that is formed is high in metallic aluminum content. Furthermore, since the density of nitrogen is comparable to air overall, there is a tendency to pick up moisture in nitrogen lines as fittings deteriorate and humid air is aspirated into the delivery lines. For this reason, all nitrogen sources should include an in-line dryer. Nitrogen purity should be at least 99.99% for any critical casting applications. Argon, while significantly more expensive than nitrogen, produces a less metal-rich dross. Argon is more inert, and, being heavier, provides a protective cover over the melt during degassing, precluding not only hydrogen absorption but further oxidation as well. It is also easier to keep the argon supply drier and cleaner as hoses and fittings deteriorate. While argon is the preferred process gas for in-line degassing treatment in the production of wrought alloys, it is not extensively used in foundry casting because of the higher cost. Active Halogens. Historically, a small percentage of chlorine (5 to 10%) was used with an inert gas when degassing with a static lance. The chlorine enhanced the reaction between the melt and the hydrogen atoms because it lowered the interfacial tension at the purge gas bubble surface (through the formation of aluminum chloride, a metastable gaseous phase) and facilitated the diffusion of the hydrogen atom into the purge gas bubble. Several commercial mixtures of halogen gases—Freon (CCl2F2) (Dupont) being the most popular—were employed in the past. With the growing concern and difficulties with the environmental impact of chlorine use, sulfur hexafluoride (SF6) gained prominence 20 years ago as a replacement for chlorine. However, subsequent experience showed that SF6 also posed environmental problems with ozone depletion, and the gas is

very corrosive, even in small amounts, to foundry metal structures. Fortunately, with the advent of rotor degassing and the resultant better kinetics of reaction and mixing, it has been found that chlorine does not enhance the degassing rate or efficiency when the purge/process gas bubbles are sufficiently small, as they are with rotor impeller degassing. Rather, when chlorine is used, there is better wetting of the nonmetallic inclusions during the degassing process, so that metal cleanliness is improved, and the burden on any subsequent particulate removal process is lessened, which in turn creates a better-quality cast product. Other Considerations for the Degassing Process. There are other issues that must be taken into account during the degassing process. Temperature. It is well understood that hydrogen absorption in the melt increases with increasing temperature. Conversely, hydrogen solubility and partial pressure of hydrogen gas in the melt decrease rapidly as the melt temperature is decreased. It follows that minimizing the temperature will result in lower hydrogen value. Given sufficient time, allowing the temperature to drop will result in natural outgassing. However, the injection of a purge gas at normal standard temperature and pressure creates an endothermic reaction, and heat will therefore be lost at an accelerated rate. A general ruleof-thumb is that degassing a metal volume of 455 kg (1000 lb) in an unheated refractory vessel for 5 to 10 min will result in a temperature drop of nearly 28  C (50  F). Whatever the actual magnitude of the temperature drop due to gas injection/treatment time, this reduction must be accommodated in relation to the necessary temperature that must be maintained as the treated melt is transferred into the mold or into the casting station(s). A degassing process should nevertheless be carried out at the lowest practical temperature possible. Grain Refining. Most aluminum melts in sand, permanent mold, and low-pressure foundry production require grain refinement to achieve the best mechanical properties in the finished casting. In most instances, grain refiner is added during the degassing process, because the increased mixing and agitation facilitates the dissolution and reaction of the grain refiner itself and provides more uniform dispersion throughout the melt or treated volume. Modification. Historically, the element sodium was used to provide modification of the eutectic silicon component in aluminum casting alloys—a necessary practice, together with grain refining, to produce desired mechanical properties. In current practice, sodium has given way to a strontium addition for many reasons, including safety in handling and in the environment, better consistency, and reduced fade, which is the loss of modification with time. Because both sodium and strontium are prone to oxidation (although strontium vapor pressure is substantially lower than sodium),

and the degassing mixing creates more convection during treatment, degassing should take place before modification, so as not to lose the benefits of the dissolved modifier due to oxidation. If chlorine is used, even in small quantities, in the degassing process, both sodium and strontium will react chemically to form solid chloride reaction products. Dross Formation and Skimming. The action of the purge gas injection process by whatever means—lance, rotor, or porous plug/disperser—will result in flotation of debris and some degree of particulate inclusion content within the melt. Naturally, the degassing process creates dross. This must be thoroughly skimmed off the surface to the greatest practical extent possible to avoid transfer into the casting mold or casting station. Determining Hydrogen Content. To control hydrogen content, one must be able to measure the hydrogen content in the melt and the subsequent result of any hydrogen-removal treatment. Laboratory evaluations are possible using vacuum fusion (Leco technique) and subvacuum fusion on previously solidified test specimens. Obviously, laboratory measurements cannot be used to make real-time decisions but rather are useful in process development. Real-time determinations are limited to the simple reduced-pressure test (RPT) and its variants, and in situ techniques, such as Alscan, continuous hydrogen analysis by pressure evaluation, and others. The RPTs are the simplest, but these tests are a measure of molten metal cleanliness and not absolute hydrogen content per se. The presence of inclusions in the melt during an RPT evaluation will create exaggerated porosity nucleated during solidification and a resultant lower density value. The Alscan test and others of this type do generate a real-time result through the principles of recirculating gas of a known standard hydrogen content and the thermal conductivity of dilute solutions.

Degassing of Magnesium Magnesium behaves similarly to aluminum in its ability to dissolve hydrogen easily in the molten state (Fig. 7). It too will suffer from porosity when cast if the hydrogen is not removed or naturally expelled during solidification. Aluminum and zinc, which are typical alloying elements in magnesium alloys, lower this solubility somewhat. Figure 8 indicates that much more gas is evolved from magnesium alloys containing zinc and aluminum than from pure magnesium upon solidification. Gas porosity occurs in solidified castings when the residual hydrogen exceeds 15 mL/100 g. Zirconium is also a major constituent of certain magnesium alloys and has a marked effect on hydrogen solubility. Hydrogen solubility decreases as the zirconium content of the alloy increases (Ref 9). Because of this low solubility even in the liquid state, hydrogen porosity is minimal in castings of magnesium-zirconium

Degassing / 189 alloys. However, excess hydrogen in the melt readily precipitates as zirconium hydride, thus depleting the alloy of zirconium in solid solution. It is therefore advisable to minimize hydrogen absorption into the melt and/or to employ degassing methods even with this alloy. Sources of gas in magnesium alloy melts are the same as those for aluminum—atmospheric conditions, fluxes, tools, and so on. Flux covers or antiburning agents do not necessarily preclude hydrogen absorption during melting. Like aluminum alloys, molten magnesium metal and alloys can be degassed with hexachloroethane, chlorine, or nitrogen. Hexachloroethane has the added advantage of supplying carbon through the decomposition of the hexachloroethane, with the carbon serving as a grain refiner (see the article “Magnesium and Magnesium Alloy Castings” in this Volume). Chlorine

or chlorine/nitrogen gas purging is optimally performed at temperatures of 725 to 750  C (1335 to 1380  F). In this range, liquid MgCl2 will also be formed, which will retard burning on the surface. Liquid MgCl2 wets the oxides and the magnesium melt surface, facilitating hydrogen escape. Pure chlorine provides for more liquid MgCl2 formation than either gas mixtures or hexachloroethane and therefore creates better degassing. Fluxes higher in MgCl2 that are used as covers during degassing also aid in hydrogen desorption at the flux/metal interface and in subsequent expulsion to the atmosphere. The degassing of magnesium alloys in crucible or pot-type furnaces should be carried out with clean, dry lances, preferably graphite, with thorough mixing for 5 to 10 min. Pressures and volumes should not exceed those required to achieve a mild, gentle roll or small bubble (25 mm, or 1 in., in diameter) on the melt surface. Rare-earth-containing magnesium alloys should only be degassed when required with nitrogen or argon, because chlorination will readily remove these expensive alloying elements.

Degassing of Copper Alloys In the melting and casting of many copper alloys, hydrogen gas absorption can occur because of the generous solubility of hydrogen in the liquid state of these alloys. The solid solubility of hydrogen is much lower; therefore, the gas must be rejected appropriately before or during the casting and solidification process to avoid the formation of gas porosity and related defects (excessive shrink, pinholing, blow holes, and blistering). These defects are almost always detrimental to mechanical and physical properties, performance, and appearance. Different copper

alloys and alloy systems have varying tendencies toward gas absorption and subsequent problems. Other gases, particularly oxygen, can cause similar problems (see the article “Copper and Copper Alloy Castings” in this Volume).

Gases in Copper The type and amount of gas absorbed by a copper alloy melt and retained in a casting depend on a number of conditions, such as melt temperature, raw materials, atmosphere, pouring conditions, and mold materials. Table 1 lists the gases that can be found in a number of copper alloy systems. Hydrogen is the most obvious gas to be considered in copper alloys. Figure 9 (Ref 11) shows the solubility of hydrogen in molten copper, tin, and copper-tin alloys. As in aluminum and magnesium alloys, solubility is reduced in the solid state; therefore, the hydrogen must be removed prior to casting or rejected in a controlled manner during solidification. Alloying elements have varying effects on hydrogen solubility (Fig. 10). There are many potential sources of hydrogen when melting copper alloys. For the most part, these are identical to the sources of hydrogen in aluminum melting. Oxygen also presents a potential problem in most copper alloys, but oxygen alone does not cause problems in copper alloys. In the absence of hydrogen, oxygen alone may not cause problems, because it has limited solubility in the melt. However, it forms a completely miscible liquid phase with the copper in the form of cuprous oxide (Fig. 11). During solidification, the combination of cuprous oxide and hydrogen can give rise to casting porosity resulting from the steam reaction (discussed later). Sulfur gases have significance in primary copper through the smelting of sulfide ores

Fig. 7

Solubility of hydrogen in molten magnesium at 101 kPa (760 torr). Source: Ref 9

Fig. 8

Effect of alloying elements on hydrogen solubility in liquid magnesium. AC, as-cast; HT, heat treated. Source: Ref 9

190 / Molten Metal Processing and Handling Table 1 Summary of gases found in copper alloys Alloy family

Pure copper Copper-tin-leadzinc alloys Aluminum bronzes Silicon brasses and bronzes Copper-nickels

Gases present

Water vapor, hydrogen Water vapor, hydrogen, CO Water vapor, hydrogen, CO Water vapor, hydrogen Water vapor, hydrogen, CO

Remarks

Approximate hydrogen/water vapor ratio of 1. Higher purity increases the amount of water vapor and lowers hydrogen. Lead does not affect the gases present. Higher tin lowers total gas content. Increased zinc increases the amount of hydrogen, with a loss in water vapor. The presence of 5 wt% Ni in alloy C95800 causes CO to occur rather than water vapor. Lower aluminum leads to higher total gas contents. Approximate hydrogen/water vapor ratio of 0.5. Increased zinc decreases hydrogen and increases water vapor. All three gases are present up to 4 wt% Ni, after which only CO and hydrogen are present. Hydrogen increases with increasing nickel up to 10 wt% Ni but is decreased at 30 wt% Ni.

Source: Ref 10

Fig. 11

Fig. 9

Solubility of hydrogen at 1 atm (760 mm Hg) in pure copper, pure tin, and copper-tin alloys. Source: Ref 11

and in the remelting of mill product scrap containing sulfur-bearing lubricants. Sulfur dioxide is the most probable gaseous product. Foundry alloys and foundry processing usually do not experience sulfur-related problems unless high-sulfur fossil fuels are used for melting. Carbon can be a problem, especially with the nickel-bearing alloys. The nickel alloys have extensive solubility for both carbon and hydrogen. Carbon can be deliberately added to the melt, along with an oxygen-bearing material such as nickel oxide. The two components will react to produce a carbon boil, that is, the formation of CO bubbles, which collect hydrogen as they rise through the melt (Ref 13). If not fully removed, however, the residual CO can create gas porosity during solidification. Nitrogen does not appear to be detrimental or to have much solubility in most copper alloys. However, there is some evidence to suggest that nitrogen porosity can be a problem in cast copper-nickel alloys (Ref 14).

Fig. 10

Effect of alloying elements on the solubility of hydrogen in copper. Source: Ref 12

Water vapor can exist as a discrete gaseous entity in copper alloys (Ref 10, 15, 16). In many copper alloys, the most important gas causing porosity is water vapor (Ref 11). Water itself is not dissolved, but dissolved atomic oxygen and hydrogen in the melt combine to form water until the oxygen becomes depleted; hydrogen is produced as a separate species. Water vapor is evolved from solidifying copper alloys, which always have some residual dissolved oxygen. Thus, gas porosity can occur even with low hydrogen content when oxygen is present. Gas porosity occurs when the gas pressure exceeds 1 atm (760 mm Hg). To prevent such porosity, the solid solubility limits must not be exceeded, and the sum of individual gas partial pressures must be less than 1 atm (760 mm Hg) through the solidification process (Ref 11). Testing for Gases. There are essentially three ways to determine the presence of gas in a copper alloy melt. The easiest and simplest is a chill test on a fracture specimen. In this

Copper-oxygen phase diagram. Source: Ref 12

method, a standard test bar is poured and allowed to solidify. The appearance of the fractured test bar is then related to metal quality standards (Ref 17). The second method is an RPT similar to the Straube-Pfeiffer test used for aluminum alloys. In recent years, considerable development and definition have been given to this test as adapted for copper alloys, because the different classes of copper alloys have different solidification characteristics that affect the response to the test (Ref 10, 18–20). The objective of the test is to use a reduced pressure only slightly less than the dissolved gas pressure so that the surface of the sample mushrooms but does not fracture (Fig. 12). The test apparatus is shown schematically in Fig. 13. The key to establishing controllable results, that is, unfractured but mushroomed test sample surfaces, is to provide wave-front freezing. This is accomplished by proper selection of vessel materials to achieve the correct solidification characteristics. Because the freezing ranges of copper alloys vary from short to wide, different materials must be used. The various material combinations that have proved successful are described in Ref 10.

Degassing Methods There are several kinds of treatments which may be employed for the improvement of melt quality in copper alloys, such as oxidationdeoxidation practice, zinc flaring, inert gas fluxing, and the use of solid degassing fluxes. Metallurgical principles for common copper alloy classes are discussed in detail in Ref. 21. Oxidation-Deoxidation Practice. The steam reaction previously mentioned is a result of both hydrogen and cuprous oxide being present in the melt. These constituents react to form steam, resulting in blow holes during solidification as the copper cools: 2H þ Cu2 O ! H2 OðsteamÞ þ 2Cu The proper deoxidation of a copper alloy melt will generally prevent this steam reaction,

Degassing / 191

Fig. 12

Effect of pressure on the appearance of copper alloy reduced-pressure test samples containing the same amount of gas. (a) Pressure of 7 kPa (55 torr) results in surface shrinkage. (b) At 6.5 kPa (50 torr), a single bubble forms. (c) Boiling and porosity occur at 6 kPa (45 torr).

Fig. 13

Fig. 14

Schematic of the reduced-pressure test apparatus used to assess amounts of dissolved gas in copper alloys

Hydrogen/oxygen equilibrium in molten copper. Source: Ref 10

Fig. 15

Influence of zinc content on boiling point or vapor pressure in copper alloys. Source: Ref 12

although excessive hydrogen alone can still cause gas porosity if it is not expelled from the casting before the skin is completely solidified. Fortunately, there is a mutual relationship between hydrogen and oxygen solubility in molten copper (Fig. 14). Steam is formed above the line denoting equilibrium concentration, but not below. Consequently, as the oxygen content is raised, the capacity for hydrogen absorption decreases. Therefore, it is useful to provide excess oxygen during melting to preclude hydrogen entry and then to remove the oxygen by a deoxidation process to prevent further steam reaction during solidification. This relationship gave rise to the Pell-Walpole oxidation-deoxidation practice for limiting hydrogen, especially when fossil-fuel-fired furnaces were used for melting. The melt was deliberately oxidized using oxygen-bearing granular fluxes or briquetted tablets to preclude hydrogen absorption during melting. The melt was subsequently deoxidized to eliminate any steam reaction with additional hydrogen absorbed during pouring and casting. Zinc Flaring. The term zinc flare applies to those alloys containing at least 20% Zn. At this level or above, the boiling or vaporization point of copper-zinc alloys is close to the usual pouring temperature, as shown in Fig. 15 (Ref 12). Zinc has a very high vapor pressure, which precludes hydrogen entry into the melt. Zinc may also act as a vapor purge, removing hydrogen already in solution. Furthermore, the oxide of zinc is less dense than that of copper, thus providing a more tenacious, cohesive, and protective oxide skin on the melt surface and increasing resistance to hydrogen diffusion. In zinc flaring, the melt temperature is deliberately raised to permit greater zinc vapor formation. Because some zinc loss may occur, zinc may have to be replenished to maintain the correct composition. Inert Gas Fluxing. With gas fluxing, an inert collector or sparger gas such as argon or nitrogen is injected into the melt with a graphite fluxing tube. The bubbling action collects the hydrogen gas that diffuses to the bubble surface, and the hydrogen is removed as the inert gas bubbles rise to the melt surface. The reaction efficiency depends on the gas volume, the depth to which the fluxing tube or lance is plunged, and the size of the collector gas bubble generated. Again, finer bubble sizes have higher surface-area-to-volume ratios and therefore provide better reaction efficiencies. Figure 16 depicts the amount of purge gas necessary to degas a 450 kg (1000 lb) copper melt. Figure 17 shows the response of an aluminum bronze alloy to nitrogen gas purging. The curve for oxygen purging is also shown. Solid Degassing Fluxes. Other materials can be used to provide inert gas purging. A tabletted granular flux such as calcium carbonate (CaCO3), which liberates CO2 as the collector gas upon heating, has been successful in degassing a wide variety of copper alloys. This type of degassing is simpler than nitrogen but must

192 / Molten Metal Processing and Handling

Fig. 16

Amount of purge gas required to degas a 450 kg (1000 lb) copper melt. Source: Ref 20

Fig. 17

Fig. 18

Comparison of the effectiveness of solid degassing flux versus nitrogen purging. Source: Ref 21

be kept dry and plunged as deep as possible into the melt with a clean, dry plunging rod, preferably made of graphite. This type of degasser may also have the advantage of forming the collector gas by chemical reaction in situ. This results in an inherently smaller initial bubble size than injected gas purging for better reaction efficiency. Such an advantage can be inferred from Fig. 18, although after 5 min of degassing, the end results are the same. Other solid degassers in the form of refractory metals or intermetallic compounds, such as calcium-manganese-silicon, nickel-titanium, titanium, and lithium, are effective in eliminating porosity due to nitrogen or hydrogen by their ability to form stable nitrides and hydrides. Again, maintaining dry ingredients, deep plunging, and stirring or mixing will enhance their effectiveness. For best results and optimum reaction efficiency, such degassers should be wrapped or encapsulated in copper materials to provide controlled melting when plunged. Alternatively, copper master alloys containing these elements are available.

Effect of nitrogen flow rate during purging on the residual gas pressure remaining in an aluminum bronze melt. The cure for oxygen purging is also shown. Source: Ref 19, 22

For both inert gas fluxing and solid degassing additions, the sparging gas reaction should not be so violent as to splash metal and create an opportunity for gas reabsorption. Furthermore, a melt can be overdegassed; an optimum amount of residual gas remaining in the melt helps to counter localized shrinkage in longfreezing-range alloys such as leaded tin bronzes. Vacuum degassing is not generally applied to copper alloys, although it can be very effective. However, the cost of the equipment necessary is relatively high, and there may be a substantial loss of more volatile elements having a high vapor pressure, such as zinc and lead. Furthermore, significant superheat (up to 150  C, or 300  F) may be required to accommodate the temperature drop during the degassing treatment, further aggravating the vapor losses. Auxiliary Degassing Methods. There are other ways to degas a copper alloy melt than using a specific treatment. The technique of lowering the melt superheat temperature, if possible, and holding the melt (in a dry, minimal gas environment) provides outgassing simply by lowering the equilibrium liquid-state solubility of the gas. During casting, the use of chills to provide directional solidification, particularly for the long-freezing-range alloys such as the red brasses and tin bronzes, results in less tendency for gas porosity. In the mold-metal reaction, previously degassed and deoxidized metal containing excess phosphorus or lithium deoxidants can react with green sand containing moisture. Hydrogen is liberated and absorbed as the metal comes into contact with the sand. This can be

minimized by correct deoxidant additions. In addition, finer facing sands or mold coatings (inert or reactive, such as sodium silicate or magnesia) can be used; these confine any reaction to the mold/metal interface area and retard hydrogen penetration into the solidifying skin of the cast metal.

REFERENCES 1. R. Monroe, Porosity in Castings, Trans. AFS, Vol 113, 2005, p 519–546 2. J.E. Gruzleski and B.M. Closset, The Treatment of Liquid Aluminum-Silicon Alloys, AFS, 1990 3. D. Apelian, Clean Metal Processing of Aluminum, Proceedings, Materials Solutions Conference ’98 on Aluminum Casting Technology, ASM International, 1998 4. G. Sigworth and T. Engh, Chemical and Kinetic Factors Related to Hydrogen Removal from Aluminum, Metall. Trans., Vol 138, 1982, p 447 5. G. Sigworth, A Scientific Basis for Degassing Aluminum, Trans. AFS, Vol 95, 1987, p 73 6. D.V. Neff, Understanding Aluminum Degassing, Mod. Cast., May 2002 7. D. Apelian and M.M. Makhlouf, Ed., High Integrity Aluminum Die Casting, NADCA, April 2004 8. M.M. Makhlouf, L. Wang, and D. Apelian, Measurement and Removal of Hydrogen in Aluminum Alloys, AFS Special Report, 1998 9. E.F. Emley, Principles of Magnesium Technology, Pergamon Press, 1966

Degassing / 193 10. P.K. Trojan, T.R. Ostrom, and R.A. Flinn, Melt Control Variables in Copper Base Alloys, Trans. AFS, 1982, p 729; “High Conductivity Copper Alloys,” Bulletin 43, Foseco, Inc., 1972 11. Casting Copper-Base Alloys, 2nd ed., American Foundry Society, 2007, p 128–131 12. R.J. Kissling and J.F. Wallace, Gases in Copper Base Alloys, Foundry, 1962, 1963 13. M.P. Renatv, C.M. Andres, G.P. Douglas, and R.A. Flinn, Solubility Relations in Liquid Copper-Nickel-Carbon-Oxygen Alloys, Trans. AFS, 1976, p 641

14. B.N. Ames and N.A. Kahn, Gas Absorption and Degasification of Cast Monel, Trans. AFS, 1947, p 558 15. T.R. Ostrom, P.K. Trojan, and R.A. Flinn, Gas Evolution from Copper Alloys—A Summary, Trans. AFS, 1981, p 731 16. T.R. Ostrom, P.K. Trojan, and R.A. Flinn, The Effects of Tin, Aluminum, Nickel, and Iron on Dissolved Gases in Molten Copper Alloys, Trans. AFS, 1980, p 437 17. Casting Copper Base Alloys, American Foundrymen’s Society, 1984 18. T.R. Ostrom, R.A. Flinn, and P.K. Trojan, Gas Content of Copper Alloy Melts—Test

19. 20. 21. 22.

Equipment and Field Test Results, Trans. AFS, 1977, p 357 T.R. Ostrom, P.K. Trojan, and R.A. Flinn, Dissolved Gases in Commercial Copper Base Alloys, Trans. AFS, 1975, p 485 R.J. Cooksey and R.W. Ruddle, New Techniques in Degassing Copper Alloys, Trans. AFS, Vol 68, 1960 D.V. Neff, Fluxing, Degassing, Deoxidation of Copper Base Alloys, Trans. AFS, 1989, p 439 P.K. Trojan, S. Suga, and R.A. Flinn, Influence of Gas Porosity on Mechanical Properties of Aluminum Bronze, Trans. AFS, 1973, p 552

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 194-205 DOI: 10.1361/asmhba0005351

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Molten-Metal Filtration D.V. Neff, Pyrotek Inc. Robert P. Pischel, FOSECO Metallurgical Inc.

FILTRATION has become a primary means of removing solid impurities from nonferrous molten metal. Aluminum alloys, especially with their tendency toward oxidation, often contain impurities (inclusions) that are deleterious to physical, mechanical, electrical, and aesthetic properties. In the handling of molten aluminum, it is fairly common to use filters as a part of the melting unit (in-furnace filtration) and in the gating and/or riser system (in-mold filtration). Zinc and copper alloy melts also are filtered but to a much lesser extent than aluminum. In almost all instances of metal casting, the presence of dross, inclusions, or particulate impurities detracts from casting performance and can be a major contributor to scrap castings and lost productivity and profitability for the foundry. For example, the presence of dross, inclusions, or particulate impurities detracts from the natural fluidity of any melt (Fig. 1) (Ref 1). Mechanical properties of the cast product can also be degraded, particularly in elongation/ductility and fatigue strength, which are most sensitive to the presence of inclusions. Inclusions also are harder than the surrounding matrix and give rise to tool chatter, tool wear, and ultimately, tool breakage during machining operations. In addition, inclusions may result in poor continuity, such as creating leakers in

pressuretight castings by nucleating porosity formation during solidification or by causing obvious discontinuities and blemishes in cosmetic surfacefinishing operations such as powder coating, plating, painting, polishing, and anodizing. Thus, inclusion-related defects should be minimized to the greatest extent possible to prevent scrap castings, lost productivity, and lower profitability. This article describes the methods of in-furnace and in-mold filtration, with emphasis on the filtration of molten aluminum. Moltenmetal filtration is an ongoing area of interest in aluminum foundry practice (Ref 2–6). For example, the following table comprises typical quality improvements in several aluminum foundries using in-furnace filtration: Foundry

Al alloy

Result from in-furnace filtering

380 319 380 319 380 518

Hardspots eliminated Leakers reduced 75% Tool life extended 600% Scrap reduced 95% Tool breakage reduced 95% Scrap eliminated

A B C D E F

Sources of Inclusions Undissolved particulate contaminants (inclusions) include slag (ferrous), dross (nonferrous), nonmetallic impurities (oxides, sulfides, nitrides), or insoluble intermetallic precipitates (see the article “Inclusion-Forming Reactions” in this Volume). Inclusions arise from many sources. Charge materials can contain nonmetallic inclusions from unfiltered primary or secondary ingot. Scrap materials may contain molding sand, debris from back-charged gates and runners, metalworking lubricants, paint and coatings, plastics, or unsorted and unwanted heavy-metal contaminant oxides (iron, manganese, chromium, and so on). Furnace tools, such as skimmers, puddlers, and rakes, can also add solidified oxide skins back into the melt if they are not cleaned before use. Factors that influence the formation of inclusions include:  Excessive

Fig. 1

Effect of metal cleanliness on fluidity of an aluminum die casting alloy. Source: Ref 1

temperatures during melting accelerate the oxidation of aluminum and alloying elements. Where fluxes are used,

incomplete skimming or separation of fluxing salts often results in contamination and salt-wetted oxide inclusions.  Grain-refiner additions can be a source of inclusions.  Intermetallic compound precipitation may also contribute to inclusions. Iron, chromium, and manganese form intermetallics with aluminum, and the formation of intermetallic sludge is a common problem in die casting alloys.  Refractory inclusions from the erosion and wear of melting and holding vessels may add alumina, spinel, silica, and silicon carbide inclusions to nonferrous alloy melts. Aluminum Inclusions. Inclusions in molten aluminum foundry processing arise from several sources. These inclusions exist as oxides present as skins on charge materials; from oxidation of the melt itself; from erosion of refractory vessels used for melting, transfer, and holding of molten aluminum; from additives such as master alloys; and grain refiners and modifiers as well. The various types of solidphase inclusions in molten aluminum alloys include:       

Oxides (Al2O3, MgO) Spinels (Mg2AlO4) Borides (TiB2, VB2, ZrB2) Carbides (Al3C4, TiC) Intermetallics (MnAl3, FeAl3) Nitrides (AlN) Exogenous refractory inclusions (oxides and/ or carbides of iron, silicon, aluminum, etc.)

The size and shape of the inclusions vary from globular (oxides and borides) to stringerlike (oxides and spinels). The latter are particularly insidious in their effect on castability and the properties of the casting. Aluminum oxide is present in virtually all aspects of aluminum alloy casting production, because aluminum has a high oxidation potential. Prime metal in either liquid or solid form, alloy additions, and so on will always have an aluminum oxide skin. Melting in nearly all processes is done in air, thus providing another source for aspirated air to enter, especially in closed or even

Molten-Metal Filtration / 195 flue-operated larger furnaces that are not maintained under positive pressure. Foundry returns— an essential part of the foundry economic charge mix—also contain not only aluminum oxide skin but also can contribute sand debris, machining lubricant residues, die casting release agents, plunger lube, and other fluids that may contain some level of solids content. In addition, alloy and metallic impurity components in excess of the solubility limit form intermetallic compounds that do not dissolve during processing and hence can become inclusions in the solidified metal. In high-pressure die casting alloys and most aluminum-silicon shape casting alloys, an intermetallic compound known as sludge can form with sufficient constituent concentration (iron, manganese, chromium) and low temperature to form such an intermetallic. Additionally, segregated heavy metals (zinc, copper, iron, manganese) can also form a segregated agglomerate in the melting furnace, which has the potential to be transferred out and into a casting. During fluxing operations in a melting furnace or transfer ladle, residual flux inclusions may not become fully skimmed or removed from the surface nor have time to either float out or sediment. If these flux inclusions do become entrapped within the casting, they create sources for galvanic corrosion to take place. Typical solid-phase inclusions that may be found in aluminum casting alloys are listed in Table 1. Oxide-based inclusions are varied and will exhibit a spectrum of densities—some with specific gravity lower than the melt (nominally 2.4 g/cm3, or 0.087 1b/in.3) and some higher than the melt. Given sufficient time, separation of the solid particulate from the liquid is possible to a degree, based on the density difference. Thus, lighter particles could float to the surface; heavier particles will sink or sediment to the bottom of a melting or casting vessel or furnace. As Fig. 2 demonstrates, however, this phenomenon is time-dependent as well as size-dependent. In casting alloys, a discrete particulate soon becomes an agglomerate. Surface tension/interfacial energies usually dictate that complexes of inclusions have a bulk density nearer that of the melt, and hence, separation purely by gravity means becomes retarded. Liquid-phase inclusions can occur in molten aluminum alloys. The most prevalent is MgCl2 arising from the chlorination of magnesium-containing melts. These inclusions are not generally removed to a great extent by conventional filtration techniques and therefore must be avoided. Preventing excessive and inefficient chlorination of the melts during degassing and thorough skimming at least help to avoid such inclusions. The nature of these and other inclusions in aluminum alloys is discussed in more detail in Ref 7 and 9 to 12. Inclusions in Magnesium Alloys. The inclusions found in magnesium melts are similar in many cases to those found in aluminum—oxides, spinels, furnace refractory particles, and so on. In addition, alloying elements (zirconium and the rare earths) lead to oxides of these elements

as well as intermetallic compound precipitates. If the fluxing process is used to clean magnesium, fluxing salt inclusions may also be present. Inclusions in Copper Alloys. In addition to the usual inclusions arising from oxides, fluxing salts, and intermetallics, copper oxide inclusions and phosphorus pentoxide (from deoxidation) may be present in copper alloys if the melt is not allowed to settle or if it is inadequately skimmed before pouring and casting. Inclusions in Zinc Alloys. In the zinc foundry and die casting alloys containing aluminum, nonmetallic oxide inclusions are

usually less important than intermetallic inclusions. Iron-zinc and iron-aluminum intermetallic compounds may also be present, especially above 480  C (900  F) in melting in iron-base vessels. These particles may be small crystals less than 5 mm (200 min.) in size or may agglomerate into millimeter-sized aggregates. Other heavy metals present, such as lead, cadmium, nickel, manganese, and chromium, can also lead to intermetallic compound precipitation (for example, Mn2Al5, Ni2Al3, and CrAl4). These compounds (often called aluminides) can form when the concentration of impurity exceeds

Table 1 Common types of inclusions in aluminum foundry melts Inclusion type

Chemical formula

Density Form

g/cm3

Size range lb/in.3

mm

min.

8–1200 400–2  105 4–200 400–2  105 4–200 400–2  105 20–200

Oxides Alumina

Al2O3

Particles, skins

3.97

0.143

Magnesia

MgO

Particles, skins

3.58

0.129

Spinel

MgAl2O4

Particles, skins

3.60

0.130

Silica

SiO2

Particles

2.66

0.096

0.2–30 (particles) 10–5000 (skins) 0.1–5 (particles) 10–5000 (skins) 0.1–5 (particles) 10–5000 (skins) 0.5–5

Salts Chlorides Fluorides

Varies Varies

Particles Particles

1.98–2.16 1.98–2.16

0.072–0.078 0.072–0.078

0.1–5 0.1–5

4–200 4–200

Carbides Aluminum carbide Silicon carbide

Al4C3 SiC

Particles Particles

2.36 3.22

0.085 0.116

0.5–25 0.5–25

20–1000 20–1000

Nitrides Aluminum nitride

AlN

Particles, skins

3.26

0.117

10–50

400–2000

Borides Titanium boride

TiB2

4.50

0.163

1–30

40–1200

Aluminum boride

AlB2

Agglomerated particles Particles

3.19

0.115

Sludge Various

Al(FeMnCr)Si

Particles

>4.0

>0.145

Source: Ref 7, 8

Fig. 2

Sedimentation/flotation of inclusions in aluminum melts

0.1–3

4–120 ...

...

196 / Molten Metal Processing and Handling the liquid solubility level of the element, which is often quite low (<0.02% for manganese, silicon, chromium, and nickel in die casting alloys). The intermetallic particles can be lighter than the base alloy and therefore rise to the top dross. Fluxing salt inclusions and exogenous refractory particles can also be present, as with the other nonferrous alloy systems. Oxide inclusions in zinc are somewhat limited because of the high positive vapor pressure of zinc. However, alloys containing aluminum can be subject to greater oxidation of the aluminum constituent. Oxide inclusions can also result during melting or casting when excessive turbulence is encountered; thus, mechanical stirring or molten metal pumping should be controlled. Other nonmetallic inclusions arise from refractory vessel erosion and include silica, alumina, and silicon carbide. There is an appreciable density difference between these oxides and the base melt, unlike in aluminum; for this reason, gravity separation and thorough removal are substantially easier.

Removal of Inclusions There are three basic methods of mechanically removing or separating inclusions from molten metal:  Sedimentation  Flotation  Positive filtration

Fig. 3

Two modes of positive filtration. (a) Depth filtration. (b) Cake filtration

The first two processes—sedimentation and flotation—are based on density differences between the liquid and solid phases. Actual separation capability is based not only on the density differences but also particle size. For example, rotary degassing is a method that removes inclusions by flotation. It is usually effective in removing inclusions larger than 40 mm (1600 min.) in size. Figure 2 is an application of Stokes’s law governing phase separation as applied to inclusions in molten aluminum. The greater the density difference from molten aluminum and the larger the particle size, the faster the rate of rise (flotation to the melt surface) or fall (sedimentation to the bottom of the furnace or holding vessel). Thus, in many metal casting operations, a holding or quiescent period within a furnace will result in much of the inclusion content being able to be removed by skimming the surface, or avoidance of the sedimented inclusions by careful pouring so as not to entrain the inclusions within the pouring stream or into the transfer ladle or casting mold. However, two factors combine to render such gravity-related processes less than efficient—at least for light nonferrous metals such as aluminum. First, there is a high surface tension/interfacial surface energy between most oxide inclusions in aluminum and the melt, which makes actual separation difficult. Secondly, many oxide inclusions, even small ones, can be detrimental to casting properties if concentrated into the casting during solidification. It can be seen that small-sized inclusions have very slow rates of separation. Additionally, many inclusions and/or inclusion complexes will have bulk

Fig. 4

densities that are very similar to the density of the melt itself and thus will also not separate easily or quickly. Consequently, positive filtration will need to be employed to remove these. Positive filtration involves the passing of molten metal through a porous device (a filter) in which the inclusions contained in the flowing metal are trapped or captured by one or more filtration mechanisms. With any filter, there are two mechanisms of positive filtration (Fig. 3):  Cake filtration  Depth-mode (or deep-bed) filtration

Cake filtration occurs when there is a buildup of particulate (filtrate) on the inlet surface of a filter. Particles that are larger sized than the inlet openings on a filter will naturally accumulate at the surface to form a cake. In most instances, however, inclusion particles are considerably smaller than the filter pore openings, but the filter body itself connecting the pores serves to slow the molten-metal flow through the filter, such that the particles begin to adhere to the surface. Fortunately, even though these inclusions begin to build up, the accumulation remains quite lacy and permeable to continued metal flow, up to an extent. Depth-mode filtration occurs within the body of a filter. In this mode, inclusion particles flowing with the liquid are subject to changes in velocity and direction created by the internal complexity or tortuosity of the flow path within the filter. The inclusions become captured or adhere to the internal surfaces of the filter by several possible mechanisms (Fig. 4) of diffusion (important for inclusions less than 1 mm, or

Transport and capture mechanisms of depth filtration. (a) Diffusion. (b) Direct interception. (c) Inertia or sedimentation. (d) Fluid dynamics effects on particulate transport

Molten-Metal Filtration / 197 40 min.), direct interception, sedimentation, and fluid dynamics. Filtration is a matter of probability, and so, the greater the flow path through the filter, the greater the opportunity for an inclusion to be subject to changes in direction and velocity and come into contact with the filter internal surface for adhesion and capture. Depth filtration is the filtering mode most prevalent in bed filtration. Depth filtration is capable of removing particles much smaller than 30 mm (1200 min.) as they are carried in a molten-metal flow stream through a very tortuous path in a deep-bed filter (Fig. 5), which is commonly used in mill product casthouses. The tortuosity provides increasing probability for a two-step particle-capture mechanism involving transport from the liquid metal flow and attachment to the filter surface itself. Bed filters composed of tabular Al2O3 are quite common in the casting of mill products, where up to 1800 Mg (2000 tons) of a single alloy may be filtered in-line between holding furnace and casting station before replacement is necessary. Bed filters provide the best source of fine particle filtration, but their size, expense, and single-alloy use preclude their economic use in multiple-alloy wrought casting (ingot) or foundry casting operations. Bed filters are capable of providing nearly the highest filtration efficiency attainable, but because the bed is not rigid, movement and flow channeling can occur, which can dump inclusions.

Filter Characteristics Molten-metal filters obviously require hightemperature integrity in terms of strength, refractoriness, thermal shock resistance, and corrosion resistance. The filters may be placed directly in either the melting or casting furnace (or sometimes both) in the cases of in-furnace applications, or the filters may be in the running system of the casting mold (in-mold filters). In either application, it is important to understand the basic characteristics of the filters. The mechanical removal of undissolved particulates from molten metal follows the same

Fig. 5

basic physics as the filtration of particulate from any liquid. The basic principles of filtration are:  Larger inclusions are more easily removed

from melts than smaller ones.

 Lower flow rates of metal through a filter

result in greater efficiency of inclusions removal.  Filters with larger inlet surface areas have greater inclusion-holding capability and longer filter life, for a given incoming inclusion concentration.  Filter bodies with greater internal tortuosity or filter depth provide a higher probability of inclusion capture. However, some basic misunderstandings persist in understanding the characteristics of filters. One misconception is that filters can catch everything unwanted in the melt. As a basic physical principle, mechanical filtration can only remove the insoluble particulates in a liquid. Mechanical filtration cannot remove contaminants that are liquid to semiliquid themselves or dissolved constituents such as hydrogen gas, various low-melting fluxes, or alloying eutectic components such as silicon. In addition, it is critical to understand that reactive molten metals (especially aluminum) react with the environment for the entire time of its molten state. Aluminum is particularly reactive with the atmosphere and reacts easily with oxygen, forming solid particulate aluminum oxide, which is filterable, and hydrogen, which is soluble and therefore not filterable. This includes the time spent traveling in the runner system and filling the mold postfilter placement. Hence, some particulate contamination may still occur if the molten aluminum is allowed to travel in a turbulent state postfiltration. This would be especially true in sand molds where lower compaction, incomplete bonding, and turbulence of metal flow will create the possibility of sand erosion. Secondly, there are significant differences in the mechanism of various types of media described as a filter. The two most common

Example of one type of bed filter used for in-line processing

media found in today’s (2008) aluminum foundries are screens and refractory foam filters (Fig. 6, 7). Screens, for all intent and purpose, are two-dimensional structures with evenly spaced round or square holes. Regardless of composition, that is, metal or fiberglass being the most common, they function primarily as sieves. Sieves can only remove particulate larger than the opening in the sieve or, in this case, the screen. Foam filters, on the other hand, have two completely different and distinct modes of filtration that allow them to act as true filters rather than just simple sieves. Because foam filters are three-dimensional and have a random pattern of openings, they exhibit two completely different behaviors. On the surface, they behave similarly to screens in that particulate larger than the openings of the pores of the foam will be trapped on the inlet side of the foam. This is often referred to in the literature as cake filtration, as previously noted. However, the metal must continue to pass through the pores and cells that make up the internal honeycomb structure of the filter. Within the structure of the filter, due to rapid changes in pressure and metal velocity, particulate much smaller than the actual pores and cells themselves may become trapped. This is commonly referred to as deep-bed or depthmode filtration. It is important to understand that typically, sieving or cake filtration can be nearly 100% under normal conditions for large particulates.

Fig. 6

In-mold ceramic foam filters

Fig. 7

In-mold screens

198 / Molten Metal Processing and Handling Unless excessive force is applied, all particulates greater than the opening size of the holes cannot pass through the holes. On the other hand, deep-bed filtration can only function as a percentage-reduction filtration mechanism dependent on many variables: metal fluidity, velocity, foam filter porosity, foam filter thickness, particle size of the contaminant to be trapped, and so on. That is, some percentage of extremely small particulate will be able to escape even the best foam filter. It therefore follows that the choice of filter medium is based at least in part on the requirement for casting quality. For those castings where very small particulates do not contribute to quality issues, screens may be adequate. However, as the quality requirement of the casting increases, a point will be reached where screens will be incapable of providing the level of melt cleanliness required to meet the casting quality demands. At that point, a ceramic foam honeycomb filter structure becomes the best option to achieve the required quality. The one characteristic sometimes missing in the discussion of aluminum filtration is the function of refractory material in the foam structure. Unlike iron, steel, and copper alloys, the majority of particulate impurities in aluminum alloys is aluminum oxide, the most prevalent impurity, which exhibits very little physical or chemical affinity for most common refractory materials. Unlike iron, for example, where sticky silicacious slags will physically stick to many suitable refractories, aluminum oxide (alumina) in aluminum is “dry” and does not exhibit a particular physical affinity to any refractory. For this reason, there has never been any reliable data presented that support a claim of one refractory being superior over another in nonmetallic reduction in molten aluminum alloys, all other attributes of the foam filter being equal. The refractory composition can be responsible for other filter behaviors, however, such as strength (both cold and hot), friability, thermal shock, wettability, thermal capacity (based on density of the material), thermal cycling (remelt behavior), and soon. For these aforementioned reasons, alumina, silicon carbide, mullite, and graphite have been used as media in foam filters for aluminum filtration. Although they offer different physical refractory performance for any given porosity and thickness, they all behave similarly in particulate-retention capability. Once the appropriate refractory is chosen, the most important physical characteristic of the foam filter is its porosity. At this time (2008), there is no true, accurate universal definition in the industry to describe the porosity of a foam filter. The nomenclature of pores per inch (ppi) is a carryover from the foaming process that is used to make all kinds of foam products, such as cushioning materials, packaging, air filtration substrates, carpet padding, and so on, as well as the reticulated polyurethane foam used as substrate for refractory foam filters.

In general, there are three basic classes of foam porosity most commonly used in the aluminum foundry industry:  Coarse (10 ppi)  Medium (20 ppi)  Fine (30 ppi)

However, the porosity of the foam substrate is only a portion of the final story. No matter what foam is used, the refractory slurry applied to the foam can have different effects on the final foam filter flow characteristics, all else being equal. Therefore, comparing one manufacturer’s 10 ppi filter to another’s 10 ppi filter or even 15 ppi filter leads to confusion on the part of the foundry. Regardless of the ppi rating, what a foundry is really purchasing is the pressure drop that the filter provides the metal stream between the entrance face of the filter and the exit into the balance of the running system. For any given porosity rating and filter thickness, the pressure drop should be consistent, with minimum expected variation. As the pressure drop increases (as is the case with finer-grade filters), the filter has a greater ability to retain a higher and higher percentage of small and smaller particulates. Conversely, as the pressure drop increases, the expected flow rate in pounds per second will decrease. This often creates a basic dilemma. A common mistake made in filter printing is not allowing enough of an increase in the area of the filter versus the runner it is placed in in order to compensate for the filter pressure drop and correspondingly to sustain the necessary flow rate. Under normal circumstances, the rule of thumb is to allow for a 4-to-1 ratio of the usable filter area (excluding the print) versus the runner area in order to compensate for the filter pressure drop. When using coarser filters (10 ppi), this ratio can be somewhat reduced. When finer filtration is required (20 or 30 ppi) along with fast flow rates, the only alternative is to increase the usable surface area of the filter. Although most people focus on the particulate-removal capabilities of refractory foam filters as their primary function, it is, in fact, only a part of the contribution to overall casting quality improvement that they provide. In particular, unlike screens, properly printed foam filters have the capability to reduce the amount of turbulence and entrained air in the liquid metal stream, further minimizing the likelihood of the formation of regenerated oxides and dissolved hydrogen postfilter. In fact, an important additional function of an in-mold ceramic foam filter is to provide flow control. Unlike foam filters, screens actually split the metal stream into up to hundreds of small streams exiting the screen (Fig. 8, 9). This is particularly important in the first few seconds of mold filling prior to steady-state flow being achieved. The formation of the small streams dramatically increases the surface-area-tovolume ratio of the liquid metal, likewise increasing the surface area exposed to the air.

The more liquid metal surface area available to the atmosphere, the greater the potential for reoxidation to occur. Additionally, it can be shown that air bubbles can be easily aspirated into the mold through screens. Unlike screens, refractory foam filters inhibit the transmission of large air bubbles through the foam structure. An aluminum foundry should treat an air bubble as the encapsulated air it truly is. As damaging as an air bubble can be if it remains trapped in the solidified casting, it is equally damaging as a source of oxygen and hydrogen, which can

Fig. 8

Screen water model, split flow

Fig. 9

Ceramic foam filter, convergent flow

Molten-Metal Filtration / 199 react, forming aluminum oxides or dissolving in solution. One way to look at an air bubble in molten aluminum is as if it were a balloon on a string, the air bubble being the balloon and the string a trail of aluminum oxides following along wherever the balloon travels. Because refractory foam filters, via their unique pressure drop, can remove a significant portion of the energy from the metal stream as well as smooth the metal flow free of air bubbles, the metal exiting the refractory foam filter will be substantially less turbulent. Once this turbulence reduction is achieved, it will not remain so unaided by the foundry. The metal stream is the most nonturbulent when exiting the refractory foam filter immediately at the exit face. If anything is done to disrupt the metal stream as it travels away from the filter to the actual mold cavity, turbulence will increase. For this reason, anything that would increase the metal stream velocity postfilter should be avoided. This can include, but is not limited to, a choke postfilter, a further metal drop, a high-aspect-ratio runner (i.e., a runner that is substantially taller than it is wide), or any configuration that would tend to induce splash.

Types of Molten-Metal Filters Filter selection often depends on ease of use. The simplest filters are the strainer cores used in down sprues in conventional foundry sand casting. While serving to choke or control metal flow, they trap some particulates by surface attachment and a degree of cake filtration. For in-mold foundry applications, ceramic foam filters are common. Ceramic foams are lighter, easier to preheat, and provide easier flow capability than bonded-particle filters. In the primary and mill products casting industry, filter selection will depend on economics, space constraints, auxiliary metal treatment capabilities, desired filtration efficiency, and end-product application. Metal or Fiberglass Screens. Steel mesh screens and fiberglass cloth are also commonly used for aluminum. Molybdenum mesh is used for copper alloys. These materials act as planar filters. They provide moderate filtration, particularly of larger particles and inclusions exceeding 100 mm (0.004 in.) in size. The mesh screen, fiberglass cloth, and strainer cores are, of course, single-use filters. Bonded-particle filters consist of a refractory grain (either Al2O3 or SiC) bonded together to form a rigid structure. Bonded-particle filters provide a relatively tortuous path for metal flow through the filter (good depth filtration). These filters are often used vertically to separate melting or holding furnace hearth areas from the dip-out well. Silicon carbide vertical gate filters are quite durable and offer the potential of low thermal gradients between the hot face and cold face because of the relatively high thermal conductivity of silicon carbide compared to other materials, such as oxide ceramics. Temperature

gradients between inlet/outlet sides of the filter may start as low as 5  C but will build with use to gradients of 20 to 40  C (40 to 75  F) have been reported for these filters, versus 35 to 70  C (60 to 130  F) for oxide ceramic foams. Bonded- particle filters possess less porosity than for a ceramic foam (38 versus 75%, respectively) and greater tortuosity; therefore, they function more as depth filters. Cartridge Filters. One interesting application of the rigid filter concept is the cartridge filter. It is simply a shell-and-tube-type construction similar to a heat exchanger. This type of filter provides a great amount of surface area and usually has very fine pore size. Therefore, although a substantial priming head (up to 0.3 m, or 1 ft) of metal may be required before flowthrough begins, and although the flow rate is not very high, filtration efficiency is excellent. Cartridge filters are now receiving attention for such premium-quality applications as aircraft plate, computer disks, and foil. Efficiencies of 95% or better in removing particles less than 5 mm (200 min.) in size have been reported. Ceramic foam filters are the most common types of filters currently used in nonferrous casting. They are produced by slurry coating a reticulated polyurethane cellular foam, followed by drying and firing to burn out the precursor foam. This results in a ceramic replica of the original organic foam structure. Suitable ceramics for these filters are alumina, zirconia, and mullite. Some ceramic filters also employ chromic oxide. These filters are approximately 75% porous and are produced in a variety of pore sizes, as measured by pores per lineal inch (ppi). Ceramic foam filters are used in flat, platelike form. In the primary aluminum industry, plate sizes may range from 300 by 300 to 585 by 585 mm (12 by 12 to 23 by 23 in.), depending on flow rate (up to 680 kg/ min, or 1500 lb/min) and casting size (up to 34,000 kg, or 75,000 lb). Smaller filters are used in the die casting and foundry industries. Dual-structure or multilayered ceramic foam filters have been employed in the past but do not find present application. These use a finer pore size for greater inclusion-capture capability, coupled with a coarser pore size to permit greater throughput before the filter becomes saturated or clogged.

In-Mold Filtration Reduction of entrapped particulates in a casting is governed first by proper melting and handling practices and then by management of the casting process as the metal is poured, fed to the mold, and solidified. In-mold filtration is an important technology in aluminum foundry casting production. Filter application in the mold must follow sound principles to balance the pressure drop accompanying the resistance of a filter within the gating system with the necessary flow rate to minimize turbulence and therefore to achieve a sound, solidified cast structure. Proper selection of the specific filter

type and grade, sizing, and placement of the filter within the gating system will achieve significant particulate removal as well as metal flow control. Filter Selection and Application in Sand Gating. To achieve maximum filtration and flow modification benefits, the filter must be properly sized, placed, and coordinated with the other elements in the gating system. A filter that is too small can impede the flow of metal, increase fill times, and result in short pours if it blocks too soon. Also, remember that a filter is not an excuse to neglect good melting, treatment, and pouring practices. The selection of a filter size and porosity will depend on a variety of factors, including the pour weight of the casting, desired mold filling rate, alloy type, initial metal cleanliness, and pouring temperature. The most basic of guidelines for sizing based on predicted filter blockage for foam filter/flow modifiers is 0.2 to 0.4 g/cm2 (15 to 30 lb/in.2) of filter face before blockage. For flow rate, using the most common porosity foam filter/flow modifiers, the typical range for a 10 ppi filter is 50 to 100 g/cm2/s (0.7 to 1.4 lb/in.2/s) and for a 20 ppi filter, 30 to 80 g/cm2/s (0.4 to 1.1 lb/in.2/s). The filter (or filters in a multifilter system) should be large enough (total exit area and porosity) that the fill time is the same as it would be in an unfiltered casting with the same gating system dimensions. Also, the system should be unpressurized, and the filter should not become the choke in the system. If the filter becomes the choke, the metal will accelerate as it exits the filter and create undesired turbulence. If the ingate is the choke (a pressurized system), the flow modification effect of the filter will be lost as the metal sprays into the casting cavity. When blockage ultimately occurs from buildup of inclusions on and within the filter, it is not gradual but almost immediate. The selected filter must be large enough to maintain normal flow throughout the pour. As a starting point, the filter frontal area should be at least four times the cross-sectional area of the runner. Depending on foundry practice, it may be possible to reduce the filter size as casting experience and confidence are acquired. Filters may be placed horizontally, vertically, or at some intermediate angle within the runner system, depending on the molding method, pattern layout, or other molding restrictions. Filters in the vertical position are best placed just beyond the sprue well that will absorb the energy of the initial falling metal, as shown in Fig. 10. Filters in the horizontal position are best placed just off of the sprue well (Fig.11). Filters placed at the base of the sprue, where direct impingement of the metal stream will force the metal through the filter at high velocity until the sprue fills, may create turbulence in the initial metal entering the casting. Although the metal can be forced through the foam structure at these higher velocities, it can be defeating of the flow-modifying capability of the foam to do so.

200 / Molten Metal Processing and Handling

Fig. 11 Fig. 10

Horizontal installation of filter off of the sprue well

Vertical installation of filter in a mold feeding system

Within the basic guidelines for in-line filter application in sand molds, there is considerable latitude in positioning and placement (Fig. 12a–c). A single centrally located filter may be used, a filter may be placed in each runner, or a filter may be placed at each ingate for maximum effectiveness. Filters may be inserted manually or with core-setting equipment. Filter Selection and Application in Permanent Mold Casting. Most of the gating system design issues critical to sand casting of aluminum apply to permanent mold casting as well. There is even greater emphasis on the need for repeatable flow and tranquil filling characteristics and precisely controlled thermal properties to assure progressive solidification in the direction of the risers and feed heads. Proper gating design is vital to minimize turbulence as the mold fills and to avoid superheating the mold at any one point. The high density of the mold and total lack of permeability, compared to sand molds, put special emphasis on proper venting. Both horizontal and vertical gating and risering systems benefit from use of a tapered sprue to avoid air entrainment, an enlarged basin at the base of the sprue to absorb the kinetic energy of the falling metal, and an extended runner beneath the mold cavity to absorb entrained air and trap initial dross entering the runner system. The choke is normally at the base of the sprue, resulting in an unpressurized system. Quality issues with permanent mold castings may arise from entrained oxides and/or the reoxidation of treated metal, as is the case with sand castings. Filter screens are commonly used in an effort to intercept inclusions in permanent mold castings. However, screens have minimal filtration effect and damage the metal by splitting the stream and creating turbulence.

Development efforts are currently underway to apply foam filters to permanent mold castings to trap inclusions and minimize turbulence and reoxidation in vertical, horizontal, tilt, and low-pressure permanent mold applications. Filter selection is based on the same criteria (pour weight, desired fill rate, alloy type, initial metal cleanliness, pouring temperature) as sand casting. Direct Pouring. The primary purpose of runners and gating systems is to deliver metal to the casting cavity in a nonturbulent fashion. However, even though gating and running system components may be recycled, gating is not free. Significant costs of gating include:  Cost of poured metal not turned into saleable  



 



castings, including melt conversion cost, handling, and melt loss Total cost of remelting removed gating metal, including energy, melt loss, alloy loss, and other treatment costs Total cost of molding, including consumption of pattern and mold space, extra sand and binder, more metal, and runner maintenance due to runner erosion Pouring temperature cost, including superheat to compensate for loss of heat in runner system and oxidation aggravated by higher temperatures and turbulence Cleaning costs, including gating-removal labor, equipment and supplies, and handling and storage of removed gating Compromised thermal profile resulting from coldest metal ending up in the riser, and unnecessarily large risers needed for adequate thermal mass to accommodate shrinkage Oxidation and air entrainment resulting from turbulence in runners (rolling-back wave generation and air bubbles)

Fig. 12

Placement of filters (shaded). Filters may be placed in (a) each ingate, (b) one central location, or (c) each runner.

Molten-Metal Filtration / 201 By pouring directly through a sprue attached to the top of the casting cavity or mounted on a side entry close to the casting, many of the costs of a conventional gating system can be reduced or eliminated. Direct pouring will certainly reduce poured weight and increase yield. However, for automotive and other castings with stringent physical requirements, some means must be used to control filling turbulence and prevent mold erosion and/or core damage from direct impingement of the molten metal. A direct-pouring unit for sand or permanent mold casting (Fig. 13) has been developed and patented (FOSECO Metallurgical Inc.) that incorporates an insulating pouring sleeve (that may also act as a riser) with a foam filter to trap inclusions and modify the flow of the metal as it enters the casting cavity. Filter media other than foam have not been proven adaptable to the direct-pouring process. It is essential that turbulence reduction in the metal flow be possible from any device used to attempt direct pouring. In direct pouring, filtration takes a back seat to reduced turbulence in terms of primary operational characteristics.

Fig. 13

Patented direct-pouring filter. Courtesy of FOSECO Metallurgical Inc.

This direct-pouring system replaces all the traditional gating system elements and reduces feeding requirements. By eliminating the runner system and ingates, ingate hot spots requiring feeders are eliminated. Also, with metal entry through the top of the casting, the hottest metal ends up in the direct-pouring unit sleeve, reducing thermal mass requirements if the sleeve serves as a feeder as well. Example of In-Mold Filtration for Scrap Reduction in Sand Casting of Automotive Intake Manifold. Prior to the introduction of filtration/flow modification, a sand cast manifold experienced unacceptably high scrap rates resulting from oxides and other inclusions. Mold-filling turbulence was identified as a contributing factor. Minor modifications to the existing gating system enabled the addition of a single foam filter/flow modifier that led to a reduction of machining scrap and other oxiderelated scrap from 4.6 to 0.8%. With the casting valued at approximately $25 (United States) when the scrap was discovered, this scrap-reduction level represents a gross savings per casting of $0.95. Based on the typical production rate of 325,000 castings per year, this results in an annual gross savings of $308,750. Example of In-Mold Filtration for Scrap Reduction in Casting Oil Cooler Extension. Machining scrap and leakers were responsible for a combined scrap rate of 4.73%. Before the part could be leak tested, it required machining. At the point of machining, the part was valued at $16.75 (United States). After leak testing, the part was valued at $20. Placing a foam filter/flow modifier in the conventional gating system reduced the machining scrap rate from 1.81 to 0.4% and the leaker rate from 2.92 to 0.28%. The combined scrap rate reduction (machining and leakers) results in a gross savings to the foundry of $0.76 per casting or $45,600 based on the annual production rate of 60,000 pieces. Example of Direct-Pour Filtration in Casting Large Commercial Pump Body. Initially, this 30 kg (65 lb) casting included a conventional runner system and had a pour weight of 42 kg (93 lb) for a yield of 70%. By replacing the components of the runner system with a direct-pouring system, the pour weight was reduced by 9 kg (19 lb) and the yield increased to 88%. At $0.10/lb (United States), this accounted for a $1.90 savings per casting. The elimination of the conventional running and gating system also resulted in a reduction in cleaning room time from 6 to 3 min (removal of risers, flash, etc.), saving an additional $1.35 per casting. The filtration capability and flow-modifying effects of the foam filter in the direct-pouring unit reduced the (machining) scrap rate from 7 to 2%. With an approximate casting value of $75 at the time scrap was identified, this resulted in an additional savings of $3.75 per casting. With a total savings of $7.00 per casting and a production rate of 3000 castings per year, this resulted in a $21,000 gross

savings. At the selling price of $110, this represents a gross profit improvement of 6.4%.

In-Furnace Filtration of Molten Aluminum As described in the preceding section, filters are also placed in the feeding system for casting. However, it may not always be possible to employ direct filtering within the mold system. Consequently, in-furnace filtering is an alternative, such as with high-pressure die casting. Other processes, such as gravity permanent mold, low pressure, and sand casting, can also make use of filters directly into a melting or casting furnace. In-furnace filtration provides a very helpful means to the foundry in removing inclusions and thereby increasing the molten-metal quality to produce castings. In-furnace filters can be placed directly in either the melting or casting furnace, or sometimes both. In-furnace filtering products can be applied to most of the typical melting and holding/casting furnaces in use today (2008), with proper evaluation. The benefits of in-furnace filtration can be demonstrated using various analytical techniques on the molten metal itself prior to casting, or by monitoring and evaluation of machining and finishing operations and overall casting production, to determine reduction in the number of scrap castings that can be shown to be inclusion-related.

In-Furnace Filter Types There are several types of filter media that have been employed in a furnace. Historically, the earliest and simplest form of filter was to use a fiberglass sock at the tap-hole or outlet spout on the furnace to catch the dross or other debris. Additionally, use of a ceramic-media bed in a box just off the outlet allowed the metal to flow through and the ceramic media to collect not only the dross or other debris but also to filter some inclusions from the pouring metal itself. While these fiberglass filters at a tap-hole of a melting furnace are still employed and are indeed important, newer furnace filtering schemes have taken over in many instances. For in-furnace applications, either a ceramic foam filter or a bonded-particle filter is often used in modern practice. Many factors tend to favor the latter. The bonded-particle filter is an aggregate of silicon carbide granular particles together with a proprietary ceramic binder in a high-temperature firing process to give body to the filter. The forming process for the bonded-particle filter allows many different sizes, grades, and three-dimensional filter shapes to be produced for specific applications. This is especially useful in that there are many variants in size, refractory construction, and configuration in foundry melting and casting furnaces employed today (2008).

202 / Molten Metal Processing and Handling This type of filter is not suitable as a singleuse, in-mold filter but rather is useful and economical only in applications that allow for sustainable usage. A bonded-particulate filter has a close-packed structure, with just 40% nominal porosity. The open-pore volume at this level is insufficient to permit normal rules of solidification (Chvorinov’s rule) to be followed without an excessively-sized print in the gating to accommodate the flow rate required through the filter to achieve proper solidification in a casting. The bonded-particle filter provides outstanding longevity—often two to three months or more—in many aluminum casting operations because of its chemical stability afforded by the binder, which possesses excellent molten aluminum resistance. The inherent higher mass/density and strength of a bonded-particle filter versus a ceramic foam filter—which replicates a reticulated polyurethane foam structure—favors the application of the former in furnaces where long filter life is desirable. The bonded-particle filter is used either as a vertical gate filter in a melting furnace in front of a tap-hole or placed in front of a transfer pump to provide first-stage filtering. This firststage filtering, which lessens the burden on downstream filtering processes, minimizes entrainment of sludge and other debris associated with melting foundry returns and other materials with a high inclusion content. The molten metal is usually further delivered by ladle for subsequent treatment, such as degassing and/or flux injection, and then transferred to casting furnaces in the permanent mold or high-pressure die casting process or direct poured into sand or precision-sand molds. Figure 14 depicts a typical application of a vertical gate filter in a melting furnace in front of a tap-hole. In the dip-well of a casting furnace, further filtration is advantageous, and this can be accomplished either by another vertical gate filter (Fig. 15) or by an all-media box filter (Fig. 16), from which the casting can be poured by using either a manual ladle or an automatic ladle dipping from the interior. The box filter provides the additional advantage that the filtered metal is always available without regard to buildup and breakaway of inclusions/accretions on the sidewalls of the refractory. Both vertical gate and box filters can be produced in many sizes and geometric shapes to meet specific furnace and ladle dimensional requirements. The long-life potential of a bonded-particle filter allows for its long-term use as an in-furnace filter, complementing any further filtering that may be performed in-mold. In addition to long life, the silicon carbide composition provides greater thermal conductivity than an oxide-based filter, which is important in maintaining temperature in most casting processes. Additionally, the multifaceted particle structure, together with a fully interconnected pore structure, provides a tortuous metal-flow path through the filter. This tortuosity, coupled with

Fig. 14

Vertical gate filter in front of a tap-hole of a reverberatory melting furnace

Fig. 15

Vertical gate filter in a holding or casting furnace. (a) Filter separates heated chamber from dip-well. (b) Installation of a vertical gate filter in a new furnace dip-well

Fig. 16 dip-well

Box filters. (a) All-media bonded-particle box filter. (b) All-media box filter employed in a

Molten-Metal Filtration / 203 the chemical and physical affinity of the binder system, enhances the probability of inclusion capture and hence contributes to enhanced filtration efficiency.

In-Furnace Filter Installation Filter installation into reverberatory furnaces and use of in-furnace filters can be accomplished with careful effort. Measurements of the dip-well dimensions, or other areas in the furnace where the filter will be installed, should be taken carefully, with emphasis on the clean refractory, removing any buildup prior to taking the measurements. For a vertical gate filter, gaskets may be used on the filter edges to provide a seal and take up some degree of irregularity in the refractory sidewalls. It is necessary to fully scrape the sidewalls and bottom of the furnace vessel to provide a good fit. It is also very advantageous to create accepting slots in the furnace refractory for the filter, which can be done during a furnace relining. For either a vertical gate filter or a box filter, it is necessary to allow sufficient space for the ladling operation. Preheat of the filter is necessary to remove any moisture condensation and to prevent thermal shock, but with a thermally-conductive silicon carbide filter, this is straightforward and only requires that the filter be preheated to a temperature of 200 to 300  C (400 to 575  F). This may be accomplished by placing the filter on the furnace sill plate, covered with an insulating blanket, against a furnace hot wall or in a nearby oven. The filter may then be placed directly into the molten metal in the furnace as quickly as safety will allow, without the need to drain the furnace. When a melting or casting furnace is being originally manufactured, relined, or rebuilt, it is often desirable to create slots in the refractory sidewalls to accommodate a filter. In this case, the face-edges of the filter may have a compressible fiber gasket attached to further create a seal in the contact area between the filter medium and the sides or shoulder of the slotted construction. Crucible Furnace with Filter Baffle. Filter installation into crucible furnaces is somewhat similar in principle to installation into a reverberatory furnace, but there are specific differences. The filter can be viewed as a baffle separating the crucible into two sections, which may or may not be equal. Filter baffles have also been placed in the pouring ladle (Fig. 17). If a crucible furnace is used for both charging and degassing (and/or other melt treatment), space must be allowed for casting. In that instance, it is important to maintain both temperature and molten-metal level stability by matching intake with take-out volume, in order to avoid sludge formation and to maximize filtering surface area effectiveness. Further, the contours of the crucible place an extra imposition on the fabricated filter shape to become

sealed to the crucible wall to avoid metal bypass from unfiltered to filtered section, usually demanding extra gasketting. In addition, if due to space/volume constraints the filter must be placed offset to the main diameter, stops are necessary to avoid filter movement into the more open diameter. Figure 18 portrays a ceramic bonded-particle filter placement in a crucible. As an alternative to a baffle filter, there may be sufficient space available in casting crucible holding furnaces, especially larger ones, to accommodate a filtering vessel, such as an allmedia filter or one constructed of refractory materials with a smaller filter that can be replaced when spent. Low-pressure furnaces readily employ filters directly in or onto the stalk or riser tube. Both ceramic foam filters and bonded-particle filters are used in this application. In the lowpressure furnace, application of air forces metal up the stalk or riser tube and into the mold cavity with each cycle. Excess metal (the riser volume) drains back through the filter at the completion of each cycle. The filter medium must not create an undue restriction to metal flow, which would affect the pressurization cycle, nor the backflow of metal from the riser

during the drainage. The lower-porosity ceramic foam filters are often found to be more advantageous in this regard, with a disc shape being easily affixed into or onto the end of the riser tube. Therefore, bonded-particle filters with substantially lower porosity tend to be supplied in more three-dimensional shapes (Fig. 19) in this application to provide more filtering surface area and hence metal-flow capability. To accommodate a larger filter, there must be a sufficient gap or space between the end of the riser tube and the floor of the furnace, as shown in Fig. 19.

Factors That Determine Filter Life There are many factors that determine filter life, including overall incoming metal quality on the inlet side, the specific filter grade, metal flow rate and total throughput through the filter, and metal-level drawdown (in which the filter loses its prime). Other than simply gaining experience and knowledge about how long an in-furnace filter can be used, there are two observations that can be made. As the filter begins to clog, there will be a metal-level difference between the upstream and downstream

Fig. 17

Filter baffles in (a) a crucible and (b) a pouring ladle

Fig. 18

Crucible baffle filter

Fig. 19

Schematic of low-pressure furnace with bonded-particle filter

204 / Molten Metal Processing and Handling sides of the filter. In addition, there will often be a measurable and increasing temperature difference between the upstream and downstream metal. This is caused by the increasing concentration of oxide inclusions that are lower in thermal conductivity than both the molten metal and the silicon carbide filter. The following factors are extremely important in achieving optimum performance of any in-furnace filtering application:  Intrinsically, the finer the filter grade and the



  



lower the filter inherent porosity, the better will be the filter particulate-capturing ability, the quicker the filter will clog, and therefore, the shorter the filter life. By contrast, the larger the filter, the more surface area is available to accommodate both metal flow and inclusion buildup, by both cake and depth-mode filtration. Any ceramic filter must be preheated to at least a minimal extent before being inserted or exposed to molten metal in order to resist thermal shock and premature failure due to mechanical/thermal damage. Excessive metal-level drawdown will lessen the filter useful life, since it will lose the filtering capability of the exposed surface. Heat will be lost as it wicks from an exposed filter surface, unless the filter is covered with an insulating material. In a casting furnace, the molten metal itself is a heat source coming through to a dipwell. When not casting, the furnace should be kept covered. Otherwise, heat will be lost, and the oxide buildup on the inlet side of the filter may not allow sufficient heat to be recovered through to the casting side when casting again commences. Naturally, the inclusion concentration in the molten metal being delivered to the filter in a given time, and the volume of molten metal passing through the filter, will affect filter life.

Fig. 20

Data comparison from K-mold technique in measuring melt quality of filtered and unfiltered molten aluminum ready for die casting

Dirtier metal will clog a filter faster and therefore require more frequent filter change-out.

Filtered Metal Quality It is often desirable to be able to demonstrate the differences in quality between filtered and unfiltered molten metal. There are several means that are possible to evaluate this, including a few real-time, shopfloor methods as well as more extensive laboratory methods. After initial molten-metal quality improvements have been demonstrated by the application of filtration, the true measure of improvement must be shown in product-quality enhancements on the cast product. Performance results of in-furnace filtering systems often show significantly reduced scrap, resulting in greater quality, productivity, and profitability. Specifically, one of the biggest benefits is improvement in machinability, since inclusion removal minimizes hardspots in castings that cause tool wear and tool breakage. Inclusions also nucleate porosity formation during solidification, which can cause leakers; filtering removes these and therefore improves

pressuretightness. Mechanical properties are also improved, particularly elongation and fatigue strength, which are quite sensitive to the presence of inclusions. Surface finish operations—polishing, anodizing, powder coating, and so on—are also enhanced, because inclusion-causing blemishes are minimized or eliminated. Beneficial effects of in-furnace filtration can be demonstrated by several different methods of measuring molten-metal quality. Figure 20 shows the difference between unfiltered and filtered molten metal using the real-time shopfloor K-mold technique. In this test, molten metal is carefully sampled several times in a notched-bar fracture test, and the fracture surfaces are examined macroscopically or with the aid of low-power magnification to observe any surface discontinuities, which will relate to inclusions, dross, or porosity. An index value can be determined; Fig. 20 shows that, of 20 sample surfaces investigated, filtration lowered the observed discontinuity count dramatically, to as little as 1 observation in 20 (one-twentieth, or 0.05). Another measurement method of melt quality is the Prefil Footprinter technique (N-Tec Ltd.). It has been used to compare quality relative to

Fig. 21

Prefil results on A356 benchmarked to other data (bars)

Fig. 22

Porous disc filtration analysis results on 380 die cast alloy. Flux injection in transfer ladle followed by settling and filtration in casting furnace

Molten-Metal Filtration / 205 data for several casting foundries using the same alloy. Prefil results are shown in Fig. 21 on benchmarked A356. Figure 21 shows the impact of in-furnace filtration, providing acceptable molten-metal quality suitable for a high-integrity casting process. In many aluminum foundry casting processes, metal is melted in a melting furnace, then transferred via transfer ladle to the casting furnace or to the mold. Metal treatment, such as fluxing, degassing, flux injection, grain refining, and modification, is often made in this transfer ladle. Figure 22 presents porous disc filtration analysis (PoDFA) (Alcan International Ltd.) results on a die casting alloy at several steps of the process. There is a proportionate improvement in metal quality, as evidenced by the metallographic inclusion average as obtained by image analysis, at each stage of the process. Even after settling time in the casting furnace, there is a further reduction in inclusion concentration after the metal has flowed through an in-furnace vertical gate filter. While the foregoing represents analytical techniques to assess the effectiveness of in-furnace filtration on molten-metal quality, it is necessary for the foundry to determine the beneficial effects and cost-effectiveness of filtration on the ultimate cast product. Destructive testing can be done on random or selected samples, especially to determine solidification structure and other microstructural features in the critical areas of a given casting, where engineering performance

criteria must be met. Separately-cast test bars do offer some measure of physical metal quality but do not reflect the cast part per se. Two mechanical properties that are most sensitive to the presence of, or absence of, inclusions are tensile elongation/ductility and fatigue strength. The presence of inclusions can lower the ductility and fatigue strength cycles to failure. However, good tensile results do not necessarily imply good casting quality, and a substantial number of test bars must be evaluated to create statistical significance. There are three foundry production methods to evaluate casting quality as affected by in-furnace filtration. Overall measurement of scrap quality and identification of those scrap castings that are due to observable inclusion-related defects, improved machining performance through elimination or minimization of hard spots, and leak testing (inclusions nucleate porosity formation) are three of the best means to ascertain the benefits of filtration. REFERENCES 1. D.E. Groteke, A Production Filtration Process for Aluminum Melts, Trans. Soc. Die Cast. Eng., 1983 2. D.V. Neff, Continuous, Sustained, ‘ReUsable’ Filtration Systems for Aluminum Foundries and Diecasters, Proceedings, Fourth International Conference on Molten Aluminum Processing, AFS, 1995, p 121

3. D.V. Neff, Using Filters and Evaluating Their Effectiveness in High Pressure Diecasting, Trans. NADCA, Sept 2002 4. D.V. Neff, The Role of Melt Treatment in the Production of High Integrity Aluminum Castings, Proceedings, International Molten Aluminum Conference, AFS, 2006 5. D.V. Neff, Ensuring Quality Prior to Pouring, Mod. Cast., June 2006, p 27 6. D.V. Neff, The Exclusion of Inclusions, Die Cast. Eng., Jan 2008, p 48 7. C.J. Simensen and G. Berg, A Survey of Inclusions in Aluminum, Aluminum, Vol 56, 1980, p 335 8. L.J. Gauckler and M.M. Wacher, Industrial Application of Open Pore Ceramic Foam for Molten Metal Infiltration, Light Metals, The Metallurgical Society, 1985, p 126 9. J. Langerweger, Nonmetallic Particles as the Cause of Structural Porosity, Heterogeneous Cell Structure and Surface Cracks in DC Cast Al Products, Light Metals, The Metallurgical Society, 1981, p 685 10. C.J. Simensen, The Effect of Melt Refining Upon Inclusions in Aluminum, Metall. Trans., 1980 11. C.J. Simensen and U. Hartvedt, “Analysis of Oxides in Aluminum by Means of Melt Filtration,” Center for Industrial Research, Trondheim, Norway, 1984 12. C.J. Simensen, Sedimentation Analysis of Inclusions in Aluminum and Magnesium, Metall. Trans. B, 1981, p 733

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 206-229 DOI: 10.1361/asmhba0005199

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Steel Melt Processing MELT PROCESSING of steels can be broadly classified as either primary steelmaking or secondary steelmaking. The two most important primary steelmaking processes are the electric arc furnace (EAF) process and the basic oxygen furnace (BOF) process or Linz-Donawitz process. The EAF process has been increasingly used, gradually overtaking the BOF as the primary steelmaking furnace in the United States. Some other processes, such as the open-hearth furnace, are still practiced in a few countries to produce special steels. Primary melt refinement also includes the uses of converters, which are furnaces in which oxygen is blown through a bath of molten metal, oxidizing the impurities and maintaining the temperature through the heat produced by the oxidation reaction. A typical converter in steel refining is the argon oxygen decarburization (AOD) vessel. In most modern steel operations, the steel is melted and only roughly refined in one furnace. Then, it goes to a secondary operation where it is further processed to yield the final desired chemistry. This secondary operation can be carried out while the steel is in the furnace, ladle, the degasser unit, or, in the case of stainless steel, the liquid metal can be poured into an AOD vessel. After secondary refining, the molten metal is transported to a tundish or a bottom pure ingot making facility for teeming. The melting furnaces used in steel foundries are essentially the same as those used for primary steelmaking, except that most foundry melting units are smaller. Although the equipment is the same, there are also important process differences. For example, one major difference between melting practice in a steel foundry and a steel mill is the higher tapping temperature used for foundry melting to attain better fluidity of the molten steel. Refractory materials in steel foundries are also different than those in primary steel production. Electric arc and induction melting are common methods of melting steels. After furnace melting, the molten steel may be refined in a converter (outside the melting unit) or in a ladle. Thus, steel melt refinements are classified as treatments by either converter metallurgy or ladle metallurgy. Converter metallurgy includes melt refinement in AOD vessels and vacuum oxygen decarburization in a converter vessel. Ladle metallurgy refers to melt treatments in a ladle refining station after melting in the

furnace/converter unit. Ladle metallurgy is used for deoxidation, decarburization, and adjustment of chemical composition, including gases. Methods of ladle metallurgy include:
Vacuum induction degassing
Vacuum oxygen decarburization
Vacuum ladle degassing

These vacuum methods of melt refinement are briefly reviewed in the section “Ladle Metallurgy” in this article. Nonvacuum processes (carried out at atmospheric pressure) for production of special and clean steel and without any supplemental reheating include various argon bubbling processes (such as AOD), electroslag remelting (ESR), and ladle injection methods. Argon bubbling methods are discussed in the section “Converter Metallurgy” in this article. Melt refinement by ESR furnaces is covered in the article “Electroslag Remelting” in this Volume.

Furnaces and Refractories Electric arc and induction melting are the most common methods of melting steels. Coreless induction furnaces are also used for melting carbon, low-alloy steels, stainless steels, high-alloy steels, or nickel-base heat-resistant alloys. The type of furnace, refractories, and melt processing equipment depends on a number of factors that include the quality of the

Fig. 1

Refractory lining of ladles

charge material to be melted, chemistry control requirements, capital equipment costs, manpower, and energy and operational costs.

Refractories Several different types of refractories can be considered for the melting furnaces, ladles, and tundishes for holding and transporting molten metal. Usually, placement of refractory linings involves a practice called zoning, where refractories with different properties are used in different areas of the same vessel to resist the attack of the steelmaking process in that specific area. For example, several wear mechanisms occur in a ladle. At the slagline, wear proceeds primarily by corrosion, and magnesia-carbon or magnesia-chrome bricks are commonly used as the hot-face liner (Fig. 1). In the bottom and lower sidewalls of the ladle, wear spots can develop by erosion of bricks or by the impingement of the molten steel stream on the refractory. It has been reported that 70% alumina, dolomite, and alumina-spinel can be used as the hot-face liner in this area (Ref 1). Monolithic refractory use has progressed in accordance with the need for more mechanized lining installation and repair. Corrosion, oxidation, and thermal cycling contribute to lining wear. Properties are derived from the major refractory constituents, additives, microstructure, grain size, and fabrication process (bricks or

Steel Melt Processing / 207 castable). Changing these factors imparts specific characteristics that include melting temperature, creep resistance, oxidation resistance, corrosion resistance, mechanical strength, and erosion resistance. Lining technology is constantly being updated by better and more costeffective refractories. Refractories are classified as basic, acid, or neutral by an index of the “basicity” of the slag formed on the metal surface. The compatibility of this basicity index with a refractory can give an indication of its potential to corrode a refractory. Basicity can be represented by different indexes, such as the B2-ratio (the weight ratio of CaO to SiO2 in the slag) and the V-ratio (the weight ratio of CaO + MgO to SiO2 in the slag) (Ref 2): CaO ðwt%Þ SiO2 ðwt%Þ CaO þ MgO V¼ SiO2

B2 ¼

Steel slags typically contain five major oxides: CaO, MgO, SiO2, FeO, and Al2O3. Slags with B2- or V-ratios less than 1 are defined as acid slags, while basic slags are more than 1. The vast majority of steel is produced with a basic slag, while most iron production uses an acid slag. That is, most steelmaking refractories are composed of MgO and/or CaO, while ironmaking refractories consist of Al2O3 and SiO2. The use of these or other slag indexes may vary from plant to plant and from heat to heat, depending on the grade of steel being made. Foundries that have access to a good grade of scrap, and therefore do not need to reduce the phosphorus and sulfur contents of the steel to meet specifications, usually prefer to use a furnace lined with silica refractories (acid lined). Foundries that do not have a good source of low-phosphorus, low-sulfur scrap use refractories such as magnesite and dolomite for furnace lining (basic lining). The compositions of certain steels, such as austenitic manganese steels, require that they be made in a furnace with a basic lining. Basic Refractories. Basic linings are principally magnesium oxide, although some linings contain chromium oxide, calcium oxide, and other oxides in controlled amounts (Ref 3). They resist temperatures several hundreds of degrees higher than acid linings. Basic refractories are magnesite-based and are either natural or synthetic magnesite, brucite, and dolomite. Dense synthetic magnesia is produced from seawater and has a purity of 95 to 99% MgO. Natural magnesite is present as a hydroxide or a carbonate. The double carbonate of dolomite (CaCO3MgCO3) is found in abundance. Hightemperature calcining (firing in kilns to dissociate the carbonates) is required in all the cases to produce a dense material with minimum impurities. Magnesia-chrome refractory spinels are also commonly used without firing in combination with magnesia refractory. Neutral refractories are typically alumina refractory bricks, as high-alumina silicates or

pure sintered alumina. Neutral refractories are very resistant to chemical attack and are more tolerant to intermittent melting than acid or basic linings.

Direct Arc Melting The direct arc furnace consists essentially of a metal shell lined with refractories (Fig. 2). This lining forms a melting chamber, the hearth of which is bowl shaped. Three carbon or graphite electrodes carry the current into the furnace. Steel, either solid or molten, is the common conductor for the current flowing between the electrodes. The metal is melted by arcs from the electrodes to the metal charge—both by direct impingement of the arcs and by radiation from the roof and walls. The electrodes are controlled automatically so that an arc of proper height can be maintained. For basic electric furnace melting, the furnace lining is a basic refractory such as magnesite or dolomite. The charge is usually composed of purchased scrap steel and foundry returns. During the melting period, small quantities of lime are occasionally added to form a protective slag over the molten metal. Iron ore is added to the bath just as melting is complete. The slag is then highly oxidizing and in the correct condition to take up phosphorus from the metal. Shortly after all of the steel has melted, this first slag is taken off (if a two-slag process is to be used), and a new slag composed of lime, fluorspar, and sometimes a little sand is added. As soon as the second slag is melted, the current is reduced, and pulverized coke, carbon, or ferrosilicon, or a combination of these, is spread at intervals over the surface of the bath. This period of furnace operation is known as

Fig. 2

the refining period, and its purposes are to reduce the oxides of iron and manganese in the slag and to form a calcium carbide slag, which is essential to the removal of sulfur from the metal. The refining slag has approximately the following composition: Element

CaO SiO2 FeO CaF

Composition, %

45–55 15–20 0.50–1.5 5–15

Adjustments are made in the carbon content of the bath by the addition of a low-phosphorus pig iron. After the proper bath temperature is obtained, ferromanganese and ferrosilicon are added and the furnace is tapped. Aluminum is generally added in the ladle as a final deoxidizer.

Induction Melting The high-frequency induction furnace is essentially an air transformer in which the primary is a coil of water-cooled copper tubing and the secondary is the metal charge. The circular winding of copper tubing is placed inside the shell. Firebrick is placed on the bottom of the shell, and the space between that and the coil is rammed with grain refractory. The furnace chamber may be a refractory crucible or a rammed and sintered lining. General practice is to use ganister rammed around a steel shell that melts down with the first heat, leaving a sintered lining. Basic linings are often preferred; a rammed lining of magnesia grain or a clay-bonded magnesia crucible can be used. The process consists of charging the furnace with steel scrap and then passing a high-frequency current through the primary coil, thus

Schematic cross section of a typical electric arc furnace showing the application of different refractories

208 / Molten Metal Processing and Handling inducing a much heavier secondary current in the charge, which heats the furnace to the desired temperature. As soon as a pool of liquid metal is formed, a pronounced stirring action takes place in the molten metal, which helps to accelerate melting. In this process, melting is rapid, and there is only a slight loss of the easily oxidized elements. If a capacity melt is required, steel scrap is continually added during the melting-down period. Once melting is complete, the desired superheat temperature is obtained, and the metal is deoxidized and tapped. Because only 10 to 15 min elapse from the time the charge is melted down until the heat is tapped, there is not sufficient time for chemical analysis. Therefore, the charge is usually carefully selected from scrap and alloys of known composition in order to produce the desired composition in the finished steel. Composition can be closely controlled in this manner. In most induction furnaces, no attempt is made to melt under a slag cover, because the stirring action of the bath makes it difficult to maintain a slag blanket on the metal. However, slag is not required, because oxidation is slight. The induction furnace is especially valuable because of its flexibility in operation, particularly in the production of small lots of alloy steel castings. Induction melting is also well adapted to the melting of low-carbon steels, because no carbon is picked up from electrodes, as may occur in an EAF. Coreless Induction Furnaces. Both channel- and coreless-type induction electric furnaces are used in the steel industry. The coreless induction furnace melts by passing an alternating current through a water-cooled coil that surrounds the refractory linings of the furnace. The current generates a magnetic field that induces currents in the charge and molten metal. Coreless induction furnaces provide operational flexibility when processing different alloys, because the furnace can be easily emptied and recharged. Vacuum Induction Melting (VIM). In the VIM remelting process, the charge material has basically the same chemical analysis as the final melt. No major refining or metallurgical work is performed other than lowering the residuals and inclusions. When remelting alloy steel, certain elements that readily vaporize or oxidize must be added under controlled conditions. As a rule of thumb, alloying elements that oxidize should be added as late as possible, and difficult-to-dissolve elements should be added early and during high stirring periods. The furnace lining is dependent on the nature of the slag and the desired finish temperature. Oxygen, hydrogen, and, to some extent, nitrogen can be removed during vacuum melting. Carbon can also be lowered to produce some types of stainless steels. Pressures as low as 5 mm of mercury are used in these furnaces. The control of pressure and composition of gas over the melt makes it possible to deoxidize the melt with carbon or hydrogen, because they produce gaseous deoxidation products, preventing the formation of nonmetallic inclusions. Use of low

pressures also eliminates nitrogen pickup by steel. The volatility of certain alloying elements such as chromium, aluminum, and manganese can result in high losses that can be minimized by replacing the vacuum with an inert gas atmosphere over the melt during additions. Vacuum induction melting is often employed as a remelting operation for very pure steels. It is also used as a first stage in conjunction with vacuum arc remelting (VAR) furnaces and ESR furnaces for duplex (VIM-VAR) or triplex (VIMESR-VAR) melting. Some modified types of VIM equipment (such as the Therm-I-Vac process) use either a vacuum induction melter under low pressure or as a stream degasser. The system can be used for cold charging of steel in the induction furnace and melting, refining, and teeming of steel under low pressure. The steel could also be hot charged under low pressure in the vacuum furnace during which stream degassing occurs, reheated to compensate for the losses during degassing, adjusted for alloy additions to meet the chemical composition specifications, and teemed, all under low pressure. Superheating of steel is not required if hot charging is performed. The steel can be partly deoxidized before charging into the vacuum induction melter. This equipment is mostly used as a vacuum induction unit to melt a cold charge.

Converter Metallurgy Converters are used in secondary metallurgy to refine melts outside the primary metallurgical melting unit. Various converters are available that apply the bottom blowing of oxygen/inert gas mixtures. These bottom-blowing converters use different types, numbers, and arrangements of injection nozzles and the following gases:
Argon as a cooling inert gas with a purity

ranging from 85 to 99.99%


Nitrogen as a cooling inert gas with a purity

of 99 to 99.9%


Argon-oxygen and nitrogen-oxygen mix-

tures in the case of the AOD converter


Dry air as a diluted reactive gas
Oxygen as a reactive gas with a purity of 80

to 99.5%


Steam as a cooling reactive gas
Carbon dioxide as a diluted reactive gas

Bubbling methods allow quick and uniform mixing of alloys, temperature homogenization, adjustment of chemical composition, and partial removal of nonmetallic inclusions. These functions are accomplished by lowering a refractory-protected lance close to the ladle bottom or by blowing through porous refractory plugs at the bottom or sidewall of the ladle or vessel. This section describes the converter designs for degassing:
Argon oxygen decarburization
Oxygen top and bottom blowing converter,

which is an extension of AOD technology


Vacuum oxygen decarburization converter

In addition to AOD in a converter unit, argon bubbling processes also include capped argon bubbling and composition adjustment by sealed argon bubbling. These bubbling methods are described with other ladle injection methods in the section “Ladle Metallurgy” of this article.

Argon Oxygen Decarburization Some foundries have installed AOD units to achieve some of the results that vacuum melting can produce. These units look very much like Bessemer converters with tuyeres in the lower sidewalls for the injection of argon or nitrogen and oxygen. They are processing units that must be charged with molten metal from an arc or induction furnace. Up to approximately 20% cold charge can be added to an AOD unit; however, the cold charge is usually less than 20% and consists of solid ferroalloys. The continuous injection of gases causes a violent stirring action and intimate mixing of slag and metal, which can lower sulfur values to below 0.005%. The gas contents approach or may be even lower than those of vacuum induction melted steel. The dilution of oxygen with inert gas, argon, or nitrogen causes the carbon-oxygen reaction to go to completion in favor of the oxidation reaction of iron and the oxidizable elements, notably chromium in stainless steel. Therefore, superior chromium recoveries from less expensive high-carbon ferrochromium are obtained compared to those of electric arc melting practices. Argon oxygen decarburization is a secondary refining process that was originally developed to reduce material and operating costs and to increase the productivity in production of chromium-bearing stainless steels. A premelt is prepared in an electric arc furnace by charging high-carbon ferrochrome, ferrosilicon, stainless steel scrap, burned lime, and fluorspar and melting the charge to the desired temperature. The heat is then tapped, deslagged, weighed, and transferred into an AOD vessel, which consists of a refractory-lined steel shell mounted on a tiltable trunnion ring (Fig. 3). As shown in Fig. 3, process gases (oxygen, argon, and nitrogen) are injected through submerged, sidemounted tuyeres. The primary aspect of the AOD process is the shift in the decarburization thermodynamics that is afforded by blowing with mixtures of oxygen and inert gas as opposed to pure oxygen. The heat is decarburized in the AOD vessel to 0.03% C in stages, during which the inert gas-to-oxygen ratio of the blown gas increases from 1-to-3 to 3-to-1. During the blowing, fluxes are added to the furnace and a slag is prepared. Following the decarburization blow, ferrosilicon is added, and the heat is argon stirred for a short period. The furnace is then turned down, a chemistry sample is taken, and the heat is deslagged. Additional alloying ele ments are added if adjustments are necessary,

Steel Melt Processing / 209

Fig. 3

Schematic of an argon oxygen decarburization vessel

and the heat is tapped into a ladle and poured into ingot molds or a continuous casting machine. With the AOD process, steels with low hydrogen (<2 ppm) and nitrogen (<0.005%) can be produced with complete recovery of chromium. In addition to its economic merits, AOD offers improved metal cleanliness, which is measured by low unwanted residual-element contents and gas contents; this ensures superior mechanical properties. The AOD process is duplexed, with molten metal supplied from a separate melting source to the AOD refining unit (vessel). The source of the molten metal is usually an EAF or a coreless induction furnace. Foundries and integrated steel mills use vessels ranging in nominal capacity from 1 to 160 Mg (1 to 175 tons). Although the process was initially targeted for stainless steel production, AOD is used in refining a wide range of alloys, including:





Stainless steels Tool steels Silicon (electrical) steels Carbon steels, low-alloy steels, and highstrength low-alloy steels
High-temperature alloys and superalloys Argon oxygen decarburization units are used in the production of high-alloy castings, particularly of grades that are prone to defects due to high gas contents. Carbon and low- alloy steels for castings with heavy wall sections may be subject to hydrogen embrittlement and are also processed in these units with good results. Fundamentals. In the AOD process, oxygen, argon, and nitrogen are injected into a molten metal bath through submerged, side-mounted tuyeres. The primary aspect of the AOD process is the shift in the decarburization thermodynamics that is afforded by blowing with mixtures of oxygen and inert gas as opposed to pure oxygen. References 4 to 6 contain

Fig. 4

Carbon-chromium equilibrium curves

detailed discussions of the thermodynamics and the dilution principle and its influence on the refining of chromium-bearing steel. To understand the AOD process, it is necessary to examine the thermodynamics governing the reactions that occur in the refining of stainless steel, that is, the relationship among carbon, chromium, chromium oxide (Cr3O4), and carbon monoxide (CO). The overall reaction in the decarburization of chromium-containing steel can be written as: / Cr3 O4 þ C Ð 3/4 Cr þ COðgÞ

14

(Eq 1)

The equilibrium constant, K, is given by: K¼

ðaCr Þ3=4 ðPCO Þ ðaCr3 O4 Þ1=4 ðaC Þ

(Eq 2)

where a and P represent the activity and partial pressure, respectively. At a given temperature, there is a fixed, limited amount of chromium that can exist in the molten bath that is in equilibrium with carbon. By examining Eq 2, one can see that by reducing the partial pressure of CO, the quantity of chromium that can exist in the molten bath in equilibrium with carbon increases. The partial pressure of CO can be reduced by injecting mixtures of oxygen and inert gas during the decarburization of stainless steel. Figure 4 illustrates the relationship among carbon, chromium, and temperature for a partial pressure of CO equal to 1 and 0.10 atm (1000 and 100 mbar, or 760 and 76 torr). The data shown in Fig. 4 indicate that diluting the partial pressure of CO allows lower carbon levels to be obtained at higher chromium contents with lower temperatures. In refining stainless steel, it is generally necessary to decarburize the molten bath to less than 0.05% C. Chromium is quite susceptible to oxidation; therefore, prior to the introduction

of the AOD process, decarburization was accomplished by withholding most of the chromium until the bath had been decarburized by oxygen lancing. After the bath was fully decarburized, low-carbon ferrochromium and other low-carbon ferroalloys were added to the melt to meet chemical specifications. Dilution of the partial pressure of CO allows the removal of carbon to low levels without excessive chromium oxidation. This practice enables the use of high-carbon ferroalloys in the charge mix, avoiding the substantially more expensive low-carbon ferroalloys. Figure 5 compares the refining steps in the two processes. Equally important, however, are the processing advantages offered by submerged blowing. These include excellent slag/metal and gas/ metal contact, superior decarburization kinetics, and 100% utilization of the injected oxygen by reaction with the bath. In addition, submerged injection allows the accurate control of endpoint nitrogen and the removal of dissolved gases (nitrogen and hydrogen) and nonmetallic inclusions. Equipment. The processing vessel consists of a refractory-lined steel shell mounted on a tiltable trunnion ring (Fig. 3). With a removable, conical cover in place, the vessel outline is sometimes described as pear shaped. Several basic refractory types and various quality levels of the refractories have gained widespread acceptance (Ref 7, 8). Dolomite refractories are used in most AOD installations; magnesite-chromium refractories are predominant in small (<9 Mg, or 10 ton) installations and are used almost exclusively in Japan. As seen in Fig. 3, process gases are injected through submerged, side-mounted tuyeres. The number and relative positioning of tuyeres are determined in part by vessel size, range of heat sizes, process gas flow rates, and the types of alloys refined. Process gases are oxygen, nitrogen, argon, and, in some cases, carbon dioxide (CO2). The most recent AOD installations include the use of top-blown oxygen (one-third of all AOD installations have this capability). Oxygen top- and bottom-blowing converters are described in the discussion “Oxygen Top and Bottom Blowing” in this section. The specific volume of AOD vessels ranges from 0.4 to 0.8 m3/Mg. The gas control system supplies the process gases at nominal rates of 0.5 to 3 m3/min/Mg. The system accurately controls the flow rates and monitors the amount of gas injected into the bath to enable the operator to control the process and keep track of the total oxygen injected. Normally, a shop has three interchangeable vessels. At any given time, one of the vessels is in a tiltable trunnion ring refining steel, a second vessel is at a preheating station, and the third vessel is being relined. The vessel in the trunnion ring can be replaced with a preheated vessel in less than 1 h. The control of process gases, vessel activities, and ancillary equipment can range from manual to fully automated. Most installations

210 / Molten Metal Processing and Handling

Fig. 5

Composition changes in refining type 304-L stainless steel using electric arc furnace practice and argon oxygen decarburization (AOD)

Fig. 6

Schematic of stainless steel refining cycle for small (<27 Mg, or 30 ton) vessels showing the relationship between carbon and chromium contents as a function of time and temperature

are equipped with a computer to assist in process control by calculating the required amount of oxygen as well as alloying additions. Some installations have computer control systems capable of sending set points and flow rates to the gas control systems. AOD Processing of Stainless Steels. Charge materials (scrap and ferroalloys) are melted in the melting furnace. The charge is usually melted with the chromium, nickel, and manganese concentrations at midrange specifications. The carbon content at meltdown can vary from 0.50 to 3.0%, depending on the scrap content of the charge. Once the charge is melted down, the heat is tapped, and the slag is removed and weighed prior to charging the AOD vessel. In the refining of stainless steel grades, oxygen and inert gas are injected into the bath in a stepwise manner. The ratio of oxygen to inert gas injected decreases (3:1, 1:1, 1:3) as the carbon level decreases. Once the aim carbon level is obtained, a reduction mix (silicon, aluminum, and lime) is added. If extralow sulfur levels are desired, a second desulfurization can be added. Both of these steps are followed by an argon stir. After reduction, a complete chemistry sample is usually taken and trim additions made following analysis. Figure 6 illustrates the relationships among carbon, chromium, temperature, and the various processing steps for refining a typical type 304 stainless steel using AOD. Another critical aspect of AOD refining is the ability to predict when to change from nitrogen to argon to obtain the aim nitrogen specification. Table 1 lists the process parameters, carbon removal efficiencies, and carbon removal rates for the AOD refining of type 304 stainless steel. The point during refining when the oxygen-to-inert-gas ratio is lowered is based on carbon content and temperature. The ratios and carbon switch points are designed to provide optimal carbon removal efficiency without exceeding a bath temperature of 1730  C (3150  F). AOD Processing of Carbon and Low-Alloy Steels. The refining of carbon and low-alloy steels involves a two-step practice: a carbon removal step, followed by a reduction/heating step. The lower alloy content of these steels eliminates the need for injecting less than a 3:1 ratio of oxygen to inert gas. Once the aim carbon level is obtained, carbon steels are processed similarly to stainless steels. Figure 7 illustrates the carbon content and temperature relationship for the AOD refining of carbon and low-alloy steels. Because the alloy content of these grades of steel is substantially lower than that of stainless steel and because the final carbon levels are generally higher, there is no thermodynamic or practical reason for using an oxygen/inert gas ratio of less than 3:1. Oxidation measurements indicate that all of the oxygen reacts with the bath and that none leaves the vessel unreacted. By monitoring and recording the oxygen consumption during

Steel Melt Processing / 211 Table 1 Processing parameters for argon oxygen decarburization refining of type 304 stainless steel Total gas flow rate: 1.5 m3/min/ton Gas ratios

Partial pressures, atm

Carbon level, %

O2

N2

Ar

Ar

N2

CO

3.0–0.7 1–0.25 0.25–0.12 0.12–0.04 0.04–0.01

4 3 1 1 1

1 1 1 ... ...

... ... ... 3 8

... ... ... 0.83 0.96

0.14 0.20 0.53 ... ...

0.86 0.80 0.47 0.17 0.04

Temperature CRE(a), %

80 65 45 30 15

CRR (b), ppm/min

1200 830 430 133 26



C

1510 1700 1700 1730 1730



F

2750 3100 3100 3150 3150

(a) CRE, carbon removal efficiency. (b) CRR, carbon removal rate

Fig. 7

Schematic of carbon and low-alloy steel argon oxygen decarburization refining

Table 2 Composition control of argon oxygen decarburization-refined high-strength low-alloy steel Element

Aim

Mean

Carbon Manganese Silicon Phosphorus Sulfur Molybdenum Chromium Nickel Aluminum

0.27 0.80 0.50 ... ... 0.28 0.50 ... 0.05

0.264 0.83 0.49 0.019 0.002 0.28 0.56 0.13 0.043

Standard deviation

0.006 0.03 0.04 0.002 0.001 0.01 0.07 0.03 0.009

refining, very close control of end-point carbon is achieved. Because the oxygen and inert gases are introduced below the bath and at sonic velocities, there is excellent bath mixing and intimate slag/metal contact. As a result, the reaction kinetics of all chemical processes that take place within the vessel are greatly improved. Composition and Property Improvement with AOD. It has been well documented that

AOD-refined steels exhibit significantly improved ductility and toughness, along with impact energy increases of over 50% (Ref 9–11). These improved properties result from a decrease in the number and size of inclusions. The capability to produce low-gas-content steel with exceptional microcleanliness, along with alloy savings, is the primary factor for the growth of the AOD process for refining stainless, carbon, and low-alloy steels. Decarburization. In both stainless and lowalloy steels, the dilution of oxygen with inert gas results in increased carbon removal efficiencies without excessive metallic oxidation. In stainless grades, carbon levels of 0.01% are readily obtained. AOD Chemistry Control. The excellent compositional control of AOD-refined steel is indicated in Table 2 for a ten-heat series of highstrength low-alloy steel. The injection of a known quantity of oxygen with a predetermined bath weight enables the steelmaker to obtain very tight chemical specifications. Desulfurization with AOD. Sulfur levels of 0.01% or less are routinely achieved, and levels less than 0.005% can be achieved with single-

slag practice. When extralow sulfur levels are required, a separate slag treatment for 3min is sufficient. Slag Reduction. During oxygen injection for carbon removal, there is some metallic oxidation. Efficient slag reduction with stoichiometric amounts of silicon or aluminum permits overall recoveries of 97 to 100% for most metallic elements. Chromium recovery averages approximately 97.5%, and the recovery of nickel and molybdenum are approximately 100%. Nitrogen Control. Degassing in AOD is achieved by inert gas sparging. Each argon and CO bubble leaving the bath removes a small amount of dissolved nitrogen and hydrogen. Final nitrogen content can be accurately controlled by substituting nitrogen for argon during refining. Nitrogen levels as low as 25 to 30 ppm can be obtained in carbon and low-alloy steels, and 100 to 150 ppm N can be obtained in stainless steels. The ability to obtain aim nitrogen levels substantially reduces the need to use nitrided ferroalloys for alloy specification, and this also minimizes the use of argon. Hydrogen levels as low as 1.5 ppm can be obtained. Oxygen top and bottom blowing (OTB), also referred to as combined blowing, is an extension of AOD technology for refining steel. During OTB, oxygen is injected into the molten steel through a top lance during carbon removal, and inert gases, such as argon, nitrogen, or carbon dioxide, are injected with oxygen through submerged tuyeres or alternate forms of gas injectors such as canned bricks, porous bricks, or thin pipes set in the refractory brick. The top-blown oxygen can react with the bath, reducing refining times, or with CO. The combustion of CO above the surface of the bath increases the thermal efficiency of the refining process and decreases the quantity of silicon and aluminum required for reduction of metallic oxides. If the top-blown oxygen system is designed so that more than 65% of the oxygen reacts with the bath, the system is referred to as hard blown; if less than 65% of the oxygen reacts with the bath, it is referred to as soft blown. As a general guideline, a top-blown oxygen system installed in AOD vessels with a nominal capacity greater than 45 Mg (50 tons) will be hard blown; smaller vessels will be soft blown. In hard-blown top oxygen systems, the flow rate of oxygen through the top lance will be between 50 and 150% of the oxygen injected through tuyeres or injectors. In soft-blown systems, the top oxygen flow rate is between 50 and 100% of the oxygen flow through the tuyeres or injectors. Because foundry AOD vessels have a nominal capacity less than 45 Mg (50 tons), the top-blown oxygen systems are designed to be soft blown. This design maximizes the thermal efficiency of the smaller AOD vessels, reducing the quantity of fuel and reduction material required during refining. Top-blown oxygen flow rates in foundry AOD vessels range from 50% of the bottom oxygen flow rate to 120%.

212 / Molten Metal Processing and Handling Figures 8 and 9 compare OTB with various refining processes. More detailed information on OTB technology can be found in Ref 12 and 13.

Vacuum Oxygen Decarburization (VODC) Vacuum oxygen decarburization can be carried out in a converter vessel (VODC) or in the ladle (VOD). The parameters that favor the choice of ladles as reaction vessels include the following:
Tapping, treatment, and teeming are done in

the same reaction vessel. Thus, there are no temperature losses due to any final transfer of the melt, and the high level of cleanliness achieved during the treatment can be preserved up to teeming.
The use of electric power as an inexpensive energy source permits the highest flexibility in the melt shop. The ladle unit can act as a time buffer between the melting unit and the casting stand. Information on VOD ladle technology can be found in the section “Ladle Metallurgy” in this article. The primary reasons for selecting the converter as the VOD reaction vessel are as follows:
Initial carbon contents are as high as possi-

ble, together with high oxygen blow rates.


Ease of deslagging and strong stirring action
Shop restrictions such as a limited number

of ladles, prohibited use of slide gates, and restricted crane capacity in the casting bay make converter technology more attractive. This section reviews both the equipment and the processing parameters for VOD, which is used for the production of low-alloy steels, stainless steels, and superalloys.

Fig. 8

Equipment. Vacuum oxygen decarburization converters are similar to AOD and OTB converters in terms of design and the tilting device used. Bottom blowing is, however, restricted to the introduction of small amounts of inert gas through simple pipes, thus avoiding the special erosion-resistant refractory material used around tuyeres. Flue gas handling is easier and is incorporated into the vacuum system. In terms of vessel design, the conical converter top is closed by a vacuum hood with an oxygen lance feedthrough and vacuum addition lock (Fig. 10). Because the VODC system is closed and no air enters the vessel, permanent control of the decarburization rate and the carbon level in the bath can be maintained and monitored with a flue gas analyzing device (Ref 14, 15). Pollution control for carbon monoxide (CO) and dust is also incorporated into the system. VODC Processing Parameters. Figures 8 and 11 show the relationship among CO partial pressure, gas blow rates, and carbon and chromium oxidation rates for the VODC, VOD, AOD, and OTB processes. As shown, low chromium losses and high carbon removal rates are possible using VODCs. The high carbon removal rates are due to the higher oxygen yield of the VODC and the fact that oxygen dilution is not required. The amount of bottom-blown inert gas is only approximately 2 to 5% of the oxygen volume. With a specific oxygen blow rate of 0.7 m3/ min/Mg, the decarburization rate exceeds 600 ppm/min at high carbon levels (Fig. 8). In smaller VODC heats, higher blow rates of the order of 10 m3/min (350 ft3/min) are maintained, leading to a carbon removal of 1500 ppm/min in 5 Mg (5.5 ton) units. Figures 8 and 11 clearly demonstrate the low chromium losses and the high ratio of carbonto-chromium removal rate due to the very low

Carbon removal rates and chromium oxidation rates for the vacuum oxygen decarburization converter (VODC), oxygen top blowing (OTB), and argon oxygen decarburization (AOD) processes

CO partial pressure achieved in the VODC process. As in any vacuum process, the lowest carbon levels (from 50 to 500 ppm) are reached in stainless steel heats very quickly (6 to 10 times the carbon removal rate of AOD) without applying excessive heat and excessive amounts of inert gas. As shown in Fig. 8, temperatures of approximately 1650  C (3000  F) are common. Inert gas volumes are generally less than approximately 40 L/min/Mg. Figure 9 shows the relationship between the oxygen blow rates and the inert gas addition for various refining techniques, as well as the resulting chromium and carbon levels achieved at specific process steps. Figure 9 also shows the extent to which decarburization is accompanied by chromium oxidation. Carbon and nitrogen levels below 100 ppm are achieved with corresponding chromium yields of over 95% before reduction and 98% after reduction (Ref 14, 17). Sulfur levels are at 10 to 30 ppm for any grade of steel without any sacrifice in the degassing effect. Hydrogen levels are below 2 ppm. The operating conditions necessary for achieving low levels of sulfur and oxygen are the same as those in all strong stirring converter processes (AOD, OTB, VODC). A disadvantage of the VODC, as with all converter processes, is that gas pickup of the melt is encountered during tapping into the teeming ladle. Optimal cleanliness and castability therefore require a subsequent ladle treatment.

Ladle Metallurgy Ladle metallurgy follows furnace melting or refining in a converter. A primary reactor is emptied into a tapping ladle, and then the steel can be transferred into other ladles, furnaces, or vessels for treatment. Alternatively, the steel can be treated in the tapping ladle before being poured into molds. Ladle metallurgy is used for deoxidation, decarburization, and adjustment of chemical composition, including gases. Refining times in the furnace can often be shortened and production rates increased with an efficient secondary steelmaking practice. Ladle metallurgy also allows the control of teeming temperature, especially required for continuous casting. Steels with as low as 0.002 wt% S can be produced free of oxide or sulfide inclusions. Inclusion shape control is also performed in ladles to improve mechanical properties. There are various types of ladle treatment. Table 3 lists the primary methods and equipment used for ladle treatment and their relative capabilities. These methods can be categorized under four major groups (Fig. 12): stirring, injection, vacuum, and heating processes (Ref 18). Accurate assessment of temperature, chemical composition, and quantity of steel in the ladle must be known for efficient secondary processes. The use of high-quality refractories in the liquid steel handling facilities are desirable to achieve lower oxygen activity when

Steel Melt Processing / 213

Total gas flow Technique

m3/min/Mg

ft3/min/ton

Inert gas, %

Electric are furnace (EAF)

1.2

4.6

Lance bubbling equilibrium (LBE)

2.5

95

1

Oxygen top blowing (OTB)

2.0

77

10

Argon oxygen decardurization (AOD)

1.0

40

20--90

Vacuum oxygen decarburization converter (VODC)

0.7

27

2--5

Vacuum oxygen decarburization (VOD-ladle)

0.4

15

0.5--4

Fig. 9

0

Processing steps for achieving proper chromium and carbon levels in stainless steel using various refining techniques with varying gas blow rates and inert gas additions

steel is contained in ladles or tundishes. Ladle preheating and temperature control is also an integral part of ladle metallurgy. Vacuum treatments can be broadly classified into vacuum stream degassing, ladle degassing, and recirculation degassing (Table 3 and Fig. 12). Vacuum processes are designed to reduce the partial pressure of hydrogen, nitrogen, and carbon monoxide gas in the ambient atmosphere to enhance the kinetics of degassing, decarburization, and deoxidation. Alloy additions can also be made during the vacuum treatment. Alloy additions susceptible to oxidation, for example, iron-niobium, are added after deoxidation. The recovery of expensive alloys is improved by ensuring complete dissolution.

Injection treatment in ladles with inert gas, inert gas/oxygen mixtures, or pure oxygen is practiced. Injection of powdered agents of CaSi, CaC2, CaAl, rare earths, and magnesium with an inert gas carrier, or of submerged cored wire with a prior synthetic slag treatment, allows for simultaneous deoxidation, desulfurization, and sulfide shape control.

Metallurgical Objectives The fundamental objective of melt processing is the removal or addition of specific elements to achieve desired steel properties. Residual solid as well as gaseous elements in steel, such as

sulfur, phosphorus, oxygen, nitrogen, and hydrogen, are present due to thermodynamic and kinetic limitations in the primary and secondary steelmaking. Except for specific grades of steel, a lowering of these elements is preferred. Other elements, such as carbon, silicon, and manganese, can be controlled (lowered or raised) to the desired level. Other alloy additions, such as chromium, nickel, titanium, copper, and so on, may be present as residuals from the scrap used in steelmaking or as intentional additions to manufacture special alloyed steels. Degassing. Liquid steel absorbs gases, which can cause embrittlement, voids, inclusions, and so on in the steel when solidified. Proper melting practice and, to a lesser degree, proper heat treatment can limit the gas content of conventionally melted steel to acceptable levels, regardless of the type of furnace equipment employed. Failure to control oxygen, hydrogen, and nitrogen contents may result in porosity or a serious decrease in ductility or both. Gas content is largely adjusted during the oxygen boil. After the cold charge is melted and the bath is in the temperature range of 1510 to 1540  C (2750 to 2800  F), oxygen is introduced into the molten metal, usually by means of a piping arrangement. The oxygen combines with the dissolved carbon in the steel to form bubbles of carbon monoxide. As the bubbles form, dissolved hydrogen and nitrogen are caught up in the bubbles in much the same way that dissolved oxygen finds its way into bubbles of boiling water. Thus, the bubbles of gas contaminants are boiled out. Oxygen is the principal refining agent in steelmaking and plays a role in determining the final composition and properties of steel and influences the consumption of deoxidizers. The production of rimmed, semikilled, or killed grades of steel essentially refers to oxygen management in liquid steel before solidification. If the oxygen content of molten steel is sufficiently high during vacuum degassing, the oxygen will react with some of the carbon to produce carbon monoxide. The evolved carbon monoxide is removed by the created vacuum, known as vacuum-carbon deoxidation. In undeoxidized steel, the carbon and oxygen contents approach equilibrium at a given temperature and pressure according to: C þ O $ CO ðgasÞ

eq K1600 C ¼

p CO ðatmÞ wt%C wt% O

(Eq 3)

(Eq 4)

where Keq is the thermodynamic equilibrium constant, and pCO is the partial pressure of carbon monoxide. Figure 13 shows the contents of carbon and oxygen in equilibrium at 1600  C (2910  F) as a function of partial pressure of carbon monoxide. As the pressure is lowered by vacuum treatments, more and more carbon reacts with oxygen to establish the new carbon monoxide partial pressure, thus making the oxygen unavailable for

214 / Molten Metal Processing and Handling

Fig. 10

Fig. 11

Schematic layout of a vacuum oxygen decarburization converter

Comparison of CO partial pressure for the argon oxygen decarburization (AOD), vacuum oxygen decarburization converter (VODC), and vacuum oxygen decarburization (VOD) processes

inclusion formation with the added deoxidizers. Strong deoxidizers, such as aluminum, titanium, and silicon, react with oxygen with greater affinity than carbon at atmospheric pressure (Fig. 14), so carbon cannot react with oxygen when vacuum degassed. If not floated properly, oxides of deoxidants can result in inclusions when solidified. Carbon is a stronger deoxidizer than aluminum, titanium, or silicon below a pressure of 0.01 atm (10 mbar, or 7.6 torr). Hydrogen causes bleeding ingots, embrittlement, low ductility, thermal flaking, and blowholes. Hydrogen is usually picked up from the moisture in the charge and the environment and can be lowered by an effective vacuum treatment, according to Sievert’s law: ½%H ¼ KðpH2 Þ1=2

(Eq 5)

The amount of hydrogen dissolved in molten steel, [%H], is related to the partial pressure of

hydrogen gas above molten steel, pH2 . The equilibrium constant K is equal to 0.0027 at 1600  C (2910  F). Hydrogen removal during vacuum degassing is affected by factors such as surface area exposed to vacuum, degassing pressure, steel composition, extent of prior deoxidation, and hydrogen pickup from alloy additions, slags, and refractories. The normal hydrogen content in acid-melted steel is in the range of 2 to 4ppm. Hydrogen content resulting from basic practice is slightly higher. Austenitizing and tempering treatments serve to lower the hydrogen content further in sections of average thickness. However, these treatments may not suffice when very heavy sections are involved. The alternative is to degas the molten steel under vacuum, particularly when pouring heavy castings that will be subjected to severe dynamic loading. Nitrogen is particularly harmful for low-carbon steels intended for drawing applications

and should be lowered as much as possible. Although nitrogen follows Sievert’s law, its removal by argon flushing or vacuum degassing is limited due to the tendency of nitrogen to form stable nitrides. Primary control of nitrogen is attempted during steelmaking practices. The lowest nitrogen contents are obtained with either acid or basic open-hearth practice. Nitrogen contents in electric arc melted steels range from 3 to 10 ppm. Deoxidation and Alloying. Steels are required to meet certain temperature and chemical requirements for proper ladle metallurgy. Toward the end of tapping, some amount of furnace slag is carried over into the ladle. Deoxidation and alloying additions can be made during tapping. However, if ultralow carbon steel grades are to be produced, deoxidation is usually not performed at the time of tapping, since dissolved oxygen is later required to react with carbon. The primary role of deoxidation is to prevent pinholes due to carbon monoxide formation during solidification. Oxygen content should be less than 100ppm to prevent porosity. Silicon and manganese are mild deoxidizers; they are added to stop the carbon boil and to adjust the chemistry. Manganese (>0.6%) and silicon (0.3 to 0.8%) additions are limited by other alloy effects and are normally inadequate alone to prevent pinholes. Aluminum is the most common supplemental deoxidizer used to prevent pinholes. As little as 0.01% Al will prevent pinholes. Aluminum is normally added at the tap—approximately 1 kg/ Mg (2 lb/ton) (0.10% added) with a recovery of 30 to 50% for a final content of approximately 0.03 to 0.05%. This is normally supplemented at the pour ladle with additional deoxidation, which could be more aluminum or calcium, barium, silicon, manganese, rare earths, titanium, or zirconium. More than the minimum amount of deoxidizer required for preventing porosity is needed to maintain sulfide inclusion shape control, but excessive amounts can cause intergranular failures or dirty metal. The second addition at the pour ladle is normally 1 to 3 kg/Mg (2 to 7 lb/ ton). The commonly used elements for deoxidation are (in order of decreasing power) zirconium, aluminum, titanium, silicon, carbon, and manganese. The primary role of aluminum as a deoxidant is to react with the dissolved oxygen and float up as alumina. However, some of the finer alumina particles are retained in the steel as inclusions. Making low-inclusion steels requires further treatment to float the fine oxide particles. Slag carry-over is minimized to control the hydrogen and nitrogen pickup by the steel as well as the phosphorus reversal due to reduction of P2O5 by deoxidizing agents. Several types of devices are available as slag stoppers. Desulfurization. Ladle processes allow for desulfurization by judicious selection of an agent. Oxygen is a deterrent in sulfur removal and therefore should be pre-deoxidized to the lowest practicable oxygen content, preferably with aluminum. Simultaneous deoxidation and

Steel Melt Processing / 215

Fig. 12

Processes for secondary steelmaking. Stirring process: (a) bottom injection and (b) lance injection. Injection processes: (c) powder injection and (d) wire feeding. Vacuum processes: (e) stream degassing, (f) Ruhrstahl-Heraeus degassing, and (g) Dortmund-Horder-Huttenunion degassing. Heating processes: (h) Vacuum oxygen decarburization process, (i) vacuum arc degassing process, and (j) ladle furnace process

Table 3

Relative efficiency of secondary steelmaking processes

0, none; 1, good; 2, better; 3, best Metallurgical functions Steelmaking processes(a)

Stream degassing D-H degassing R-H degassing RH-OB degassing Ladle refining furnace, reheat only Ladle refining furnace, reheat and vacuum degassing Argon oxygen decarburization Argon bubbling, CAS Argon bubbling, CAB Lance powder injection Cored-wire injection REM canister

Composition control

Temperature control

Deoxidation (O2)

Degassing (H2)

Decarburization

Desulfurization

Microcleanliness

Inclusion morphology

1 2 2 2 3 3

0 2 2 2 3 3

1 2 2 2 3 3

3 2 2 2 0 2

1 2 2 3 0 2

0 0 0 0 1 3

1 2 2 2 3 3

0 1 1 1 0 1

1 2 2 1 2 1

2 1 1 0 0 0

2 1 2 2 2 1

1 0 0 0 0 0

3 0 0 0 0 0

2 0 2 3 1 1

1 1 2 2 2 1

1 0 2 3 3 3

(a) D-H, Dortmund-Horder-Huttenunion; R-H, Ruhrstahl-Heraeus; RH-OB, Ruhrstahl-Heraeus oxygen blowing; CAS, composition adjustment by sealed argon bubbling; CAB, capped argon bubbling; REM, rare earth metals

desulfurization is also commonly performed. For specialty steels, the ladle furnace is used to desulfurize the melt after vacuum degassing. Efficiency of desulfurization is lowered by the carry-over of oxidizing furnace slags, since the oxides are unstable during the sulfur removal treatment. In addition, FeO lowers the desulfurization efficiency. Similarly, oxides in the ladle lining can affect the efficacy of sulfur removal. Because desulfurization involves slagmetal reactions, intimate mixing of steel and

desulfurizing agents is a prerequisite. In addition to the previous requirements, the initial and final levels of sulfur are critical according to the law of mass action. Thus, steels with less than 0.006% S are difficult to make. Decarburization. It is difficult to produce steel by conventional steelmaking with <0.03% C. Carbon can be lowered to 0.01% from 0.04 to 0.06% levels by simply exposing it to a vacuum, as shown in Fig. 13. Particular attention must be paid to the amount of carbon

in alloying additions made after decarburization of steel. Sulfur and Phosphorus Control. It has been emphasized that the primary removal of sulfur is effected during ironmaking in the blast furnace, where reducing conditions prevail. However, in steelmaking, direct oxidation at the gas-metal interface in the jet impact zone causes 15 to 25% of dissolved sulfur to directly oxidize into the gaseous phase due to the turbulent and oxidizing conditions existing in the impact zone.

216 / Molten Metal Processing and Handling

Fig. 13

Carbon-oxygen equilibrium relationship in liquid steel at 1600  C (2910  F) for various partial pressures of CO gas above liquid steel. If carbon and oxygen levels are at 0.04 and 0.05 wt%%, respectively, at the start of vacuum treatment (region A), then enough oxygen is available to lower the carbon content to 0.01 wt% at pCO of 0.01 atm. However, for carbon levels greater than 0.05 wt% (region B), an external oxygen source is necessary before vacuum treatment to raise the oxygen level to region O, if 0.01 wt% C is desired in the steel at pCO of 0.01 atm. Note that regions A and B are above and below the CO gas stoichiometric line, respectively.

with carrier gas has been described using a reactor model (Ref 20). Ladle refining to produce ultraclean steels and efficient desulfurization are achieved when steel is treated under a basic, nonoxidizing slag. These synthetic slags have low oxygen potential, low melting point, moderate fluidity, and high solubility for alumina and sulfur. If the synthetic slag used in ladle metallurgy has a high sulfide capacity, sulfur removal can be accomplished to ultralow levels of 10 ppm S. If the activity of alumina in the slag is low, then the deoxidation product alumina can be absorbed in the top slag. Thus, calcium-aluminate-type synthetic slags (typically 50 CaO:7 SiO2:43 Al2O3) can deoxidize and reduce total oxygen in the steel as well as desulfurize, simultaneously. Several empirical relationships have been developed to relate the slag composition to the sulfide capacity (Ref 21). Melting and mixing of slag is as important as composition. Premelted slags are commonly used. Vigorous argon stirring of the metal-slag emulsion is necessary for proper deoxidation and desulfurization. Top slag heating is also performed in practice to prevent a thermal gradient in the slag layer. Alternatively, an exothermic mixture of burned lime, hematite, and aluminum powder is used, which helps lower the sulfur content of steel to 0.003 wt% or less, when strongly stirred with argon. The partitioning of phosphorus between the slag and metal is very sensitive to process conditions in steelmaking. An empirical relationship is used to define the phosphorus partitioning parameter, kps (Ref 22, 23): kps ¼ ð%P2 O5 Þ  ð1 þ ð%SiO2 ÞÞ=½%P ¼ fðð%FeOÞ; BÞ

Fig. 14 Equilibrium relationship between the contents of total oxygen and various deoxidizing elements in a steel bath Regression equations based on operational data are employed to predict the end-point sulfur within acceptable limits (Ref 19): ð%SÞ=½S ¼ 1:42B  0:13ð%FeOÞ þ 0:89

(Eq 6)

where B is the slag basicity defined as the ratio of the weight percents of basic to acidic slag components. For example, (CaO + MgO)/ (SiO2 + Al2O3), (%FeO), and (%S) are FeO and sulfur concentrations in the slag, respectively, and [S] is the sulfur concentration in the metal. Desulfurization and inclusion morphology control are accomplished through injection metallurgy in secondary steelmaking. In essence, a powdered reagent, such as calcium-silicon, CaC2, magnesium, CaO + CaF2, or CaO + Al2O3, is introduced into the liquid steel with the help of a cored wire or a carrier gas. Desulfurization during powder injection

(Eq 7)

where B is the slag basicity; (%P2O5), (%SiO2), and (%FeO) are the respective oxide contents in the slag; [%P] is the level of phosphorus in the metal; and f is the function of FeO and B. The parameter kps is independent of basicity for B > 2.5. The distribution of phosphorus is also linked to the turn-down carbon level. In general, high basicity and low temperature favor dephosphorization. The addition of fluxes, for example, ilmenite, fluorspar, lime, and soon, to lower slag viscosity, also helps to lower the metal phosphorus content. Phosphorus can be essentially captured in the slag during early stages of steelmaking through judicious adjustment of both the blowing practice and hot metal silicon level (Ref 24). Thus, phosphorus control in secondary steelmaking refers to the control of phosphorus reversal occurring through the use of synthetic slags, which is dependent on the activity of P2O5 in the slag and the amount of aluminum and silicon in the metal. The Perrin process was developed for ladle dephosphorization using appropriate synthetic slags. It is important to do a slag-free tapping if synthetic slag-aided deoxidation with aluminum or silicon is carried out. The kinetics of

deoxidation by aluminum in an argon-stirred vessel or a Ruhrstahl-Heraeus degasser has been reported in literature (Ref 25). Inclusion Shape Control. Nonmetallic inclusions that form during solidification depend on the oxygen and sulfur contents of the casting. Deoxidation decreases the amount of nonmetallic inclusions by eliminating the oxides, and it affects the shape of the sulfide inclusions. High levels of oxygen (>0.0012% in the metal) form FeO, which decreases the solubility of manganese sulfides and causes the manganese sulfides to freeze early in solidification as globules. The use of silicon deoxidation alone normally causes the formation of globular (type I) sulfides. Decreasing the oxygen to between 0.0008 and 0.0012% in the metal, through the use of aluminum, titanium, or zirconium, increases the solubility of the manganese sulfides so that they solidify last as grain-boundary films. These grain-boundary sulfide films (type II inclusions) are detrimental to the ductility and toughness of the steel and cause increased susceptibility to hot tearing. If the oxygen content is very low (<0.0008% in the metal), then the deoxidizer forms complex sulfides that are crystalline and form early in solidification. These type III inclusions are less harmful than type II but are more harmful than type I. The actual sulfide shape depends on the type and amount of deoxidizer. There are complex interactions, so care must be taken to avoid porosity and type II sulfides. If a high level of rare earth metals (RE/S > 1.5) is added, then galaxies of sulfides (type IV inclusions) form; these are only for rare earth additions. The most common inclusion modifier is calcium. Methods of addition include injection (see “Ladle Injection Methods” in this article).

Argon Stirring Clean steel melt processing can be carried out at atmospheric pressure without any supplemental reheating by various methods that include ESR, ladle injection methods, and various argon bubbling processes such as AOD, capped argon bubbling (CAB), and composition adjustment by sealed argon bubbling (CAS). The CAB and CAS methods for argon bubbling were developed to make controlled additions to the ladle as well as to improve the steel refining capability. Stirring treatment with argon gas, also known as argon rinsing, can be done to homogenize the bath and to promote decarburization by lowering the carbon monoxide partial pressure. Bubbles agitate the bath and help in alloy dissolution, deoxidation, quick and uniform mixing of alloys, temperature homogenization, adjustment of chemical composition, and partial removal of nonmetallic inclusions. These functions are accomplished by either blowing argon through a refractory-protected lance lowered to within 300 mm (12 in.) of the ladle bottom or by blowing argon through porous refractory plugs inserted in the bottom or sidewall of the ladle.

Steel Melt Processing / 217 Methods include argon bubbling, stirring, trimming, and rinsing:
Argon bubbling often supplements other

secondary steelmaking operations by promoting bulk movement of steel in the ladle for chemical and thermal homogeneity, enhancing the flotation of inclusions, and promoting intimate metal-slag mixing for refining operations, such as desulfurization and deoxidation.
Argon stirring is the most vigorous of the bubbling treatments, injecting the highest flow rates of up to 0.3 m3/min (10 scfm) through liquid steel. Stirring, which usually follows tapping, is used to mix the slag and metal and for temperature and chemical homogenization. The high flow rate breaks through the top synthetic slag layer and creates an opening for alloy and deoxidant additions. Sometimes, radiant heat losses are promoted during stirring to prepare the heat for continuous casting. Carry-over of slag from the converter should be minimized.
Argon rinsing follows the stirring to help float the inclusions into the slag, and the gas flow rate is below 0.15 m3/min (5 scfm). During argon rinsing, the gentle flow rate of argon prevents the formation of new heatradiating surfaces. If the steel composition requires adjustment through ferroalloy additions, it is carried out during argon trimming, which occurs between the stirring and rinsing steps. Just enough gas flow (0.15 to 0.3 m3/min, or 5 to 10 scfm) during the trimming period allows the ferroalloys to mix in the steel and not become lost in the slag layer. The specified rates in this description are somewhat dependent on the ladle size. Stirring is simple purging or rinsing of liquid steel by inert argon gas passed either through a porous plug or a submerged lance (Fig. 15) to achieve bath homogenization and oxide flotation. Deoxidation and desulfurization by enhancing slag-metal reactions can also be achieved by argon purging if a synthetic slag is used. Argon rinsing also helps in lowering the nitrogen and hydrogen content, as well as enhancing alloy dissolution and slag-metal reactions due to stirring effects. Exchange treatments with synthetic (low-FeO) slags (Ref 26) or conditioned (high-CaO) slags (Ref 27) can be carried out during or before tapping, followed by argon rinsing to promote deoxidation, desulfurization, and inclusion modification. Methods for argon rinsing are shown schematically in Fig. 15. Although a stopper-rod arrangement is shown in Fig. 15 for teeming the ladle, use of slide-gate systems is more prevalent now. The CAB process uses a conventional ladle with a cover and requires a synthetic slag over the steel surface after tapping (Fig. 16a) to act as a sponge for the absorption of nonmetallic

Fig. 15

Schematic illustrating methods for using the argon stirring process. (a) The lance method. (b) The porousplug method

Fig. 16

Schematic arrangement of equipment used in argon bubbling processes. (a) Capped argon bubbling process. (b) Composition adjustment by sealed argon (CAS) bubbling process

inclusions. The introduction of argon into the covered ladle through a porous bottom plug stirs the metal vigorously and creates a slagmetal mix, unlike other argon bubbling methods, which require an intact slag layer to protect the melt from oxidation. The cap on the ladle prevents any air from affecting the metal. The slag-metal emulsion is useful in enhancing microcleanliness, chemical homogenization, desulfurization, and deoxidation. The CAS process (Fig. 16b) uses a refractory-lined snorkel, which is lowered inside the melt during argon stirring so the steel inside the snorkel is slag free. This allows the addition of ferroalloys and deoxidizers without any slag interference and therefore is an effective secondary steelmaking process for achieving compositional control.

Ladle Injection Methods Alloy additions or deoxidation reagents are introduced either in a powder form through a

submerged lance with the help of a carrier gas or by encasing the materials in a wire by injection metallurgy. The primary objectives of introducing powdered reagents into liquid steel are for desulfurization, deoxidation, alloy dissolution, and inclusion shape control. Powders are mechanically mixed or premelted and granulated for injection with an inert gas carrier. Submerged cored wire with a prior synthetic slag treatment is the other injection method. If ultralow carbon steel grades are to be produced, deoxidation is usually not performed at the time of tapping, since dissolved oxygen is later required to react with carbon. A wide variety of powders (CaSi, CaC2, CaAl, rare earths, and magnesium) are used in industry for simultaneous deoxidation, desulfurization, and sulfide shape control. Calcium silicide treatment is most prevalent for simultaneous sulfur and inclusion shape control. Lime, alumina, and fluorspar mixtures are also injected to form the synthetic slag. Alloy additions that are usually made through injection include iron-silicon,

218 / Molten Metal Processing and Handling CaCN2, graphite, NiO, MoO2, lead, tellurium, and ferroalloys of iron-boron, iron-titanium, iron- zirconium, iron-tungsten, and so on. Several commercial systems are available for powder injection through a lance as well as for cored-wire feeding. Ultralow levels of sulfur (<0.005%) can be obtained with this technique. Powder injection and wire injection operations are shown schematically in Fig. 17. The basis of reactive metal injection desulfurization is the chemical combination of the reactive metal with sulfur dissolved in the liquid steel. Because the reactive metals are strong deoxidizers, it is necessary to have a low oxygen level in the liquid steel before the start of the injection process. Otherwise, the reactive metal would combine with the dissolved oxygen and not the dissolved sulfur. Aluminum is used to obtain the necessary low oxygen levels. It is also necessary to have a slag present to hold the sulfur removed by the reactive metal. If such a slag is not present, the metal sulfide will be oxidized, and the sulfur will return to the liquid metal. When a powder is fed through the lance in the steel melt, particles are accelerated by the high velocity of the carrier gas upon entering the gas stream. For example, in gas-borne powdered injection with the help of a submerged lance, 0.6 to 1.0 mm (0.02 to 0.04 in.) calcium-silicon particles are injected at a carrier gas pressure of 5 to 15atm. At the lance tip, the gas expands as a result of a pressure drop and a rise in temperature. In most steelmaking applications, jetting usually occurs during the injection, which is in the form of a gas-jet cone. On leaving the nozzle, the particles travel in a straight line downward due to their high momentum, until the jet breaks up into a swarm of rising gas bubbles. The majority of the powder particles are trapped in the rising gas bubbles. If the powder particles are wetted by liquid steel, they can break away and be recirculated in the melt. A slagmetal-gas emulsion thus forms near the top surface. Wire feeding allows the introduction of additions at the bottom of the ladle and therefore a better alloy recovery. Injection metallurgy is also used for addition of deoxidants, such as aluminum, in the form of wire feeding as well as lance injection. The initial capital investment for

Fig. 17

a powder injection installation is much more than that for a wire feeder, but cored wires are more expensive than ordinary metal or alloy powders. Figures 12(c) and (d) show a schematic of the injection processes. The calcium treatment, primarily carried out for sulfur control, also has a modifying effect on inclusion morphology. A complete modification of type II manganese sulfide inclusions takes place when the sulfur content is <0.07%. In practice, 0.5 to 0.7 kg per metric ton of steel is necessary for inclusion shape control (Ref 26). These reagents could be wire fed or powder injected. However, for wire feeding, recovery of calcium is three to six times greater than powder injection. Rare earth metals (REM) are very efficient in morphology control (stringer type to globular type), although they only have a minor role in desulfurization. Rare earth metal additions are usually accompanied by a magnesium addition to provide the agitation needed for uniformly distributing the REM throughout the heat.

Ladle Heating Processes Steadily increasing requirements for steel cleanliness have led producers to adopt several novel refining technologies and process routes, many of which involve increased use of the ladle as a refining vessel. Such procedures require longer holding times in the ladle, which necessitate either increased superheats in the furnace or external heating in the ladle in order to avoid early solidification. Higher superheat, in addition to requiring excessive energy expenditure, can contribute to the problem of dirty steel. The preferred solution is to supply heat to the ladle, maintaining the steel at minimal superheat during refining. The loss in temperature during secondary treatment can be compensated by an additional reheating process. The ladle furnace is normally used to heat up, to hold, and to finish all kinds of metal melts (see the section “Ladle Furnace and Vacuum Arc Degassing (LF/VD-VAD)” in this article). Depending on metal, ladle size, and expected heating rate, electrical power of up to several megawatts may be applied. Heating can be done by graphite electrodes, plasma torches, or induction coils. However, graphite electrodes

Schematics showing powder injection and wire injection methods used for desulfurization. (a) Powder injection. (b) Wire injection

cannot be used for numerous steel grades, especially those with extremely low carbon contents, because of the risk of carbon pickup. High-power plasma torches have been considered as an attractive nonconsumable heat source for ladle design. Prior to use in ladle metallurgy stations, plasma torches had been used primarily for scrap melting and heat loss compensation before and during continuous casting. Ladle heating with plasma torches is used for horizontal and vertical continuous casting. It also has application for ladle treatment in foundries. In principle, the plasma ladle furnace operates with arcs in exactly the same way as the conventional graphite-heated ladle furnace, but with two major differences:
The electrodes produce no carburization.
Arc stabilization with argon results in the

active creation of an inert furnace atmosphere. Depending on the application, the systems that can be operated with direct or alternating current power are equipped with either watercooled metal torches or special graphite electrodes. The plasma arcs are stabilized with argon. Where several torches are employed, the floor electrode can normally be eliminated. Torches can be positioned vertically or at an angle. An inclined torch arrangement also permits contact-free heating with nontransmitted arc. Because arcs can also be operated without melt contact, the plasma ladle furnace can be used for slag treatment, because even electrically nonconducting alloy elements can be melted without contact and superheated under total control. It is therefore well suited for desulfurization. A plasma ladle reheater with a graphite return electrode is shown in Fig. 18. It consists of a refractory-lined ladle cover with one or

Fig. 18

Schematic of a plasma ladle reheater

Steel Melt Processing / 219 more plasma arc torches mounted on top. Also mounted on top of the unit is a return electrode mechanism. The return electrode can be a graphite bar that is immersed in the melt from the top, or it can be another plasma arc torch. In an alternate design, the return electrode consists of a conductive layer of carbon-magnesite bricks (Fig. 19). When the cover is placed on the ladle, a copper plate makes contact with the electrically conducting bricks and serves to conduct the plasma current to the grounded cover. Induction furnaces equipped with a gasstabilized graphite electrode have also been investigated to control oxygen and inclusion levels and enhance desulfurization afforded by the transferred arc plasma. A schematic example of plasma torch heating with argon stirring is shown in Fig. 20 from a joint development effort by the Electric Power Research Institute Center for Materials Production and Maynard Steel Casting Company. Molten steel from the arc furnace is refined by the combined action of superheating (which increases the slag-metal reactivity) and electric currents at the slag-metal interface, which assist the electrochemical reactions. In addition, the entire bath of molten metal is stirred by argon passing through a porous plug to improve chemical and temperature homogeneity. The high heat capacity of the arc allows close control of the molten metal temperature and superheating of the slag cover. Ladle refiners with plasma heating can refine commercial grades of cast steel to very low levels of sulfur and oxygen. Oxygen can be reduced on the order of 30 ppm, with reduction of sulfur to 10 ppm. Additional advantages of ladle refining include reduced melting energy use, precise chemistry control, and improved shop logistics.

Vacuum Ladle Degassing Vacuum degassing processes expose liquid steel to a low-pressure environment to decrease gaseous content. The major gases to be eliminated are oxygen, hydrogen, and nitrogen. The effectiveness of these methods is directly linked to

the surface area of molten steel exposed. As described earlier, hydrogen removal is dependent on the partial pressure and diffusion, whereas oxygen removal primarily relies on the chemical reaction between carbon and oxygen and the partial pressure of carbon monoxide gas. These processes also accomplish composition and temperature control, decarburization, microcleanliness, and inclusion morphology control. Vacuum degassing is accomplished primarily by three methods:
Stream degassing
Recirculation degassing
Vacuum degassing in the ladle

Because heat losses during degassing occur through radiation and absorption of heat by ladles or vacuum vessels, and because heat losses vary for different processes, the tapping temperature of the steel must be adjusted for compensation of temperature losses. Ways of conserving heat during vacuum processing are the retention of a slag layer on the molten steel surface, use of radiation shields, vessel preheating, supplemental heating arrangements, and optimization of processing time. The vacuum in most modern degassing systems is created by steam ejectors installed in series to provide multistage pumping capability. A greater number of pumping stages are used for low-carbon steels in the partly deoxidized condition than for deoxidized high-carbon steels when treated at the same rate. A reheating-type vacuum induction melting process uses up to six stages, whereas a recirculating-type Ruhrstahl-Heraeus (R-H) or Dortmund-Horder-Huttenunion (D-H) degasser uses five. A four-stage steam ejector is common for other vacuum degassing systems. Moreover, vacuum degassing processes allow the addition of deoxidizers and alloy additions to the molten steel while under reduced pressure. These additions are optimized for the capability of the system to manage the heat and mixing. Stream degassing methods include ladle-toladle, ladle-to-mold, and tap degassing. Ladle-

to-ladle degassing achieves hydrogen removal as well as oxygen removal by carbon deoxidation. A preheated teeming ladle with either a slide-gate mechanism or a nozzle and stopperrod arrangement is placed inside the tank, and the opening in the tank is covered with an aluminum disk. The tank is evacuated to 0.01 kPa (0.1 mbar, or 0.1 torr), and a superheated, partly deoxidized steel in the tapping ladle is placed on top of the evacuated tank. The stream of molten steel from the tapping ladle melts the aluminum disk cover and flows into the teeming ladle while other deoxidants are added through the addition hopper under vacuum. The pressure stabilizes at  0.11 kPa (1.1 mbar, or 0.8 torr) during degassing. Nitrogen is used to bring the system back to atmospheric pressure after degassing. In the ladle-to-mold degassing, an ingot mold is placed on a base, fitted with a hot top, and preheated using burners. The vacuum tank is placed around the base and makes a vacuumtight seal. A transfer or “pony” ladle, fitted with stopper rod and nozzle assembly, is placed in line with the mold inside the tank. A preheated refractory collar is fitted into an opening in the top of the tank to direct the pouring stream from the pony ladle nozzle into the mold. The tank is evacuated to 0.01 to 0.03 kPa (0.13 to 0.27 mbar, or 0.10 to 0.20 torr), and the tapping ladle containing the superheated steel is placed above the pony ladle for bottom pouring of steel. Once the pony ladle is two-thirds full, the stream of steel is allowed to enter the tank through the collar. The stream breaks up into tiny droplets after entering the vacuum chamber and attains a very large surface area, enhancing the degassing process. Several tapped steel heats can be vacuum cast into large ingots by this method. Nitrogen or air is allowed into the tank to break the vacuum.

Fig. 20 Fig. 19

Schematic of conductive ladle and cover with carbon-magnesite bricks as the counterelectrode

Direct-current plasma ladle refiner with argon flowing through the electrode to form and stabilize the plasma arc

220 / Molten Metal Processing and Handling Tap degassing consists of a special ladle with a flange at the top and an O-ring seal plate for the nozzle opening at the bottom, a ladle cover, and a stoppered tundish having an aluminumdisk seal just below the nozzle. The nozzle is sealed, and the special ladle is covered with the lid. An alloy addition hopper is inserted through the ladle cover. The preheated stoppered tundish is placed on top of the ladle cover, and the entire assembly is positioned such that the tapping stream from the steelmaking furnace can be captured in the tundish. The ladle is evacuated to 0.03 to 0.07 kPa (0.3 to 0.7 mbar, or 0.2 to 0.5 torr), and the stopper rod of the tundish is pulled when the tundish is full with the superheated steel tapped from the furnace. The required deoxidizers are added through the hopper during degassing. After degassing, the ladle is restored to atmospheric pressure and teemed. Recirculating degassing methods use atmospheric pressure to force the liquid steel from a ladle into an evacuated chamber, where it is exposed to low pressure and then flows back into the ladle. The metal can recirculate 40 to 50 times to achieve degassing, where the vacuum is used to recirculate the metal stream as well as to degas. A refractory-lined vacuum vessel with a “snorkel” tube, an electric-resistance graphite heating rod, an alloy addition hopper, a mechanism to lower or raise the vacuum vessel or ladle, and a steam-ejector system comprise the D-H degassing method (Fig. 12g). The vacuum vessel is internally preheated to the desired temperature, and the alloying additions are placed in the hopper. The superheated steel is tapped in an unoxidized or partially oxidized condition in a ladle and placed beneath the vacuum vessel. Either the ladle or the vacuum vessel are moved vertically. The snorkel is immersed in the molten steel, and the vacuum vessel is evacuated, causing the steel to be sucked into the vessel. Once the steel is degassed, the snorkel is raised, causing the steel to partially run back into the ladle. The gases are sucked out of the vessel at this stage. Several cycles of this operation degas the steel to the desired level. Alloy additions are subsequently made before the vacuum vessel is returned to atmospheric pressure. Some snorkels are designed to allow the passing stream of steel to be exposed to argon or oxygen. Two legs or snorkels are used in the R-H process, where one leg is fitted with a gas-inlet pipe (Fig. 12f ). The preheated vessel, similar in construction to the D-H degasser, is lowered in the molten steel ladle containing the superheated steel until 150 mm (6 in.) of both the legs are immersed. The vacuum chamber is evacuated, and an inert gas is introduced through the gas-inlet pipe. The lower density of the gas-liquid mix causes the steel to rise and flow into the vacuum vessel. The metal is deoxidized and returned to the ladle through the other leg under gravity. Several cycles of this process at a pressure of 0.04 to 0.08 kPa (0.4 to 0.8 mbar, or 0.3 to 0.6 torr) degas steel

to the desired level before conventional casting. A variation of the R-H process, known as the Ruhrstahl-Heraeus oxygen blowing (RH-OB) process, is used primarily to remove hydrogen and carbon from steel. The difference between the R-H and the RH-OB processes lies in the gases injected into the system during the process of degassing. The RH-OB process also introduces oxygen, in addition to an inert gas, into the system to enhance decarburization. Inert gas also helps in stirring the melt. Vacuum ladle degassing processes include induction stirring and vacuum oxygen decarburization (VOD). In the induction stirring process, induction coils are used to induce eddy currents in the molten steel to produce a stirring effect. Steel is contained in a ladle with shell sections made of nonmagnetic austenitic stainless steel and lined with high alumina or basic refractory. Superheated, fully deslagged, nondeoxidized steel is tapped at a relatively higher temperature than other vacuum degassing processes. The ladle is placed inside the induction coil positioned inside the vacuum chamber, a heat radiation and splash shield is placed on the vacuum tank, and the chamber is evacuated slowly to 0.01 kPa (0.1 mbar, or 0.1 torr). During degassing, the induction coils are energized to stir the steel. Alloy additions, primarily ferrosilicon, aluminum, and carbon, are added through a hopper toward the end of degassing. Stirring is continued for a few minutes to mix the alloys. After the vacuum tank is returned to atmospheric pressure, a synthetic slag is placed on top of the steel surface to conserve heat and prevent reoxidation. The VOD method was designed for the production of stainless steel. A stainless steel heat is prepared by charging and melting ferrochromium, stainless steel scrap, ferrosilicon, burned lime, and fluorspar and then heating it to the tapping temperature. The stainless steel is tapped in a preheated basic-lined ladle equipped with porous plugs for blowing argon. The ladle is placed inside a vacuum chamber and evacuated to low pressures while oxygen is lanced above the steel bath and argon is bubbled through the bottom plugs in the ladle (Fig. 12h). After the oxygen blowing period, further decarburization and deoxidation can be accomplished by additional vacuum time. Deoxidants could also be added during vacuum if desired. A desulfurizing slag can be added to the heat after the vacuum treatment, enabling the removal of up to 50% of the initial sulfur. Argon blowing through bottom plugs can be replaced with induction stirring.

The ladle, containing argon injection or induction coils, can be placed into an independent vacuum vessel (Ref 28), or, if argon injection/ induction coils are needed to stir the melt (Fig. 21), a vacuum ladle with a directly applied vacuum hood can be used. The vacuum vessel can be either stationary with a vacuum connection (Fig. 22a) or movable on a buggy (Fig. 22b). If additional heating of the melt is needed in the ladle, ladle furnaces or vacuum arc degassing are used (see the section “Ladle Furnace and Vacuum Arc Degassing (LF/VD-VAD)” in this article). Plasma arc torches are also used for heating during ladle refinement. The vessel is equipped with a ladle stool for quick and safe placement of the ladle into the vessel. The vessel also has an inert gas connecting device for the ladle. The ladle has the required freeboard (Fig. 23) and is covered by a ladle hood, which retains metal splashes and heat. There is enough space left below the ladle bottom to receive the molten metal in the event of a breakthrough of the ladle. The vacuum vessel cover is equipped with the necessary control devices for adding slag constituents and alloying and cooling material and for conducting temperature measurement, sampling, and observation to ensure uninterrupted molten metal processing. For optimal metallurgical results, all of these activities must be executed without breaking the vacuum. Devices are also available for the addition of bulky cooling material and for the injection of aluminum, calcium, calcium-silicon, graphite, and ferroalloys. Vacuum degassing is used to melt heats ranging in weight from 15 to 350 Mg (16.5 to 390 tons). Its success depends on treatment

Vacuum Ladle Degassing (VD) Static ladle vacuum degassing technology has developed from argon injection devices and the progress in ladle metallurgy with porous plugs, tuyeres, refractory material, heating technologies, slide gates, and so on. Two basic setups can be used during ladle degassing in vacuum.

Fig. 21

Flow pattern of inductive and inert gas injection stirring in ladles. Left: regular flow (no particle agglomeration). Right: irregular flow (increased particle agglomeration)

Steel Melt Processing / 221 reactions. Thus, chromium is protected against oxidation. This applies, to a limited extent, to manganese and silicon if the oxidation rates are kept sufficiently low. Modern VOD plants consist of the following equipment:
Basic-lined ladle with a freeboard of 1 to













Fig. 22

Schematics of (a) stationary and (b) movable vacuum vessels

time, which in turn depends on the previous soaking of the ladle lining and the surface-tovolume ratio of the melt. The mean temperature decrease during vacuum treatment varies from 6  C/min (11  F/min) for 15 Mg (16.5 ton) heats to 0.7  C/min (1  F/min) for 300 Mg (330 ton) heats during the time the ladle lining is in thermal equilibrium. It has not been possible to achieve satisfactory degassing results with the presence of slag. However, this has changed because of:
The progress made in the fabrication of

basic linings and in the corresponding lining and heating technology
The availability of low-silica active slag mixtures
The development of vacuum technology that prohibits foaming of the slag and permits the injection of large quantities of inert gas up to 10 L/min/Mg to optimize stirring Such improvements have led to simultaneous desulfurization rates of 95%, hydrogen removal of over 85%, nitrogen removal of 60%, and aluminum contents below 200 ppm. Figure 24 (Ref 29, 30) shows the removal of sulfur, hydrogen, and nitrogen as a function of gas circulation rate for a 180 Mg (200 ton) VD treatment. Vacuum treatment results for several low-alloy steel heats are given in Table 4. Vacuum Degassing Stirring Procedures. There are two vacuum degassing stirring procedures. In the first, hydrogen, nitrogen, and sulfur are eliminated by strong stirring, that is, with 5 to 10 L/min/Mg. In this case, the oxygen level is decreased only to 30 to 50 ppm total

oxygen, with the major portion of this being the suspended oxidic inclusions. In the second procedure, the total oxygen level is reduced to 10 to 25 ppm by a soft stirring phase in combination with a previously conducted aluminum deoxidation procedure. This requires a very low stirring rate, a strongly basic slag, and sufficient temperature reserve for an extended treatment. The distribution of stirring energy should be uniform. For strong stirring, inductive stirring combined with inert gas stirring is required. Inductive stirring alone is sufficient for gentle stirring if the flow pattern is irregular enough to create turbulences that will produce inclusion agglomeration. The differences in the flow patterns created by inductive and inert gas stirring are shown in Fig. 21. Vacuum oxygen decarburization differs from vacuum degassing, which is conducted under reducing conditions, in that decarburization is an oxidizing process that not only increases dissolved oxygen content in the melt by solid oxygen carriers but also introduces gaseous oxygen through a top-blowing lance that permits higher decarburization rates. The VOD process is primarily used for decarburizing stainless steel melts or other high-chromium-bearing alloys; however, a VOD plant can be used for all types of steel grades and alloys when decarburization, degassing, desulfurization, or chemical heating with oxygen is required (Ref 31–33). Chemical heating with oxygen is also called vacuum oxygen heating. The VOD process is based on the fact that decarburization is strongly favored under reduced pressure with respect to other oxidizing

1.5 m (3 to 5 ft), a porous plug in the bottom for inert gas injection, a slide gate, and refractory-lined ladle hood tightly sealed on the ladle Vacuum vessel with ladle stool, an inner lining, a safety basin for receiving a metal breakthrough, a vacuum connection, and a vessel cover with accessories and watercooled splash protection Top-blowing oxygen lance Probing device for in-process temperature measurement and sampling Movable guide pipe for feeding either aluminum wire or wire filled with such materials as calcium, calcium-silicon, titanium, graphite, and fluxes Observation port with television camera Vacuum pump set (steam ejectors, water ring pump, or mechanical pumps) with a suction capacity such that during oxygen blowing the vacuum is held below 10 kPa (100 mbar, or 75torr), and oxidizing as well as reducing vacuum treatments can be performed below 0.1 to 1 kPa (1 to 10 mbar, or 0.75 to 7.5 torr). The pump must also be adjusted to handle carbon monoxide gas safely and to remain operational under a heavy dust load of the carried gas of up to 0.4 kg/m3 (0.4 oz/ft3).

Oxygen blow rates vary from 5 to 25 m3/min (175 to 900 ft3/min), depending on heat size and initial carbon content. The specific blow rate ranges from 0.2 to 1.4 m3 O2/min/Mg, the highest specific blow rates are achieved with the smallest heat sizes. The inert gas blow rates vary strongly with the process and the desired results; that is, absolute rates range from 0.1 to 1.2 m3/min (3.5 to 45 ft3/min), and specific rates are 2 to 7% of oxygen flow. According to equilibrium data for the Fe-Cr-C-O system, the lower the chromium losses, the lower the partial pressure of the formed CO + CO2 (Fig. 4). However, the main oxygen-blowing period of the VOD process is performed at an intermediate vacuum pressure range of 3 to 8 kPa (30 to 80 mbar, or 20 to 60torr), which is governed by the suction capacity of the installed vacuum pump and the freeboard available in the ladle. The chromium loss at given carbon levels and temperatures depends on the prevailing kinetic conditions at the blowing spot. With an intense inert gas stirring from the ladle bottom, the chromium loss can be reduced to nearly zero by the strong stirring VOD process, as described in Ref 34. In most VOD plants, the chromium yield at the end of blowing is 90 to 95% and is increased to 98 to 99% during the subsequent reducing phase by the combination of low

222 / Molten Metal Processing and Handling vacuum pressure and the addition of reducing agents such as aluminum and silicon. The yield of chromium also depends on the initial and the final carbon contents of the melt. The limited freeboard of VOD ladles and the strongly exothermic effect of the process dictate that the starting carbon content and the oxygen blow rate be limited. A typical sequence of operations during a VOD treatment is shown in Fig. 25. Achievable results reported from several plants that use VOD capability are summarized in Table 5.

Ladle Furnace and Vacuum Arc Degassing (LF/VD-VAD)

Fig. 23

Freeboard requirements in secondary metallurgy

Fig. 24

Degree of refining as a function of the circulation rate in a 185Mg (205 ton) vacuum degassing plant. Source: Ref 29, 30

As noted in the section “Ladle Heating Processes,” secondary metallurgy techniques are energy-consuming, and any extended ladle treatment leads to heat losses. As a result, the goal has always been to perform such operations as vacuum degassing, desulfurization, and alloying with simultaneous heating so that the ladle treatment could be carried out with no evident heat loss. Simultaneous heating and degassing is difficult and previously was possible only by vacuum induction degassing or partially by vacuum arc degassing (VAD); however, the combination of an arc-heating ladle furnace (LF) with a vacuum degassing (VD) station meets operational requirements and produces excellent metallurgical results, as is summarized in this section. For vacuum vessels, there are two principal designs. The first is the stationary vessel, and the second is the track-mounted movable vessel. The stationary vessel is enclosed in a mobile vacuum cover or a swiveling ladle furnace cover. For tandem stationary vessels, each is enclosed in a moving vacuum cover, while both are served by a common swiveling ladle furnace, a common vacuum pump, and common alloying equipment. This design permits brief tap-to-tap times of approximately 30 to 45 min. A schematic of a combined LF/VD plant is shown in Fig. 26. The movable vacuum vessel moves with the vacuum ladle car to several treatment platforms that contain the vacuum cover, arc-heating ladle furnace, plasma torches (in the case of ladle plasma heating units), injection stands, and alloying and wire feeding station. There are single-stand layouts that permit both arc heating and vacuum degassing in addition to other secondary metallurgy treatments, such as alloying, injection, and stirring, at the same treatment station. The movable vacuum vessel is transported to this treatment station, where arc heating and vacuum degassing can be performed by ladle furnace vacuum degassing or vacuum arc degassing. For combined ladle furnace and vacuum degassing, arcing is carried out at the pressure created by a pollution-control device; this underpressure is 0.5 kPa (5 mbar, or 3torr). The protective atmosphere is created under a partially closed vessel lid, leaving only the openings for three electrodes, alloying, sampling, process control, and flue gas suction. The atmosphere composition is

Steel Melt Processing / 223 Table 4

Parameters and results of vacuum ladle degassing of low-alloy steel

EAF, electric arc furnace; see Fig. 23 for explanations of other abbreviations. Parameter

Initial hydrogen, ppm Final total hydrogen, ppm Initial nitrogen, ppm Final nitrogen, ppm Oxygen activity, ppm Final total oxygen, ppm Total aluminum, ppm Dissolved calcium, ppm Initial sulfur, ppm Final sulfur, ppm Temperature loss,  C (  F) Alloying additions, kg/Mg (lb/ton) Slag additions, kg/Mg (lb/ton) Total amount of slag, kg/Mg (lb/ton) Argon volume, m3/Mg Duration of treatment, min

EAF/VD, 35–40 Mg (38–44 ton)

LF/VD, 40–42 Mg (44–46 ton)

LF/VD, 75–80 Mg (82–88 ton)

3–14 0.4–1.0 100–150 30–60 5–10 20–25 300–450 ... 100–200 40–50 50–60 (90–110) 2–6 (4–12) 20–25 (40–50) 20–25 (40–50) 0.03–0.05 12–16

5–10 0.6–1.5 120–150 50–100 ... 15–25 200–400 ... 300–400 100–200 30–40 (55–70) 0–0.5 (0–1.0) 0 (0) 20–25 (40–50) 0.02–0.03 15–16

5–6 1.0–1.5 80–120 45–90 ... 20–40 100–300 ... 130–200 20–30 30–40 (55–70) 0.5–1.0 (1.0–2.0) 1 (2) 18 (30) 0.15–0.18 20–30

BOP/VD, 180–190 Mg (200–210 ton)

3–7 0.9–1.6 50–70 25–35 1–2 10–20 100–400 20–30 150–200 5–10 35–50 (65–90) 1–3 (2–6) 7–9 (14–18) 15–22 (30–44) 0.12–0.15 25–30

step at 60 to 90 kPa (600 to 900 mbar, or 450 to 700 torr), followed by a vacuum procedure carried out at below 100 Pa (1 mbar, or 0.7 torr) with the electrodes lifted. Ladle Furnace Design. The basic design and equipment of any arc-heated ladle furnace are similar to those of electric arc melting furnaces, except that no tilting movement is required, and ladle furnaces are frequently in a gantry design (Fig. 27). Electrode regulation is accomplished by either electromechanics or hydraulics. The hydraulic fluid is a water emulsion, water glycole, or oil. Direct-current ladle furnaces are not manufactured, but if very small ladle furnace units are required, a monophase alternating-current design with two electrodes is available. Stirring, by inductive coils or inert gas, is necessary in any ladle furnace design. However, the ladle furnace has requirements that are very different from those of the electric arc melting furnace. Ladle furnace requirements include the following:
Controls

governing electrode movement must be more sensitive because of the inert gas bubbling through the ladle bottom.
The arc must be immersed in the foaming slag; this requires short arcs and therefore a low voltage, a low voltage- to-current ratio, and a low total inductivity.

Fig. 25

Typical sequence of operations during vacuum oxygen decarburization (VOD) ladle treatment. VCD, vacuum carbon deoxidation; VD, vacuum degassing

controlled by the carbon monoxide formed by the reaction of graphite electrodes with oxidic slag and the inert gas introduced through porous plugs in the ladle bottom or under the ladle furnace roof. For vacuum treatment, the electrodes are lifted above the vessel lid, and the lid opening is closed by a vacuum cover. When vacuum arc degassing is used, complete protection of the electrodes is provided by telescopic water jackets and inflatable top seals that permit simultaneous arcing, vacuum treatment, and inert gas stirring. However, arcing under reduced pressure is restricted to approximately 30 kPa (300 mbar, or 225 torr).

Below this pressure, the gas atmosphere would be ionized, resulting in severe damage to the equipment. It should be noted that 30 kPa (300 mbar, or 225 torr) arcing is feasible only if the arc voltage is kept below 100V and the electrode regulation is reliable. The arc length increases with the arc voltage and with the reduced pressure. Therefore, pressures are not applied below 80 kPa (800 mbar, or 600 torr), because the extreme arc length cannot be fully immersed in the slag. Submerged arcing is one precondition for the high thermal efficiency of this process. Therefore, the VAD process is also divided into two steps: one arcing

The refractory wear of immersed arcing is concentrated at the wall/slag zone. It is proportional to the amount of specific arc power brought into the slag and the intensity of slag cooling during treatment. For ladle furnaces, secondary voltages of 180 to 280 V, combined with a 2 to 3 mO furnace plus transformer reactance, lead to arc voltages below 100 V. The resulting relatively high current density is 10 A/mm2 (6500 A/in.2) in rigid bars and pipes, 7 A/mm2 (4500 A/in.2) in flexible water-cooled cables, and 0.40 A/mm2 (250 A/in.2) in graphite electrodes. The response time resulting from electrode control is defined as the time required to dissolve a short circuit by lifting the electrode approximately 10 mm (3/8 in.). This response time may vary between 100 and 400 ms, according to the electrode control system applied. This value is important because shorter response times result in less carbon pickup and higher thermal yields. The heating rate is also affected by heat transfer, particularly during the first 10 to 30 min of treatment. Heat losses from the melt to the different furnace components are illustrated in Fig. 28, which shows that the heating rate depends on the heat flow from the arc (area 1) to the different parts surrounding the melt (area 2). Even in the case of thermal equilibrium with the lining (area 3), permanent thermal and electrical losses (area 4) are present such that the best thermal yield under a steady state of heat transfer is only 50 to 60%. The thermal balance reached at a near steady state for arc heating (VAD/LF) and induction heating (vacuum induction melting and degassing, or VIM/VID) is shown in Fig. 29. Figure 30 shows

224 / Molten Metal Processing and Handling Table 5 Process parameters and metallurgical results in vacuum oxygen decarburization (VOD) ladle plants EAF, electric arc furnace; see Fig. 23 for explanations of other abbreviations. EAF-AOD-VOD Parameter

Chromium, % Nickel, % Molybdenum, % Initial carbon, % Final carbon, ppm Initial nitrogen, ppm Final nitrogen, ppm Chromium yield, % Titanium yield, % Manganese yield, % Oxygen blow rate, m3/min (ft3/min) Specific oxygen rate, m3/min/Mg Inert gas rate, m3/min (ft3/min) (ft3/min) Ar/O2 ratio, % Ladle freeboard, m (ft) Slag addition, kg/Mg (lb/ton) Alloy addition, kg/Mg (lb/ton)

EAF/VOD, 15 Mg (16.5 ton)

EAF/VOD, 27 Mg (30 ton)

EAF/VOD, 30 Mg (33 ton)

LF/VOD, 40 Mg (44 ton)

18 10 ... 1.1–1.9 200–600 ... ... ... ... ... 5–8 (175–280) 0.53

18 ... 2 0.6–1.1 80–180 300–430 40–80 97.6 40–60 60–80 5–14 (175–495) 0.52

18 11 2 0.8–1.0 300–400 ... 140–240 96.6 74.5 85.0 5–10 (175–355) 0.34

12 ... ... 0.5–0.6 50–100 150–200 50–85 96.3 ... ... 5–15 (175–530) 0.37

18 10 ... 0.6–0.8 10–20 400–800 50–70 99.2 ... 70.0 10–15 (355–530) 0.23

17 ... ... 0.7–0.9 10–30 300–600 50–80 99.0 ... 80.0 10–16 (355–565) 0.25

0.03

0.05–0.25

0.05–0.10

0.1–0.2

0.3–0.6

0.4–0.7

(1.0) 0.5 0.8–1.0 (2.5–3.0) 20–25 (40–50) 12–20 (24–40)

(1.75–9.0) 0.3–5.0 1.0–1.2 (3.0–4.0) 10–12 (20–24) 10–15 (20–30)

(1.75–3.5) 0.5–2.0 1.2–1.5 (4.0–5.0) 30–40 (60–80) 15–25 (30–50)

(3.5–7.0) 0.7–4.0 1.5 (5.0) 10 (20) 5–8 (10–16)

(11–21) 2.0–6.0 1.6 (5.2) 25 (50) 10–12 (20–24)

(14–25) 2.5–7.0 1.6 (5.2) 25 (50) 7–10 (14–20)

65 Mg (72 ton)

65 Mg (72 ton)

Equipment and Processing. Vacuum induction degassing units consist of a simple induction furnace, designed for vacuum operation, that is equipped with a bell-type vacuum cover (Fig. 31). The furnace crucible is either rammed or lined with brick and has a porous plug in the bottom. A frequency converter power supply provides the energy for the induction coil to control the temperature of the melt. With the vacuum cover removed, the furnace is charged with either liquid or solid. For vacuum treatment, the vacuum cover is placed on the furnace body by crane. A mechanical pumping system is used to achieve an operating pressure of approximately 0.01 to 0.1 kPa (0.1 to 1 mbar, or 0.075 to 0.75 torr). The process is primarily applied to tool and die steels, bearing steels, chromium steels, nickel-chromium steels, and nickel-base alloys to remove dissolved gases such as hydrogen and nitrogen and to remove oxide inclusions from the melt. Figure 32 shows the results of VID treatment on the hydrogen, nitrogen, and oxygen contents of a die steel. In Fig. 32(a), the use of VID lowered the hydrogen content from more than 2.2 ppm to 0.60 ppm (hydrogen removal of 50%). With proper metallurgical treatment, the nitrogen content can also be decreased. Figure 32(b) shows that the nitrogen content was lowered from more than 370 ppm to approximately 50 ppm (nitrogen removal of 50%). With VID, oxide inclusions can be removed by applying metallurgically active slags as an absorbing buffer. In addition, VID units are equipped with argon bubbling systems, which accelerate and improve the removal of nonmetallic inclusions. Changes in oxygen content as a result of VID, with and without argon bubbling, are shown in Fig. 32(c). The final total oxygen content when using VID with argon bubbling is 30% lower than VID without argon bubbling, which demonstrates the benefits of soft argon rinsing.

ACKNOWLEDGMENTS Contents adapted from:

Fig. 26

Schematic of a combined ladle furnace (LF) and vacuum degassing (VD) plant. VOD, vacuum oxygen decarburization

a typical treatment cycle and associated parameters for a 75Mg (80 ton) VAD-heated melt. Independent of thermal yield, the physical boundary for heat transfer by arc, plasma, electron beams, and oxygen gas streams is that the melt evaporates at the heat-transfer spots. Table 6 (Ref 28, 35, 36). compares the various types of heating processes described in this section, in addition to vacuum induction degassing, plasma heating, and vacuum oxygen heating, which are described elsewhere in this article.

Vacuum Induction Degassing (VID) Thermal balance is extremely critical for the secondary metallurgical treatment of small

heats. In most cases, excessive chemical heating through aluminum additions is the only method of compensating for temperature losses during the secondary metallurgical treatment of melts below 15 Mg (16.5 tons). This technique, however, has various disadvantages, such as increased aluminum oxide (Al2O3) inclusion content in the final product and the high equipment costs associated with aluminum heating and oxygen (O2) blowing. In such cases, VID, a new secondary steelmaking technology, also permits economical vacuum treatment for small heat sizes ranging from 1 to 15 Mg (1.1 to 16.5 tons). Table 7 compares the metallurgical applications of VID with other heating and melting operations.


W. Burgmann, Vacuum Ladle Degassing

(VD), Leybold AG, West Germany


G. Kienel, Degassing Processes (Converter

Metallurgy), Casting, Vol 15, ASM Handbook, ASM International, 1988, p 426–431
B. Mishra, Steelmaking Practices and Their Influence on Properties, Metals Handbook Desk Edition, 2nd ed., ASM International, 1998

REFERENCES 1. J.P. Bennett, K.-S. Kwong, G. Oprea, M. Rigaud, and S.M. Winder, Performance of Refractories in Severe Environments, Corrosion: Materials, Vol 13B, ASM Handbook, ASM International, 2005

Steel Melt Processing / 225

Fig. 27

Schematic of a vacuum arc degassing plant layout. 1, transformer; 2, bus bars; 3, flexible cables; 4, electrode arms; 5, temperature-measuring and sampling device; 6, alloying lock; 7, electrode housing; 8, electrode masts; 9, operating platform; 10, lifting device for platform; 11, vacuum cover; 12, ladle heat shield; 13, electrodes; 14, slag; 15, melt; 16, argon plug; 17, vacuum vessel; 18, vessel drive; 19, emergency pit

4.

5.

6.

7.

Fig. 28

Thermal flow pattern of a ladle furnace. 1, arc; 2, melt; 3, slag, electrode tip, and lining; 4, vessel, cover, busbars, and environment

2. V.D. Eisenhu¨ttenleute, Ed., Slag Atlas, 2nd ed., Verlag Stahleisen, GmbH, D-Du¨sseldorf, Germany, 1995 3. M. Blair and S. Thomas, Steel Casting Handbook, 6th ed., Steel Founders Society

8.

9.

of America and ASM International, 1995, p 14–2 F.D. Richardson and W.E. Dennis, Effect of Chromium on the Thermodynamic Activity of Carbon in Liquid Iron-Chromium-Carbon Metals, J. Iron Steel Inst., Nov 1953, p 257–263 R.J. Choulet, F.S. Death, and R.N. Dokken, Argon-Oxygen Refining of Stainless Steel, Can. Metall. Q., Vol 10 (No. 2), 1971, p 129–136 R.B. Aucott, D.W. Gray, and C.G. Holland, The Theory and Practice of the ArgonOxygen Decarburizing Process, J. W. Scot. Iron Steel Inst., Vol 79 (No. 5), 1971–1972, p 98–127 D.A. Whitworth, F.D. Jackson, and F.F. Patrick, Fused Basic Refractories in the Argon-Oxygen Decarburization Process, Ceram. Bull., Vol 53 (No. 11), 1974 D. Brosnan and R.J. Marr, “The Use of Direct Burned Dolomite Brick in the AOD Vessel,” Paper presented at the Electric Furnace Conference (Detroit, MI), Iron and Steel Society, Dec 1977 P.A. Tichauer and L.J. Venne, AOD, A New Process for Steel Foundries, Promises Better Properties, Less Repair, 33 Magazine, April 1977, p 35–38

10. D. Heckel, J.P. Wiencek, and N. Netoskie, “Metallurgical Aspects of the AOD,” Paper presented at the Electric Furnace Conference (Detroit, MI), Iron and Steel Society, Dec 1977 11. F.J. Andreini and S.K. Mehlman, “Quality Aspects of Foundry AOD,” Paper presented at the Electric Furnace Conference (Detroit, MI), Iron and Steel Society, Dec 1979 12. S.K. Mehlman, Ed., Mixed Gas Blowing, Proceedings of the Fourth Process Technology Conference, Iron and Steel Society of AIME, American Institute of Mechanical Engineers, 1984 13. L.G. Kuhn, Ed., Mixed Gas Blowing—A New Era of Pneumatic Steelmaking, Iron Steelmaker, Aug 1983 14. R. Heinke and S. Ko¨hle, Computer Control of the VOD Process, Proceedings of the Seventh International Conference on Vacuum Metallurgy (Tokyo), Section 34–2, Iron and Steel Institute of Japan, 1982, p 1356–1363 15. L. Tolnay, I. Sziklavari, and G. Karoly, Method of Regulating the Endpoint of the Vacuum Decarburization Process Developed for the Production of Acid Resistant Steel Grades with Extra Low Carbon Content in ASEA-SKF Ladle Metallurgy Plant, Proceedings of the Eighth International Vacuum Metallurgy Conference (Linz), M1.3, Eisenhu¨tte Osterreich (Vienna), 1985, p 606–615 16. P. Tennilae, M. Kaivola, and M. Walter, Systems Technology and Operating Results of VODC Converter, Proceedings of the First Electr. Steel Congress (Aachen), F4, Verein Deutscher Eisenhu¨tterleute (VDEh, Du¨sseldorf), 1983, p 1–12 17. W. Burgmann, O. Wiessner, and G. Reese, Vacuum Converter Technology for the Production of Superalloys, Stainless and Low Alloy Steels, Proceedings of the Vacuum Metallurgy Conference on Special Metals and Melting Process (Pittsburgh, PA), Iron and Steel Society, 1985, p 3–10 18. B. Mishra, Steelmaking Practices and Their Influence on Properties, Metals Handbook Desk Edition, 2nd ed., ASM International, 1998 19. B. Deo, Models for Predicting Sulfur and Phosphorus Distribution Ratios in Oxygen Steelmaking, Trans. IIM, Vol 41, 1988, p 475–479 20. L.E.K. Holappa, Ladle Injection Metallurgy, Int. Mater. Rev., Vol 27, 1982, p 53–76 21. E.T. Turkdogan, Physicochemical Aspects of Reactions in Ironmaking and Steelmaking Processes, Trans. ISIJ, Vol 24, 1984, p 591–611 22. F.D. Richardson, Physical Chemistry of Melts in Metallurgy, Vol 2, Academic Press, 1974 23. A. Bergman, The Application of Optical Basicity to Dephosphorization Equilibria, Steel Res., Vol 61, 1990, p 347–352

226 / Molten Metal Processing and Handling

Fig. 29

Thermal energy balances for arc-heated and induction-heated processes. VAD/LF, vacuum arc degassing/ ladle furnace; VID/VIM, vacuum induction degassing/vacuum induction melting

24. H. Ono et al., Removal of Phosphorus from LD Converter Slag by Floating Separation of Dicalcium Silicate During Solidification, Trans. ISIJ, Vol 21, 1981, p 135–144

25. K.H. Karl, H. Abratis, H. Maas, and M. Wahlster, Reaktionen Sauerstoffasffiner Begleitelemente des Eisens mit Eisenoxidhaltigen Basischen Schacken, Arch. Eisenhuttenwes., Vol 45, 1974, p 9–16

26. K.W. Lange, Thermodynamic and Kinetic Aspects of Secondary Steelmaking Processes, Int. Mater. Rev., Vol 33, 1988, p 53–89 27. T.H. Bieniosek, B.S. Kupchik, and K.C. Ahlborg, Ladle Slag Deoxidation with Aluminum-Limestone Mixtures, Steelmaking Conf. Proc., ISS-AIME, Vol 72, 1980, p 409–411 28. R. Schumann and W. Schwarz, Ladle Metallurgy Comparison of Vacuum Ladles and Vacuum Vessels, Metall. Plant Technol., Vol 4, 1985, p 10–18 29. K.H. Bauer and H. Wagner, Operational Experience with a Vacuum Ladle Degassing in a EOF Shop, Stahl Eisen, Vol 107, 1987, p 426–430 30. T. Hsiao, T. Lehner, and B. Kjellberg, Scand. J. Metall., Vol 9 (No. 3), 1980, p 10 31. H.G. Bauer, K. Behrens, and M. Walter, Review on VOD/VAD-Treatments, Proceedings of the Seventh International Conference on Vacuum Metallurgy (Tokyo), Section 33–1, The Iron and Steel Institute of Japan, 1982, p 1314–1331 32. W. Burgmann, J.-S. Yie, and M. Kaivola, Operational VOD Results in Ladles and Converters, Proceedings of the Seventh International Conference on Vacuum Metallurgy (Tokyo), Section 34-1, The Iron and Steel Institute of Japan, 1982, p 1348–1355 33. G. Pietschmann, G. Scharf, W. Burgmann, and M. Velikonja, The Freital Ladle Metallurgy Plant: Equipment and Results, Proceedings of the Eighth International Conference on Vacuum Metallurgy (Linz/ ¨ sterreich, 1985, Austria), Eisenhu¨tte O p 582–605 34. H. Kaito, T. Othani, and S. Iwaoka, Production of Super Ferritic Stainless Steels by S.S.-VOD Process, Tetsu-to-Hagane´ (J. Iron Steel Inst. Jpn.), 1977, p 139–156 35. R. Vasse, H. Gaye, J.P. Motte, and Y. Fautrelle, Thermal and Metallurgical Efficiencies in Ladle Induction Furnace Treatments, Proceedings of the Seventh International Vacuum Metallurgy Conference (Tokyo), Section 25–2, The Iron and Steel Institute of Japan, 1982, p 1125–1132 36. C. Lechevalier, J.P. Motte, and G. Forestier, Proce´de´ de Chauffage par Induction de L’acier Liquide en Poche, Rev. Me´t., Vol 10, 1980, p 791–797

Steel Melt Processing / 227

Fig. 30

Treatment cycle of a 75Mg (80 ton) vacuum arc degassing-heated melt. 1, tapping; 2, additions of C + Al + 0.5% CaO; 3, temperature check; 4, sampling and temperature check; 5, addition of 1.5% alloys + 0.5% CaO; 6, temperature check; 7, additions + 0.5% CaO; 8, degassing; 9, venting

Table 6

Data comparison for various ladle heating operations

VID, vacuum induction degassing; Vott, vacuum oxygen heating; see Fig. 23 for explanations of other abbreviations. Parameter

Method of heating Method of stirring Working pressure, mbar (torr)

VAD

LF

Arc Gas injection 500–800 (375–600) 10–20 (7.5–15)

Arc Gas injection 1020 (770)

ASEA/SKF(a)

VID

Arc Induction Induction plus gas Induction plus gas 1020 (770) 0.1–1 (7.5  102 to 7.5  101) O2 partial pressure, mbar (torr) 20–50 (15–38) 20–50 (15–38) 0.02–0.2 (1.5  102 to 1.5  101) Heating rate, K/min 3–5 3–5 2–4 5–15 Melt weight, Mg (ton) 30–150 15–300 15–250 0.5–15 (0.55–16.5) (33–165) (16.5–330) (16.5–275) Installed power, MVA 5–20 5–40 5–25 0.5–3 Lining thickness Bricked, mm (in.) 200–350 (8–14) 200–300 150–250 (6–10) 80–100 (3–4) (8–12) Rammed, mm (in.) ... ... ... ...

(a) Source: Ref 28. (b) Source: Ref 35, 36

VID, Calidus/IRSID(b)

Induction Induction 1–10 (7.5  101 to 7.5)

Plasma

Plasma Gas 1020 (770)

VOH

Oxygen + Al Gas 1–100 (7.5  101 to 75) 0.2–20 (1.5  101 to 15) 20–50 5–300 (5.5–330)

0.2–2 (1.5  101 to 1.5) 10–20 (7.5–15) 2–7 0.2–1 0.3–25 (0.33–27.5) 50–300 (55–330) 0.3–5 1–6 ... ... 200–300 (8–12) 200–300 (8–12) 100–150 (4–6)

...

...

228 / Molten Metal Processing and Handling Table 7 Comparison of applications and operational characteristics of various heating and melting processes Process(a) Benefit/application

VD

LF/VD

VOD

Melt improvement Hydrogen removal Nitrogen removal Deoxidation Vacuum carbon deoxidation Desulfurization Inclusion removal Extralow carbon Temperature adjustment

VODC

P P P P P P ... ...

Savings Basic materials Alloy material Operational capabilities Relief of primary melting system Increased production capacity Buffer function Collection of molten metal Split melts

VID

P P P P P P ... P

S P S P P S P ...

S P S P P S P ...

P P P P P P P P

S P

S P

P P

P P

S P

P P ... ... ...

P P P P P

P P ... ... ...

P P ... ... ...

P P P P P

(a) P, primary benefit or application; S, secondary benefit or application. VD, vacuum degassing; LF/VD, ladle furnace vacuum degassing; VOD, vacuum oxygen decarburization (ladle metallurgy); VODC, vacuum oxygen decarburization (converter metallurgy); VID, vacuum induction degassing

Fig. 31

Schematic of a vacuum induction degassing unit. (a) Top view. (b) Side view. 1, mold, die, or ladle; 2, charging device; 3, filter; 4, vacuum pumping system; 5, melt current supply; 6, vessel with vacuum bell; 7, gas-purging set; 8, cooling water manifold; 9, control cabinet; 10, hydraulic unit; 11, lid station; 12, transformer

Steel Melt Processing / 229

Fig. 32

Effect of vacuum induction degassing (VID) processing on the (a) hydrogen, (b) nitrogen, and (c) total oxygen contents of X 38 CrMoV 51 die steel (Fe-0.38C-1.0Si- 0.40Mn5.2Cr-1.3Mo-0.40V)

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 230-239 DOI: 10.1361/asmhba0005300

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Aluminum Fluxes and Fluxing Practice Rafael Gallo, Foseco Metallurgical David Neff, Pyrotek

IT IS WELL KNOWN that molten aluminum casting alloys have two inherent characteristics: the tendency to absorb hydrogen gas, and the ability to readily oxidize. Hydrogen is made available at the surface of molten aluminum alloys through the reaction of the molten bath with the water vapor present in the melting environment (Ref 1–4). The reaction between water vapor and molten aluminum yields not only hydrogen gas but also the formation (within milliseconds) of an amorphous aluminum oxide (Al2O3) film on the surface bath, which acts as a protective oxidation barrier for the molten metal underneath the film. The amorphous films have been referred to as young films (Ref 5). The two oxidation reactions that occur when molten aluminum contacts air proceed according to the following: 2Al þ 3H2 O ! Al2 O3 þ 3H2 4Al þ 3O2 ! 2Al2 O3

The hydrogen produced through the reduction of water in turn dissociates into its atomic form at the melt surface and then diffuses through the amorphous aluminum oxide film, from which it is quickly dissolved by the molten bath. The aluminum oxide films are an intrinsic part of the melting process; they protect the metal underneath the film from further oxidation. However, in actual foundry operations, the surface of the molten bath always has some movement due to one or more of the following melting practices:      

Charging Skimming Cleaning Degassing Transferring Ladling

Any of these melting practices causes the thin aluminum oxide films to break and reoxidize, causing rapid aluminum oxide film thickening (oxide buildup). The constant metal movement and breaking of the aluminum oxide films cause the films to crumble, thicken, and encapsulate unoxidized molten aluminum, generating what is known as wet dross. A typical representation of this

phenomenon is well observed during the filling process of transfer ladles, such as the one shown in Fig. 1. As soon as the molten metal is tapped from the melting furnace into the transfer ladle, not only does the molten stream become disturbed (from the very beginning to the end), but also a significant amount of splashing between the molten stream and the rising metal surface promotes aluminum oxidation, causing the surface of the bath to thicken with wet dross. The aluminum content of wet drosses is typically reported to be on the order of 60 to 85%, while the remaining 40 to 15% is aluminum. The amount of trapped liquid metal in the dross varies according to the melting and/or handling molten practice. The aluminum oxide is a very stable compound that cannot be reduced to aluminum under ordinary melting conditions. However, the amount of suspended liquid metal could be reduced from the 60 to 85% range to 30% by proper fluxing and drossing techniques. Formation of dross is an intrinsic process when melting aluminum alloys. The dross is considered to be the main contributor in influencing the total metal loss during melting. Depending on the efficiency of the melting furnace, and melting practices, the amount of dross generated may be from 1 to 10% of the total metal melted. In addition to the formation and elimination of dross, another problem that aluminum foundries face in the molten metal bath is the nonmetallic and the metallic impurities that are suspended and floating in the bath. Impurities and aluminum oxides will remain suspended and floating in the liquid bath because they are porous and contain some gas adhering to and trapped in the pores. In this case, the oxide density is similar to that of the aluminum alloy. Nonmetallic and metallic impurities are introduced into the melt during the charging process, the molten metal treatment, and the handling operations. Even when melting primary ingot, nonmetallic impurities such as hydrogen and aluminum oxide are introduced. However, as would be expected, the majority of the nonmetallic and metallic impurities come from charging returns. The term returns means scrap castings, gates, risers, trimmings, and so on. According to the type and origin of the return

Fig. 1

Metal being tapped from a holding furnace into a transfer ladle, causing aluminum oxide films to crumble, thicken, and encapsulate unoxidized molten aluminum, generating wet dross

metal and the conditions of raw material storage, the metal can contain considerable amounts of both nonmetallic and metallic impurities. Hydrogen absorption, formation of dross, generation of bifilms, metallic and nonmetallic inclusions, and oxide buildup are inherent characteristics when melting and handling molten aluminum, regardless of the melting and/or holding furnace design or the energy used (gas or electricity). Therefore, it is of vital importance to understand and control such inherent characteristics, because they will greatly affect the quality of the molten metal, which in turn impacts the final casting quality with respect to porosity, shrinkage, oxides, and inclusions. While the existing types of inclusion material that would be present in melting and holding furnaces and/or transfer ladles will vary from foundry to foundry, their removal is essential for proper molten metal cleanliness. A number of commercially accepted melt treatment techniques are being used by aluminum foundries to remove and separate inclusions from the molten aluminum alloy prior to casting. These include various methods of fluxing, degassing, and filtration in the furnaces and the gating systems (Ref 3, 6–11). Any of these techniques will have an impact on the melt cleanliness of the molten aluminum alloy. However, the effectiveness of evaluating their removal would rely on the melt cleanliness measurement technique being used.

Aluminum Fluxes and Fluxing Practice / 231 The strengths and weaknesses of these methods with regard to equipment requirements, sampling, sensitivity, timing, and practical means of assessing inclusion levels on the foundry floor have been fully discussed and published in the literature (Ref 12–29). Technical articles, which specifically cover fluxing as a factor for ensuring metal cleanliness, have also been published (Ref 7, 30–35). In addition to aluminum, fluxing is also commonly done to some extent with virtually all nonferrous mill and foundry melting operations as the first step in obtaining a clean melt. Examples include:  Fluxing of magnesium alloys using salt

fluxes or inert gas as a cover (see the article “Magnesium and Magnesium Alloy Castings” in this Volume).  Fluxing of copper alloys to remove gas or prevent its absorption into the melt, to reduce metal loss, and to remove specific impurities and nonmetallic inclusions (see the article “Copper and Copper Alloy Castings” in this Volume)  Fluxing of zinc alloys with chloride-containing fluxes that form fluid slag covers, which can be used to minimize melt loss if they are carefully skimmed from the melt before pouring (see the article “Zinc and Zinc Alloy Castings” in this Volume)

particulate (i.e., aluminum oxide inclusions) from the liquid metal. The principal function of gas fluxing is hydrogen removal, not dross treatment or recovery. Gas fluxing does not reduce dross but rather increases dross.

Solid Fluxes Solid fluxes can be broadly categorized as passive or active fluxes. Passive fluxes protect the surface of the molten aluminum from oxidation and prevent hydrogen pickup by the melt. Active fluxes react chemically with the aluminum oxides and clean the melt more effectively. Foundrymen recognize the importance of furnace type; furnace design; furnace maintenance; quality grade of smelter ingot; quality of the inhouse returns; condition of the skimmers, ladles, and the like; and the charging and melting practice on the starting quality of the molten metal to make a casting. Nonetheless, it seems that the foundrymen still have some concerns over the lack of general accepted information on conventional solid fluxes.

Aluminum Fluxing Fluxing is a term commonly used in foundries, especially within the melting department work force, to refer only to the addition of chemical compounds to clean molten aluminum alloy baths in either the furnaces (melting or holding) or the transfer ladles. Fluxing is the first step in obtaining clean molten metal by preventing excessive oxide formation, removing nonmetallic inclusions from the melt, and preventing and/or removing oxide buildup on furnace walls. In general, fluxes may be grouped into two classes: gaseous or solids. Gaseous fluxes may be a blend of an inert and a chemically active gas that is injected into the molten bath. Solid fluxes are blends of salts, which, at the present time (2008), are the most preferred type of fluxes used in foundries, especially since gaseous mixtures with chlorine have been almost completely eliminated because of environmental concerns.

Gas Fluxing Gas fluxing is the application of a gas treatment to the melt. The gas may be inert, such as argon or nitrogen, and the latter is more common because of economics. Gas mixtures may also be used, usually adding an active halogen component—chlorine or fluorine compounds— in smaller amounts, typically only 5 to 10%, to the base inert gas. The purpose of this addition is to provide a better separation of solid

Fig. 2

Equilibrium diagram of cryolite

Even though there are many different commercial fluxes on the market, the basic ingredients of such commercial fluxes are based on the very early work, conducted approximately 100 years ago, to improve the electrolytic production of aluminum by reducing the aluminum oxide using molten cryolite (Na3AlF6). Since then, both natural and synthetic cryolite and fluorides have been evaluated. Figure 2 shows the equilibrium diagram of cryolite. The removal of aluminum oxide by halides has its foundations in basic research and development on the systems NaF-AlF3-Al2O3 and NaF-AlF3. Studies on these systems have shown that the solubility of aluminum oxide increases with the sodium fluoride content. The reaction between cryolite and aluminum oxide in the presence of a surplus of NaF takes place according to the following reactions: 4NaF þ 2Al2 O3 ! 3NaAlO2 þ NaAlF4

or 2NaF þ Al2 O3 ! NaAlO2 þ NaAlOF2

232 / Molten Metal Processing and Handling In addition, halide and fluoride salts have been thoroughly researched by secondary (recycled) aluminum producers during the melting and processing of aluminum scrap and dross. The recovered aluminum results from the effectiveness with which nonmetallic and metallic impurities are removed or reduced to acceptable levels. In these operations, low-melting-point fluxes are typically used. They are basically sodium and potassium chloride with additions of fluorides and some other chlorides, such as calcium chloride and magnesium chloride. Salts are used in fluxes because of the following general characteristics:       

Cost-effective Easy to use Combine easily with other ingredients Serve as fillers for active ingredients Have lower density than aluminum Give ability to cover the molten surface Allow a low-melting-point, high-fluidity product  Have capability to absorb oxides and reaction products from the fluxing action Solid fluxes are basically blends of sodium chloride and potassium chloride salts, with or without the addition of fluorides. In addition, small quantities of oxidizing compounds, such as carbonates, sulfates, and nitrates, are added to promote exothermic chemical reactions (Ref 7, 29–31). Exothermic reactions are important because they prompt the coalescence of the trapped liquid aluminum particles in the dross. The mechanism of how salt fluxes work has been attributed to thermodynamic chemical reactions and surface tension effects between the aluminum oxide and the flux, the aluminum oxide and the molten metal, and the molten metal and the flux. This interaction between aluminum oxides, flux, and molten metal has been explained in earlier papers (Ref 11, 12, 31, 32). In addition, the effect of the flux on the liquid metal will depend on the chemistry of the flux used, morphology of the flux, total amount of flux added, molten metal temperature, flux contact time, rabbling (stirring) technique, and so on. Nevertheless, from the point of view of flux chemistry, it is important to understand that different combinations and proportions of ingredients will impart different flux properties, such as flux density, flux fluidity, flux wettability, and flux reactivity. These four different flux properties are responsible for the characterization of any particular flux. To better understand how these flux properties are accomplished, it is necessary to recognize that essentially each different ingredient provides different effects that directly influence the final property of a flux. Despite the fact that many ingredients are available, not all of them are necessary or found in each flux. Table 1 lists the characteristics of the materials used in fluxes. Generally, ingredients can be classified into four major groups based on their primary influence over the mixture.

Table 1

Characteristics of materials used in fluxes Melting point

Chemical

Molecular mass, g/mol

Solid density, g/cm3

LiCl NaCl KCl CaCl2 MgCl2 AlCl3 BaCl2 LiF NaF KF CaF2 MgF2 AlF3 Na3AlF6 LiNO3 NaNO3 KNO3 Li2SO4 Na2SO4 K2SO4 CaSO4 MgSO4 Li2CO3 Na2CO3 K2CO3 MgCO3 CaCO3

43.39 58.44 74.56 110.99 95.22 133.34 208.25 25.94 41.99 58.10 78.08 62.31 83.98 209.94 68.94 84.99 101.11 109.94 142.04 174.27 136.14 120.37 73.89 105.99 138.21 84.32 100.09

2.068 2.165 1.984 2.150 2.320 2.440 3.920 2.635 2.558 2.480 3.180 3.180 2.882 2.900 2.380 2.261 2.109 2.221 ... 2.660 2.610 2.660 2.110 2.532 2.420 2.960 2.710



C

605 801 770 782 714 190 963 845 993 858 1423 1261 ... 1010 264 307 339 859 897 1069 1450 ... 723 851 894 ... 1339



Boiling point F

1121 1474 1418 1440 1317 374 1765 1553 1819 1576 2593 2302 ... 1850 507 585 642 1578 1647 1956 2642 ... 1333 1564 1641 ... 2442



C

1325 1413 1500 1600 1412 177.8 1560 1676 1695 1505 2500 2239 1291(a) ... 600(b) 380(b) 400(b) High ... 1689 High 1124(b) 1310 High High 350(b) 850



F

2417 2575 2732 2912 2574 352 2840 3049 3083 2741 4532 4062 2356(a) ... 1112(b) 716(b) 752(b) ... 3072 2055(b) 2390

662(b) 1562

(a) Sublimes. (b) Decomposes. Source: Ref 32

Chlorides. Examples are aluminum chloride (AlCl3), barium chloride (BaCl2), calcium chloride (CaCl2), lithium chloride (LiCl), magnesium chloride (MgCl2), potassium chloride (KCl), and sodium chloride (NaCl). The melting point of these compounds, in their pure state, may range from 190 to 965  C (375 to 1765  F). However, they form low-temperature eutectics in combinations. In addition, it is very typical to have at least three compounds in a given flux, so that none of these compounds would be used as a single compound in a flux recipe. Chlorides are mainly used because of their fluidizing effects and because they are used as fillers. Fluxes based only on chloride salts should not react with molten aluminum, or at least the reaction should be negligible. In addition, these salts provide negligible effects on surface tension as compared to fluorides. Fluorides. Examples are simple fluorides such as aluminum fluoride (AlF3), barium fluoride (BaF2), calcium fluoride (CaF2), lithium fluoride (LiF), magnesium fluoride (MgF2), potassium fluoride (KF), sodium fluoride (NaF), and double-fluoride compounds such as sodium silicofluoride (Na2SiF6) and potassium silicofluoride (K2SiF6). The melting point of these compounds, in their pure state, may range from 845 to 1425  C (1553 to 2600  F). The high melting point of these compounds provides thickening effects on a flux. Fluoride salts have limited solubility for oxides. However, no fluoride salts can dissolve massive aluminum oxides. Fluoride salts act as surfactants, affecting surface tension forces

between flux, liquid metal, and aluminum oxides. As the flux wets the interface between the aluminum oxide particles and the liquid metal, the adhesion force between the liquid aluminum and the oxides decreases, promoting oxide separation and metal coalescence. Fluorides are still the most effective compounds being used in fluxes to improve aluminum recovery from wet dross. Oxidizing Compounds. Examples are:  Nitrates such as potassium nitrate (KNO3)

and sodium nitrate (NaNO3) such as calcium carbonate (CaCO3), potassium carbonate (K2CO3), and sodium carbonate (Na2CO3)  Sulfates such as potassium sulfate (K2SO4) and sodium sulfate (Na2SO4)  Carbonates

The melting point of the nitrates ranges from 307 to 339  C (585 to 642 F); the melting point of the carbonates ranges from 851 to 1339  C (1564 to 2442  F); and the melting point of the sulfates ranges from 859 to 1450  C (1578 to 2642  F). Oxidizing compounds are used to promote exothermic chemical reactions. They react with the smallest molten aluminum particles that are present in the dross, yielding aluminum oxides as well as considerable heat. The purpose of the exothermic reaction is twofold: 1) It allows larger pockets of entrapped aluminum to coalesce and fall back into the molten bath, and 2) it facilitates reactions between aluminum and fluorides. The exothermic reaction continues until all of the fine aluminum particles are burned.

Aluminum Fluxes and Fluxing Practice / 233 Solvents of Aluminum Oxides. Examples are:  Borax (Na2B4O7)  Potassium borate (K2B2O4)  Sodium cryolite (Na3AlF6)

Both the Hall and Heroult patents covered the electrolysis of aluminum oxide in a bath of molten halide salts. Since then, dissolution of aluminum oxides (Al2O3) by cryolite has been documented.

Flux Morphology The manufacturing of fluxes requires compliance with strict general quality requirements for the raw ingredients (commercial purity, grain size uniformity, and less than 1% moisture content) and for the manufacturing process (blending of the ingredients). Control over these variables assures consistent products and consistent performance. For most of the aluminum foundry history, powder fluxes have been used. Powder fluxes have been in commercial use for nearly 60 years. However, the basic formulations and compositions of the fluxes have not changed dramatically over this period. Present-day research and development in flux chemistry is still based on trial-and-error procedures. In the last ten years, research and development for possible new formulations and variations of traditional mixes has more closely evaluated the use of fluorides because of environmental emission concerns. It has been reported that if fluorides were removed from the formulation, it would render a powerless

Fig. 3

Powder flux being thrown inside the furnace

Fig. 4

Granular flux being thrown inside the furnace

and inefficient flux (Ref 7, 33). At the same time, it has been reported that emissions from fluxes containing fluorides could be reduced by at least half only if the morphology of the flux changed from powder to granular. It is still prudent to mention that emissions from powder fluxes are within standard environmental limits. Fluoride-containing salts are being subjected to stricter environmental regulations and constraints. Therefore, fluoride-free fluxes are available. However, it is important to realize that fluoridefree fluxes will never perform as efficiently as fluxes containing fluorides. On the other hand, fluoride-free fluxes are the best option to eliminate internal smoke and odor in the plant, as well as to treat metal in electrical resistance furnaces without damaging the electrical elements. A granular flux may or may not have the same formulation as a powder flux. However, because of the different grain morphology, granular fluxes offer other operational and product advantages over powder fluxes. Granular fluxes are less polluting to the atmosphere; emissions are reduced by at least 50%. Furnace tenders appreciate the fact that granular fluxes greatly reduce smell and smoke, facilitating working closer to the furnaces during drossing operations. Granular fluxes are easier to apply and to spread over the molten surface, since they are fines and dust free. Therefore, application rates can often be reduced. Additional product benefits are the better consistency in chemistry from grain to grain and the fact that there is no segregation during flux transportation or handling. Figures 3 and 4 depict the difference, which is very noticeable, when applying a powder or a granular flux. Figure 3 shows that, as the

Fig. 5

Powder flux grains

furnace tender throws the powder flux, the flux falls down in a big cloud as soon as it leaves the cup. On the other hand, Fig. 4 shows that the granular flux will travel farther before falling down into the melt. That difference in traveling distance facilitates spreading and directing the granular flux inside a furnace. Consistency in chemistry is based on the fact that, since powder fluxes are blends of salts, each grain will represent only one of the different chemical compounds from which the flux is made. On the other hand, and because of the manufacturing process, granular fluxes have 100% grain uniformity in chemistry, since each grain will represent the same chemistry as the flux recipe. Figures 5 and 6 are used to illustrate this fact.

Categories of Fluxes Years of casting experience have demonstrated that proper fluxing is a key element to cost reduction and quality improvement in the liquid melt. Fluxing is the best means of obtaining clean metal by preventing excessive oxide formation, removing nonmetallic inclusions from the aluminum melt, and preventing and removing oxide buildup from furnace walls. Historically, solid fluxes have been classified into four categories, depending on their use and function at the foundry operation. These categories are:    

Cover fluxes Drossing fluxes Cleaning fluxes Furnace wall cleaner fluxes

Fig. 6

Granular flux grains

234 / Molten Metal Processing and Handling Cover, drossing, and melt-cleaning fluxes are usually manually spread, shoveled, or thrown onto the melt surface. However, drossing and melt-cleaning fluxes can also be added via rotary flux injection and/or lance flux injection. Furnace wall cleaner fluxes are typically blown (with a gunning device) onto the furnace walls at the melt line. The flux injection process introduces the powder flux in a finely dispersed stream to the bottom of the melt. To ensure that melt-cleaning and drossing fluxes will do their job effectively (regardless if they are either the powder or granular type), it is essential that, after manual addition, the flux be stirred into the melt to achieve as much contact as possible with the molten metal and dross layer. Based on the size of the furnace, the stirring can be accomplished manually by the furnace tender or mechanically by a forklift or mechanized vehicle. Typically, agitation and stirring of the flux is completed between 1 and 3 min in crucibles or transfer ladles and between 5 and 10 min in reverberatory furnaces with up to 41,000 kg (90,000 lb) of liquid capacity. If flux injection is used, there is no need for manual activation, since the mixing and activation would take place during the injection treatment cycle. It is also important to emphasize that the flux reaction efficiency depends on three interrelated factors: molten metal temperature, stirring, and activation time. Improper control of these factors will result in unreacted flux floating on top of the dross, as shown in Fig. 7. Cover fluxes are used, as the name implies, to cover the melt. These fluxes, which are comprised of salt compounds that form a liquid at normal aluminum melting temperatures, provide a molten barrier “blanket” on the surface of the metal to protect it against oxidation and hydrogen gas adsorption. Cover fluxes are mainly a blend of NaCl and KCl salts with small quantities of fluorides. Fluxes having these three compounds are suitable for almost all types of aluminum alloys, excluding hypereutectic aluminum alloys and aluminummagnesium alloys with over 7% Mg. Over the years, a great number of different commercial cover fluxes have been developed

with preferred additions and proportions of NaF, KF, Ca2F, and other compounds, such as CaCl2, for lowering the melting point of the flux, as well as for providing cleaning effects by agglomerating dirt and oxides. A typical base composition is 47.5% NaCl, 47.5% Kcl, and 5% Na3 AlF6, or cryolite. Other composition components are listed in Table 2. Cover fluxes, for smelting application may have slightly lower melting points than those used in foundry applications. Foundries use these types of fluxes during the melting of heavily oxidized foundry returns and machining chips, as well as when metal holding temperatures exceed the 770 to 790  C (1420 to 1450 F) range. In smelting operations, cover fluxes are mixed in the rotary furnace during the melting of fine scrap, turnings, sawings, fines, and so on. Typically, cover fluxes for smelting applications have lower melting points than those used for foundry applications (e.g., 424 versus 665  C, or 795 versus 1229  F). Drossing fluxes are designed to assist in separating the entrapped aluminum from the adherent oxide skin envelope, which surrounds the aluminum in the dross layer. A high percentage of useful metallics can be lost when wet drosses are removed from the furnace; thus, many operations attempt to reduce the metallic content by in-furnace treatment of the drosses with exothermic fluxes. These contain oxidants and fluorides to generate temperature and help promote a separation of the oxides from the metallics. The exothermic reaction does consume aluminum values while generating heat and can be the source of a loss of up to 20% of the contained metallics in the dross, if the flux is applied too liberally (see the article “Dross, Melt Loss, and Fluxing of Light Alloy Melts” in this Volume for more information on drossing fluxes). Drossing fluxes may be based on salt blends of chlorides, simple and/or double fluorides, and oxidizing compounds. Therefore, drossing fluxes are able to react exothermically, generating heat and improving flux wettability. The wetting action of the flux promotes coalescence, which brings the fine aluminum drops tighter to form larger drops that are much easier Table 2 Compound

Fig. 7

Unreacted flux

AlF3 CaCl2 MgCl2 MnCl2 KF NaF NaCl KCl CaF2 Na3AlF6 Na2SiF6 KNO3 C2Cl6 K2CO3 Na2CO3 K2TiF6 KBF4

to recover. The work of drossing fluxes is considered to be due to both the surface tension effects and the dissolution of aluminum oxides. There are wide commercial ranges of different flux compositions. Even though a drossing flux will always be considered as “hot or reactive flux” in furnace-tender terms, it is important to realize that the reactivity in a drossing flux is due to a combination of the oxidizers and the double-fluoride compounds. This is an important concept to understand, because even without double fluorides, the flux may look too reactive due to an excess of oxygen-bearing compounds that may be burning excessive amounts of good metallic aluminum without actually dissolving aluminum oxides in the melt. Some manufacturers may decrease the use of double fluorides, since they are more expensive compared to other compounds. At the same time, it is important to note that a lower-grade compound, which is cheap, may also decrease the effectiveness of the flux. The same concept will be true for any other type of flux; the quality of the chemical compounds will influence the price of the flux. Thus, a good drossing flux must be designed to reduce the rich metallic aluminum content of the dross. As the dross is treated with the drossing flux, it changes from a wet dross appearance (bright, shiny metallic color) to a dry dross appearance (dark, powdery). Proper flux treatment could reduce the amount of metallic aluminum content of the dross to 30%. Figure 8 shows the difference in appearance between untreated dross and dross treated with flux. Cleaning Fluxes. Melt-cleaning fluxes are designed to remove aluminum oxides from the melt. Melt-cleaning fluxes usually have similar chlorides and oxidizing compounds as the ones used in the drossing fluxes, but in different proportions. In addition, the composition of a meltcleaning flux will typically include only simple fluorides, compared to the simple and double fluorides that are present in drossing fluxes. Since a melt-cleaning flux is less reactive than a drossing flux, it will yield less dry dross than a drossing flux.

Typical fluxing compounds employed Fluidizer (F)/thickener (T)

Oxide surfactant

Chemically active

Exothermic

Gas released

Element added

F F F F F F F F T T T ... ... ... ... ... ...

... ... ... ... ... ... ... ... X X X ... ... ... ... ... ...

X ... ... X ... X ... ... ... ... ... X X X X X X

... ... ... ... ... ... ... ... ... ... ... X ... ... ... ... ...

... ... ... ... ... ... ... ... ... ... ... ... Cl2, AlCl3 CO2 CO2 ... ...

... ... ... ... ... Na ... ... ... ... ... ... ... ... ... Ti B

Aluminum Fluxes and Fluxing Practice / 235 Table 3

Typical manual addition rates for powder and granular fluxes

Application for:

Crucibles and ladles Reverberatory

Fig. 8

Untreated (left) and treated (right) dross

The work of metal-cleaning fluxes is considered to be only due to the surface tension effects, as previously stated. It is very important not to confuse a metal-cleaning flux with a furnace wall-cleaning flux. Furnace Wall-Cleaning Flux. Wall-cleaning fluxes are specifically designed for the softening and removal of excessive aluminum oxide buildup that occurs on the walls of melting furnaces, especially along the melt line. This type of flux helps keep crucible and furnace walls free of oxide buildup above and below the melt line. They are usually applied with a lance or flux gun or may simply be broadcast manually onto the walls. Slightly exothermic, these fluxes, under closed-door and high-fire conditions, increase the localized temperature of the oxide buildup and thereby soften or loosen the attached oxides, permitting easier mechanical removal and subsequent skimming. Wall-cleaning fluxes contain the highest amounts of double-fluoride compounds, such as Na2SiF6 and Na3AlF6. The exothermic reactions that occur because of the oxidizing compounds and the double fluorides enhance more penetration of the flux into the oxide buildup, facilitating the removal of the oxide buildup at the furnace wall. Fluoride-free fluxes cannot be used effectively as a wall-cleaning flux. If the oxide buildup has been converted over time to the insidious form of aluminum oxide called corundum, it cannot be removed chemically. However, in its early stages, sufficient loosening of the mildly adherent product can be effected with the combination of fluxing salts and hard work.

Foundry Practices Powder and granular fluxes can be added manually, either by weight percentage of the total metal charged or melted or by weight in pounds based on the molten surface area to cover. Manufacturer recommendations are the starting point. Typical recommendations are shown in Table 3. However, the exact quantity of flux will depend on a variety of operating conditions, such as furnace type, charging and melting practices, initial metal cleanliness,

Addition based on:

Powder flux

Granular flux

Percentage of total metal charged or melted Pounds of flux per square foot of molten surface area

0.25–0.50% 0.35–0.50 lb

0.20% 0.30 lb

molten metal temperature, and final metal cleanliness required. The rotary flux-injection process is typically used in crucible furnaces and transfer ladles, while lance flux injection is typically used for reverberatory furnaces. In both manual and flux injection processes, either powder flux or granular flux may be employed, provided that the composition is suitable for the required temperature/reactivity. In flux injection, the flux material must be flowable in the injection device. Typical recommendations for flux-injection rates are from 0.009 to 0.32 kg/min (0.02 to 0.70 lb/min). The amount of flux added is determined by the injection time, which will also depend on a similar variety of operating conditions mentioned previously. Molten metal cleaning with fluxes varies from foundry to foundry, with some performing a systematic and very careful fluxing and drossing procedure, and others skimming molten metal or cleaning furnaces without any flux at all. The difference in the furnace conditions and the scrap defects related to molten metal quality is very obvious. Although often overlooked, the furnace tender plays a key role in the success or failure of the use of any type of flux.

Flux Injection Foundry fluxing practice involves the use of either a manual flux addition, followed by mixing, stirring, or otherwise creating a reaction, or alternatively using a mechanical flux-injection process. Flux injection is becoming a common melt treatment for improving the cleanliness of an aluminum melt, and in a more efficient manner than manually adding flux, which can be very irregular and subjective (Ref 36). The general process involves the use of a specific fluxing salt that is applied through a delivery gas such as argon or nitrogen through an injection device. While chlorine has often been added to an inert gas in historical practice, today (2008), virtually all flux injection uses just an inert gas as the flux carrier. The purpose of flux injection is threefold in most foundry applications: hydrogen removal, a partial removal of inclusions by flotation, and, with the proper flux composition, treatment of the dross in situ for improved metal recovery. Static or stationary lances submerged into a furnace are used primarily in small melting furnace batch-type applications, such as small reverberatories, crucible furnaces, and holders. For larger furnace vessels, a rotary flux-injection device is used. The flux-injection method addresses the major drawback of conventional fluxing practices, namely the limited contact of flux salt surface with the unwanted impurities in the melt. Flux injection

overcomes this limitation by delivering predetermined amounts of powdered or granulated flux beneath the melt surface. The flux melts into small droplets that expose a large total specific surface area to the melt as they float to the surface, which accelerates flux-induced melt cleaning, that is, partial removal of inclusions. These inclusions then rise to the surface of the melt, where this residue or debris may be skimmed. When halogen-containing salts decompose in the metal, additional wetting occurs between the gas bubbles and the insoluble particulate within the melt. Inclusions are partially removed by attachment to the rising bubble to the melt surface, where they may be skimmed. Hydrogen is removed as well, because the dissolved monatomic hydrogen diffuses into the insoluble carrier gas bubble, then forms a hydrogen molecule. This double-gas bubble species, containing both the carrier gas plus the hydrogen, then continues to rise through the melt and reaches the top surface. Certain solid fluxing compound components themselves may be employed in the flux composition, which will decompose within the melt to generate an insoluble gas that further assists the carrier gas in this task. A typical static flux-injection system includes a dry powder or granulated material feeder that mixes the flux particulate into an inert gas stream, carrying it through a lance immersed in the melt, as shown schematically in Fig. 9. This can be done either into small melting furnaces or into melt transfer/treatment ladles. Flux injection provides many advantages, including delivering the flux into the body of the melt where it can truly react, minimizing the exposure of fluxing salt to the atmosphere and providing for better handling, and controlling the amount of flux to be used. However, there is one major drawback. The area of influence of the flux reaction in a static delivery system is confined to a very small area around the flux-delivery lance itself; very little twophase mixing between the melt and the carrier gas/flux mixture is feasible unless the operator moves the lance around. Consequently, static flux injection is principally suitable only for small furnaces. For larger furnaces, further evolution of the flux injection process has resulted in hybrid equipment that combines the best traits of flux injection and spinning nozzle rotor degassing, shown schematically in Fig. 10. Rotary flux injection is now commonly employed in many foundries and diecasters in furnaces as well as in batch-mode application, such as treatment in a ladle upon tapout from a melting furnace and before delivery to casting units. There are several manufacturers who offer this hybrid equipment; one example is depicted in Fig. 11.

236 / Molten Metal Processing and Handling

Fig. 9

Schematic of the flux-injection process

Fig. 11

Mobile unit for flux injection/rotor dispersion. Courtesy of Pyrotek Inc.

Fig. 12

Fig. 10

Metallic aluminum recovery during ladle flux injection with rotor dispersion. FIP, fluxinjection process

Schematic of flux injection plus rotor dispersion

The combination of flux injection coupled with rotor dispersion offers the following advantages:  Fluxing salts are contained and protected from

the atmosphere, thereby minimizing personal exposure and health issues during use.  Only measured or controlled amounts of fluxing salts are used, resulting in minimal spillage and waste.  By introducing the flux in the submerged position within the molten metal, any emissions are minimized.  Rotor dispersion increases the kinetics of mixing and two-phase contact, very frequently resulting in reduced treatment time.

 By selecting the proper flux composition

being introduced, the dross resulting from the process is treated in situ, achieving increased recovery of the metallic aluminum contained in the dross.  When performed regularly in a melting furnace or transfer ladle, the flux-injection process provides a strong cleaning action on the refractory walls when used wisely. The flux-injection process will provide efficient metal recovery in a batch-treat ladle process, with results analogous to furnace fluxing. Generally, a 455kg (1000lb) ladle may generate 7kg (15lb) of dross during a degas-alone melt treatment, and this dross will average 80 to

85% metallic aluminum content entrapped within a lacy matrix of aluminum oxide skins. Much of this metal is indeed recoverable. By treatment with the appropriate flux plus rotor dispersion, the resultant dross after the process may average just 3 to 4kg (6 to 8lb), and the metallic aluminum content may be reduced to 40% or less. Typical results are demonstrated in Fig. 12. Practical Application. Flux-injection results can be maximized or optimized with good process “housekeeping.” The operator must thoroughly skim the resultant flux-injection process dross from the bath in order to transfer good metal further down the line into casting furnaces. However, this treated dross in itself

Aluminum Fluxes and Fluxing Practice / 237 may still contain recoverable and economicallyviable metal units. Therefore, further opportunity exists to use an auxiliary dross-recovery process with other in-plant equipment. In many instances, it may not be worthwhile or a good use of foundry time to do this internally, but such residues can be sold off to an outside dross processor. It should also be noted that after a fluxing or flux-injection process, it is virtually impossible to remove all flux residues. In the first place, a certain residence time is required for all fluxing salt debris to be fully transported from within the melt to the surface. In a melt-transfer operation, where the treated melt must then be transported and delivered into holding furnaces, temperature loss must be minimized. Therefore, the ladle is usually emptied as quickly as possible to avoid too much temperature loss and the possibility of sludging, in which heavy metal segregation on intermetallic compound formation creates potential hard spots if they ultimately become entrained into a casting. Secondly, it is practically impossible to remove all fluxing debris from a ladle. Consequently, if a subsequent practice of good settling or sedimentation time in the holding/casting furnace is not used, or if final filtration is not employed, flux-based inclusions, such as those shown in Fig. 13, can become embedded into the casting, creating possible surface finish, corrosion, porosity, and mechanical property difficulties. Recent studies (Ref 35) have shown that rotary flux injection is the most effective process to reduce aluminum content in the dross and that manual fluxing is the worst process, leaving 84% metallic aluminum in the dross as compared to 6%. The second-best process to reduce metallic aluminum content in the dross is rotary degassing with a mechanized

Fig. 13

Flux inclusion entrapped in aluminum casting

flux-delivery system, yielding 17% Al in the dross. In addition, these studies showed that if flux were used during the degassing cycle, the aluminum content could be reduced by 77%. Combination degassing/flux-injection practice is gaining favor as a focus not only for improved metal quality but also for improved metal recovery from dross. Assessment and evaluations of fluxes and degassing equipment have shown that the reactivity and molten surface coverage by the powder or granular fluxes is greatly affected by the design of the rotor, operating condition of the equipment, and adherence to operating procedures.

Fluxing Applications Crucible Furnaces. Crucible furnace melting is commonly used at small foundries as well as at medium- and high-volume-production jobbing foundries in which melting versatility is a must. The most popular material for the crucible is silicon carbide. In these foundries, it is typical that the metal skimmings or dross be placed in a small metal bin located by the crucible furnace. Fluxing procedures in these shops are batch type and basically depend on the management philosophy about fluxing. It is very common that after the melting and alloying procedures, some degassing takes place. It is around the degassing stage when the fluxing and skimming operations occur. Typically, there are four variations in the use and application techniques of fluxes:  Melting, degassing, and skimming without

any flux. With this procedure, it is common that the skimmings contain very rich metallic aluminum content, sometimes up to 85%.

The dross coming from the dross bin sometimes looks like an aluminum ingot.  Melting, degassing, fluxing, and skimming. A very small number of foundries flux after the degassing operation. Proper fluxing can be accomplished after the degassing operation without introducing hydrogen gas. However, it is imperative that the added flux be dry, the skimmers be cleaned and dried, and the stirring of flux and the dross skimming be performed in a gentle manner. Nevertheless, the major drawbacks are that the metal temperature is usually at its lowest operating temperature, so the flux takes more hand stirring to become activated, and the metal treatment process takes longer. With this technique, strict discipline is a must.  Melting and degassing while flux is applied on the metal surface, then skimming at the end of the degassing cycle. This is a better method than the first two described previously. If done properly, the operator could take advantage of the mixing capabilities and oxide removal of the rotary degassing equipment. However, with this procedure, it is very important to pay close attention to the degassing technology being used. Although this appears to be the preferred method used in the industry, it is not the most effective for metal cleanliness (Ref 35).  Rotary flux injection. In this process, in which the flux is introduced (through the shaft) during the degassing process, the flux is also dispersed at the bottom of the melt by the rotary action. As the flux leaves the shaft, it liquefies, favoring the wetting of the inclusions, which in turn are carried to the melt surface by the degassing bubbles. Transfer Ladles. Transfer of molten metal by a ladle is a typical operation during the handling of molten aluminum in low-pressure permanent mold foundries and in diecasting foundries. At preset time intervals, the ladle is taken to the melting furnace, from which metal is tapped into the ladle. If degassing is part of the process, then the ladle is taken to the degassing station. If degassing is not part of the process, then the ladle is usually moved away from the furnace before performing the skimming operation. In any case, the molten metal cleanliness and drossing operations performed on the metal in the transfer ladle may include any of the four variations previously described for fluxing the molten metal in crucible furnaces. In addition, there are two supplementary variations with respect to degassing and fluxing at the same time. One variation is to add the flux at the bottom of the ladle before tapping metal into it. The other variation is to add the flux by directing it at the molten metal stream during the tapping of the furnace into the ladle. In both cases, not all the flux reacts when making contact with the stream of liquid aluminum. In the

238 / Molten Metal Processing and Handling majority of the cases, the flux tends to be pushed away from the molten stream. When the flux is added to the bottom of the ladle before tapping metal into the ladle, the only flux that reacts is that which becomes trapped below the initial surge of the molten metal cascading on top of it. The remaining flux is unreacted and tends to float and stay on the surface of the bath. Once the ladle is full, the flux must be stirred and rabbled to activate it before the degassing operation. Even in ladles in which an inert gas is introduced at the bottom of the ladle through a porous plug, the problem of unreacted flux exists. Some unreacted flux will float to the surface. The only way to react the flux is to create an excessive gas bubbling turbulence coming from the porous plug. Unfortunately, this would cause more molten metal oxidation, so that the benefits of fluxing are defeated. A technique not often used is to introduce the flux into the molten stream coming from the tapping of the reverberatory furnace. Again, this technique has the same negative effects with respect to activating flux. Only flux that becomes trapped in the molten stream is activated, while the flux falling down inside of the ladle remains unreacted. As mentioned previously, the unreacted flux will float to the surface of the molten bath. In addition, this process of adding flux is not a very safe practice, since splashing molten aluminum may burn the furnace tender more easily. Reverberatory furnaces are usually used as breakdown furnaces or holding furnaces. In some cases, melting and metal ladling are performed from the same furnace. In addition, some furnaces may have a dip-out and a degassing well at the front of the furnace. These wells may be divided by a ceramic baffle or a refractory wall. Basic use of fluxes in reverberatory furnaces is concentrated into two areas: molten metal and furnace sidewalls. Fluxing of the molten metal bath is typically accomplished after manual additions of powder and/or granular fluxes or by using the lance flux-injection process to deliver the flux. In this particular case, the powder flux is injected via a steel pipe submerged in the molten metal. Furnace sidewalls are susceptible to aluminum oxide buildup, especially along the sidewall areas that are exposed to combustion furnace atmospheres during the expected molten metal-level fluctuations inside the furnace. As the oxide buildup continues over time, it eventually will convert to corundum (the hardest phase of Al2O3. When this occurs, two things happen as the oxide builds up: Furnace efficiency and melting capacity are reduced, and it becomes too late for a wall-cleaning flux to be effective. Eventually, the furnace must be shut down for relining. However, having a routine maintenance schedule for cleaning furnace sidewalls can minimize the problem. Operational procedures for skimming and drossing of the molten metal and cleaning of the furnace sidewalls are normally divided into

exclusive sections according to the design of the furnace. It is very typical to have a specific procedure and frequency for the melting chamber, the charging well, the dip-out well, and the degassing well. Furnace sidewall cleaning should be done once a week, but it is normal to do it three times per week. Often, a good and conscientious furnace tender determines the frequency of using the wall-cleaning flux. Even though different foundries have different furnace-cleaning procedures, it is very common that good fluxing practices be based on well-written procedures and that furnace tenders adhere to proper flux addition, proper sequence by which the burners must be turned on and turned off and the furnace doors opened and closed, minimum flux activation time, and thorough scraping of the furnace sidewalls. An important consideration in the selection of fluxes for reverberatory furnaces may be derived from the fact that, as the furnace capacity increases, the furnace tender’s tolerance for discomfort decreases. A furnace tender who is subject to incessant heat, noise, and other working-related discomforts is less likely to be as effective and efficient as possible, especially if he has to deal with dust, smoke, and fumes from the flux. In such cases, a typical flux manufacturer should be able to recommend an environmentally-friendly flux for the application. Holding/Casting Furnaces. Generally speaking, foundries that employ furnaces used for holding or direct production of castings must take care to use flux and flux practices wisely. Fluxing salts will react and degrade refractory materials used in the furnace construction, and, as shown previously, flux debris can become entrained within castings that are poured directly from fluxed furnaces. If metal has been adequately processed previously, there is usually no need to flux a casting furnace, such as a holder from which metal is poured by automatic or manual ladle. The dross

Fig. 14

buildup or skim on such furnaces is considered to be new oxides, and periodic skimming can prevent these new oxides from becoming entrained into the casting. If a casting furnace is fluxed, flux must be used sparingly, and it is crucial to remove the fluxing debris as completely as possible. Other Considerations. Good management philosophy will make the furnace tender’s comfort an important consideration when selecting a flux to use. A furnace tender who is happy with the working conditions usually has the right attitude toward proper flux use. Perhaps, the final consideration that a foundry must take into account when dealing with fluxes and metal cleanliness concerns is the fact that, despite the weaknesses of commercial cleanliness techniques, each technique has a strength that could be used to assess the molten metal quality. Despite operational anomalies, analysis costs, and/or production constraints to use present-day metal cleanliness equipment, aluminum foundries could benefit from their use to set and monitor their melting and fluxing practices (Ref 34, 35). For example, the latest molten metal evaluations have shown, among other findings, that flux injection provides the cleanest metal; the dirtiest molten metal is obtained with no flux in the process. Any other process would produce molten metal qualities between curves 5 and 8, as shown by the pressure filtration results in Fig. 14. In addition, molten metal quality assessment by ultrasound techniques showed that the molten metal in the as-melted condition typically had average particle (inclusions) sizes between 200 and 500 mm (0.008 and 0.020 in., or 0.2 and 0.50 mm). After any rotary treatment, with and without flux, the particle sizes ranged from 30 to 160 mm (0.0012 to 0.006 in., or 0.03 to 0.16 mm). The great majority were in a range of 40 to 90 mm (0.0016 to 0.0035 in., or 0.04 to 0.09 mm) (Ref 35).

Typical pressure filtration curves representing metal cleanliness values as a function of different fluxing treatments. Dirtiest molten metal is given by curve 5 (rotary degassing without any flux usage). Cleanest metal is given by curve 8 (rotary flux injection).

Aluminum Fluxes and Fluxing Practice / 239 REFERENCES 1. R. Fuoco et al., Characterization of Some Types of Oxides Inclusions in Aluminum Alloy Castings, Trans. AFS, Vol 107, 1999 p 287–294 2. P. Crepeau, Molten Aluminum Contamination: Gas, Inclusions and Dross, Aluminum Alloy Castings, AFS Fourth International Conference on Molten Aluminum Processing, 1995, p 1–13 3. D. Apelian and S. Shivkumar, Molten Metal Filtration — Past, Present, and Future Trends, AFS Second International Conference on Molten Aluminum Processing, 1989, p 14–1 to 14–36 4. C. E. Eckert, The Origin and Identification of Inclusions in Foundry Alloys, AFS Third International Conference on Molten Aluminum Processing, 1992, p 17–50 5. J. Campbell, Castings, 2nd ed., Butterworth-Heinemann, Oxford, 1988, p 11–26 6. D. Apelian, How Clean Is the Metal You Cast? The Issue of Assessment: A Status Report, AFS Third International Conference on Molten Aluminum Processing, 1992, p 1–15 7. R. Gallo, Development, Evaluation, and Application of Granular and Powder Fluxes in Transfer Ladles, Crucible, and Reverberatory Furnaces, AFS Sixth International Conference on Molten Aluminum Processing, 2001, p 56–70 8. L.C.B. Martins and G.K. Sigworth, Inclusion Removal by Flotation and Stirring, AFS Second International Conference on Molten Aluminum Processing, 1989, p 16.1–16.28 9. D.V. Neff, Continuous, Re-Usable Filtration Systems for Aluminum Foundries, AFS Fourth International Conference on Molten Aluminum Processing, 1995, p 121–137 10. J.R. Schmahl, L.S. Aubrey, and L.C.B. Martins, Recent Advancements in Molten Aluminum Filtration Technology, AFS Fourth International Conference on Molten Aluminum Processing, 1995, p 71–102 11. L.C.B. Martins, Chemical Factors Affecting the Separation of Inclusion from Molten Aluminum, AFS Third International Conference on Molten Aluminum Processing, 1992, p 79–91 12. P. Crepeau et al., Characterization of Oxide Sludge, Dross and Inclusions in Aluminum Melting and Holding Furnaces, AFS Third

13. 14.

15.

16.

17.

18.

19.

20. 21.

22.

23.

24.

International Conference on Molten Aluminum Processing, 1992, p 51–77 T.L. Mansfield and C.L. Bradshaw, Ultrasonic Inspection of Molten Aluminum, Trans. AFS, 1985, p 317–322 D. Doutre et al., Aluminium Cleanliness: Methods and Applications in Process Development and Quality Control, Light Met., 1985, p 1179–1195 R.I. Guthrie and J.E. Gruzleski, Quantitative Measurement of Melt Cleanliness in Aluminum Silicon Casting Alloys, AFS Fourth International Conference on Molten Aluminum Processing, 1995, p 15–28 A. Simard, F.H. Samuel, et al., Capability Study of the Improved PREFIL-Footprinter to Measure Liquid Aluminum Cleanliness and Comparison with Different Techniques, AFS Sixth International Conference on Molten Aluminum Processing, 2001, p 93–112 J.L. Roberge and M. Richard, Qualiflash Apparatus for Testing the Inclusion Quality of Aluminum Alloy Baths, AFS Fourth International Conference on Molten Aluminum Processing, 1995, p 29–42 E.L. Rooy and E.F. Fisher, Control of Aluminum Casting Quality by Vacuum Solidification Tests, Trans. AFS, Vol 76, 1968, p 237–240 D. Dispinar and J. Campbell, Critical Assessment of Reduced Pressure Test, Part 1: Porosity Phenomena, Int. J. Cast Met. Res., Vol 17 (No. 5), 2004, p 280–286 D.E. Groteke, Improve Your Vacuum Test, Mod. Cast. Mag., Aug 2007, p 28–31 N.D.G. Mountford, I.D. Sommerville, et al., Sound Pulses Used for On-Line Visualization of Liquid Metal Quality, Trans. AFS, 1997, p 939–946 N.D.G. Mountford, I.D. Sommerville, et al., Laboratory and Industrial Validation of an Ultrasonic Sensor for Cleanliness Measurements in Liquid Metals, Proceedings of International Conference on Cast Shop Technology, TMS-AIME, 2000, p 304–310 R. Gallo et al., Ultrasound for On-Line Inclusion Detection in Molten Aluminum Alloys: Technology Assessment, AFS International Conference on Structural Casting of Aluminum, 2003, p 29–42 R. Gallo, Measurement of Hydrogen Content in Molten Aluminum Alloys Using Crucibles and Reverberatory Furnaces for Sand and Permanent Mold Castings, AFS

25.

26.

27.

28.

29.

30. 31.

32. 33.

34. 35.

36.

Fourth International Conference on Molten Aluminum Processing, 1995, p 209–225 F. Painchaud and J.P. Martin, The New Alscan Analyzer: Easy to Use, Reliable, On-Line Measurement of Hydrogen in Liquid Aluminum Alloys, AFS Second International Conference on Molten Aluminum Processing, 1989, p 20.1–20.21 C. Schwandt et al., A Novel Electrochemical Hydrogen Analyzer for Use in Molten Aluminum and Its Alloys, AFS Sixth International Conference on Molten Aluminum Processing, 2001, p 149–157 D.E. Groteke et al., Foundry Methods for Assessing Melt Cleanliness, AFS Fourth International Conference on Molten Aluminum Processing, 1995, p 55–69 D. Dispinar and J. Campbell, “A Comparison of Methods Used to Assess Aluminum Melt Quality,” Shape Casting: TMS, The Second International Symposium, 2007 D. Apelian and S. Dasgupta, Interaction of Initial Melt Cleanliness, Casting Process and Product Quality: Cleanliness Requirements Fit for a Specific Use, AFS Fifth International Conference on Molten Aluminum Processing, 1998, p 234–258 K. Strauss, Applied Science in the Casting of Metals, Pergamon, 1970, p 253–256 M.H. Kogan, Design and Development of Fluxing Agents for the Aluminum Foundry Alloys, AFS Second International Conference on Molten Aluminum Processing, 1989, p 25–1 to 25–12 T.A. Utigard et al., The Properties and Uses of Fluxes in Molten Aluminum Processing, J. Met., Nov 1998, p 38–43 S.R. Sibley Granular Fluxes for Aluminum Alloys, Environmental and Technological Advances, AFS Fourth International Conference on Molten Aluminum Processing, 1995, p 417–430 D. Neff, Ensuring Quality Prior to Pouring, Mod. Cast., June 2006, p 27–29 R. Gallo, Cleaner Aluminum Melts in Foundries: A Critical Review and Update Transactions, 112th AFS Casting Congress, May 17–20, 2008 (Atlanta, GA) D. Neff, The Benefits of Flux Injection Combined with Rotor Dispersion for Producing Clean Aluminum Casting Alloys, Proceedings, Third International Conference on Molten Aluminum Processing, Nov 1992 (Orlando, FL), AFS

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 240-254 DOI: 10.1361/asmhba0005301

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Modification of Aluminum-Silicon Alloys* G.K. Sigworth, Alcoa Primary Metals

ALUMINUM-SILICON ALLOYS are widely used in components where good strength and light weight are required or where corrosion resistance and good castability are needed. However, the commercial application of these materials depends on the control of the structure of the silicon phase. This control is usually called modification. Because of its commercial importance, modification has been the subject of several hundred studies since its discovery in 1921 (Ref 1). This article does not attempt to review the full extent of the available literature. Instead, the focus is on aspects important for the commercial production of castings. An examination of the literature shows that modification is a complex and controversial subject. Many conflicting theories have been proposed. Hence, this article by necessity reflects the personal opinions and interpretations of the author. The reader desiring more information and other viewpoints on this fascinating subject is referred to a number of excellent surveys available in the literature (Ref 2–12). This article considers the modification process in hypoeutectic and eutectic alloys, which differ only in the relative volume fraction of primary aluminum and aluminum-silicon eutectic. The refinement of hypereutectic alloys is not considered. In hypereutectic alloys, phosphorus is added to produce finer primary silicon particles, but the eutectic is not modified (Ref 3, 13–17). An unmodified alloy contains large flakes of brittle silicon, which cause the casting to have poor ductility. Unmodified alloys often have elongations no more than a few percent, and the fracture surface is primarily brittle. With a successful modification treatment, the silicon assumes a fine, fibrous structure. These fibers appear to be small, individual particles on a polished surface, but etching some of the aluminum from the surface shows that silicon is connected in a seaweed- or coral-like structure. The elongation of the modified alloy is improved significantly, and an examination of

the fracture surface reveals a dimpled texture normally associated with ductile failure. Figures 1 and 2 show the structure of modified and nonmodified A356 alloy, as observed on polished and deep-etched surfaces (Ref 18). Fracture surfaces can be seen in Ref 19.

The effect of modification on as-cast tensile properties is given in Table 1 for several alloys. Modification has a significant effect on ductility. This is the basis for a test that “old timers” sometimes use in the foundry. It is possible to pour a small bar casting. One end of the bar is

50 mm (a)

Fig. 1

(b) Structure of nonmodified A356 alloy solidified at 1.5 to 2
C/s (2.7 to 3.6
F/s). (a) Polished surface. Original Magnification: 400. (b) Deep-etched surface. Original Magnification: 1000. Source: Ref 18

50 mm (a)

Fig. 2

(b) A356 alloy modified with 156 ppm Sr and solidified at 1.5 to 2
C/s (2.7 to 3.6
F/s). (a) Polished surface. Original Magnification: 400. (b) Deep-etched surface. Original magnification: 1000. Source: Ref 18

*Used with permission from American Foundry Society (www.afsinc.org). Copyright 2008 from the following: International Journal of Metalcasting (ISSN 1939-5981), Vol 2 (No. 2), 2008, p19–41, and AFS Transactions (ISBN: 978-0-87433-338-1), Vol 116, 2008, p115–139 of the article by G.K. Sigworth, “The Modification of Al-Si Casting Alloys: Important Practical and Theoretical Aspects.”

Modification of Aluminum-Silicon Alloys / 241

Yield strength, 0.2% Alloy

Time, hr

Reported mechanical properties in the as-cast condition Ultimate tensile strength

Cast product

Modification

MPa

Ksi

MPa

Ksi

Elongation, %

Reference

13% Si

Sand-cast test bar Permanent mold test bar

... ... ... ... ... ... 112 108 125 125 126

... ... ... ... ... ... 16.2 15.6 18.1 18.1 18.3

124 193 193 221 180 210 137 159 168 191 193

18.0 28.0 28.0 32.1 26.1 30.5 19.9 23.1 24.4 27.7 28.0

2 13 3.6 8 5.5 12.0 1.8 8.4 6.0 12.0 12.8

20

13% Si

None Na-modified None Na-modified None 0.007% Sr None Sr-modified None Sr-modified 0.06% Sr

359

Permanent mold test bar

A413

Sand-cast test bar

A413

Permanent mold test bar

A413

Sample from auto wheel

20 21 22

Heat Treated Castings An examination of Fig. 1 and 2 and Table 1 makes it abundantly clear that the mechanical properties of castings depend strongly on the size and shape of silicon crystals. One way to improve ductility is to employ a chemical modification, as described previously, by adding a small amount of sodium or strontium. Another way is to employ a thermal modification by subjecting castings to heat treatment. Few foundrymen understand that a thermal modification is possible or that standard heat treatment schedules are designed for nonmodified alloys. Consider the schedules recommended for the solution heat treatment of A356 alloy by the American Foundry Society (Ref 26):  Permanent mold castings: 12 h at 538
C

(1000
F) for T6 temper; 6 to 12 h for T61 temper




4

2

1

Perm

anen

t mo

ld

400

Sand

350

Sr-modified Nonmodified

22 22

8

450

300 0

0.1

0.2

cast

0.3

t,−1/3

Fig. 3 clamped in a vise, and the free end is hit with a hammer. The energy required to break the bar and the appearance of the fracture surface indicate when the alloy is modified. Sodium was the modifier first used by foundrymen. It was employed for many years, especially in aluminum-silicon alloys low in magnesium. These alloys provide good elongation with modest strength in the as-cast condition and were used in applications requiring reasonable ductility, such as automotive wheels. However, it is difficult to add sodium, which has low solubility in molten aluminum and a high vapor pressure, so its recovery is erratic. Sodium is also rapidly lost to oxidation after the addition (Ref 23). These problems made it difficult to control sodium modification. In the 1970s, strontium began to replace sodium as the preferred modifier. Strontium is easily added via master alloys with nearly 100% recovery, and its loss to oxidation is slow (Ref 23). With strontium, the control of microstructure is easier and the mechanical properties in castings are more consistent (Ref 24, 25). Today (2008), the use of sodium as a modifier has almost completely disappeared. For this reason, the primary focus of this article is on strontium modification.

20

500

Q, MPa

Table 1




 Sand castings: 12 h at 538 C (1000 F) for

Mechanical properties (Q) versus solution time (t) in A356-T6 alloy castings. Source: Ref 27

T6 temper A detailed study of the solution heat treatment process (Ref 27) has shown that only 60 min is needed to place the magnesium and silicon in solution in permanent mold castings. The additional 11 h recommended for the T6 solution heat treatment are used to modify (spheroidize) the shape of silicon crystals. The practical consequence is important: When aluminum-silicon alloys are chemically modified with sodium or strontium, the solution heat treatment time may be reduced. This observation was first noted by Levy in 1978 (Ref 28), who described steps undertaken at Reynolds Metals to increase ductility in lowpressure die-cast aluminum wheels. Prior to this time, most aluminum wheels in the United States were sold to the automotive aftermarket. The original equipment manufactures wanted an elongation of 10%, whereas 4% was the typical value in cast wheels. Levy found that 10% elongation could be achieved by:

for the quality index follow a straight line when plotted against the inverse of the cube root of solution time (minutes). The corresponding solution time (in hours) is also indicated at the top of Fig. 3. It is useful to consider what happens to the silicon phase during solution heat treatment and how the change in structure influences the quality of aluminum castings. A study of nonmodified alloys under a hot stage microscope (Ref 31) showed that silicon plates (in the ascast material) first break up into smaller pieces. This breakup happens rapidly, in an hour or two. There then follows a period where the small silicon pieces gradually coarsen and become more spherical. The same process occurs in modified alloys (Ref 27, 29). Boeing engineers conducted extensive studies of highquality aerospace castings. They examined the effect of solution heat treatment time along with other process variables. They finally concluded (Ref 32) that the quality of A357 alloy castings was well represented by the equation:

 Maintaining a low iron content in the metal  Modifying the metal (first with sodium, then

Q ¼ 101  N 0:5  129

later with strontium)  Using a shorter aging time These procedures are now standard practice, and aluminum castings are used extensively for wheels in new automobiles. Levy also noted that the solution time with modified metal could be reduced from 8 to 4 h. The reduced solution time possible with strontium modification is now also a common practice. These considerations led to detailed studies (Ref 27, 29) on the solution heat treatment of nonmodified and strontium-modified A356 alloy castings. Some of these results are plotted in Fig. 3 for permanent mold (ASTM B108) and sand mold test bar castings. The mechanical properties of the castings in Fig. 3 have been characterized by the quality index developed empirically by Drouzy et al. (Ref 30): Q ¼ UTS þ 150 log Ef

(Eq 1)

where the quality (Q) and the ultimate tensile strength (UTS) are in MPa, and Ef is the elongation to fracture in a tensile test. The data points

þ

N 0:5 AR

1100  98  P 0:5  570 AR

(Eq 2)

where Q is the quality index (in MPa), N is the cell count (aluminum cells per 0.0001 in.2), AR is the average aspect ratio of the silicon particles, and P is the area percent of porosity found on the polished surface of test specimens. The magnitude of the coefficient of the third term (1100 MPa) shows the importance of silicon particle shape (AR) on casting quality and gives credence to theoretical studies that suggest that cracking of silicon particles controls tensile elongation and quality in aluminum-silicon alloy castings (Ref 33). Studies made by the author establish that the cost for solution heat treatment is approximately 0.5 cents/lb-h. With strontium modification, the solution time required can be cut in half. For permanent mold castings, this represents a saving of at least 4 h, or 2 cents/lb. This amount is significant in today’s (2008) competitive market and is nearly ten times the cost of the strontium addition. Moreover, halving the

242 / Molten Metal Processing and Handling solution time doubles the throughput of heat treatment furnaces, removing a production bottleneck in many foundries. These are important benefits that often justify the modification of heat treated castings. The aforementioned results show that, when long solution times are employed, non- and strontium-modified permanent mold castings have almost the same tensile properties. In fact, when examining the microstructure of castings subjected to a standard solution heat treatment, it is difficult to distinguish any difference in the silicon phase. (Consider, for example, Fig. 1 of Ref 8 or Fig. 10 of Ref 29.) Hence, one may ask: “Why modify with strontium?” One answer is that modification usually changes the distribution of porosity and shrinkage in a casting, so that strontium additions may cure shrinkage problems. This topic has been the subject of a great deal of controversy over the years and is considered in detail as follows.

Porosity and Shrinkage Formation In most aluminum foundries, one-half to threequarters of all scrap is caused by the undesirable formation of porosity or shrinkage or by problems directly related to these defects. Lack of pressure tightness, poor quality on machined surfaces, low elongation and loss of tensile strength, and poor fatigue strength are problems related to porosity and/or shrinkage formation. Unfortunately, there is no single recipe that foundrymen can use to produce good castings. A number of factors must be

considered and controlled, and the balance point or “sweet spot” that produces the best results in one casting may give poor results in another. The situation is complex, poorly understood, and subject to a good deal of controversy. This section first considers porosity and how modification changes porosity formation in a casting. Then, shrinkage is considered as well as how modification changes feeding and shrinkage formation. The amount of porosity in a casting depends on several factors, including:      

Solidification rate Gas content Metal cleanliness Modification Grain refinement Pressure in the casting

Four of these factors were studied by Fang and Granger (Ref 34). They poured directionally solidified castings of A356 alloy, which had a water-cooled copper chill at the bottom and were solidified directionally upward from this chill. Because of the directional solidification, these castings had more than adequate feeding, and shrinkage formation was not an issue. The amount of porosity and the average size of the pores were measured by quantitative metallography at various distances from the chill. The solidification rate at these locations was determined by thermocouples placed in the mold during solidification. The gas content of the molten metal was measured directly by the Telegas instrument prior to solidification.

The porosity of the castings produced as a function of solidification rate and gas content is plotted in Fig. 4. Several important observations can be made from these results: 1. Solidification rate has an overriding influence on the amount of porosity in a casting. Castings that freeze quickly can tolerate high contents of hydrogen gas. Slowly cooled castings, however, easily form significant amounts of porosity. This is why low gas levels are needed for best quality in large sand castings, but gas is sometimes added to permanent mold castings. 2. Grain refinement tends to reduce the amount of porosity. (Compare the results for castings 1 and 3 in Fig. 4b). Grain refinement also reduces the size of pores. This occurs because hydrogen gas pressure does not usually become sufficiently high for pores to form until the last 5 to 10% of solidification. Hence, pores must fit into the spaces left between primary aluminum grains. Smaller pores produce a beneficial effect on fatigue life. 3. Modification tends to increase the amount of porosity formed, at least at the relatively high levels of gas (0.31 cm3/100 g) present in this study. (Compare the results for castings 2 and 5 in Fig. 4b). The question of porosity distribution in sodiumand strontium-modified castings was studied by Pechiney and Montupet engineers (Ref 25).

0.8 0.85 ⬚C/s 1 0.7

A356 Alloy

0.5

Castings 1, 2, 4 Nonmodified Grain refined

0.4

0.3

2.6 ⬚C/s

Pore volume fraction, %

Pore volume fraction, %

0.6

10−1

(5)

(2)

10−2

(5) (2) (1) (3) (4)

0.2

0.1

7.5 ⬚C/s

10−3 10−1

0 (a)

Fig. 4

0.2

Hydrogen content, cm3/100 g Pore volume fraction in A356 alloy castings. Source: Ref 34

0.31*** 0.31** 0.25* 0.25** 0.11**

(4)

No grain refiner Grain refined Grain refined and modified

28 ⬚C/s 0.1

(1) (3)

Casting No. H2content (cm3/100g)

0.3 (b)

1

10 Cooling rate, ⬚C/s

102

Modification of Aluminum-Silicon Alloys / 243 As part of their study, they poured an end-chilled sand mold bar casting. The design of the casting is shown in Fig. 5. The porosity was determined in this bar by measuring the density (specific gravity) in sections cut along its length. These measurements were made in sodium-, strontium- and nonmodified A356 alloys. In the sodium- and strontium-modified alloys, there was an increased amount of porosity near the riser. The results for the strontium-modified alloy are shown in Fig. 5. (The results for sodium-modified alloys were similar but are not duplicated here.) The increase in porosity is most pronounced at higher addition levels of strontium. In practice, this means that one should use the minimum amount of strontium needed to produce an acceptable level of modification. This is generally approximately 0.008 to 0.012% or 80 to 120 ppm Sr. There are a number of actions that foundrymen can take to minimize the increase in porosity associated with strontium-modification. Aside from keeping strontium levels at a minimum, one should consider:  Lower gas in the metal reduces the amount of

porosity. Longer degassing times and low metal temperatures will help. The influence of gas content is illustrated well in a paper by Emadi and Gruzleski (Ref 35). They produced directionally solidified castings (similar to Fang and Granger in Fig. 4). Some of their results have been reproduced in Fig. 6. At a low gas content (0.13 cm3/100 g), the increase in porosity with modification is negligible, except at the slowest cooling rates (compare curves 1 and 2). At a higher gas content (0.26 cm3/100 g), there is a significant increase in porosity when strontium is added (curves 3 and 4). The importance of foundry degassing practices should be obvious from these results. Ingate 250

 Increasing solidification rate always reduces

porosity, but this is especially true for strontium-modified alloys. If excessive porosity is found in part of a casting, this can often be removed by using chills in sand castings or conductive die coatings (or increased water cooling) in permanent mold castings.  Grain refinement is beneficial and tends to counteract the porosity formation caused by strontium additions. For this reason, it is a good idea to use grain refinement in combination with strontium modification. Grain refinement is best accomplished by an addition of 10 to 20 ppm B in the form of 5% Ti-1%B or 3%Ti-1%B rod.  Mold design is also an important consideration in new castings. It is important to keep sources of heat away from areas where porosity is not desired. For example, in a wheel casting, one would not want to gate metal into an area that will be machined to a high polish and coated. With the increased availability of powerful casting simulation programs, it is possible to determine with great precision where porosity may form before machining patterns or molds. In fact, when a casting is designed properly, an xray examination will show that the increased porosity formed by strontium additions is located in the base of the riser(s). This represents the ideal case.  Filtration will remove inclusions, making it more difficult for gas pores to nucleate during solidification.

Modification Mechanisms At this point, it is useful to consider the mechanisms responsible for silicon modification and that cause the changes in porosity noted previously. When these mechanisms are understood, it is easier to see how strontium modification produces other effects:

37.5

Chills

1.2 Unmodified 1 3 Sr-modified 2 4

Riser

Specific gravity

2.68

0

Testpiece

Length along testpiece, in. 4 6 8 2

10

2.67

No Sr

2.66

0.01% Sr

2.65 0.07% Sr 0.10% Sr

2.64 2.63

0

Chill end

Fig. 5

0.8

3

Gas content 0.13 0.26 0.13 0.26

castings

 Increased segregation of iron compounds  Increased segregation of copper as blocky

CuAl2 in copper-containing alloys Cooling curves recorded during the solidification of aluminum-silicon casting alloys are obtained by pouring a liquid metal sample into a small crucible or sand cup, into which a thermocouple is placed. The interpretation of the temperature profile recorded during freezing is called thermal analysis. Thermal analysis can be a powerful tool, because the cooling curve of the alloy acts like a fingerprint to identify the phases forming during solidification. Thermal analysis has a long history as a laboratory tool for the study of aluminum-silicon alloys. Two outstanding early examples are the 1950 work by Thall and Chalmers (Ref 20) and the 1966 paper by Crosley and Mondolfo (Ref 3). The latter experiments, in particular, helped to lay the foundation for the present understanding of aluminum-silicon alloys. In the early 1980s, two versions of commercial thermal analysis equipment became available (Ref 36, 37), one of which was subsequently used widely as a foundry tool to evaluate grain refinement and modification treatments in the cast shop (Ref 24, 37). Unfortunately, no improvements have been made since that time, and the tool has fallen into disuse. It is rare to see thermal analysis equipment operating today (2008) in foundries. Also, researchers have proposed a number of advanced analytical techniques to determine silicon modification (Ref 18, 36, 38, 39), but none of these ideas has been transferred onto the foundry floor. There appears to be a useful opportunity for someone willing to convert this knowledge into a modern, inexpensive, and easy-to-use computer-based thermal analysis system. Researchers in Australia poured castings with a base Al-Si-Mg alloy and then made additions of strontium and phosphorus to the same metal (Ref 40). Thermal analysis curves for the three alloys are given in Fig. 7. Important clues to operating mechanisms are available from these curves. Before extracting the information, however, it is necessary to introduce a

0.6 2 0.4 1 0.2

0.12% Sr

2.62 2.61

Pore volume fraction, %

25

1.0

 Improved feeding from risers  A changed distribution of shrinkage in

200 50 100 150 Length along testpiece, mm

250

0 0.3

Distribution of porosity in an end-chilled sand casting. Source: Ref 25

0.5

1

2

3

4 5

Average cooling rate, ⬚C/s

Riser end

Fig. 6

Porosity versus freezing rate in an A356 alloy. Source: Ref 35

Fig. 7

Effect of strontium and phosphorus on cooling curves in an A356 alloy. Source: Ref 40

244 / Molten Metal Processing and Handling few definitions and some basic concepts. For this purpose, the cooling curve for a modified A356 alloy is employed, as shown in Fig. 8 (Ref 41). Figure. 8(a) shows the full curve. TC is the temperature at the center of a small crucible during freezing. dT/dt is the first derivative of the cooling curve, TC, and shows the instantaneous cooling rate in degrees per second. (The scale for the dT/dt curve appears on the right side of Fig. 8a.) Figure. 8(b) shows an enlarged version of the cooling curves for the solidification of the aluminum-silicon eutectic. (This is the area inside the box outlined by dashed lines in Fig. 8a.) Before the first eutectic forms, the cooling rate (dT/dt) is nearly constant at 0.6
C/s (31
F/s) (Fig. 8). At a temperature of approximately 570
C (1060
F), the first derivative starts to increase. This happens because the aluminum-silicon eutectic has begun to freeze, releasing latent heat into the sample. This point is called the eutectic nucleation temperature, TN. As more eutectic grains form and grow, the amount of the latent heat released becomes larger, until the cooling stops and a minimum temperature occurs, TMin. The sample ⬚C 630

1.5⬚C/s 1.3

620

1.1

Tc

0.9

610

0.7 0.5

600

0.3 590

0.1 −0.1

dT/dt

580

−0.5 −0.7 −0.9

560

−1.1 550 0

50

(a)

100

150

−1.3 200

Time, s

584

1

582

0.8

580

 The eutectic nucleation temperature (TN) is

reduced.

 The eutectic growth temperature (TG) is

reduced.

 The nucleation temperature is lowered with

respect to the growth temperature (that is, TN  TG). Phosphorus additions can be seen to have the opposite effects of those produced by strontium.

Eutectic Nucleation How does an addition of a few parts per million of strontium (or sodium) change the nucleation of the silicon eutectic? This is an interesting and controversial question that has been the subject of dozens of research papers and academic theses. In the author’s opinion, the most likely explanation is that strontium (and other modifiers) react with to remove or somehow neutralize AlP, which is an effective nucleant for the silicon phase. The supporting evidence for this hypothesis is outlined briefly as follows:  Phosphorus is added to hypereutectic alloys

−0.3

570

then begins to heat up. (This is called recalescence.) The sample continues to heat until the steady-state growth temperature for the eutectic, TG. From the cooling curves in Fig. 7, strontium additions are demonstrated to produce the following effects:

Al-10%Si alloy. The sample was quenched from the semisolid state. The fine eutectic surrounding the larger silicon crystals formed during this quench. (The dashed lines in Fig. 9 indicate the outline of primary dendrites forming prior to eutectic solidification.) Some researchers have reported finding Al-P-O particles, but this observation appears to be an artifact. AlP is a reactive compound that combines readily with oxygen in the air and perhaps with water used in polishing media. The oxygen is therefore the result of contamination. This was shown by elegant (and complicated) ion milling techniques used to remove the contaminated oxygen-containing layers. The same study (Ref 43) also showed there is a definite epitaxial crystallographic relationship between AlP and the silicon crystal.  Modifiers react with phosphorus. The elements that have been reported to produce modification (Ref 7, 41, 44) are indicated by the shaded areas in Fig. 10. These are all reactive metals that form compounds with phosphorus. As a consequence, they all have the capability to neutralize AlP particles in aluminum melts. This interpretation appears to be supported by thermal analysis data (Ref 41).  Phosphorus additions poison modification. A number of researchers have shown that high phosphorus levels make it difficult to modify aluminum-silicon alloys. Some of the maps produced by these studies are shown later, when the effects of elemental additions on the modification process are

to reduce (or refine) the size of primary silicon. In 1966, Crosley and Mondolfo (Ref 3) found phosphorus-containing particles inside primary silicon crystals. Based on crystallographic data, they proposed that AlP has an epitaxial relationship with silicon and would be an effective nucleus.  AlP nucleates silicon in hypoeutectic alloys. This has been observed by a number of researchers. An example is shown in Fig. 9. This particle was observed in a nonmodified

I

IIA

IIIB

H

Li

Be

Na

Mg

K

Ca

Sc

Rb

Sr

Y

Cs

Ba

La

0.6

Tc

0.4

578 dT/dt 576

0.2

574

0

572

−0.2

570

TG

TN

−0.4 −0.6

568 Tmin

566 564 80 (b)

Fig. 8 Ref 41

−0.8 −1

100

120

140

160

Time, s Thermal indicators in eutectic solidification. See text for an explanation of parameters. Source:

Fig. 9

Phosphorus-rich particle inside silicon crystal in an Al-10%Si alloy. Source: Ref 42

Fig. 10

Elements that have been observed to produce modification in aluminum-silicon alloys

Modification of Aluminum-Silicon Alloys / 245 discussed. One example of this poisoning effect is shown in Fig. 11.  Strontium poisons AlP in hypereutectic alloys. The interaction between AlP and strontium is reciprocal and also occurs in hypereutectic alloys. In these alloys, phosphorus is normally added to nucleate and refine primary silicon. When strontium is added in sufficient quantities, and some time is allowed for the reaction between AlP and strontium to occur, the size of primary silicon increases significantly and the number of primary silicon particles is reduced (Ref 46, 47).  Modifier additions change eutectic grain size. When a modifier is added, nucleants for the silicon eutectic are removed. The most significant consequence is a larger eutectic grain size and the change in porosity distribution discussed previously. Due to the importance of this subject, it is worthy of discussion in a separate section.

Fig. 12(b). In this case, no silicon eutectic grains nucleated in the liquid. Instead, the eutectic first formed at the outside of the casting (in contact with the mold) and grew toward the center as a planar solidification front. Unfortunately, no thermal analysis was conducted in this study. From the structures observed, however, it would be reasonable to assume that for nonmodified alloys, the eutectic nucleation temperature is greater than the growth temperature (that is, TN > TG). This may be inferred from the appearance of eutectic grains in the liquid phase ahead of the solid/liquid interface. Conversely, for sodium-modified

alloys, the nucleation temperature is lower than the growth temperature (TN < TG). These differences have an important effect on the formation of porosity and shrinkage. Inferences can also be made about the eutectic grain structure by anodizing a polished surface and looking at the sample in a microscope with polarized light. This procedure reveals the grain structure (size and relative orientation) of the aluminum phase in the eutectic. This procedure was used in the 1975 study by Jenkinson and Hogan (Ref 49). More recently, a number of studies have been undertaken by researchers in Australia.

Eutectic Grain Structure Most popular casting alloys are hypoeutectic. For example, A356 alloy contains 7% Si, compared to the eutectic composition of 12.7% Si. Consequently, the as-cast structure contains 50 to 60% primary aluminum grains surrounded by a layer of aluminum-silicon eutectic. It is difficult to discern the eutectic grain structure in this material by simple metallographic observation. One way to observe eutectic grains is to partially solidify the alloy and subject a sample to a rapid quench. This procedure is usually called an interrupted quench. (That is, solidification is interrupted by a rapid quench.) This procedure was used in the 1963 research paper by Kim and Heine (Ref 44). It is possible that they were the first to observe the change in the eutectic grain structure of aluminum-silicon alloys that occurs with modification. Unfortunately, their observation was not connected with the important issues at hand (porosity and shrinkage distribution in castings), and it was forgotten. In 1981, Flood and Hunt used interrupted quench experiments to study sodium modification of eutectic alloys (Ref 48). Their results clearly showed that sodium prevented the nucleation of eutectic grains. Figure. 12 is taken from their study. Figure. 12(a) shows the structure of the nonmodified alloy. The lighter (white) portion of the ingot was allowed to solidify normally. The darker-gray areas were quenched after approximately one-third of solidification was completed. It can be seen that numerous eutectic grains nucleated in this nonmodified alloy and grew into small, spherical colonies of silicon crystals. The arrows in this figure show some of these eutectic grains. It is noteworthy that these grains are isolated, or completely surrounded by liquid metal. The sodium-modified alloy is shown in

Fig. 11

Effect of strontium and phosphorus additions and holding time. Mechanical properties are for an as-cast Al-12%Si alloy. Source: Ref 45

Nonmodified alloy (a)

Fig. 12

Na-modified alloy (b)

Structure of aluminum-silicon eutectic alloy observed by an interrupted quench. Arrows show individual eutectic grains. Source: Ref 48

246 / Molten Metal Processing and Handling They developed special techniques to see eutectic grains in aluminum-silicon alloy castings. For the researcher interested in experimental detail, the new procedures are reviewed in Ref 50 and 51. For the purpose of this article, however, it is sufficient to extract the most important observations that have practical significance for foundrymen. The eutectic structures observed in aluminum-silicon alloy castings (Ref 42, 50–55) are summarized in Fig. 13. This figure shows schematically the structure of three Al-7%Si alloy castings at a point during solidification when the primary aluminum grains have formed and the eutectic solidification is just beginning. Figure 13(a) shows the nonmodified structure. The black areas indicate regions where aluminumsilicon eutectic has solidified. There is a shell of eutectic in contact with the mold wall, and numerous small, solid grains of eutectic are in contact with primary aluminum. In this case, eutectic grains nucleate and grow on existing aluminum dendrites. Figure 13(b) shows the strontium-modified alloy. In this case, the number of eutectic grains is smaller, and the eutectic grain size is much larger. In experiments conducted by McDonald et al. (Ref 53–55), the eutectic grain size increased by an order of magnitude when strontium was added. The eutectic grains were a few hundred microns in diameter in nonmodified alloys, and a few thousand microns in strontium-modified castings. Note also that the solid shell of eutectic is thinner when strontium is present. In fact, in large sand or investment castings there may be little solidified eutectic at portions of the mold surface. These changes have important ramifications for shrinkage formation. The structure found in sodium-modified alloys is shown in Fig. 13(c). In this case, no nucleation occurs in the liquid, and eutectic growth is from the mold walls toward the center. (This case was also shown in Fig. 12b for an alloy of eutectic composition.) In the discussion of Fig. 4, it was noted that grain refinement (of the primary aluminum

(a)

Fig. 13

(b)

phase) had a beneficial effect on the amount of porosity. The size of the pores was also reduced. From Fig. 13, it can now be seen that strontium additions have the opposite effect: The size of eutectic grains is increased. This change is almost certainly responsible for the increase in porosity found in most modified castings. To one familiar with the technical literature in this area, this simple conclusion may be surprising. There has been a good deal of controversy over the last 25 years, and nearly a dozen theories have been proposed to account for the change in porosity observed in modified alloys. Surprisingly, eutectic grain size was never identified as a possibility, presumably because no one could measure it. For those interested in the scientific history and background, the other theories and the literature are reviewed elsewhere (Ref 52, 54, 56–58). It may be useful to present a simple calculation. Fuoco and co-workers (Ref 59) duplicated the experiments of the Pechiney researchers summarized in Fig. 5 and also conducted interrupted quench experiments in small crucibles. Their melts were maintained at a gas content of 0.13 cm3/100 g (standard temperature and pressure, or STP). When strontium was added as a modifier, the amount of porosity increased, and the pore size increased to an average diameter of 300 mm. Now, the question is this: From what volume of liquid metal (eutectic grain size) did this pore receive its hydrogen gas? It is assumed that all of the gas originally present in the melt finds its way into a pore during solidification. If this is true, how much porosity forms? One hundred grams of metal, at a liquid density of 2.35 g/cm3, has a volume of 43 cm3. The gas volume at standard temperature and pressure is 0.13 cm3. However, at the eutectic temperature (and 101 kPa, or 1 atm, pressure), this gas will have a volume of 0.13 cm3  (843/ 273) = 0.4 cm3. Thus, the amount of porosity that corresponds to a gas content of 0.13 cm3/100 g (at STP) is approximately:

(c)

Structure of aluminum-silicon eutectic grains in three semisolid Al-7%Si alloys. (a) Unmodified. (b) Strontium modified. (c) Sodium modified. Source: Ref 42

% porosity ¼ Volume of gas=Volume of metal ¼ 0:4 cm3 =43 cm3 ¼ 0:0093 or 0:93%

(Eq 3)

For the material to have that amount of porosity, each pore must be surrounded by a volume of solid (pore-free) metal having an average size 107 times larger than the pore ([1  0.0093]/0.0093 = 107). From this is derived: ffiffiffiffiffiffiffiffi Dgrain p ffi 3 107 or 4:75; thus Dpore Dgrain ffi 1425 mm

(Eq 4)

In nonmodified alloys, the eutectic grains are approximately 250 mm in size. Using the same calculation as in Eq 4, it is found that the pore size is much smaller—approximately 50 mm. From this simple calculation, one can see how a large eutectic grain size in modified alloys may contribute to a larger pore size in castings. The previous calculation assumed that all hydrogen was able to diffuse to the pore, and it corresponds to the case of slow solidification. During rapid freezing, much of the hydrogen is frozen or trapped in the solid metal. In this case, the amount of porosity is less and the pores are smaller. Unfortunately, this simple picture is not the complete story. The experiments of Fuoco et al. (Ref 59) and Anson et al. (Ref 56, 57) show that, at medium to high gas levels, pores nucleate or form earlier during the solidification process when strontium is present. The change in eutectic grain size cannot account for this observation. One possibility is a different nucleus for the pores. It is well known that oxide films nucleate porosity in aluminum castings (Ref 60, 61). Strontium additions have been observed to change the composition of the oxide on the surface of a eutectic aluminum-silicon alloy and to increase the oxidation rate of the melt (Ref 62). Strontium-containing oxides have also been observed to nucleate pores in strontium-modified castings (Ref 63, 64). So, it is possible that a strontium-containing oxide film is a better nucleant for porosity. The discussion now shifts to other important changes that occur when an aluminum-silicon alloy is modified. Feeding and Shrinkage Formation. Considering once more the structures indicated in Fig. 13, it is evident that relatively large channels of liquid eutectic will exist in the semisolid, strontium-modified castings. These open channels change the feeding patterns. Because there is a thin shell of eutectic at the surface, these channels sometimes come to the surface. The author has seen large, strontium-modified sand castings that exhibited a “wormhole” on the surface. These holes may go deep into the casting where an unfed thick section sucked liquid away from the surface. This left an empty hole where the eutectic feeding channel was located. More common is the problem of a rough surface, usually called an “orange-peel” defect, sometimes seen in investment castings. Here again, poor feeding in a thick section pulls liquid away from the surface, leaving

Modification of Aluminum-Silicon Alloys / 247 a rough surface of exposed, dry aluminum dendrites. This defect has been described in the literature (Ref 65). Strontium has also been shown to promote surface slumping in unfed sections of permanent mold castings (Ref 52, 66). Sodiummodified castings did not exhibit surface slumping, presumably because the solid outer shell of solidified eutectic is thicker. The literature regarding the effect of modifiers on feeding is contradictory. Iwahori and co-workers (Ref 67) conducted an elegant series of experiments that showed feeding stopped in nonmodified alloys when the aluminum-silicon eutectic began to freeze. In Fig. 13 (a), it is tempting to say that the many small eutectic grains in this structure act to close off liquid feeding paths between aluminum dendrites. Iwahori and his colleagues later (Ref 68) repeated the same experiments in modified alloys. Sodium-modified alloys fed longer (later into solidification), which meant that smaller risers could be employed without creating shrinkage in their casting. On the other hand, strontium-modified alloys were observed to behave in the same manner as nonmodified melts. Unfortunately, these experiments were conducted with 1.7 kg (3.7 lb) melts, so the results cannot be scaled up directly to melt practices found in operating foundries. A third study (Ref 69) reported that strontium made feeding worse in a simple bar casting, whereas three other papers (Ref 70–72) reported that strontium modification improved feeding. The author also has experienced many instances of strontium effects on shrinkage. In the early 1980s, the author began to work for a company that made and sold aluminumstrontium master alloys. During one foundry visit, they were having problems with shrinkage located on the machined deck face of lost-foam cylinder heads in a 319-type alloy. When they modified the alloy with strontium, the shrinkage defect disappeared. Since that time, the author has seen the same thing occur in many other castings, and others (Ref 42, 66, 73, 74) have

(a)

Fig. 14

(b)

reported the same effect of strontium on shrinkage. An example of the change possible is given in Fig. 14. This shows the internal structure at a hot spot in a permanent mold casting. In the nonmodified alloy (Fig. 14a), a large, concentrated shrink is found. When strontium is added (Fig. 14b and Fig. 14c), the shrink disappears and is replaced by dispersed microporosity. The author thinks that strontium helps to cure shrinkage by a combination of better feeding and increased porosity. Both tend to displace shrinkage. By looking carefully at the tops of risers during solidification, it will become apparent that feeding usually improves when strontium is used as a modifier. (The shape of the cavity at the riser top changes, showing that feeding occurs longer when the metal is modified.) Grain refinement also assists feeding, which is another reason to use it in combination with strontium. In the sodium-modified casting (Fig. 14d), a concentrated shrink is found. However, it is smaller in size than in the nonmodified alloy. Perhaps the shrink was reduced by improved feeding. Copper and Iron Segregation. The larger eutectic grain size in modified alloys tends to increase copper and iron segregation in the same way it increases segregation of gas. Several studies have noted that strontium promotes the formation of large, blocky CuAl2 particles (Ref 55, 75–77). These particles take longer to dissolve during solution heat treatment (Ref 75). There also is an observable change in the copper eutectic cooling curves (Ref 75, 77), so thermal analysis could be used to help control CuAl2 morphology in modified Al-Si-Cu casting alloys. The effect of strontium on copper segregation in an Al-10%Si-2%Cu alloy is shown in Fig. 15. Grain refinement of the primary aluminum phase helps to disperse the blocky CuAl2 particles (Ref 75). A similar segregation effect may also occur with iron intermetallic compounds, depending

(c)

on the iron content of the alloy. Figure 16 shows the distribution of b plates in an Al10%Si alloy that contains 0.11% Fe and 140 ppm Sr. The iron-containing plates are found at the boundaries of the eutectic grains.

Recommended Foundry Practices Consider how strontium is best used in the foundry. Each process step is considered individually. Adding Strontium. Making additions of strontium is relatively easy. There are several master alloy compositions and different products available for this purpose. Their recovery and speed of dissolution have been studied in detail (Ref 24, 39, 78–80). Waffle products tend to be slow to dissolve. Higher metal temperatures generally promote more rapid dissolution and higher recoveries. The exception is the 90Sr-10Al product, which is exothermic and goes into solution best when the melt temperature is low. Strontium recoveries are generally high with all products—90% or more. In past years, an Al-14Sr-10Si master alloy was marketed by some companies, but this was a terrible product. It was hard to dissolve, and it always gassed the metal. This is perhaps because the product contained high levels of calcium and barium (Ref 78). This master alloy has not been available in North America for many years, but the author was recently told that an overseas company is once again making this product. Foundrymen should be aware of potential problems with this strontium-silicon master alloy. Several workers have noted that strontium is slow-acting, showing an incubation time of 1 or 2 h (Ref 10, 24, 39, 45, 72). Modification was observed to improve with time after the addition, even though part of the added strontium may be lost to oxidation. An example of this effect was shown in Fig. 11. More recent research has established that most of the delay

(d)

Cross section of hot-spot area (spoke-rim junction) of permanent mold Al-10%Si-Mg alloy casting. (a) Unmodified alloy. (b) 50 ppm Sr. (c) 100 ppm Sr. (d) Na-modified. Source: Ref 42

248 / Molten Metal Processing and Handling

Distribution of b-phase in an Al-10%Si alloy. Arrows show b plates located at eutectic grain boundaries. Source: Ref 55

Fig. 16

100 µm

(a)

(b)

Fig. 15

Binary image showing distribution of CuAl2 phase in an Al-10%Si alloy casting. (a) 2 wt% Cu, 0 ppm Sr. (b) 2 wt% Cu, 100 ppm Sr. Source: Ref 55

Fig. 17

Quality index of A356-T6 alloy castings at three cooling rates. Curve 1, 1.5
C/s; curve 2, 0.5
C/s; curve 3, 0.08
C/s. Source: Ref 80

is caused by the time needed for intermetallic particles in waffle products to dissolve. Strontium-containing rod products usually have small particles of SrAl4 and are fast dissolving and fast acting (Ref 24, 39). With strontium master alloy rod additions, no more than approximately 10 to 15 min is needed to reach full modification.

The melt loss of strontium is slow (Ref 23), so it will be convenient for many foundrymen to have strontium added to their alloys by their ingot supplier. How Much to Add. It is generally accepted that the best silicon modification in A356 alloy occurs at a level of approximately 0.012% Sr, or 120 ppm. An example of how mechanical

properties have been observed to vary with strontium content is shown in Fig. 17 (Ref 80). This plot shows the quality index of A356 alloy at three different cooling rates. In practice, however, good modification can be obtained at significantly lower concentrations, depending on the purity of the alloy. In primary metal, the phosphorus level is usually below 10 ppm, and other bad impurities are absent, so full modification can be obtained with strontium contents as low as 60 to 70 ppm. Good modification can also be obtained at higher levels of strontium, although this tends to result in more porosity. The reason for the small drop in quality as strontium becomes more than approximately 0.03% is that higher levels result in coarser silicon particles (Ref 18, 81). Gas Pickup and Degassing. It is generally agreed that adding strontium does not result in any appreciable gas pickup (Ref 82, 83). Sodium additions, however, do add gas to the melt (Ref 84). The evidence regarding the effect of modifiers on gas pickup from the atmosphere is less clear, however. Experiments with small melts held under atmospheres having a controlled humidity (Ref 62) show that strontium additions do increase the gas by changing the composition of the oxide on the melt surface. Other laboratory experiments (Ref 83) suggest that strontium has no effect. The author’s experience in the foundry is in agreement with the observations offered by Hurley and Atkinson (Ref 24). As long as large strontium additions and high melt temperatures (>760
C, or 1400
F) are avoided, there is no significant increase in gas pickup from the atmosphere. Degassing of modified melts should be done using nitrogen or argon, preferably with a rotary impeller. No chlorine or other reactive gas (nor salts or fluxes) should be employed, because these will tend to burn out dissolved strontium. Contrary to an observed loss of sodium (Ref 84), degassing with inert gases does not result in a significant loss of strontium. Quality-Control Testing. Under normal conditions, a spectrographic analysis for strontium is sufficient to control modification in a

Modification of Aluminum-Silicon Alloys / 249 foundry using primary metal. As shown subsequently, however, some elements can poison the effect of modifiers. These may become a concern when secondary metals or scrap are incorporated into the furnace charge. In this case, it would be wise to check modification levels by the routine use of thermal analysis. This measurement technique has been found to be an acceptable quality-assurance tool (Ref 24, 37). Mold Reactions. The addition of strontium has been observed to increase hydrogen gas pickup in green sand mold castings (Ref 84, 85). This can become a serious problem in large, complicated castings. To a certain extent, this gas pickup can be avoided by careful gate and runner design and by use of mold coatings.

Growth of Silicon Eutectic A careful examination of the literature makes it evident that the crystallization of silicon in aluminum-silicon alloys is an exceedingly complex affair. In particular, the presence (or absence) of impurity elements has a significant effect on the structure formed. In the earlier discussion of cooling curves, it was noted that modifiers change the nucleation of silicon. This effect was considered in some detail, to explain effects ancillary to silicon modification (changes in porosity formation, feeding, etc.). There is also, however, a change in the growth of the silicon phase during chemical modification with sodium or strontium. This is evidenced by a 6 to 10
C (11 to 18
F) depression in the eutectic growth temperature, TG. The formation of silicon crystals in liquid aluminum-silicon melts is influenced by the high latent heat (and entropy) associated with freezing of silicon (Ref 86). This complicates the kinetics of the silicon crystals growing from the melt. In fact, depending on the impurities present and the solidification conditions, silicon can form an amazing range of structures. The papers by Fredriksson et al. (Ref 87) and Day and Hellawell (Ref 88) show the many chameleon-like forms that silicon can assume. A number of theoretical and experimental studies have been offered to explain the mysterious growth of silicon. A detailed review of this work is beyond the scope of this article, whose focus is on practical aspects of modification. Instead, the reader is referred to a number of key papers and reviews that provide an introduction to this field of inquiry (Ref 9, 11, 12, 55, 86–92). There are also directional solidification studies that show how the aluminum-silicon eutectic grows from liquid melts (Ref 49, 93, 94). Nearly 20 years ago, a committee of foundrymen (within the American Foundry Society, or AFS) considered the range of silicon structures commonly observed in commercial castings. They established a modification rating system (Ref 95). It is available as a wall chart that

may be purchased from AFS. A number from one to six is assigned, depending on the silicon structure observed. For the sake of discussion here, an abbreviated form of the rating system is given in Fig. 18. Each micrograph in Fig. 18 is for a polished, unetched sample of A356 alloy. The silicon phase has a dark-gray color. Iron-containing intermetallic phases are light gray, and aluminum is white. The following is a brief description of each rating. Nonmodified Structure (1). The fully nonmodified aluminum-silicon eutectic consists of large, acicular silicon flakes. This is the structure observed in normal commercial alloys when no modifiers are present. However, very rapid solidification can produce some refinement and a structure similar to rating number 2 or 3. Lamellar Structure (2). Small amounts of strontium produce some refinement of the large, acicular plates, so that fine lamellae of silicon are present in the eutectic. This structure is preferable to rating number 1, since sufficiently long solution heat treatments break up the lamellae and produce rounded silicon particles. Transitional Lamellar Structure (3). At slightly higher strontium levels, the lamellar plates of silicon become even finer and start to break down into smaller segments. The arrows in Fig. 18(3) indicate two plates that are starting to transform into an aligned row of individual fibers. The reason for this change in structure is perhaps best seen in studies of directionally solidified aluminum-silicon alloys (Ref 49, 93, 94). In nonmodified alloys, the silicon plates grow faster than aluminum, so that new cells of aluminum nucleate on the protruding silicon plates. As strontium is added, the growth of silicon is retarded, until the aluminum phase starts to overgrow the silicon. (The discussion by Major and Rutter in Ref 94 on this point is particularly lucid.) This overgrowth is starting to occur in this structure, as indicated by the arrows. The result is that some of the lamellar silicon plates are starting to break down into smaller, fibrous segments. Mostly Modified Structure (4). At this point, sufficient strontium has been added to completely transform the silicon from acicular/ lamellar forms into a modified, fibrous structure. However, some larger silicon crystals are still found in the structure, as indicated by the arrows in Fig. 18(4). The larger silicon crystals in this micrograph appear to have formed at a location where two eutectic grains have collided. Between two eutectic cells, the local growth rate will slow considerably, and when the strontium level is marginal, there will be a local loss of modification. (See the modification maps in Ref 49, 53, and 55 for a more detailed explanation.) Fully Modified Structure (5). This is the fibrous silicon that is normally considered to be the fully modified structure. Super modified Structure (6). On rare occasions, an extremely fine, fibrous eutectic was observed. The individual fibers are so small that they cannot be seen in an optical

micrograph. (Figure 18(b) is shown at 800 magnification.) They are, however, distinguishable with a scanning electron microscope at higher magnifications. Overall Ratings. It is fairly common to find a mix of structures in a casting. The usual way to assign a rating to a mixed structure is to multiply the area percentage of each structure by the rating number and average the values. For example, in a casting that is approximately 20% rating 3 and 80% rating 4, the overall rating for the casting would be: 0:2  3 þ 0:8  4 ¼ 3:8 Similarly, a casting that is 50% structure 4 and 50% structure 5 would have an overall modification rating number equal to 4.5.

Effect of Other Elements A number of elements have a significant effect on the silicon structure, even when present in small quantities. These effects are not usually of concern in primary alloys. For the foundry using secondary alloys, however, it is important that impurity elements be considered. Each element is briefly considered as follows. Phosphorus is present in most primary alloys at a level of 5 to 10 ppm. Even this very small amount has a significant effect on the silicon structure in foundry alloys. Studies have been made with very high-purity alloys (Ref 4, 96). A zone-refined alloy containing 0.6 ppm P exhibited a refined eutectic structure similar to a strontium-modified alloy (Ref 4), whereas an alloy containing 6 ppm P had the normal nonmodified structure shown in Fig. 18(1). Directional solidification studies by Major and Rutter (Ref 94) showed that phosphorus changes the relative growth kinetics of aluminum and silicon during eutectic solidification and the morphology of the solid/liquid interface. A number of other researchers have considered the effect of phosphorus in aluminum alloys (Ref 4, 22, 40, 43, 96–100). It is well known that phosphorus poisons the modifying effect of strontium (Ref 4, 22, 97, 100) and sodium (Ref 4, 99, 100). The effect of phosphorus on strontium modification is shown in Fig. 19. This figure is a map of the strontiumphosphorus interaction in A356 alloy sand castings. The “F” represents the fibrous or fully modified structure. The “L” represents the region where lamellar silicon is found, and the “A” stands for acicular, nonmodified silicon. The results in this figure indicate that, for each 10 ppm increase in phosphorus, one must increase the strontium by 30 ppm to offset the poisoning effect of phosphorus. Antimony additions will produce a lamellar structure similar to the one indicated in Fig. 18 (2). This was the basis for the Calypso series of antimony-modified foundry alloys developed by Pechiney approximately 30 years ago (Ref 4, 25). Antimony does not increase porosity, as strontium and sodium sometimes do (Fig. 5).

250 / Molten Metal Processing and Handling

Fig. 18

Fig. 19

Modification rating system proposed by the American Foundry Society. See text for explanations of rating system. Source: Ref 95

Strontium-phosphorus interaction in sand-cast A356 alloy. F, fibrous or fully modified structure; L, lamellar silicon region; A, acicular, nonmodified silicon region. Source: Ref 100

For many years, antimony was added to aluminum-silicon casting alloys by French and Japanese wheel makers to produce a partial refinement (lamellar structure). For this reason, antimony sometimes turns up in secondary alloys made from scrap wheels. Antimony resides underneath phosphorus in group VA of the periodic table, so its chemical properties are similar in many respects. A number of researchers have documented that antimony poisons the modification effect of both strontium and sodium (Ref 100–105). The deleterious effects of antimony can often be counteracted with increased levels of strontium, as shown in Fig. 20. Unfortunately, once antimony is added, it remains in the scrap cycle. It cannot be removed by oxidation with oxygen or chlorine. Improved properties and shorter solution times are possible with strontium. The

widespread use of rotary impellor degassers now makes it easy to obtain a low hydrogen content in melts. Antimony is a heavy metal, and many are concerned about possible health risks associated with its use in the foundry. For all of these reasons, antimony is now rarely used. Bismuth is an element that is added to freemachining alloys. This scrap sometimes finds its way into secondary alloys such as 319. Because it is also a group VA element, it poisons the effect of strontium modification when present in significant quantities (Ref 97, 106, 107). Researchers and foundrymen agree that the strontium-bismuth ratio must be greater than 0.45 in order to produce a fully modified silicon eutectic in 319 alloy. Magnesium. When magnesium is added to aluminum-silicon alloys, there is a coarsening of the silicon eutectic (Ref 4, 108). This is perhaps because magnesium changes the phase relationships to produce a eutectic mushy zone. Boron. Some researchers have experimented with boron additions to grain-refine aluminumsilicon alloys. Boron forms the compound AlB2, which is an effective nucleant in foundry alloys, but boron also forms the compound SrB6. Thus, boron additions interfere with strontium modification (Ref 109). For this reason, grain refinement is best accomplished by additions of TiB2, via Al-Ti-B master alloys. Calcium. There have been many studies on the use of calcium as a modifier (Ref 41, 44, 62, 106, 110, 111), either alone or in combination with sodium or strontium. When calcium is added to strontium-modified melts, there is a coarsening of the silicon eutectic, similar to that which occurs with magnesium or excess quantities of strontium. The action of calcium is similar to strontium as a modifier, except calcium does not work as well at slow cooling rates. Calcium changes the surface oxide when dissolved in aluminum melts, making the metal extremely susceptible to gas pickup. Calcium sometimes finds its way into the foundry via silicon additions used to make up the alloy. It is best to keep calcium at less than approximately 30 ppm to avoid excessive gas pickup. When necessary, calcium is easily removed by additions of reactive salts or by degassing with a few percent of chlorine in the inert gas. Salts and chlorine also remove strontium, however, so it is best to first remove calcium, skim the melt to remove any reactive salts, and then make the strontium master alloy addition.

Summary In writing this article, the author was amazed by the volume of information available in the technical literature. However, there is a relative paucity of information in some areas, with possible future advances in two areas of modification technology. For these reasons, a few suggestions are made for areas where future research and development appear to be useful.

Modification of Aluminum-Silicon Alloys / 251 These nucleants contained titanium and vanadium. They then added titanium, vanadium and boron (in unspecified proportions) and were able to reduce the eutectic grain size. (Note Fig. 6 in Ref 51.) This is an exciting result. If it could be translated into commercial practice, it would be the most significant advance in modification technology since strontium was introduced more than 30 years ago. It is to be hoped that this work will continue.

500 F

450

Acicular (A) 400 Lamellar (L)

Sr analysis, pmm

350

Partial modification (Pm) Fibrous (F)

PM

300

ACKNOWLEDGMENT The author gratefully acknowledges Alcoa Primary Metals, whose support has made this study possible.

350 A

200

REFERENCES

150 L + PM 100

50

0

0

200

400

600

800

1000

1200

1400

1600

1800

2000

2200

2400

Sb analysis, pmm

Fig. 20

Strontium-antimony interactions in A356 alloy. Source: Ref 102

There have been studies on the effect of phosphorus or strontium additions on the formation of iron-containing compounds in foundry alloys (Ref 112–115). Unfortunately, the available information is conflicting and contradictory. Since iron is always a problem in aluminum castings, it would be worth exploring any method that offers promise of controlling or limiting the loss of elongation and tensile strength normally associated with higher iron contents. A detailed fundamental study would be extremely useful. Perhaps the best way to develop better information would be with directional-solidification experiments. Careful thermal analysis studies may also be helpful, and studying the effect of freezing rate on particle formation would also be informative. In Fig. 18(6), a supermodified structure was shown. If it were possible to produce a structure like this consistently in commercial casting processes, this would be a significant advance in casting technology. The mechanical properties of the castings should be superior to anything that is produced today, and it may allow for further reduction of solution heat treatment times. Two recent laboratory studies suggest ways in which this may be accomplished. The first is the excellent study of aluminum-silicon alloys produced by semisolid metal processing (Ref 18). In melts that were subjected to stirring during solidification, an extremely fine supermodified eutectic was produced. In the second

study (Ref 116), ultrasound produced the same structure. This question arises: What level of vibration or ultrasound (or metal movement) is required to produce a supermodified structure? It would be worth trying to answer this question. It also raises profound questions about the mechanisms operating to produce the modified structure. There was only one paper (Ref 117) that studied the effect of modification practice on fatigue properties. This study suggested that best fatigue properties are obtained when a grain refiner is used in combination with strontium. In view of the importance of fatigue life, further study appears to be justified. Likewise, there was only one paper on machinability (Ref 118). This paper suggested that sodium and strontium modification improves surface finish and reduces tool wear. An increase in shear angle was recommended in modified alloys for best results. Machinability is another subject that merits further study. Finally, there is the foundryman’s dream to somehow modify aluminum-silicon alloys without the undesirable increase in porosity. This may be possible. Researchers at the Cooperative Research Center for Cast Metals Manufacturing (CAST) in Australia have been studying nucleants for the silicon eutectic in aluminum-silicon alloys (Ref 51). In nonmodified alloys, they found AlP, but in strontiummodified alloys, other nucleants were observed.

1. A. Pacz, U.S. Patent 1,387,900, Aug 16, 1921 2. R.W. Smith, Modification of Aluminum-Silicon Alloys, Solidification of Metals, Iron and Steel Inst., London, 1968, p 224–237 3. P.B. Crosley and L.F. Mondolfo, The Modification of Aluminum-Silicon Alloys, Mod. Cast., March 1966, p 89–100 4. S. Bercovici, “Control of Solidification Structures and Properties of Al-Si Alloys,” Paper presented at the 45th International Foundry Congress (Budapest), Oct 1978 5. R.J. Kissing and J.F. Wallace, Refinement of Aluminum-Silicon Alloys, Foundry, Vol 91 (No. 4), April 1963, p 74–79 6. S.-Z. Lu and A. Hellawell, Modification of Al-Si Alloys: Microstructure, Thermal Analysis, and Mechanisms, J. Met., Feb 1995, p 38–40 7. G.K. Sigworth, Theoretical and Practical Aspects of the Modification of Al-Si Alloys, AFS Trans., Vol 91, 1983, p 7–16 8. M. Adachi, Modification of Hypoeutectic and Eutectic Al-Si System Casting Alloys, J. Jpn. Inst. Light Met., Vol 34 (No. 6), 1984, p 361–373 9. J.E. Gruzleski and B. Closset, The Treatment of Aluminum-Silicon Alloys, American Foundry Society, 1990, p 25–105 10. L. Backerud, G. Chai, and J. Tamminen, Solidification Characteristics of Aluminum Alloys—Volume 2: Foundry Alloys, AFS/ Skanaluminium, 1990, p 25–38 11. J.E. Gruzleski, The Art and Science of Modification: 25 Years of Progress, AFS Trans., Vol 100, 1992, p 673–683 12. M.M. Makhlouf and H.V. Guthy, The Al-Si Eutectic Reaction: Mechanisms and Crystallography, J. Light Met., Vol 1, 2001, p 199–218 13. M. Adachi, Modification of Hypereutectic Al-Si System Casting Alloys, J. Jpn. Inst. Light Met., Vol 34 (No. 7), 1984, p 430–436 14. A.P. Bates, Hypereutectic Aluminum-Silicon Alloys: A Review of Published Information, Metallurgia, Vol 61 (No. 364), Feb 1960, p 70–79

252 / Molten Metal Processing and Handling 15. J.L. Jorstad, The Hypereutectic Al-Si Alloy Used to Cast the Vega Engine Block, Mod. Cast., Vol 60, Oct 1971, p 59–64 16. Y.P. Telang, Process Variables in Al-21 Si Alloys Refinement, Mod. Cast., Vol 43 (No. 5), 1963, p 232–240 17. J.L. Jorstad and D. Apelian, Hypereutectic Al-Si Alloys: Practical Processing Techniques, Die Casting Eng., Vol 48 (No. 3), May 2004, p 50–58 18. S. Nafisi and R. Ghomashchi, Effects of Modification during Conventional and Semi-Solid Metal Processing of A356 AlSi Alloy, Mater. Sci. Eng. A, Vol 415, 2006, p 273–285 19. M. Helmanova, The Influence of Solidification Rate on the Spatial Distribution of Eutectic Silicon in Technical Silumins, Prakti. Metallogr., Vol 12, 1975, p 59–68 20. B.M. Thall and B. Chalmers, Modification in Aluminum-Silicon Alloys, J. Inst. Met., Vol 77, 1950, p 79–97 21. Z. Poniewierski, Effect of the Type of Modification on the Microstructure of Al-Si Eutectic, Rev. Int. Hautes Temp. Re´fract., Vol 14, 1977, p 253–260 22. F. Fommei, Modification Treatments of AlSi Alloys, Aluminio, Vol 46, 1977, p 121–135 23. J. Gobrecht, The Influence of Alloying Elements on the Duration of Modification of Na and Sr in Al-Si Cast Alloys, Giesserei, Vol 65, 1978, p 158–164 24. T.J. Hurley and R.G. Atkinson, Effects of Modification Practice on Aluminum A356 Alloys, AFS Trans., Vol 93, 1985, p 291– 296 25. M. Garat, G. Laslay, S. Jacob, P. Meyer, P.H. Gaerin, and R. Adam, State of the Art Use of Sb-, Na- and Sr-Modified AlSi Casting Alloys, AFS Trans., Vol 100, 1992, p 821–832 26. J.J. Jorstad and W.M. Rasmussen, Aluminum Casting Technology, 2nd ed., American Foundry Society, 1993, p 76–77 27. S. Shivkumar, S. Ricci, B. Steenhoff, D. Apelian, and G. Sigworth, An Experimental Study to Optimize the Heat Treatment of A356 Alloy, AFS Trans., Vol 97, 1989, p 791–810 28. S.A. Levy, Techniques for Improving Ductility of Low Pressure Die Cast Aluminum Wheels, AFS Trans., Vol 86, 1978, p 135–138 29. F. Paray and J.E. Grusleski, Factors to Consider in Modification, AFS Trans., Vol 102, 1994, p 833–842 30. M. Drouzy, S. Jacob, and M. Richard, Interpretation of Tensile Results by Means of a Quality Index, AFS Int. Cast Met. J., Vol 5, 1980, p 43–50 31. P.Y. Zhu, Q.Y. Liu, and T.X. Hon, Spheroidization of Eutectic Silicon in Al-Si Alloys, AFS Trans., Vol 91, 1983, p 609–614 32. D.L. McLellan and M.M. Tuttle, Aluminum Castings—A Technical Approach, AFS Trans., Vol 91, 1983, p 243–252

33. Q.C. Wang, C.H. Caceres, and J.R. Griffiths, Damage by Eutectic Particle Cracking in Aluminum Casting Alloys A356/ A357, Met. Mater. Trans. A, Vol 34, 2003, p 2901–2912 34. Q.T. Fang and D.A. Granger, Porosity Formation in Modified and Unmodified A356 Alloy Castings, AFS Trans., Vol 97, 1989, p 989–1000 35. D. Emadi and J.E. Gruzleski, Effects of Casting and Melt Variables on Porosity in Directionally Solidified Al-Si Alloys, AFS Trans., Vol 102, 1994, p 307–312 36. J. Charbonnier, Microprocessor Assisted Thermal Analysis Testing of Aluminum Alloy Structures, AFS Trans., Vol 92, 1984, p 907–922 37. D. Apelian, G.K. Sigworth, and K. Whaler, Assessment of Grain Refinement and Modification of Al-Si Foundry Alloys by Thermal Analysis, AFS Trans., Vol 92, 1984, p 297–305 38. N. Tenekedjiev and J.E. Gruzleski, Thermal Analysis of Strontium Treated Hypoeutectic and Eutectic Al-Si Casting Alloys, AFS Trans., Vol 99, 1991, p 1–6 39. G. Chai and L. Ba¨ckerud, Factors Affecting Modification of Al-Si Alloys by Adding Sr-Containing Master Alloys, AFS Trans., Vol 100, 1992, p 847–854 40. K. Nogita and A.K. Dahle, Eutectic Growth Mode in Sr, Sb and P Modified Hypoeutectic Al-Si Foundry Alloys, Mater. Trans., Vol 42 (No. 3), 2001, p 393–396 41. A. Knuutinen, K. Nogita, S.D. McDonald, and A.K. Dahle, Modification of Al-Si Alloys with Ba, Ca, Y and Yb, J. Light Met., Vol 1, 2001, p 229–240 42. A.K. Dahle, S.D. McDonald, and K. Nogita, Eutectic Solidification in Hypoeutectic Al-Si Alloys and Its Effect on Porosity, Advances in Aluminum Casting Technology, M. Tiryakioglu and J. Campbell, Ed., ASM International, 2002, p 1–10 43. K. Nogita, S.D. McDonald, K. Tsujimoto, K. Yasuda, and A.K. Dahle, Aluminum Phosphide as a Eutectic Grain Nucleus in Hypoeutectic Al-Si Alloys, J. Electron Microsc., Vol 53 (No. 4), 2004, p 361–369 44. C.B. Kim and R.W. Heine, Fundamentals of Modification in the Aluminum-Silicon System, J. Inst. Met., Vol 92, 1963–1964, p 367–376 45. Y. Tsumura, A. Sakakibara, K. Toyoda, and M. Ishikawa, The Mechanism of Incubation Period of Modification Effect of AlSi Alloy, J. Jpn. Inst. Light Met., Vol 30, 1980, p 239–245 46. N. Tenekedjiev, D. Argo, and J.E. Gruzleski, Sodium, Strontium and Phosphorus Effects in Hypereutectic Al-Si Alloys, AFS Trans., Vol 97, 1989, p 127–136 47. G.K. Sigworth, Observations on the Refinement of Hypereutectic Al-Si Alloys, AFS Trans., Vol 95, 1987, p 303–314

48. S.C. Flood and J.D. Hunt, Modification of Al-Si Eutectic Alloys with Na, Met. Sci., Vol 15, 1981, p 287–294 49. D.C. Jenkinson and L.M. Hogan, The Modification of Al-Si Alloys with Strontium, J. Cryst. Growth, Vol 28, 1975, p 171–187 50. A.K. Dahle, K. Nogita, S.D. McDonald, C. Dinnis, and L. Lu, Eutectic Modification and Microstructural Development in Al-Si Alloys, Mater. Sci. Eng. A, Vol 413–414, 2005, p 243–248 51. K. Nogita, S.D. McDonald, and A.K. Dahle, New Approaches to Understand Modification and Nucleation Mechanisms of Hypoeutectic Al-Si Alloys, Shape Casting: Second International Symposium, The Metals Society, 2007, p 51–58 52. A.K. Dahle, J.A. Taylor, and D.A. Graham, The Role of Eutectic Growth Mode in Porosity Formation in Al-Si Alloys, Alum. Trans., Vol 3 (No. 1), 2000, p 17–30 53. S.D. McDonald, A.K. Dahle, J.A. Taylor, and D.H. St. John, Eutectic Grains in Unmodified and Strontium-Modified Hypoeutectic Al-Si Alloys, Metall. Mater. Trans. A, Vol 35, 2004, p 1829–1837 54. S.D. McDonald, A.K. Dahle, J.A. Taylor, and D.H. St. John, Modification-Related Porosity Formation in Hypoeutectic Al-Si Alloys, Metall. Mater. Trans. B, Vol 35, 2004, p 1097–1106 55. S.D. McDonald, “Eutectic Solidification and Porosity Formation in Unmodified and Modified Hypoeutectic Al-Si Alloys,” Ph.D. thesis, Division of Materials Engineering, University of Queensland, Brisbane, Australia, July 2002 56. J.P. Anson, M. Stucky, and J.E. Gruzleski, Effect of Sr Modification on Nucleation and Growth of Microporosity During Solidification of Al-7% Si Foundry Alloy, AFS Trans., Vol 108, 2000, p 419–426 57. J.P. Anson, M. Stucky, and J.E. Gruzleski, Effect of Sr Concentration on Microporosity Formation in A356 Aluminum Alloy, AFS Trans., Vol 109, 2001, p 1–16 58. R. Fuoco, H. Goldenstein, and J.E. Gruzleski, Evaluation of Effect of Modification Induced Undercooling on Microporosity Formation in A356 Aluminum Alloy, AFS Trans., Vol 102, 1994, p 297–312 59. R. Fuoco, E.R. Correa, and A.V.O. Correa, Effect of Modification on Microporosity Formation in 356 Aluminum Alloy, AFS Trans., Vol 103, 1995, p 379–386 60. K.J. Brondyke and P.D. Hess, Interpretation of Vacuum Gas Test Results for Aluminum Alloys, Trans. TMS-AIME, Vol 230, 1964, p 1542–1546 61. E.L. Rooy and E.F. Fischer, Control of Aluminum Casting Quality by Vacuum Solidification Tests, AFS Trans., Vol 76, 1963, p 237–240 62. K.E. Ho¨ner and Z. Youling, Influence of Ca and Sr on the Hydrogen Pickup in Aluminum Alloy Melts by Example of

Modification of Aluminum-Silicon Alloys / 253

63.

64.

65.

66. 67.

68.

69.

70.

71.

72.

73.

74.

75.

76.

G-AlSi12, Giessereiforschung, Vol 39 (No. 11), 1987, p 34–48 L. Liu, A.M. Samuel, F.H. Samuel, H.W. Doty, and S. Valtierra, Influence of Oxides on Porosity Formation of Sr-Treated Al-Si Casting Alloys, J. Mater. Sci., Vol 38, 2003, p 1255–1267 D. Emadi, J.E. Gruzleski, and M. Pekguleryuz, Melt Oxidation Behavior and Inclusion Content in Unmodified and Sr-Modified A356 Alloy—Their Role in Pore Nucleation, AFS Trans., Vol 104, 1996, p 763–768 S. McDonald, M. Dargusch, G. Song, and D. St. John, Unintentional Effects of Sr Additions in Al-Si Foundry Alloys, Shape Casting: Second International Symposium, The Metals Society, 2007, p 59–66 D. Argo and J.E. Gruzleski, Porosity in Modified Aluminum Alloy Castings, AFS Trans., Vol 96, 1988, p 65–74 H. Iwahori, K. Yonekura, Y. Sugiyama, Y. Yamamoto, and M. Nakamura, Behavior of Shrinkage Porosity Defects and Limiting Solid Fraction of Feeding on Al-Si Alloys, AFS Trans., Vol 93, 1985, p 443–451 H. Iwahori, K. Yonekura, Y. Yamamoto, and M. Nakamura, Occurring Behavior of Porosity and Feeding Capabilities of Sodium- and Strontium-Modified Al-Si Alloys, AFS Trans., Vol 98, 1990, p 167–173 R. Fuoco, E.R. Correa, and H. Goldenstein, Effect of Modification on Microporosity in 356 Aluminum Alloy, AFS Trans., Vol 104, 1996, p 1151–1157 . Nielsen and S.O. Olsen, Experiment for Quantification of Feedability and Permeability in Industrial Aluminum Alloys, Int. J. Cast Met. Res., Vol 19 (No. 1), 2006, p 32–37 J.M. Kim, H.W. Kwon, and C.R. Loper, Jr., Feeding Behavior of Modified and Unmodified Al-Si Alloys, AFS Trans., Vol 104, 1996, p 743–749 P.D. Hess and E.V. Blackmun, Strontium as Modifying Agent for Hypoeutectic Aluminum-Silicon Alloys, AFS Trans., Vol 92, 1975, p 87–90 S.D. McDonald, K. Nogita, A.K. Dahle, J.A. Taylor, and D.H. St. John, Eutectic Solidification and Porosity Formation in Al-Si Alloys: Role of Strontium, AFS Trans., Vol 108, 2000, p 463–470 C.M. Dinnis, A.K. Dahle, and J.A. Taylor, Eutectic Morphology and Porosity in Unmodified and Sr-Modified Al-Si Alloys, AFS Trans., Vol 113, 2005, p 99–106 Z. Li, A.M. Samuel, F.H. Samuel, C. Ravrindran, and S. Valtierra, Effect of Alloying Elements on the Segregation and Dissolution of CuAl2 Phase in Al-Si-Cu 319 Alloys, J. Mater. Sci., Vol 38, 2003, p 1203–1218 F.H. Samuel, A.M. Samuel, and H.W. Doty, Factors Controlling the Type and Morphology of Cu-Containing Phases in 319 Aluminum Alloy, AFS Trans., Vol 104, 1996, p 893–901

77. M. Djurdjevic, T. Stockwell, and J. Sokolowski, Effect of Strontium on Microstructure of Al-Si and Al-Cu Eutectics in 319 Aluminum Alloy, Int. J. Cast Met. Res., Vol 12, 1999, p 67–73 78. B. Closset and S. Kitaoka, Evaluation of Sr Modifiers for Al-Si Casting Alloys, AFS Trans., Vol 85, 1987, p 233–240 79. M.O. Pekguleryuz and J.E. Gruzleski, Conditions for Strontium Master Alloy Additions to A356 Melts, AFS Trans., Vol 96, 1988, p 55–64 80. B. Closset and J.E. Gruzleski, Mechanical Properties of A356 Alloy Modified with Pure Strontium, AFS Trans., Vol 90, 1982, p 453–464 81. R. DasGupta, C.G. Brown, and S. Marek, Analysis of Overmodified 356 Aluminum Alloy, AFS Trans., Vol 96, 1988, p 297–310 82. J. Gruzleski, N. Handiak, H. Campbell, and B. Closset, Hydrogen Measurement by Telegas in Sr-Treated A356 Melts, AFS Trans., Vol 94, 1986, p 147–154 83. F.C. Dinayuga, N. Handiak, and J.E. Gruzleski, The Degassing and Regassing Behavior of Sr-Modified A356 Melts, AFS Trans., Vol 96, 1988, p 83–88 84. X.-G. Chen, F.J. Klinkenberg, R. Ellerbrok, and S. Engler, Efficiency of Impeller Degassing and Regassing Phenomena in Aluminum Melts, AFS Trans., Vol 102, 1994, p 191–197 85. G.K. Sigworth, C. Wang, H. Huang, and J.T. Berry, Porosity Formation in Modified and Unmodified Al-Si Alloy Castings, AFS Trans., Vol 102, 1994, p 245–261 86. B. Gallois and G.K. Sigworth, An Analysis of Silicon Eutectic Modification, Thermal Analysis of Molten Aluminum, American Foundry Society, 1985, p 101–119 87. H. Fredriksson, M. Hillert, and N. Lange, Modification of Aluminum Silicon Alloys by Sodium, J. Inst. Met., Vol 101, 1973, p 285–299 88. M.G. Day and A. Hellawell, The Microstructure and Crystallography of Aluminium-Silicon Eutectic Alloys, Proc. R. Soc. (London) A, Vol 305, 1968, p 473–491 89. S.-Z. Lu and A. Hellawell, Mechanisms of Silicon Modification in Aluminum-Silicon Alloys: Impurity Induced Twinning, Metall. Trans. A, Vol 18, 1987, p 1721–1731 90. A. Hellawell, How Does a Modifier Work?, Proc. Second International Conference on Molten Aluminum Processing, American Foundry Society, 1989, p 9–1 to 9–17 91. R. Elliott, Eutectic Solidification Processing, Butterworth, London, 1983 92. R. Elliott, Growth Kinetics in Aluminum Silicon Eutectic Alloys, Proc. Second International Conference on Molten Aluminum Processing, American Foundry Society, 1989, p 10–1 to 10–24 93. L. Clapham and R.W. Smith, Partial Modification in Unidirectionally Solidified Al-Si Eutectic Alloys, Acta Metall., Vol 37 (No. 1), 1989, p 303–311

94. J.F. Major and J.W. Rutter, Effect of Strontium and Phosphorus on Solid/Liquid Interface of Al-Si Eutectic, Mater. Sci. Technol., Vol 5 (No. 7), 1989, p 645–656 95. G.K. Sigworth, Determining Grain Size and Eutectic Modification in Aluminum Alloy Castings, Mod. Cast., Vol 77, July 1987, p 23–25 96. S.D. McDonald, K. Nogita, and A.K. Dahle, Eutectic Nucleation in Al-Si Alloys, Acta Mater., Vol 52, 2004, p 4273–4280 97. C.R. Loper, Jr., H.-G. Seong, and J.-I. Cho, Interaction of Phosphorus and Bismuth (with Sr) in A356.2 Alloy, AFS Trans., Vol 109, 2001, p 1–13 98. C.R. Loper, Jr. and J.-I. Cho, Influence of Trace Amounts of Phosphorus in Aluminum Castings, AFS Trans., Vol 108, 2000, p 667–672 99. B. Gu¨nther and H. Ju¨rgens, Influence of Minor Impurities on Mechanical Properties, Structure and Solidification Form of Eutectic on Near-Eutectic Al-Si Casting Alloys, Giesserei, Vol 67, 1980, p 8–13 100. S. Bercovici, M. Garat, and R. Scalliet, “A Review of Recent French Casting Alloy Developments,” Report 13691, Pechiney Ugine Kuhlmann, April 1978 101. N. Handiak, J.E. Gruzleski, and D. Argo, Sodium, Strontium and Antimony Interactions During Modification of AS7G03 (A356) Alloys, AFS Trans., Vol 95, 1987, p 31–38 102. E.N. Pan, Y.C. Cherng, C.A. Lin, and H.S. Chiou, Roles of Sr and Sb on Silicon Modification of A356 Aluminum Alloys, AFS Trans., Vol 102, 1994, p 609–629 103. B.L. Tuttle, D. Twarog, and E. Daniels, The Effect of Trace Amounts of Antimony on the Structure and Properties of Aluminum Alloy A356.2, AFS Trans., Vol 99, 1991, p 7–16 104. W. Wang and J.E. Gruzleski, Interactive Effects During Sodium or Strontium Treatment of Antimony-Containing A356 Alloy, AFS Trans., Vol 98, 1990, p 227–234 105. Incompatibility of Sodium and Antimony in the Modification of Al-Si Alloys, Fondeur, Vol 272, 1976, p 12–13 106. S. El-Hadad, A.M. Samuel, F.H. Samuel, H.W. Doty, and S. Valtierra, Effect of Bi and Ca Additions on the Characteristics of Eutectic Silicon Particles in Sr-Modified 319 Alloys, Int. J. Cast Met. Res., Vol 15, 2003, p 551–564 107. C.J. Machovec, G.E. Byczynski, J.W. Zinel, and L.A. Godlewski, Effect of BiSr Interactions on Si Morphology in a 319 Type Aluminum Alloy, AFS Trans., Vol 108, 2000, p 439–444 108. G. Heiberg et al., Effect of Mg, Fe and Cu on Eutectic Solidification of Hypoeutectic Al-Si Alloys, AFS Trans., Vol 110, 2002, p 347–358 109. B. Golhahar, A.M. Samuel, F.H. Samuel, H.W. Doty, and S. Valtierra, Effect of Grain Refiner-Modifier Interactions on

254 / Molten Metal Processing and Handling the Microstructure and Mechanical Properties of A356 Alloys, AFS Trans., Vol 115, 2007, paper 07–012 110. A. Abdollahi and J.E. Gruzleski, Evaluation of Calcium as a Eutectic Modifier in A357 Alloy, Int. J. Cast Met. Res., Vol 11, 1998, p 145–155 111. C.R. Loper, Jr. and J.-I. Cho, Influence of Trace Amounts of Calcium in Aluminum Castings, AFS Trans., Vol 108, 2000, p 585–592 112. L. Lu and A.K. Dahle, Iron-Rich Intermetallic Phases and Their Role in Casting Defect Formation in Hypoeutectic Al-Si Alloys, Metall. Mater. Trans. A, Vol 36, 2005, p 819–835

113. W. Kalifa, A.M. Samuel, F.H. Samuel, H.W. Doty, and S.Valtierra, Metallographic Observations of b-AlFeSi Phase and Its Role in Porosity Formation in Al7%Si Alloys, Int. J. Cast Met. Res., Vol 19, 2006, p 156–166 114. B. Kulunk, S.G. Shabestari, and J.E. Gruzleski, Beneficial Effects of Strontium in A380 Alloy, AFS Trans., Vol 104, 1996, p 1189–1193 115. S.G. Shabestari and J.E. Gruzleski, Modification of Iron Intermetallics by Strontium in 413 Aluminum Alloy, AFS Trans., Vol 103, 1995, p 285–293 116. X. Jian, C. Xu, T. Meek, and Q. Han, Effect of Ultrasound Vibration on the

Solidification Structure of A356 Alloy, AFS Trans., Vol 113, 2005, p 131–138 117. P.D. Lee and T.C. Lindley, The Influence of Sr and TiB2 Additions on the Formation of Porosity in Aluminum Castings and Their Effect on Fatigue Properties, Shape Casting: The J. Campbell Symposium, The Metals Society 2005, p 225– 234 118. E. Sato, T. Saga, and S. Nagai, Effects of Additional Elements on the Machinability of Al-Si Casting Alloys, J. Jpn. Inst. Light Met., Vol 33 (No. 11), 1983, p 49–654

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 255-262 DOI: 10.1361/asmhba0005302

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Grain Refinement of Aluminum Casting Alloys* G.K. Sigworth and T.A. Kuhn, Alcoa Primary Metals

GRAIN REFINEMENT of aluminum was first accomplished in the early 1930s by making titanium additions to the melt (Ref 1–4). The improvement in castability was considerable, and this quickly became the established practice. The resulting change in the structure of the casting is remarkable. This may be seen by cutting a piece from the casting and polishing the cut surface with a series of progressively finer sandpapers, until a mirror-smooth surface is obtained. After the polished surface is etched with acid, the underlying grain structure is revealed. An example of the transformation obtained by grain refinement is shown in Fig. 1. Grain refinement has a significant effect on the mechanical properties of castings, primarily because the distribution of second-phase materials is changed. Consider the casting shown in Fig. 1. The nonrefined casting (top in Fig. 1) has a small area near the chill where equiaxed grains were formed, but the rest of the metal has elongated, feathery grains. All brittle intermetallic phases and any porosity formed will be located between these large grains, so the elongation normal to the grains will be poor. The material shown at the bottom of Fig. 1 has a much better structure. The grain size is small (or fine, which is the basis for the term refinement). Hence, mechanical properties are isotropic, and the material is stronger. As is shown subsequently, grain refinement in casting alloys tends to reduce the amount of porosity and the size of the pores. This improves mechanical properties, especially fatigue strength. Feeding is also improved. It is for these reasons that most cast aluminum is grain refined.

procedure is more convenient with large samples, for example, when cutting through the middle of a casting. The sample is placed on a milling machine, and metal is cut away until a flat surface is obtained. Then, if a final light cut with a sharp tool is made, the resulting surface may be etched directly without polishing. For aluminum alloys low in copper, the test sample is placed in hot, nearly boiling water for a few minutes. When the sample is hot, it is removed from the water, and the surface to be etched is immersed briefly in a solution of Poulton’s etch. (This etchant contains the following ingredients: 60% HCl, 30% HNO3, 5% HF, and 5% H2O.) This acid is held at room temperature. Alternatively, the surface may be etched by wiping it with a piece of cotton wool or absorbent cloth that has been soaked in the etchant. For alloys that contain copper, it is more difficult to determine the grain size. One procedure that works reasonably well is to etch the sample at room temperature in a solution of 10% HF. This results in a black smudge on the sample, produced by the copper in the alloy. This smudge may be cleaned off by rubbing the

   

Average intercept distance Calculated average grain diameter ASTM grain size number Grains per unit area

Table 1 gives values in different units of measurement, for the sake of comparison. The easiest procedure is usually to measure the average intercept distance. All grain sizes in this article are reported in this way. The units employed are microns (mm, or 10 6 m).

Mechanisms of Grain Refinement Grain refinement dates from the early 1930s, when foundrymen began to use titanium additions to improve the structure of their castings. To understand this action of titanium, one must

Chill

Measurement of Grain Size To determine grain size, one obtains a sample and produces a smooth surface on one side. Thismay be done in the laboratory using standard metallographic techniques. However, another

surface of the sample under running water or by rinsing it in a dilute solution of nitric acid. There are various procedures established to measure grain size on the etched samples. These are outlined in ASTM E 112. The grain structure can be reported in one of several units:

Chill

Fig. 1

Grain structure in 3004 alloy castings. Top: no refinement; bottom: 10 ppm boron added as 5Ti-1B alloy

*Republished with the permission of American Foundry Society (www.afsinc.org). Copyright 2007 from the following: International Journal of Metalcasting (ISSN 1939–5981), Vol 1 (No. 1), 2007, p 31–40, and AFS Transactions (ISBN 978-0-87433-338-1), Vol 115, 2007, p 177–188

256 / Molten Metal Processing and Handling Table 1

Comparison of various grain size measurements (ASTM E 112)

1

Average intercept distance in.

mm

mm

Average diameter, mm

ASTM grain No.

Grains/cm2

Grains/in.2

0.008 0.015 0.031 0.047 0.079

0.2 0.4 0.8 1.2 2.0

200 400 800 1200 2000

252 504 1008 1512 2520

14.5 12.5 10.5 9.5 8

1890 496 112 56 20

12,200 3200 724 362 128

2

3 1000

1830

900

1650 L

4

L + TiAl 3

Temperature, ⬚C

700

1290

665

0.15

1.15

660.1 α + TiAl 3

α

600

Temperature, ⬚F

1420

800

5

1110

930

500

6 400

900 0

0.5

1.0

1.5

2.0

Titanium, wt%

Fig. 2

The aluminum-rich side of the aluminum-titanium phase diagram. Source: Ref 5

7 of the base alloy. Hence, the first nucleation (or formation) of solid aluminum will be at the surface of the aluminide particle, as shown by “3”. The aluminum crystal then grows around the surface of the aluminide (“4” and “5”). In the process, it consumes the dissolved titanium in the vicinity of the particle, and growth stops. As the metal cools further, dendritic growth begins (“6”) and continues (“7”) as solidification proceeds. This course of events may be confirmed in two ways. The first is to pour a sample of metal into a small cup and record the temperature during solidification with a thermocouple. The resulting cooling curves of refined and nonrefined metal are shown in Fig. 4. When titanium is added, the nucleation temperature, Tn, is often above the growth temperature, Tg. When no titanium is present, the metal cools several degrees below the growth temperature before nucleation occurs. Another confirmation of the mechanism proposed in Fig. 3 may be found by a careful examination of the microstructure of aluminum castings. When anodized in the proper way and viewed under polarized light, the titanium-rich areas corresponding to “6” and “7” of Fig. 3 are revealed. An example is presented in Fig. 5 (Ref 8). The darker areas located in the

Fig. 3

TiAl3 nucleates aluminum grain. Source: Adapted from Ref 6

1 Grain refiner added

2 3,4 Temperature

consider the aluminum-titanium phase diagram (Ref 5), shown in Fig. 2. The important thing to note is that titanium raises the melting point of aluminum. Pure aluminum melts at 660.1  C (1220.2  F); adding titanium increases the melting point to 665  C (1229  F). This effect has important ramifications for the nucleation and growth of solid aluminum grains. The best explanation was given in 1983 by Backerud (Ref 6). Figure 3 is taken from his paper and shows schematically what happens. Titanium additions are made via aluminum master alloys, which contain between 5 and 10% Ti. These materials contain numerous crystals of the titanium aluminide compound TiAl3. (Pictures of aluminide crystals are given in Ref 7.) When a master alloy is added, typically a few minutes before casting, millions of these microscopic particles are released into the melt. The presence of one is indicated by “1” in Fig. 3. When the TiAl3 crystal comes in contact with liquid aluminum, it starts to dissolve. This means the liquid metal at the surface of the particle becomes enriched in titanium, as shown by “2” in Fig. 3. From the relationships shown in the aluminum-titanium phase diagram, the titanium-rich metal in contact with the aluminide can begin to solidify at a temperature that is above the melting point

5 Tn

6

7

Tg Tn No grain refiner

Time

Fig. 4

Cooling curves during solidification. Tn, nucleation temperature; Tg, growth temperature

Grain Refinement of Aluminum Casting Alloys / 257

100 mm

Fig. 5

Titanium-rich areas in dendrites of Al-0.06%Ti alloy. Source: Ref 8

10,000

100

5000 2000

200

1000 400

500

1 2

Grain size, mm

Number of grains per cm3

20,000

3

200 100

800 0

0.10

0.20

0.30

Titanium, wt% 1 2 3

Fig. 6

Delamore and Smith Grebe and Grimm Crossley and Mondolfo

Refinement of 99.7% Al by titanium. Source: Ref 9

center of the grains were shown by microprobe measurements to be high in titanium. The aforementioned is a clear explanation of how titanium acts to grain refine aluminum, but there is an additional factor that must be considered: the dissolution of TiAl3 particles. A normal addition level of titanium is approximately 100 ppm (0.01% Ti). When this small amount of titanium is added to relatively pure aluminum at normal casting temperatures, it dissolves. When the added particles have dissolved, the grain-refining effect disappears. This loss of refining ability with time is called fade. The fade time depends on a number of factors. The most important are the titanium content of the metal, the metal temperature, and the type of master alloy used. The latter determines the TiAl3 particle size. Aluminum-titanium rod products tend to have smaller particles and dissolve faster—typically in 10

to 15 min. Waffle products are cast from a higher temperature and have larger aluminide particles. These may take 30 to 40 min to dissolve. The practical consequence in the foundry is relatively simple: If the titanium in the metal is significantly less than the solubility (0.15% Ti at 665  C or 1229  F), then it is not possible to grain refine consistently. Results from three early studies of grain refinement in high-purity aluminum are shown in Fig. 6 (Ref 9). The best refinement was obtained when the titanium added was more than 0.15%, the solubility limit of TiAl3 in aluminum. The picture painted thus far is fairly attractive and relatively easy to understand. This understanding of grain refinement has been around since 1930. During that time, it has entered deeply into the consciousness of foundrymen, much like Michelangelo’s Mona Lisa has entered the mainstream of consciousness forming our popular culture. For this reason, it needs to be well understood. This picture is essentially the source of, and basis for, most of the grain-refinement practices used today (2008) in aluminum casting alloys. The maximum limits for titanium are quite high in most alloys, and, in some cases, a minimum is specified, for example, 201 alloy. It is vital to grain refine this alloy to prevent hot cracking during casting. The specification for 201 alloy published by the Aluminum Association permits up to 0.35% Ti, and a minimum of 0.15% Ti is called for. Many of the other 2xx and 7xx casting alloys also have a minimum specified (0.10 or 0.15% Ti). With a few exceptions, aluminum casting alloys permit a maximum of at least 0.20% Ti. The reason for these large quantities of titanium may as well be engraved in stone. As shown subsequently, it should be a tombstone. In other words, the epitaph for this 75-year-old grain-refining practice should read:

High quantities of titanium are necessary to prevent dissolution of aluminum-titanium refiners. From an historical point of view, the aforementioned accurately describes the first 40 years of aluminum grain refinement. Then, in the early 1970s, the first modern Al-Ti-B grain refiners became available. The role of boron was studied in the laboratory in the early 1950s (Ref 10, 11). This work showed that boron in combination with titanium gave much better grain refinement, but the use of boroncontaining salts in the casthouse was an extremely messy and haphazard business. Finally, in the late 1960s, Kawecki discovered how to produce boron-containing master alloys by reacting a mixture of KBF4 and K2TiF6 with molten aluminum. The result was the first commercial Al-Ti-B master alloy—a fast-acting rod product that could be fed continuously into the metal transfer launder or holding furnace by wire feeders. The new boron-containing grain refiners were much more powerful and economical. They therefore replaced the use of the earlier aluminum-titanium products in most aluminum alloys and still form the basis for accepted commercial practices to this day. How do the modern refiners differ from the old ones? There are many differences, but the most important is this: Most of the titanium in modern Tibor refiners is present as TiB2 particles, and titanium diboride is virtually insoluble in molten aluminum. The consequence: High quantities of titanium are NOT necessary to prevent dissolution of Al-Ti-B grain refiners. However, that is only part of the story. In addition to low solubility, borides are extremely effective nuclei in foundry alloys. In fact, boron is better than titanium as a grain refiner for aluminum casting alloys.

Boron as a Grain Refiner In 1981, Lu, Wang, and Kung published a study of the grain refining of A356 alloy in the Journal of the Chinese Foundrymen’s Association (Ref 12). The results shown in Fig. 7 are taken from their study. Three different master alloys were added to A356 alloy: Al-5%Ti, Al-5%Ti-1%B, and Al-4%B. The results clearly showed that boron is more powerful than titanium as a grain refiner. This result ran completely counter to the prevailing viewpoint described previously. Everyone believed that titanium was needed for grain refinement, but the Chinese study showed titanium was only a weak grain refiner. It also suggested that boron is much better than titanium. Intrigued by these results, Sigworth and Guzowski (Ref 13, 14) began their own study. Laboratory tests confirmed that aluminumboron master alloys containing AlB2 particles gave excellent grain refinement. They then went to Stahl Specialty Company in Missouri, where foundry trials confirmed that aluminum-

258 / Molten Metal Processing and Handling With this understanding, it is now possible to define an optimized grain-refining practice for each of the common casting alloys.

2000

Best Practices for Grain Refinement Grain size, mm

1500

Before defining the best grain-refinement practices, it is necessary to clearly distinguish between the two forms of titanium.

Al-5%Ti 1000

Forms of Titanium 5%Ti-1%B 500 Al-4%B

0

0

0.02

0.04

0.06

0.08

0.10

Addition level, %Ti or %B

Fig. 7

Grain refining of A356 alloy by three master alloys. Source: Ref 12

1000 A356 alloy No TiB 2

Grain size, mm

800 600 400

10 ppm B 200 0

0

0.02

0.04

0.06

0.08

0.10

0.12

Soluble titanium, wt%

Fig. 8

Grain size as a function of titanium content. Source: Ref 16

boron master alloys were good refiners. However, the added boron reacted with dissolved titanium to produce a sludge in holding furnaces. Also, boron apparently reacted with strontium, since boron additions resulted in a loss of modification in some heats modified with strontium. A more recent study has confirmed the interaction between strontium and boron (Ref 15). This presented an interesting problem. If one compares the nucleation potency and other characteristics of the different crystals in Al-7%Si alloys, one finds:  TiAl3 crystals are poor nuclei. TiAl3 crystals

also have a relatively high solubility in aluminum. For both reasons, a large amount of titanium must be added to produce consistently small grain sizes.  TiB2 particles are excellent nuclei. Titanium diboride has almost no solubility in liquid aluminum. Thus, TiB2 particles produce good

refinement at small addition levels. The refinement is also long lasting when the particles are not allowed to sediment from the melt.  AlB2 is the best nucleus. It produces the smallest grain size. However, AlB2 dissolves readily in aluminum, where it reacts with titanium and strontium in the melt. Longterm use would produce an undesirable sludge in furnaces. Thus, in spite of its potency as a grain-refining nucleant, AlB2 cannot be used in the foundry. Would it be possible to develop a grain refiner with the potency of AlB2 and without the problems associated with boron additions? After reflecting on this question for some time, Sigworth and Guzowski produced grain refiners containing a mixed boride (Ref 13, 14). The compounds AlB2 and TiB2 have the same crystal structure, and x-ray studies showed the two crystals form a continuous solid solution. Thus, a mixed boride has a composition intermediate between AlB2 and TiB2 and can be represented by the formula (Al,Ti)B2. Mixed borides appear to have a nucleation potency close to AlB2, with a low solubility similar to TiB2. Thus, the problems associated with straight boron additions were avoided. The first mixed boride refiner was an Al2.5Ti-2.5B alloy. More recently, it has been produced as an Al-1.7%Ti-1.4% master alloy, marketed under the name Tibloy. This is an excellent refiner but somewhat slow acting, requiring approximately 5 min to produce the best grain size. In 356 alloy, its performance does not usually justify the additional cost, but in 319 alloy, Tibloy refines significantly better than Al-5Ti-1B. (See Fig. 17 and 18 in Ref 13.) One could summarize the previous discussion by saying: Boron is the most powerful grain refiner in aluminum casting alloys.

Soluble Titanium. This is the TiAl3 compound, which is present as 10 to 20 mm-sized particles in aluminum-titanium master alloys. It is also present in Al-Ti-B alloys that have an excess of titanium required to produce TiB2. (The stochiometric ratio is 2.22 wt% Ti for each 1 wt% B.) The TiAl3 compound dissolves quickly in alloys that contain less than approximately 0.15% Ti. As noted previously, this particle is a poor refiner in aluminum-silicon casting alloys. Insoluble Titanium. This is the titanium present as TiB2, which has such a low solubility in aluminum that, for all practical purposes, it can be considered to be insoluble. (TiC is another form of insoluble titanium. The carbide is also a good grain refiner, but because it offers no advantage in casting alloys, Al-Ti-C refiners are not considered further in this article.) In the past, the single word titanium has been used to describe both its soluble and insoluble forms. This terminology appears to be the principal reason for confusion by foundrymen when trying to understand grain refinement. Thus, to establish a linguistic foundation for an improved understanding, the following terminology is proposed: The word titanium shall refer to the amount of titanium dissolved in an alloy; the word boron shall refer to the quantity of insoluble boride particles present. With this terminology in mind, the question can be posed: What combination of titanium and boron gives the best grain refinement? The answer is this: It depends on the alloy family under consideration. For this reason, each is considered separately.

Practices Aluminum-Silicon Casting Alloys. The AlSi and Al-Si-Mg casting alloys, such as 356 and 357, offer excellent castability and provide a range of attractive mechanical properties. Hence, they find widespread commercial application. These alloys have also been studied extensively. The grain refining of the (copperfree) aluminum-silicon casting alloys has been reviewed in a separate publication (Ref 16). Figure 8 is taken from this study. These results are for grain refinement of an Al-7%Si-0.35% Mg (A356) alloy. This alloy was modified with approximately 0.02% Sr and grain refined with a 10 ppm boron addition of Al-5Ti-1B master alloy. Castings were also poured without the

Grain Refinement of Aluminum Casting Alloys / 259 500 A356 alloy

Grain size, μm

400 5% Boral

2.5Ti-2.5B

300

200 5Ti-1B 100

0

0

0.05

0.10

0.15

0.20

Soluble titanium, wt%

Fig. 9

A356 alloy grain size from Aluminum Association test. Source: Ref 17

10,000

Grain size, μm

%Ti 0.0 0.05

1000

0.10 0.15 0.20

300 0

Fig. 10

5 10 15 Boron added as 3Ti - 1B, ppm

20

Grain refinement of 319 alloy. Source: Ref 20

boron addition. All samples were cut from a plate casting at a distance of 35 mm (1.4 in.) from a chill, where the cooling rate was similar to that obtained in permanent mold cast wheels. Grain sizes were measured in accordance with ASTM E 112. The average and standard deviations of the grain sizes are plotted. It is readily seen that when TiB2 particles are added via commercial grain refiners, the grain size of the casting is essentially constant and independent of the dissolved titanium-content in the alloy. This result was also confirmed in other experimental studies. A plot of results obtained with the Aluminum Association grain size test is given in Fig. 9. Three different boron-containing grain refiners were employed. It appears a small amount of titanium (approximately 0.02%) is desired when using Tibloy or aluminum-boron master alloys, but this is not necessary for TiB2-containing refiners, such as Al-5Ti-1B.

These results clearly point to improved grainrefining practices for the aluminum-silicon casting alloys, such as 356 and 357. Best grain refinement is obtained when 10 to 20 ppm of boron are added in the form of Al-5Ti-1B or Al-3Ti-1B rod. With this boron addition, it is possible to reduce the large amounts of soluble titanium now added to these alloys. Lower titanium does not appear to result in any loss of mechanical properties or castability, as determined by grain size (Ref 16). Lower titanium contents also offer the advantage that the tendency for sludge formation is eliminated. Sludge can become a significant problem when the titanium content is held over 0.10% for long periods of time (Ref 18, 19). Aluminum-Silicon-Copper Casting Alloys. The grain refinement of 319 alloy was studied in detail by Pasciak and Sigworth (Ref 20). A number of castings and solidification conditions were studied. The grain size observed in test castings, whose freezing rate was similar to that found in permanent mold castings, has been plotted in Fig. 10. Both the (dissolved) titanium and (insoluble) boron contents were varied in these trials. For some reason, the presence of copper in the alloy completely changes the grain-refining response to titanium. In this case, it is beneficial to have a minimum of approximately 0.10% Ti dissolved in the alloy. Without this quantity of titanium, it was not possible to produce a small grain size. The best grain size is obtained when 10 to 20 ppm of boron are added. The boron addition can be made in the form of Al-5Ti1B or Al-3Ti-1B rod. A Tibloy alloy can also be used and may be cost-effective, but, in this case, at least 5 min should be allowed for the refiner to become active. Although they have not been studied, it is reasonable to presume that other copper-containing aluminum-silicon casting alloys (such as 355 and 308) will behave similarly to 319. For this reason, it would be advisable to have approximately 0.10% Ti present in these Al-

Si-Cu casting alloys before making an addition of 10 to 20 ppm of boron. Aluminum-Copper Casting Alloys. The first casting alloy used in the United States was an Al8%Cu composition, known as No. 12 alloy. In 1909, Alfred Wilm discovered that an Al-4.5% Cu-0.5%Mn alloy would strengthen by aging after a quench from an elevated temperature. This alloy was called Duralumin and formed the basis for the aluminum-copper family of alloys used today (2008). These alloys have excellent mechanical properties, high-temperature strength, and fatigue life. They are difficult to cast, however. Their long freezing range makes them extremely susceptible to hot cracking. It has long been well known that grain refinement reduces the tendency for hot cracking in net-shaped castings. The study that shows this most clearly is the 1970 paper by Davies (Ref 21), who found the total length of hot tears in a casting to be proportional to the cast grain size. In other words, a good grain-refining practice is essential for the production of crack-free 2xx alloy castings. The specifications for 2xx alloys date from the time when titanium was used as a grain refiner. For this reason, a minimum content of 0.15% Ti is specified for 201, 202, 203, 204, and 206 alloys. More recently, however, it has been shown that lowering the titanium content reduces the grain size of the casting and improves hot crack resistance (Ref 22). The best grain size and hot crack resistance are obtained when the titanium is less than 0.05% and 10 to 20 ppm of boron are added as Al-5Ti-1B or Al-3Ti-1B rod. Aluminum-Zinc-Magnesium Casting Alloys. A number of aluminum casting alloys are based on the ternary Al-Zn-Mg system. These alloys naturally age to high strengths at room temperature. A high-temperature solution and aging treatment is not required. Consequently, there is a potential cost savings that can be realized by eliminating the heat treatment, especially with complex shapes that would tend to distort when quenched from a high temperature. Castings from these alloys may also be welded or brazed immediately after casting, and the assemblies will age to high strength at room temperature. This combination of properties makes the alloys promising candidates for a number of applications. In spite of their excellent properties, the 7xx casting alloys are seldom used, mainly because of their propensity for hot cracking. As a consequence, the Aluminum Association chemical composition limits established for titanium in these alloys have relatively high upper limits; in three alloys (712, 771, and 772), a minimum titanium content (0.10 or 0.15%) is specified. A recent study has shown that low titanium contents in these alloys reduce the grain size and improve hot crack resistance (Ref 23). The best grain size and hot crack resistance are obtained when the titanium is 0.02 to 0.05% and 10 to 20 ppm of boron are added as Al-5Ti-1B or Al-3Ti-1B rod.

260 / Molten Metal Processing and Handling Table 2

Composition of aluminum-magnesium alloys tested

Table 3 Grain sizes observed in 535 alloys

Composition, wt% Alloy

Cr

Cu

Fe

Mg

Mn

Ni

Si

Ti

Zn

535 535L

0.0 0.0

0.004 0.004

0.15 0.15

6.98 7.04

0.18 0.18

0.002 0.002

0.05 0.05

0.196 0.036

0.0 0.0

1000 mm

Fig. 11

Grain structure in an A356 alloy casting

Aluminum-Magnesium Casting Alloys. The 5xx casting alloys provide extremely good corrosion resistance and moderate strength without heat treatment. They are commonly used in marine and architectural applications. The alloys are not easy to cast. The high magnesium content produces a thick, sticky oxide film on the melt, and castings are susceptible to hot cracking and shrinkage. The Aluminum Association chemical composition limits established for these alloys have relatively high upper limits for titanium; in three alloys (516, 520, and 535), a minimum content of 0.10% Ti is specified. Sigworth (Ref 24) conducted a study of the grain refinement of Al-7%Mg (535) alloy. Two alloys were prepared, having the same composition except for the soluble titanium. The chemistry of the two alloys is given in Table 2. To each alloy, an addition of 30 ppm boron was made in the form of Al-3%Ti-1%B master alloy. Grain size samples were taken by using the test specified by the Aluminum Association. The resulting grain size, as measured by the average intercept distance, is shown in Table 3. Since the statistical (1s) error associated with the determination of grain size is approximately 10%, for all practical purposes, the two alloys have the same grain size. Thus, it appears the soluble titanium has little effect in 5xx alloys. This result is similar to that found previously for Al-Si and Al-Si-Mg alloys, such as 356 and 357. In other words, it should be

possible to reduce the titanium content of aluminum-magnesium alloys without affecting castability when boron is used as a refiner.

Benefits of Grain Refinement Grain refinement tends to reduce both the amount of porosity and the size of the pores in casting alloys. The reason for this may be seen by careful examination of the structure in a casting (Fig. 11). In 356 alloy, the first 50 or 60% of the solidification process is taken up by the formation and growth of aluminum grains. Then, a film of aluminum-silicon eutectic forms around the aluminum dendrites. Gas porosity typically forms during the last 5 to 10% of the solidification process. Thus, gas must fit into the spaces available between the aluminum dendrites and solidified eutectic. (The pores are seen as black areas between dendritic grains in Fig. 11.) When the grain size is reduced, the size of the spaces available for pores is also reduced. The consequence is a smaller pore size. Figure 12 shows a plot of measured pore size in directionally solidified A356 alloy castings (Ref 25). By comparing curves 1 and 3 of this figure, one can see that grain refinement reduces the size of pores formed. (Rapid solidification also reduces pore size.) The result is an improved fatigue life for the casting. The other major benefit of grain refinement is improved feeding and a reduced tendency for

Alloy

Grain size, mm

535 535L

82 86

shrinkage formation in the casting. The mechanism for this effect was first established by Guocai Chai (Ref 26), who placed a slowly rotating paddle in a solidifying cylindrical casting. By measuring the torque on the paddle, he was able to measure the point during solidification when the semisolid “mush” started to jell, or became stiff. He called this transition point dendrite coherency. Coherency indicates the time when solid grains start to connect with each other to reduce feeding. Chai’s measurements showed that grain-refiner additions delay the onset of dendrite coherence. Figure 13 presents results obtained for an Al-4%Cu alloy solidified at two different cooling rates. Without grain refinement, dendrite coherency occurs at 25% solid; with grain refinement, coherency occurs at 50% solid. This means that a grainrefined riser behaves like a liquid nearly twice as long as a nonrefined riser. The measurements of Chai were later duplicated in other alloys by Arnberg and Backerud (Ref 27).

Concluding Remarks This review of grain refinement in aluminum casting alloys has shown that the large titanium additions commonly used, and often specified, are primarily an historical artifact. Titanium was used by itself for nearly 40 years, and with this practice, large additions are needed for good results. Now that powerful Al-Ti-B refiners are available, however, there is no longer any reason (with one exception) for large titanium additions. Lower titanium contents eliminate the tendency for sludge formation, and, in some alloys, a smaller grain size and improved resistance to hot cracking are found. The first key to an improved understanding of grain refinement is to realize that titanium can be present in two different forms: as soluble and insoluble particles. The soluble form is called titanium in this article. The insoluble form is called boron, since insoluble titanium is present as titanium diboride. The second key is to realize that boron (boride particles) is an extremely powerful nucleant in all casting alloys. Boron is much more effective than titanium. Thus, it makes more sense to say that grain refinement is accomplished with boron. It then becomes a question of which combination of titanium and boron gives the best results. The answer depends on the alloy system. A brief summary is offered as follows:  Low titanium is best: The Al-Cu and Al-Zn-

Mg alloys give best results when the soluble

Grain Refinement of Aluminum Casting Alloys / 261 As noted previously, titanium diboride particles are virtually insoluble in molten aluminum. They are more dense than the metal, however, and will settle from the melt. Sedimentation can be rapid if there is no stirring (Ref 28). The result will be a loss of grain refinement, usually called fade. For this reason, it is useful to provide a small amount of gentle stirring in holding furnaces. When stirring is present to prevent sedimentation, borides will provide grain refinement for long times in foundry alloys. In this article, the advantages and disadvantages of the various commercial Al-Ti-B master alloys have not been considered. The interested reader is referred instead to the excellent review by Spittle (Ref 29). He concluded that, when different refiners are compared on the basis of the amount of boron added, the popular Al-5Ti-1B alloy usually gives the most consistent refinement.

100 (5) (2) (1)

Equivalent average pore diameter (dp), μm

(3)

(4) 10 A356 alloy Casting #

H2 content (cc/100g)

(5) (2) (1) (3) (4)

0.31*** 0.31** 0.25* 0.25** 0.11**

1 0.1

* Non-grain-refined ** Grain-refined *** Grain-refined and modified

1

10

100

Cooling rate, ⬚C/s

Fig. 12

ACKNOWLEDGMENT The authors wish to acknowledge Alcoa Primary Metals, whose support made this study possible.

Pore size in A356 alloy castings. Source: Ref 25

REFERENCES Coherency fraction solid (fs*), %

60 55 50 45 40 35 30 25

0.5⬚C/s O 1⬚C/s

20 15 0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

0.18

0.20

0.22

Grain refiner added

Fig. 13

Dendrite coherency in an Al-4%Cu alloy. Source: Ref 26

titanium content is kept low, preferably less than approximately 0.05% Ti. The grain size is reduced and the resistance to hot cracking is improved in most castings. This runs counter to the accepted view, since many of these alloys call for a relatively high minimum titanium level. Example alloys are 201, 203, 204, 206, 240, 242, 710, 712, 713, 771, and 772.  Titanium has little effect: In many alloys, the titanium content has little effect on grain size when a powerful boron-containing refiner is employed. The Al-Si, Al-Si-Mg, and Al-Mg alloys fall into this category. Example alloys are 356, 357, 358, 359, 512, 520, and 535.

 High titanium is best: Tests in 319 alloy have

shown that it is necessary to have a minimum of approximately 0.1% Ti in order to produce a small grain size. By analogy, other Al-Si-Cu alloys presumably would also benefit from high titanium additions. Example alloys are 308, 319, 354, and 355. In all of the aforementioned cases, the best grain-refinement practice is to make an addition of 10 to 20 ppm of boron, preferably in the form of Al-5Ti-1B or Al-3Ti-1B rod. A Tibloy alloy can also be used and may be cost-effective in hard-to-refine alloys (such as 319), but at least 5 min should be allowed for the refiner to become active.

1. J.A. Nock, Jr., Method of Making and Casting Aluminum Alloys, U.S. Patent 1,912,382, 1934 2. A. Pacz, Aluminum Alloy Casting and Process of Making the Same, U.S. Patent 1,860,947, 1932 3. J. Suhr, Casting Alloy AP-33, Rev. Aluminium, Vol 9, 1932, p 1669–1680 4. W. Rosenhain, J. Grogan, and T. Schofield, Gas Removal and Grain Refinement of Aluminum Alloys, Foundry Trade J., Vol 43, 1930, p 177–180 5. G.K. Sigworth, The Grain Refining of Aluminum and Phase Relationships in the AlTi-B System, Met. Trans. A, Vol 15, 1984, p 277–282 6. L. Backerud, How Does a Good Grain Refiner Work?, Light Met. Age, Oct 1983, p 6–12 7. M.M. Guzowski, G.K. Sigworth, and D.A. Sentner, The Role of Boron in the Grain Refinement of Aluminum, Met. Trans. A, Vol 18, 1987, p 603–619 8. M.A. Easton and D.H. St. John, The Partitioning of Titanium during Solidification of Aluminum Alloys, Mater. Sci. Technol., Vol 16 (No. 9), 2000, p 993–1000 9. G.S. Cole, J. Cisse´, H.W. Kerr, and G.F. Bolling, Grain Refinement in Aluminum and Aluminum Alloys, AFS Trans, Vol 80, 1972, p 211–218 10. A. Cibula and R.W. Ruddle, The Effect of Grain Size on the Tensile Properties of High Strength Cast Aluminum Alloys, J. Inst. Met., Vol 76, 1949–1950, p 361–376

262 / Molten Metal Processing and Handling 11. A. Cibula, The Grain Refinement of Aluminum Alloy Castings by Additions of Titanium and Boron, J. Inst. Met., Vol 80, 1951–1952, p 1–16 12. H.T. Lu, L.C. Wang, and S.K. Kung, Grain Refining in A356 Alloys, J. Chin. Foundrym. Assoc., Vol 29, June 1981, p 10–18 13. G.K. Sigworth and M.M. Guzowski, Grain Refining of Hypo-Eutectic Al-Si Alloys, AFS Trans., Vol 93, 1985, p 907–912 14. G.K. Sigworth and M.M. Guzowski, Grain Refiner for Aluminum Containing Silicon, U.S. Patents 5,055,256 and 5,180,447; U.K. Patent GB2174103A 15. K. Nogita, S.D. McDonald, and A.K. Dahle, Effects of Boron-Strontium Interactions on Eutectic Modification in Al-10 mass% Si Alloys, Mater. Trans., Vol 44, 2003, p 692–695 16. G.K. Sigworth, M.A. Easton, J. Barresi, and T.A. Kuhn, Grain Refining of Al-Si Casting Alloys, Light Met., 2007, p 691–697 17. D.K. Young, B.T. Dunville, W.C. Setzer, and F.P. Koch, A Survey of Grain Refiners

18.

19.

20.

21. 22.

23.

for Hypoeutectic Al-Si Alloys, Light Met., 1991, p 1115–1121 J.H. Sokolowski, C.A. Kierkus, B. Brosnan, and J.W. Evans, Formation of Insoluble Ti(Al,Si)3 Crystals in 356 Alloy Castings and Their Sedimentation in Foundry Equipment: Causes, Effects and Solutions, AFS Trans., Vol 108, 2000, p 491–495 X.-G. Chen and M. Fortier, Formation of Primary TiAlSi Intermetallic Compounds in Al-Si Foundry Alloys, Mater. Forum, Vol 28, 2004, p 659–665 K. Pasciak and G.K. Sigworth, Role of Alloy Composition in Grain Refining of 319 Alloy, AFS Trans., Vol 109, 2001, p 567–577 V. de L. Davies, The Influence of Grain Size on Hot Tearing, Br. Foundryman, Vol 63, April 1970, p 93–101 G.K. Sigworth and F. DeHart, Recent Developments in the High Strength Aluminum-Copper Casting Alloy A206, AFS Trans., 2003, p 341–354 G.K. Sigworth, M. Kaufman, O. Rios, and J. Howell, Development Program on

24.

25.

26.

27.

28. 29.

Natural Aging Alloys, AFS Trans., Vol 112, 2004, p 387–407 G.K. Sigworth, Grain Refining of Aluminum Casting Alloys, Sixth International AFS Conference on Melt Treatment of Aluminum, Nov 11–13, 2001 (Orlando, FL), p 210–221 Q.T. Fang and D.A. Granger, Porosity Formation in Modified and Unmodified A356 Alloy Castings, AFS Trans., Vol 97, 1989, p 989–1000 G. Chai, Dendrite Coherency during Equiaxed Solidification in Aluminum Alloys, Chem. Commun., (No. 1), Stockholm University, 1994 L. Arnberg and L. Backerud, Solidification Characteristics of Aluminum Alloys, Vol 3: Dendrite Coherency, American Foundrymen’s Society, 1996 P.L. Schaffer, A.K. Dahle, and J.W. Zindel, Grain Refiner Fade in Aluminum Alloys, Light Met., 2004, p 821–826 J.A. Spittle, Grain Refinement in Shape Casting of Aluminum Alloys, Int. J. Cast Met. Res., Vol 19, 2006, p 210–222

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 263-265 DOI: 10.1361/asmhba0005247

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Refinement of the Primary Silicon Phase in Hypereutectic Aluminum-Silicon Alloys John L. Jorstad, JLJ Technologies, Inc.

PRIMARY SILICON in hypereutectic Aluminum-silicon alloys is very hard, imparting improved wear resistance but also decreasing tool life during machining. To minimize machining issues while fully utilizing outstanding wear properties, primary silicon crystals should be controlled to uniform small size and have uniform spatial distribution. This can be accomplished by adding phosphorus to the melt, thus creating an abundance of AlP3 nuclei suitable for initiating primary silicon crystal growth, and by solidifying cast product as rapidly as feasible. The conventional high-pressure die casting process normally solidifies product rapidly enough that artificial nuclei are not needed, but even there, added AlP3 nuclei may offset the less-than-ideal melt temperature and melt flow conditions that sometimes occur. Phosphorus added for refinement and the elements added to modify eutectic silicon react together such that primary silicon refinement and eutectic silicon modification in hypereutectic alloys cannot be accomplished simultaneously, except to the extent that both are affected by solidification rate. Only the additive in excess of that reacting together is effective in accomplishing refinement (phosphorus) or modification (strontium, for instance).

Importance of Primary Silicon Refinement The most notable feature of a hypereutectic aluminum-silicon alloy microstructure is the primary silicon phase. Silicon, in either the primary or eutectic form (Fig. 1), is present in aluminum alloys as essentially pure silicon crystals. (Only a miniscule amount of aluminum is dissolved in the silicon, and other alloying elements dissolve to an even lesser degree.) Silicon crystals are harder than any other phase typically found in aluminum casting alloys. On the Mohs scale, silicon is 6.5 versus corundum at 7. (The corundum form of Al2O3 is a

common medium for abrasively cutting and grinding hardened metals.) On the Knoop microhardness scale, silicon particles have been measured at 1100 and above, whereas sludge (the complex Al15(MnFe)3Si2 phase often referred to in die castings as hard spots) typically measures less than 600, and common forms of Al2O3 and MgAl2O4 (spinel) skim and particulates measure even less. On the Knoop scale, even the fully hardened Al-Cu, Al-Mg, or AlCu-Mg alloy alpha matrices typically measure well under 200. The hardness of silicon generally provides two very dramatic effects in aluminum-silicon casting alloys: superior wear resistance and reduced tool life during machining (Ref 1). Both are affected by the volume fraction, size, morphology, and distribution of silicon crystals; thus, hypereutectic alloys (containing a much larger total volume fraction of silicon crystals as well as larger average silicon size) have the greatest impact. To minimize machining issues while maintaining favorable wear performance, the primary silicon must be reasonably small and uniformly distributed throughout a cast matrix, hence the importance of primary silicon refinement.

causing eutectic aluminum grains to nucleate first and forcing silicon to then grow between those aluminum grains, thus acquiring a fibrous, broomlike morphology. This controlled growth phenomenon is commonly termed modification.

Accomplishing Primary Silicon Refinement Primary silicon size is controlled both by nucleation and solidification cooling rate. Absent any artificial nuclei, primary silicon nucleates after relatively deep undercooling and, once initiated, can then grow rapidly to large size and very irregular shape (Fig. 2). However, providing an abundance of suitable nuclei on which primary silicon crystals can initiate and grow dramatically changes that overall microstructure (Fig. 3). One of the most effective means to assure abundant nucleation sites is to add phosphorus to the melt. Phosphorus is added to hypereutectic aluminum-silicon alloys in an amount sufficient to retain 0.0005 to 0.0025% (less that 50% of that added is usually retained). Phosphorus combines with aluminum to form tiny, insoluble AlP3

Refinement versus Modification Primary Silicon Phase Refinement. Controlling the size, shape, and distribution of the primary silicon phase in hypereutectic aluminumsilicon alloy castings (alloys with >12% Si) is based on maximizing the number of nucleation sites, augmented with rapid solidification cooling rates. This nucleation phenomenon is commonly termed refinement. Eutectic Silicon Modification. Controlling the size and morphology of the eutectic silicon in hypoeutectic or eutectic aluminum-silicon alloy castings (alloys with 12% Si or less) is, on the other hand, accomplished by chemically inhibiting the usual nucleation of eutectic silicon,

Fig. 1

Primary (the larger, more equiaxed dark phase) and eutectic (the smaller, more elongated dark phase) silicon crystals in a T-6 hypereutectic aluminumsilicon alloy cast microstructure. Both forms/phases are essentially pure silicon. Original magnification: 800

264 / Molten Metal Processing and Handling particles (Fig. 4) (Ref 2). Because of very similar crystal habits (both are diamond cubic) and lattice constants, AlP3 particles are effective nuclei on which primary silicon crystals initiate. Sources of Phosphorus. Over the years, a variety of means emerged for introducing phosphorus to melts. In the timeframe of 1960 to 1980, several salts containing phosphorus were

200 µ

Fig. 2

Irregular growth of primary silicon crystals after deep undercooling and rapid solidification

Fig. 3

More uniform size, shape, and distribution of nucleated primary silicon crystals. Original magnification: 200

popular. Another method in that timeframe was vaporized PCl5 bubbled into the melt entrained in N2 as a carrier gas. While they accomplished the desired refinement of primary silicon, those products fell out of favor because the fumes associated with breakdown of salts or vaporizing of PCl5 were unpleasant if not unhealthy. Another additive researched and reported in the 1960s (Ref 2) and continuing in use even today (2008) is the alloy of phosphorus and copper. Commonly called phos-copper, it is a product readily available in the form of extruded brazing rod or as shot used to deoxidize copper alloy melts. Brazing rod is usually eutectic phosphorus-copper containing nominally 8% P. Deoxidizing shot is more often hypereutectic phosphorus-copper and contains 15 to 20% P. Both are effective additives, but the eutectic phosphorus-copper is easier to use. Phosphorus-copper must dissolve in the melt, although the melting point of the eutectic grade is close to that of many hypereutectic aluminum-silicon alloy melts. Hypereutectic phosphorus-copper often has a tough oxide surface that renders it slow and perhaps difficult to dissolve, whereas the eutectic phosphoruscopper dissolves quickly and completely. The eutectic phosphorus-copper is therefore the preferred product, and manufacturers now produce eutectic phosphorus-copper in shot form, too. It is less expensive than extruded rod and handy to precisely weigh for charging in the relatively small quantities usually required. An AlCuP master alloy rod for primary silicon refinement was introduced by VAW in Germany in the late 1980s. This rod introduces preformed AlP3, and the product has proven quick and reliable for primary silicon refinement. Effect of Solidification Rate. Solidification rate alone can be sufficient to refine primary silicon, although only at the extreme rates often

(a)

Fig. 4

AlP3 nucleus in primary silicon. Original magnification: 1500

(b)

experienced during conventional high-pressure die casting (HPDC) does AlP3 seem unnecessary. Conventional high-pressure die casting probably requires no artificial nucleation. As noted in Fig. 5, HPDC can achieve very small primary silicon size even without benefit of specifically-added AlP3. This is likely the result of the exceptional solidification rates normal to HPDC but may be aided by turbulent flow affecting nucleation, too. Regardless of the reason, controls are still required to assure reasonable uniformity of primary silicon size and spatial distribution. Figure 6, for instance, compares a well-made HPDC compressor body with one in which the die was filled either from melt that was too cold (already below the alloy liquidus entering the die) or was flowing too slowly. Primary silicon should ideally nucleate in the die cavity, influenced by pressurized intimate contact between melt and die steel. However, if the melt is held too close to liquidus temperature, or if the shot is delayed too long in the shot sleeve, or if flow ahead of the die cavity is too slow, primary crystals can initiate and grow too early in the casting cycle and under less-than-ideal conditions existing in the part cavity. Because the ideal scenario can be difficult to assure, many die casters require phosphorus in their hypereutectic alloy melts and claim that somewhat better primary silicon microstructures result. The ideal refined primary silicon size thus depends on many factors, including process conditions and specific product requirements. The type of casting process itself is also a factor, because solidification rate plays a role. In general, the nominal primary silicon size that is desirable in many sand castings is 35 to 50 mm; in permanent mold, 25 to 40 mm; and in diecasting, where refinement may or may not be employed, 15 to 25 mm is sufficient. Ingot pretreated with phosphorus can be purchased and is quite suitable; however, melts

(c)

Effect of solidification rate. (a) Unrefined high-pressure die casting chain saw crankcase, >150  C/s (300  F/s). (b) Gravity permanent mold piston, 10  C/s (50  F/s). (c) Sand cast engine prototype, <1  C/s (34  F/s). (b) and (c) AlP3 refined. Original magnification: 75

Fig. 5

Refinement of the Primary Silicon Phase in Hypereutectic Aluminum-Silicon Alloys / 265

(a)

Fig. 6

(b) (a) Well-made high-pressure die cast microstructure versus (b) one where primary silicon nucleated both outside (large crystals) and within (tiny crystals) the die cavity. Original magnification: 100

(a)

(b)

Fig. 7

(a) AlP3 inclusion on a machined surface. (b) Agglomeration of tiny AlP3 particles. The phosphide reacts with atmospheric moisture to become the phosphate, which then protrudes from the machined surface. Original magnification: 1000

often contain a significant fraction of returns, and refinement fades over time. Thus, a more robust and reliable method is to treat each melt with phosphorus according to its specific need.

Loss of Refinement Over Time. The AlP3 particles have an affinity for each other and seem to lose effectiveness largely by agglomerating into clusters that are no longer effective as nuclei. Clusters migrate to melt surfaces,

oxidize, and are removed in dross. Agglomeration and clustering are hastened by agitation of melts, meaning that melt size (relative volume affected at any given time by stirring or ladling frequency, for instance) can influence the rate of loss of AlP3 effectiveness. As a rule, large melts (tons) lose the effectiveness of phosphorus refinement over a matter of many hours and even days, whereas small melts (a few kilograms) can lose it within 4 to 5 h. The AlP3 clusters are easily floated into the dross and removed during fluxing and degassing, especially when “active” (de-wetting) fluxes such as those containing chlorine are employed. Agglomerated particles can eventually collect into relatively large masses, and, if not removed during fluxing/degassing or filtering, AlP3 can become entrapped in castings to appear as inclusions (Fig. 7). While such inclusions are rare, their mere possibility is strong argument for not going overboard with refinement frequency or phosphorus amount and for properly maintaining melt surface cleanliness. Primary Silicon Refinement and Eutectic Modification Are Not Compatible. Artificial nucleation of primary silicon with phosphorus and modification of the eutectic with sodium, strontium, or other modifiers cannot be achieved simultaneously. Phosphorus and the elements used for modification react to form compounds that accomplish neither primary refinement nor eutectic modification. Only if one or the other of the intended additives is in excess of that needed to form the compounds will refinement or modification take place, but not both.

REFERENCES 1. J.L. Jorstad, The Hypereutectic AluminumSilicon Alloys 390 and A390, Trans. TMS/ AIME, Vol 242, 1968, p 1217–1222 2. F.L. Arnold and J.S. Prestley, Hypereutectic Aluminum-Silicon Alloy Phosphorus Refinement, Trans. Am. Foundrymen’s Soc., Vol 69, 1961, p 129–137

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 269-275 DOI: 10.1361/asmhba0005206

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Thermodynamics and Phase Diagrams Ursula R. Kattner, National Institute of Standards and Technology

THERMODYNAMICS is a branch of physics and chemistry that covers a wide field, from the atomic to macroscopic scale. In materials science, thermodynamics is a powerful tool for understanding and solving problems. The part of thermodynamics that concerns the physical change of state of a chemical system following the laws of thermodynamics is called chemical thermodynamics. The thermodynamic properties of individual phases can be used for evaluating their relative stability and heat evolution during phase transformations or reactions. Traditionally, the most common application of thermodynamics is in the form of phase diagrams. Phase diagrams are visual representations of the state of a material, generally equilibrium, as a function of temperature, pressure, and, in multicomponent systems, the concentrations of the constituent components. They are therefore frequently hailed as basic blueprints or roadmaps for alloy design, development, processing, and basic understanding. The importance of phase diagrams is also reflected by the publication of many handbooks (Ref 1). Thermodynamic quantities of pure substances are tabulated in a number of compilations (Ref 2), while only a few compilations of thermochemical data for solution phases exist (Ref 3, 4). Most of these compilations provide the data in tabular form, although some provide them in the form of analytical functions or in electronic form. The Scientific Group Thermodata Europe compilation of binary systems (Ref 4) provides selected thermochemical values in tabulated and graphical form. This compilation also includes phase diagrams calculated from the same thermodynamic functions as the tabulated thermochemical values. It is the first compilation that makes use of the correlations between thermodynamics and phase diagrams established by Gibbs.

Thermodynamics The thermodynamic quantities that are most frequently used in materials science are the enthalpy, in the form of the heat content of a phase, the heat of formation of a phase, or the latent heat of a phase transformation; the heat

capacity, which is the change of heat content with temperature; and the chemical potential or chemical activity, which describes the effect of composition change in a solution phase on its energy. All of these thermodynamic quantities are part of the energy content of a system and are governed by the three laws of thermodynamics. Detailed physical derivations of the relationships of these quantities can be found in many textbooks (Ref 5, 6). A system is considered to be in equilibrium when its energy is at a minimum. Under constant pressure conditions for a closed system, this energy is the Gibbs energy, G: G ¼ H  TS

(Eq 1)

where H is the heat content or enthalpy, T is the temperature, and S is the entropy. The entropy is a rather abstract entity and can be envisioned as a measure of the disorder of atoms in the system. The temperature dependence of the Gibbs energy of a pure substance is usually expressed as a power series of T: G ¼ a þ bT þ cT lnðT Þ þ dT 2 þ eT 3 þ fT 1 þ gT 7 þ hT 9

(Eq 2)

where a . . . h are coefficients (Ref 7). Pressure and magnetic states result in additional contributions to the Gibbs energy. However, pressure dependence for condensed systems at nearatmospheric pressures is usually ignored. For multicomponent systems, it has proven useful to distinguish three contributions from the concentration dependence to the Gibbs energy of a phase, Gj: Gj ¼ G0 þ Gideal þ Gxs

(Eq 3)

The first term, G0, corresponds to the Gibbs energy of a mechanical mixture of the constituents of the phase: G ¼ 0

X

xi Gi ðT Þ

(Eq 4)

i

where xi is the mole fraction of each of the components, and Gi(T) is the Gibbs energy of the pure component. The second term in Eq 3, Gideal, is the contribution from the

configurational entropy from atoms substituting for each other on equivalent sites in a random mixture. In substitutional solutions, such as liquid or disordered solid solutions, it is given as: Gideal ¼ RT

X

xi ln xi

(Eq 5)

i

The third term in Eq 3, Gxs, is the so-called excess term. The excess Gibbs energy or Gibbs energy of mixing of a binary regular solution is given as: Gxs ¼ xA xB 

(Eq 6)

In a regular solution, the parameter O is a measure of the enthalpy of mixing of the phase of interest and represents the deviation of the Gibbs energy from ideal behavior. It is negative if A-B neighbors are favored in the solution and positive if A-A and B-B are favored in the solution. The solution is called subregular when O is a function of the composition and quasi-regular when O is also a function of temperature. In general, these solutions are referred to as regular-type solutions. The formalisms needed for the description of Gxs of ordered phases with extended homogeneity ranges or liquid phases with short-range order are generally more complex, since an internal order parameter must be taken into account. The order parameter is zero for the completely disordered state and is at maximum for the perfectly ordered state. Under equilibrium conditions, the Gibbs energy must also have a minimum with respect to this internal parameter. For a given temperature and concentration, the equilibrium (minimum Gibbs energy) may be single phase or a mixture of phases. The Gibbs energy of a mixture of phases is the weighted sum of Gibbs energies of the individual phases: G¼

X

f j Gj

(Eq 7)

j

where fj is the phase fractions of the individual phases. The requirement of minimum Gibbs energy for a system in equilibrium requires that the chemical potential, mji , of component i in phase j must be equal in all phases:

270 / Principles of Solidification maA maB .. . maZ

¼ ¼ ¼

mbA mbB .. . mbZ

¼ ¼ ¼

mgA mgB .. . mgZ

¼ ¼ ¼

... ... .. _ .

(Eq 8)

...

The chemical potentials of the two elements in a binary regular solution phase are (Ref 6): mA ðT; xA Þ ¼ GA ðT Þ þ ð1  xA Þ2  þ RT ln xA mB ðT; xB Þ ¼ GB ðT Þ þ ð1  xB Þ2  þ RT ln xB ðEq 9Þ

The Gibbs energy for a given composition can also be calculated from the chemical potentials: G¼

X i

xi mi

(Eq 10)

The chemical activity is related to the chemical potential by: mi ¼ G0i þ RT ln ai

Pi Pi0

G

T1

T2

β

α

T

l

l

l+β

T2 β

T3

α

T4

B

T3

α

TB

α XB

G

T1

l+α

l

α l+α A

β

l

TA

α+β A

XB

B

l+α

l

l+β

β

XB

A G

B

T4

β

α

(Eq 12)

The vapor pressure of a component in solution, Pi, is reduced in comparison to that of the pure component, Pi0 . The vapor pressure of an individual component is the same over all phases in equilibrium.

Fig. 2

G

(Eq 11)

where G0i is the reference state of the pure component. In an ideal solution, the activity of a component is equal to its molar fraction, ai = xi. This relation is known as Raoult’s law. Real solutions, such as regular-type solutions, deviate from Raoult’s law. The deviation of the activity from the molar fraction of the component is described by the activity coefficient, gi, ai = gi xi; this relation is known as Henry’s law for dilute solutions. The activity, ai, of component i is defined as: ai ¼

(representing the line of equal chemical potential for all components) and the equilibrium compositions of the two phases are given by the tangent points. The line in the phase diagram connecting these two points is called a tie line. The relative position of the Gibbs energy curves varies with temperature and causes the concentrations of the tangent points to change as the temperature changes. The phase boundaries plotted in a phase diagram are the locii of these tangent points as a function of temperature. A special temperature can exist where three phases may occur in equilibrium, for example, T3 in Fig. 1. Although a phase diagram is usually determined from direct measurement of concentrations and temperature, it is

The requirement that the chemical potential of a component must be the same for all phases in equilibrium directly links the Gibbs energy to the phase diagram, as illustrated in Fig. 1. The Gibbs energy functions of three phases, liquid (l), a, and b, are plotted as functions of the concentration, xB (xA + xB = 1 for a binary system) for various fixed temperatures. The minimum of the Gibbs energy is given by the lowest of the individual Gibbs energy curves and tangents to the Gibbs energy curves. The equilibrium is a single phase if the minimum Gibbs energy is given by the curve of this phase. The equilibrium is a twophase mixture if the minimum Gibbs energy is given by the common tangent of two curves

l

β

α+β

β

α α A

Fig. 1

l+α

l+β XB

β

α B

A

XB

B

Relation between the Gibbs energy curves of the liquid (l), a, and b phases

Calculated phase diagrams for a liquid (L) and solid (S) phase using the regular solution model with various regular solution parameters OL and OS. (a) OL = OS = 0 (ideal solution). (b) OL = 10 kJ/mol; OS = 20 kJ/mol. (c) OL = OS = 20 kJ/mol. The transformation entropies of the pure components are 10 J/(mol K).

Thermodynamics and Phase Diagrams / 271 important to understand the underlying thermodynamic basis for these diagrams, since it creates rules that experimentally determined phase diagrams must obey. The Gibbs energy functions can be used to illustrate how the magnitude and particular shape of the excess functions affect the resulting phase diagram. Figure 2 shows that just by varying the parameter of regular solution of a liquid and a solid phase, OL and OS, lensshaped isomorphous (Fig. 2a), eutectic (Fig. 2b), or pertitectic (Fig. 2c) phase diagrams can be obtained. A positive value of Oj results in the formation of a miscibility gap (a region of phase separation) at lower temperature. However, since the contribution of Gideal is negative, the miscibility gap closes at the critical point at sufficiently high temperature. For a regular binary solution, the miscibility gap is symmetric and closes at xjA ¼ xjB ¼ 0:5 with the critical temperature, Tcj (Ref 5): Tcj ¼

j 2R

(Eq 13)

The relationship between the Gibbs energy functions and phase diagrams is the basis for the calculation of phase diagrams (CALPHAD) method (Ref 8). The calculation of phase diagrams has a number of advantages: (1) the asobtained phase diagrams and thermochemical data are self-consistent because they are calculated from the same set of Gibbs energy functions, (2) prediction of higher component systems is possible, (3) efficient storage of the information in the form of a thermodynamic database instead of a multitude of diagrams is possible, and (4) input data for other simulation software are provided. The Gibbs energy functions employed by the CALPHAD method are relatively simple mathematical functions that are based on physical models for atomic interactions in each phase. The adjustable parameters of these functions are obtained from fitting experimental thermochemical data (enthalpies, chemical potentials, etc.), experimental phase diagram data, or predictions from first principles calculations that are based on density functional theory (Ref 9). Phase diagrams for ternary and higher-order systems can be generated by the CALPHAD method. First, the thermodynamic descriptions of the constituent binary systems are derived. Thermodynamic extrapolation methods are then used to extend the thermodynamic functions of the binaries into ternary and higher-order systems. The results of such extrapolations can then be compared to experimental data, and, if necessary, interaction functions are added to the thermodynamic description of the higher-order system. In principle, this strategy is followed until all 2, 3, . . . n constituent systems of an ncomponent system have been assessed. However, experience has shown that, in most cases, no, or very minor, corrections are necessary for

reasonable prediction of quaternary or highercomponent systems (Ref 8). Since true quaternary phases are rare in metallic systems, assessment of most of the ternary constituent systems is often sufficient to describe an n-component system. This method has been successfully applied to many multicomponent systems and a number of commercial alloy systems.

Phase Diagrams Multicomponent phase diagrams are usually two-dimensional maps of phase stability regions for ranges of temperature and composition. Phase boundaries identify the location of phase changes. The liquidus boundary separates phase regions containing only liquid phase(s) and phase regions containing liquid and solid phase(s). This boundary represents the limit where the fraction of solid is zero. The solidus boundary separates phase regions containing only solid phase(s) and phase regions containing solid and liquid phase(s). This boundary represents the limit where the fraction of liquid is zero. The solvus is the boundary between phase regions containing only one solid phase and phase regions containing two or more solid phases. The solvus line represents the limit where the fraction of the other solid phase(s) is zero, that is, the limit where the solvent phase is saturated with the solute constituents, and higher concentrations result in the precipitation of other solid phases. The degrees of freedom, F, of a phase region are governed by the Gibbs phase rule: þF ¼Cþ2

(Eq 14)

where F is the number of phases in equilibrium, and C is the number of components in the system. The degrees of freedom are the externally controllable conditions of temperature, pressure, and, in a multicomponent system, composition. Most materials processes take place at constant pressure, which often is the atmospheric pressure, removing one of the degrees of freedom: þF ¼Cþ1

(Eq 15)

Equilibria with no degrees of freedom, that is, temperature and compositions of all fixed phases, occur where C + 1 phases simultaneously exist. These are important features of a phase diagram and are called invariant equilibria or invariant reactions. For example, in a binary system, an invariant three-phase equilibria exists where a liquid is in equilibrium with two solid phases. The invariant three-phase equilibria in a binary system can be divided into two groups: decomposition or eutectic-type reactions and formation or peritectic-type reactions. In a binary eutectic (Fig. 2b), the liquid decomposes on cooling into two solid phases, and in a peritectic (Fig. 2c),

the liquid reacts with a solid phase on cooling to form a second solid phase. Phase regions with C phases in equilibrium have, according to the phase rule, one degree of freedom (monovariant equilibrium). Only one of the variables, temperature or phase composition, can be chosen freely. An invariant binary equilibrium becomes monovariant in a ternary system because of the additional degree of freedom and generally retains its eutectic-type or peritectic-type character, although it is possible for its character to change at a certain temperature. An invariant equilibrium occurs at the intersection of C + 1 monovariant equilibria. The phase fractions, fj, in a multiphase equilibrium can be determined from: x0i ¼

X j

f j xji

(Eq 16)

where x0i is the concentration of component i in the alloy, and xji is the concentration of component i in phase j. In a binary system where xA + xB = 1, the phase fractions are: fa ¼

xbB  x0B xbB



xaB

and f b ¼

x0B  xaB xbB  xaB

(Eq 17)

and the ratio of the two phase fractions becomes: f a xbB  x0B ¼ f b x0B  xaB

(Eq 18)

The same ratio exists by imagining a lever balanced at x0B , with weights replaced by the phase fractions and the lever arm lengths replaced by the concentration differences; the amount of a phase becomes larger the closer the alloy composition is to the phase boundary. This technique for determining phase fractions has come to be known as the lever rule.

Metastable Phase Diagrams Metastable equilibria occur when a phase is missing, for example, if the phase fails to nucleate. The failure of an equilibrium phase to form may result in the formation of a metastable phase. In the Gibbs energy diagram, the Gibbs energy curve of a metastable phase always is above the minimum Gibbs energy envelope of the equilibrium phases. The most well-known example for such a case is the iron-carbon system, shown in Fig. 3. Metastable cementite, Fe3C, results if graphite, or carbon, fails to form. The phase diagram shown in Fig. 3 is actually the superimposition of two phase diagrams, the stable equilibrium iron-carbon diagram (solid lines) and the metastable Fe-Fe3C diagram (dashed lines). In general, the absence of an equilibrium phase results in extended solubility in the remaining equilibrium phases. The liquidus of

272 / Principles of Solidification the metastable phase is always below the liquidus of the equilibrium phase that is absent.

Solidification Phase diagrams provide useful information for understanding the solidification of alloys. In addition to the liquidus and final freezing temperatures, important quantities for the mathematical treatment of solidification processes are obtained from the phase diagram. The ratio of the concentrations of the liquidus, xLi , and the solidus, xSi , at the same temperature is described by the partition coefficient or distribution coefficient, ki: xS ki ¼ Li xi

called the Scheil path), no diffusion in the solid and complete diffusion in the liquid are assumed. This case, where thermodynamic equilibrium exists only as local equilibrium at the liquid/solid interface, produces microsegregation with the lowest possible final freezing temperature and therefore represents the most extreme case of solidification with microsegregation. An alloy that solidified following the Scheil path is not in equilibrium after solidification. An expression similar to Eq 20 can be obtained for the Scheil path by equating the solute rejected during the formation of a small amount of solid, resulting in an increase of solid in the liquid (Ref 10): 

   xLB  xSB df S ¼ 1  f S dxLB

(Eq 21)

(Eq 19)

Phase diagram information can also be used in two simple models to describe the limiting cases of solidification behavior. For solidification obeying the lever rule at each temperature during cooling, complete diffusion is assumed in the solid as well as in the liquid. Thus, all phases are assumed to be in thermodynamic equilibrium at each temperature during solidification. For the solidification of a single solid phase in a binary alloy, rewriting Eq 17 using f S = 1  f L and assuming a constant partition coefficient gives: xLB 1 ¼ x0B 1  ð1  kB Þf S

(Eq 20)

In comparison, for solidification following the Scheil-Gulliver path (for convenience, here

0

5

1600 1538⬚C (dFe)

10

Integration from xSB ¼ kB x0B at f S = 0 gives the so-called Scheil-Gulliver equation (Ref 10): xLB 1 ¼ x0B ð1  f S Þð1kB Þ

(Eq 22)

While the Scheil-Gulliver equation can also be derived for ternary and higher-component alloys, it requires knowledge of the tie lines. An alloy that is not in equilibrium after solidification can be the result of a phase failing to nucleate, incomplete diffusion, or solute trapping. The description of such solidification processes with equilibrium phase diagram information alone is not possible. However, for most alloys where the solidifying phase is a substitutional solid solution, predictions of the Scheil-Gulliver model are closer to reality than those of the lever model, while for

Carbon, at.% 15

20

25

0.09 0.53 L

1493⬚C 1400 0.16 1394⬚C

L + C(graphite) 1252⬚C 2.1

(gFe)

4.2

1153⬚C 2.14 1147⬚C

4.3

1000

Fe3C

Temperature, ⬚C

1200

912 ⬚C 800

0.65 0.021 0.022 0.76

740⬚C 727⬚C

600 (aFe)

400

Fig. 3

0 Fe

1

2

3 4 Carbon, wt%

5

6

Superimposed stable and metastable iron-rich part of the iron-carbon system. The metastable phase boundaries are shown with dashed lines.

interstitial solid solutions where the interstitial element diffuses rapidly, predictions of the lever model are closer to reality. Modeling of real solidification behavior requires knowledge of the temperature dependence of the partition coefficients and liquidus slopes and a kinetic analysis of microsegregation and back diffusion at each temperature. The solidification of a binary alloy is schematically shown in Fig. 4 (the sketches are oversimplified to emphasize the concentration gradients). The solidification of the alloy labeled with “x” begins at TL. In the case of lever solidification, the composition of the growing a phase changes homogeneously during cooling according to the tie lines (horizontal lines) until the temperature TP, the peritectic formation of the b phase, is reached. At this temperature, b phase is formed from the a phase and the remaining liquid phase, and solidification is complete. On further cooling, the compositions and phase fractions of the a and b phases continue to change according to the lever rule. In the case of Scheil solidification, only the composition of newly growing a phase changes according to the tie lines during cooling; the composition of a phase formed at higher temperatures remains unchanged because of the absence of diffusion in the solid phase, and the remaining liquid becomes enriched in solute. On reaching the peritectic temperature, TP, solidification of the b phase begins. The amount of a becomes frozen at this point. No formation of b phase from a phase reacting with the liquid phase occurs because of the absence of diffusion. Solidification of the b phase continues until all liquid is consumed or when the eutectic temperature, TE, is reached, where the remaining liquid solidifies in a eutectic reaction. An A-rich alloy that freezes as single a phase in lever solidification may show the formation of b phase in Scheil solidification. A B-rich alloy that freezes as two phases, b + g, in lever solidification may actually have all three solid phases, a + b + g, in Scheil solidification if its composition is on the A-rich side of the peritectic point. In some alloys, annealing at elevated temperatures but below the solidus temperature can be used to remove nonequilibrium phases and concentration gradients from microsegregations. Lever solidification in a ternary alloy is, in general, similar to that of a binary alloy, but, due to the added degree of freedom, a threedimensional representation is necessary. The phase boundaries that are crossed during cooling and the sequence of phase formation can be read from a temperature-concentration section (e.g., an isopleth, a vertical section through Fig. 5a or b). However, such a section provides only very limited information on the compositions of the individual phases or phase fractions, since tie lines are rarely in the same plane as the section. A temperature-concentration section is ill suited to obtain information for Scheil solidification; more information can be obtained from the liquidus projection, which

Thermodynamics and Phase Diagrams / 273 is a topological map of the liquidus surface, and isothermal sections. To illustrate Scheil solidification for different topologies of ternary systems, three-dimensional phase diagrams and liquidus projections of two hypothetical systems are shown in Fig. 5. Figure 5(a) shows a simple ternary system. The topmost surfaces represent the liquidus surface where various phases first solidify, depending on the alloy composition. Intersections of these primary solidification surfaces occur along lines resembling valleys. These lines are also referred to as lines of twofold saturation because the liquid is simultaneously in equilibrium with two solid phases. According to the phase rule, such a three-phase equilibrium is monovariant and therefore is also called a monovariant eutectic. At the intersection of three monovariant eutectics, a ternary invariant eutectic is formed.

During Scheil solidification of the alloy labeled with “y” in Fig. 5(a), primary b phase is formed first, and the liquid becomes enriched in solutes as b is formed until composition of the liquid coincides with one of the valleys. Solidification proceeds with descending temperature following the valley and the formation of two solid phases, b + g, until final freezing occurs at the temperature of the ternary eutectic equilibrium, shown as a dot on the liquidus projection in Fig. 5(c). The path of the Scheil solidification is marked with the thick line and arrows in Fig. 5. The hypothetical system shown in Fig. 5(b) is more complex. Not all intersections of the primary solidification surfaces resemble valleys; only a slope change occurs at the intersection of the L + g and L + d surfaces. The character of such a monovariant line is peritectic, and, as in the case of the binary

L TL T1

Temperature

TP

T2 T3 T4

α

TE

β γ

A

Lever solidification

Solid fraction TL

TL Total solid fraction

T1

T1

T2

T2

T3

T4

TE Fraction β

T3

T4

TE

B

Composition

X

L

α

Scheil solidification α

L

α

L

L

α

α L

L

α

α

α

β

β

β

β

α

Solid fraction TL

T1

Total solid T1 fraction

T2

T2

T3

T3

L

β

T4

α L

β

α

Fraction β

TE Eutectic (β+γ)

Fig. 4

TL

T4

TE Fraction γ

Hypothetical binary system. Phase fractions and oversimplified microstructures are shown for lever and Scheil solidification.

peritectic equilibrium, solidification “switches” from the formation of one solid phase to the formation of another solid phase. The solidification path of an alloy encountering such a monovariant peritectic is shown for alloy y in Fig. 5(b). The sequence of phase formation, shown on the liquidus projection in Fig. 5(d), during the Scheil solidification of this alloy is L!L+g!L+d!L+d+b!d+b+ a. However, the solidification of alloy z is different; its phase formation sequence is L ! L + g ! L + g + b ! L + d + b ! d + b + a. The solidification also begins with the formation of g until the liquid composition intersects with the valley of the L + b + g monovariant eutectic. Solidification proceeds with decreasing temperature following the valley until the invariant transition reaction occurs. In the transition reaction, the valley switches from the L + b + g monovariant eutectic to the L + b + d monovariant eutectic. The character of the invariant transitions reaction is closer to that of a peritectic reaction, and it is easy to understand why this invariant equilibrium is also called quasi-peritectic. Both alloys follow the L + b + d monovariant eutectic until final freezing occurs at the ternary eutectic. Although the methodology for obtaining solidification paths for lever and Scheil solidification is straightforward, the information needed for accurate predictions is, in many cases, incomplete. Due to their multidimensionality, the interpretation of diagrams for more complex systems can be quite cumbersome. For systems with more than three components, the graphical representation of the phase diagram in a useful form becomes a very challenging task. However, the difficulty of graphically representing systems with many components can be overcome by using thermodynamics for understanding reactions and phase transformations in multicomponent systems.

Calculation of Phase Diagrams and Solidification In recent years, the application of phase diagram information obtained from thermodynamic calculations to practical processes has increased significantly. Thermodynamic descriptions have become available for a large number of alloy systems and allow the calculation of the phase diagrams of multicomponent alloys. In addition to the phase equilibria data, the thermodynamic calculation also provides parameters that are needed for the simulation of casting processes, such as enthalpy data or driving forces for phase formation, as well as thermodynamic information that is needed for simulations considering kinetic factors. The calculation of the lever and Scheil path is a straightforward task. The change in temperature during cooling is approximated by a series of small decrements in temperature. For the lever path, the temperature is stepped, with the overall composition of the alloy being constant. The calculation of the Scheil path is based on

274 / Principles of Solidification z

y T

y

T TL γ

TL

β

δ

γ C α

α

β

B

C

A A

B

(a)

(b)

z

y γ

β

γ

δ

α

α (c)

Fig. 5

β

(d)

Solidication paths in hypothetical ternary systems. (a) Phase diagram of a system with a ternary eutectic. (b) Three-dimensional phase diagram of a system with a ternary transition reaction and a ternary eutectic. Solidification of the alloys begins at TL. (c) Liquidus projection of (a). (d) Liquidus projection of (b)

Fig. 6

the assumption that local equilibrium exists at the interface for each temperature step. The method used for calculating the Scheil path is illustrated in Fig. 4. At each temperature step, the fraction of solid phases formed is removed from the calculation, and the composition of the remaining liquid phase is used as the composition of the freezing alloy for the next temperature step. For example, the equilibrium is calculated at temperature T1 for alloy composition x, the amount of a that has formed is removed from the calculation, and the composition of the remaining liquid becomes the alloy composition for the calculation at the next temperature, T2. Simultaneously, the fractions of the solid phases that have been removed are summed to give the total fraction of solid formed during solidification: j j fnew ¼ fold þ f j

L j with f j ¼ fold fstep

(Eq 23)

j where fnew is the total accumulated fraction of j phase j, fold is the total fraction of phase j j is the fraction of from previous steps, and fstep phase j formed from the liquid in the current step. Stepping is usually stopped after a eutectic equilibrium has been encountered or a predetermined fraction of solid phase has formed. However, calculation of the enthalpy may be continued beyond solidification if results are needed for other calculations, such as process simulations. The calculation of the enthalpy versus temperature during Scheil solidification requires that

Phase fraction versus temperature curves for solidification of alloy Al-4.44Cu-1.56Mg-0.55Mn-0.23Fe-0.21Si-0.05Zn (mass fraction  100) using (a) equilibrium calculation and (b) nonequilibrium calculation

Thermodynamics and Phase Diagrams / 275 the variations in the compositions within the accumulated solid phases be considered. Instead of calculating the enthalpy for all compositions and increments of solid-phase fraction accumulated, the enthalpy can be calculated for the average composition of each individual solid at each temperature step (Ref 11). The average composition of each phase is obtained from: xji;new ¼

j xji;old fold þ xji;step f j j fold þ f j

(Eq 24)

The results obtained from the lever and Scheil calculations can be conveniently presented in graphical form in a phase fraction versus temperature diagram. Such diagrams are shown in Fig. 6 for lever and Scheil calculations for an alloy that is close in composition to the commercial aluminum alloy 243. The composition (in mass fraction  100) used for this calculation is 0.21Si-0.23Fe-4.44Cu0.55Mn-1.56Mg-0.05Zn, with the remainder being aluminum. For these calculations, the Al-DATA* database (Ref 12) was used. The difference between lever and Scheil solidification becomes most noticeable toward the end of solidification. For both solidification paths, the solidification begins with the precipitation of (Al) and Al6Mn. The lever solidification, shown in Fig. 6(a), continues with the formation of the a phase, Al20Cu2Mn3, Al7Cu2Fe, decomposition of Al6Mn and the a phase, and the formation of Mg2Si. Finally, after solidification is complete, precipitation of Al2CuMg begins. The final microstructure con-

*Trade names are used in this paper for completeness only, and their use does not constitute an endorsement by the National Institute of Standards and Technology.

sists mainly of (Al) and small amounts of Al20Cu2Mn3, Al7Cu2Fe, Mg2Si, and Al2CuMg. The Scheil solidification, shown in Fig. 6(b), continues with the formation of Al7Cu2Fe, Mg2Si, Al20Cu2Mn3 (this phase is not shown in Fig. 6b since the amount is extremely small), Al2CuMg, and Al2Cu. The microstructures obtained from these solidification paths are quite different and probably result in different mechanical properties. The actual microstructure found in castings can be compared to diagrams such as those shown in Fig. 6, and typical casting parameters, such as cooling rate or mold design, can be adjusted to obtain the desired microstructure by following a solidification path somewhere between these extremes. Examples of using the calculation of phase equilibria in solidification and casting simulations can be found in various chapters of this Handbook. REFERENCES 1. U.R. Kattner, The Thermodynamic Modeling of Multicomponent Phase Equilibria, JOM, Vol 49, Dec 1997, p 14–19 2. N. Jacobson, Use of Tabulated Thermochemical Data for Pure Compounds, J. Chem. Ed., Vol 78, June 2001, p 814–819 3. R. Hultgren, P.D. Desai, D.T. Hawkins, M. Gleiser, and K.K. Kelley, Selected Values of the Thermodynamic Properties of Binary Alloys, American Society for Metals, 1973 4. Thermodynamic Properties of Inorganic Materials Compiled by SGTE, Landolt-

5. 6.

7. 8.

9.

10. 11.

12.

Bo¨rnstein, New Series, Group IV, Vol 19, Subvol B, Springer, Berlin, Germany, 2002, onward C.H.P. Lupis, Chemical Thermodynamics of Materials, North Holland, New York, NY, 1983 M. Hillert, Phase Equilibria, Phase Diagrams and Phase Transformations: Their Thermodynamic Basis, Cambridge University Press, Cambridge, U.K., 1998 A.T. Dinsdale, SGTE Data for Pure Elements, CALPHAD, Vol 15, Dec 1991, p 317–425 N. Saunders and A.P. Miodownik, CALPHAD (Calculations of Phase Diagrams): A Comprehensive Guide, Pergamon, Oxford, U.K., 1998 J. Hafner, C. Wolverton, and G. Ceder, Toward Computational Materials Design: The Impact of Density Functional Theory on Materials Research, MRS Bull., Vol 31, Sept 2006, p 659–665 W. Kurz and D.J. Fisher, Fundamentals of Solidification, 3rd ed., Trans Tech Publications, Aedermannsdorf, Switzerland, 1989 W.J. Boettinger, U.R. Kattner, and D.K. Banerjee, Analysis of Solidification Path and Microsegregation in Multicomponent Alloys, Modeling of Casting, Welding and Advanced Solidification Processes VIII, B. G. Thomas and C. Beckerman, Ed., TMS, 1998, p 159–170 N. Saunders, Phase Diagram Calculations for Commercial Al-Alloys, Mater. Sci. Forum, Vol 217–222, 1996, p 667–672

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 276-287 DOI: 10.1361/asmhba0005207

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Nucleation Kinetics and Grain Refinement J.H. Perepezko, University of Wisconsin—Madison

NUCLEATION PROCESSES play a key role in the solidification of castings by controlling to a large extent the initial structure type, size scale, and spatial distribution of the product phases. During many solidification processes, the size scale of critical nucleation events is too small and the rate of their occurrence too rapid for accurate observation by direct methods. Nonetheless, nucleation effects in the solidification microstructure exert a strong influence on the grain size and morphology as well as the compositional homogeneity. The final microstructure is also modified by the crystal growth, fluid flow, and structural coarsening processes that are important in the later stages of ingot freezing. In large bulk castings, the solidification temperature corresponding to the onset of freezing is often close to but slightly less than the melting point or equilibrium liquidus temperature. The offset of the solidification temperature with respect to the equilibrium temperature is called the undercooling or supercooling, DT, and it plays a vital role in the overall description of the initial stage of the solidification that is controlled by nucleation. The level of melt undercooling at the onset of solidification is important to consider in developing an understanding of the variety of structural modifications and grain-refining practices in common casting alloys and is the basis of more recent solidification processing technologies using rapid solidification methods. In this article, selected highlights of thermodynamic relationships during solidification and nucleation kinetics behavior are discussed in connection with the basis of nucleation treatments, such as grain refinement and inoculation, to provide a summary of nucleation phenomena during casting.

and to account for the free energy changes involved in various crystallization processes. In most castings, a full global equilibrium associated with a uniform phase composition does not occur throughout the ingot. However, it is often valid to take the liquid and solid compositions at the solidification front to be represented by the liquidus and solidus compositions determined by the tie line end points at a given temperature on the phase diagram. This concept is called local interfacial equilibrium, and it has been successfully applied to the description of casting processes in which the rate of diffusion, particularly in the solid, is too sluggish to achieve uniform phase compositions (Ref 1). The consideration of nucleation and the level of melt undercooling introduce another type of departure from full equilibrium that is known as metastable equilibrium. Only when there is a departure from full liquid-solid equilibrium will solidification be possible. For solidification to occur, this departure brings the liquid into an undercooled state in which it is metastable because of the absence of one or more stable solid phases. The change of the liquid from a stable to a metastable state occurs in a continuous manner without an abrupt change in the physical properties, such as molar volume or heat capacity. Therefore, metastable states can exhibit a true reversible equilibrium. The thermodynamic description of solidification can be quantified by recalling that for a pure metal at the equilibrium temperature, Tf, the free energy change is:

Thermodynamics of Solidification

so that:

Macroscopic Solids. Throughout the analysis of solidification, thermodynamics is used to judge the alloy phase constitution, to describe the solidification path and composition changes in terms of partition coefficients and the slopes of the liquidus and solidus phase boundaries,


Gf ¼


Hf ðTf
T Þ
Hr
T ¼ Tf Tf

(Eq 3)

when the small correction due to heat capacity terms is neglected. Based on Eq 3, there is a direct relationship between DGf and the undercooling DT. The relationship in Eq 3 is represented in Fig. 1, in which the relative stability of the two phases (stable a and metastable b) that form from the liquid without composition change is shown as a function of temperature. The behavior is also approximately followed for stable and metastable eutectics such as Al/ Al3Fe and Al/Al6Fe and Fe/C and Fe/Fe3C (Ref 2). An important feature illustrated in Fig. 1 is that the melting of a metastable phase always occurs at a lower temperature than the stable-phase melting. For alloys, the free energy of a phase is a function of both temperature and composition. When components are mixed together to form an alloy, the molar free energy change, DGm, can be estimated from thermodynamic solution models (Ref 3). The simplest approach is to assume that the heat of mixing, DHm, is zero and that the entropy change upon mixing, DSm, is given by the ideal entropy of mixing. This is the description of an ideal solution, and it is often a useful approximation for dilute alloys. For an ideal solution of composition CB:

GS
GL ¼ ðHS
HL Þ
Tf ðSS
SL Þ ¼ 0

or
Gf ¼
Hf
Tf
Sf ¼ 0


Sf ¼ ð
Hf Þ=Tf

(Eq 1)

(Eq 2)

where GS and GL, HS and HL, and SS and SL are the molar free energy, enthalpy, and entropy for the solid, S, and liquid, L, respectively. At T, a temperature other than Tf, DGf 6¼ 0, and Eq 1 and 2 can be combined to yield:

Fig. 1

Molar free energy versus temperature curves for a single component with a stable a-phase and a metastable b-phase

Revised and updated from J.H. Perepezko, Nucleation Kinetics, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 101–108

Nucleation Kinetics and Grain Refinement / 277
Gm ¼
Hm
T
Sm ¼ RT ½CB ln CB þ ð1
CB Þ lnð1
CB Þ

(Eq 4)

With more concentrated alloys, a regular solution model is often applied in which the heat of mixing is given by the symmetrical term OCB(1
CB), where O reflects the net interatomic interaction. For regular solutions, DSm is taken as the ideal solution term in Eq 4. The description of equilibrium between solid and liquid phases over a range of alloy composition and temperature is represented in terms of the component chemical potentials mA and mB for each phase. The chemical potential is related to the molar free energy at a given composition by the standard thermodynamic definition (Ref 3): mA ¼ Gm
CB


 @Gm @CB T;P


 @Gm mB ¼ Gm þ ð1
CB Þ @CB T;P

(Eq 5a)

(Eq 5b)

As illustrated in Fig. 2(a), the definitions in Eq 5(a) and Eq 5(b) indicate that a tangent to a molar free energy curve at composition Co intersects the pure component axes at mA and mB. With this description, alloy phase equilibrium is represented by an equality of chemical potentials for each of the components in the equilibrated phases so that mLA ¼ mSA and mLB ¼ mSB , as shown in Fig. 2(b). The geometric interpretation of alloy phase equilibrium is expressed in terms of the common tangent construction, which is also shown in Fig. 2(b). The molar free energy of the liquid or the solid phase is also related to the chemical potential by the expression: Gm ¼ ð1
CB ÞmA þ CB mB

(Eq 6)

and provides a useful form in representing the free energy changes involved in the formation of solid phases from an alloy liquid solution. Free energy versus composition diagrams provide a useful method of analyzing the free

Fig. 2

energy relationships for alloy solidification reactions. Some of the relationships are shown in Fig. 3 for a temperature below the eutectic. For temperatures below the eutectic, the (a + L) and (b + L) equilibria are metastable with respect to the stable (a + b) phase mixture. In the phase diagram, the metastable two-phase conditions can be examined in terms of an extension of the respective solidus and liquidus phase boundaries below the eutectic, as noted in Fig. 3. For large excursions below the eutectic, the phase boundary extrapolations may not follow a simple linear extension. The relationships illustrated in Fig. 3 also show the common characteristic for a metastable phase to exhibit a higher solubility for solute than the equilibrium stable-phase solution. Similar behavior is found for the cases of peritectic and monotectic solidification reactions (Ref 2). When one or both of the stable solid phases in a eutectic or peritectic reaction are suppressed, metastable solid phases may develop and yield metastable eutectics and peritectics (Ref 2). Figure 4 shows the modified phase diagram for a metastable peritectic. The stable Al/ Al3Ti and metastable Al/AlxTi peritectic reactions may be relevant to the operation of grain refining in aluminum alloys (Ref 4, 5). To provide a description of compositional segregation plus the influence of constitutional supercooling and solute redistribution during two-phase eutectic growth, phase diagram information such as the partition coefficient, k, or the liquidus slope, mL, is often used in alloy solidification analysis (Ref 6, 7). For dilute solutions, the van’t Hoff relation for liquidsolid equilibrium holds and relates k and mL as: k¼1


mL
HfA RðTfA Þ2

Fig. 3

(Eq 7)

where
HfA and TfA are the latent heat and melting point, respectively, of pure solvent, A. The partition coefficient can be evaluated separately for a dilute solution by: 

HfB 1 1 k ¼ exp
R TfB TfA

where
HfB and TfB are the latent heat and melting point, respectively, of component B. Microscopic Solids. The preceding discussion of solidification thermodynamics applies to cases in which the solid phases are of macroscopic size. There are several important situations during nucleation, dendritic solidification, or eutectic growth at high velocity in which the solid is of microscopic size or

Eutectic phase diagram (a) illustrating the metastable extensions of the liquidus and solidus phase boundaries to a temperature T1 below the eutectic. (b) The molar free energy versus composition diagram at T1 shows the metastable (a + L) and (b + L) equilibria by dashed common tangents and the stable (a + b) equilibrium by the solid common tangent.

(Eq 8)

Alloy free energy versus composition diagrams showing the chemical potentials of the liquid and solid components. (a) Variation of the molar free energy of mixing with composition for a liquid alloy solution. The chemical potentials, mLA and mLB , for the solution at composition Co are given by the intercepts on the pure component axes. (b) Molar free energy versus composition diagram for two-phase liquid-solid equilibrium in an alloy of composition Co. Solid of composition CS is in equilibrium with liquid of composition CL, and the chemical potentials of A and B are equal in the solid and liquid.

Fig. 4

Modified phase diagram for a metastable peritectic. When a stable A3B phase is suppressed, the stable peritectic reaction involving A and A3B may be replaced by a metastable peritectic reaction between A and AxB. The peritectic temperature Tp and a-phase solubility for the metastable peritectic are higher than for the stable peritectic case.

278 / Principles of Solidification exhibits a sharply curved boundary with the liquid. The free energy of a small particle increases inversely with its size or radius of curvature. The increased free energy for the microscopic solid compared to the macroscopic solid results in a decrease in melting temperature, as expected for metastable equilibrium behavior. For a pure component, the free energy increment due to curvature K is given by:
Gk ¼ sKVm

(Eq 9)

where Vm is the molar volume, and s is the interfacial energy. For a sphere of radius, r, K = 2/r. In terms of Eq 9, the melting of a microscopic particle of pure component with radius r, Tf(r), is depressed from the melting point of a macroscopic particle by: Tf ðrÞ ¼ Tf


2sVm ðr
Sf Þ

(Eq 10)

For alloys, the influence of curvature can be treated by including a term based on Eq 9 to the chemical potential for the solid (Eq 6). This will shift the liquidus and solidus phase boundaries, but in a manner that results in a negligible change in k and mL (Ref 6).

Nucleation Phenomena Nucleation during solidification is a thermally activated process involving a fluctuational growth in the sizes of clusters of solids (Ref 8, 9). Changes in cluster size are considered to occur by a single atom addition or by removal exchange between the cluster and the surrounding undercooled liquid. At small cluster sizes, the energetics of cluster formation reveal that the interfacial energy is dominant, as can be observed by noting that the ratio of surface area to volume of a sphere is 3/r. For the smallest sizes, clusters are called embryos; these are more likely to dissolve than grow to macroscopic crystals. In fact, the excess interfacial energy due to the curvature of small clusters is the main contribution to the activation barrier for solid nucleation. This accounts for the kinetic resistance of liquids to crystallization and is manifested in the frequent observation of undercooling effects during solidification (Ref 10, 11).

Homogeneous Nucleation The principal features of nucleation phenomena and the kinetics of the rate process during solidification can be illustrated in the simplest terms by using the capillarity model to evaluate the kinetic factors (Ref 8). With this approach, it is useful to examine first the case of homogeneous nucleation in which solid formation occurs without the involvement of any extraneous impurity atoms or other surface sites in contact with the melt. As a further simplification, only the case of isotropic interfacial

energy is treated, but it should be recognized that anisotropic behavior can yield faceted cluster shapes. The energetics of cluster formation for a spherical geometry can be expressed in terms of a surface and a volume contribution as:
GðrÞ ¼ 4pr2 s þ 4/3pr3
GV

(Eq 11)

where DG(r) is the free energy change to form a cluster of size r, and DGV = DHf DT/TfVm. The relationship in Eq 11 is characterized in Fig. 5, in which the activation barrier for nucleation, DGcr, is reached at a critical size rcr (that is, dDG(r)/dr = 0), as given by: 2s 2sTf Vm ¼

Hf
T
GV

rcr ¼




(Eq 12)

and

Fig. 5


Gcr ¼

3

16ps ¼ 3
G2V

16ps Tf2 Vm2 3
Hf2
T 2 3

(Eq 13)

At increasing values of undercooling, rcr is reduced (rcr a DT
1), and Gcr is reduced more rapidly (DGcr a DT
2). A cluster is often considered to reach the stage of a nucleus capable of continued growth with a decreasing free energy when the size rcr is achieved, but in fact, stable nucleus growth ensues when the cluster size exceeds rcr by an amount corresponding to (DGcr
kT) in Fig. 5 (Ref 8). The relationship between cluster size and the number of atoms in a cluster, ncr, is expressed by (ncr Va Þ ¼ 4pr3cr =3, where Va is the atomic volume. To relate cluster energetics and fluctuational growth to the rate of nucleation, it is necessary to describe the cluster population distribution. As a liquid is undercooled, solid clusters begin to evolve over a transient period until the cluster size distribution attains a steady state. Because the mixture of clusters in an undercooled melt is a dilute solution, the metastable equilibrium concentration of clusters of a given size, C(n), can be expressed by (Ref 8):


GðnÞ CðnÞ ¼ C‘ exp kT

(Eq 14)

where C‘ is the number of atoms per cubic meter in the liquid, and DG(n) is given by Eq 11 when r is converted to n, as noted previously. If solid nucleation is regarded as the growth of clusters past the critical size, then the resulting cluster flux or the nucleation rate I (in nuclei m
3 s
1) can be represented kinetically by the product:





I ¼ nSL Scr ZCðncr Þ

(Eq 15)

where nSL is the jump frequency associated with atom jumps from the liquid to join the cluster and can be estimated from the liquid diffusivity, DL, and jump distance, a, as in (DL/a2); Scr is the number of atoms surrounding a cluster that is approximately ð4pr2cr=a2 Þ; Z is the Zeldovich factor that accounts for the decay of

Free energy change for cluster formation as a function of cluster size. The surface contribution to DG(r) is 4pr2s, and the volume contribution is 4pr3 DGv/3; rcr occurs at
2s/DGv where DG(r) = DGcr.



supercritical clusters and is usually of the order of 0.1 to 1.0 (Ref 8); and C(ncr) is the concentration of critical clusters. The full expression for the steady-state nucleation rate is then: I¼



2 
 DL 4prcr

Gcr ZC exp ‘ a2 a2 kT

(Eq 16)

For typical metals, C‘  1028 m
3 , DL  10
9 m2/s, and ao  0.3  10
9 m so that:
I ’ 1040 exp


16ps3 Tf2 Vm2 3k
Hf2 T
T 2





 (Eq 17)

and shows a rather steep temperature dependence, as illustrated in Fig. 6. At high temperature, the temperature dependence of I is dominated by the driving free energy term, which is contained within the exponential dependence on the activation barrier, and I can vary by approximately a factor of 5 per degree Celsius. Equation 16 or 17 indicates that the maximum nucleation rate occurs at (1/3) Tf. Over a wide temperature range, the liquid diffusivity cannot be taken as constant but can instead be represented as an Arrhenius function, DL = Do exp (
Q/kT), where Do is a constant, and Q is the activation energy for liquid diffusion. At large undercooling or low temperatures, and especially for glass-forming alloys, the diffusivity is often obtained from the viscosity in the form DL = (kT/3paZo) exp
[A/(T
To)], where Zo, A, and To are constants. Often, To can be estimated by Tg, the glass transition temperature. In evaluating nucleation rates, accurate values for the activation barrier are important because of the exponential dependence of I on DGcr. The evaluation of DGV is based on thermodynamic data or reasonably accurate models (Ref 12). However, separate, independent measurements of liquid-solid interfacial energy, sLS, at the nucleation temperature are not available. Instead, estimates for sLS

Nucleation Kinetics and Grain Refinement / 279

Fig. 6

Steady-state nucleation rate as a function of undercooling below the melting point. At low undercooling, the nucleation rate is primarily controlled by driving free energy; at high undercooling, as the temperature approaches the glass transition temperature, Tg, nucleation is limited by diffusional mobility.

The interfacial energy, s, relationships among a planar nucleant substrate (n), a spherical sector solid (S), and the liquid (L). The interfacial regions are designated by the subscripts LS (liquid-solid), nL (nucleant-liquid), and nS (nucleant-solid).

Fig. 7

are obtained from correlations (Ref 13), calculations based on structural models (Ref 14), or, more recently, from ab initio atomistic computation analysis (Ref 15–17).

Heterogeneous Nucleation Homogeneous nucleation is the most difficult kinetic path to crystal formation because of the relatively large activation barrier for nucleus development (DGcr). To overcome this barrier, classical theory predicts that large undercooling values are required, but in practice, an undercooling of only a few degrees or less is the common observation with most castings. This behavior is accounted for by the operation of heterogeneous nucleation in which foreign bodies such as impurity inclusions, oxide films, or crucible walls act to promote crystallization by lowering DGcr. Because only a single nucleation event is required for the freezing of a

liquid volume, the likelihood of finding a heterogeneous nucleation site in contact with a bulk liquid is great. Indeed, it has been estimated that even in a sample of high-purity liquid metal there is a nucleant particle concentration of the order of approximately 1012 m
3 (Ref 10). Only by using special sample preparation methods to isolate the melt from internal and external nucleation sites by subdivision into a fine droplet dispersion has it been possible to achieve undercoolings in the range of 0.3 to 0.4 Tf (Ref 11). The action of heterogeneous nucleation in promoting crystallization can be visualized in terms of the nucleus volume that is substituted by the existing nucleant, as illustrated schematically in Fig. 7. For a nucleus that wets a heterogeneous nucleation site with a contact angle y, the degree of wetting can be assessed in terms of cos y = (snL
snS)/sLS, where the interfacial energies are defined in Fig. 7. As y approaches 0 , complete wetting develops; as y approaches 180 , there is no wetting between the nucleus and the nucleant (which is inert), and the conditions approach homogeneous nucleation. The energetics of heterogeneous nucleation can be described by a modification of Eq 11 to account for the different interfaces and the modified cluster volume involved in nucleus formation. In terms of the cluster formation shown in Fig. 7, the free energy change during heterogeneous nucleation is expressed by:

in applying the continuum model described previously for heterogeneous nucleation. As the contact angle decreases below approximately 20 to 30 , h approaches atomic dimensions, so that the continuum description of wetting is not valid. This issue is of importance in the analysis of heterogeneous nucleation during grain refinement where the observed undercoolings are small and the implied contact angles are below 20 . Following an analysis similar to that developed for homogeneous nucleation based on Eq 15, an expression can be obtained for the rate of heterogeneous nucleation. In this case, the value of the surface area, Scr, for a spherical cap geometry is estimated by ½2pr2cr ð1
cos yÞ=a2 . In addition, the concentration of critical clusters is represented in terms of the number of surface atoms of the nucleation site per unit volume of liquid, Ca, which is of the order of 1020 m
3. The heterogeneous nucleation rate expression Ihet (m
3 s
1) is given by:


GðrÞhet ¼ VSC
GV þ ASL sSL þ AnS snS
AnL snL

where the prefactor represents an average value that can vary by approximately 1 order of magnitude for specific cases. The comparison between heterogeneous and homogeneous nucleation kinetics is illustrated in Fig. 9 by a schematic time-temperature transformation diagram. Although only a single transformation curve (that is, C-curve) is shown in Fig. 9 for heterogeneous nucleation, in reality there will be as many curves as the number of heterogeneous nucleation sites. Each curve for heterogeneous nucleation will be distinguished by a catalytic potency, that is, f(y), and a site density (Ref 11). To attain homogeneous nucleation conditions, it is clear that all heterogeneous nucleation sites must be removed or bypassed kinetically, such as during rapid quenching.

(Eq 18)

where VSC is the spherical cap volume, and ASL, AnS, and AnL are the solid-liquid, nucleant-solid, and nucleant-liquid interfacial areas, respectively. When the volume and relevant interfacial areas are expressed in terms of the geometry of Fig. 7, the evaluation of DGcr for heterogeneous nucleation yields:   16ps3LS 2
3 cos y þ cos3 y 4 3
G2V (Eq 19) ¼
Gcr ðhomÞ½fðyÞ


Gcr ðhetÞ ¼

so that the barrier for homogeneous nucleation is modified by f(y), the shape factor, during heterogeneous nucleation. Over the variation in y ranging from complete wetting (y = 0 ) to nonwetting (y = 180 ), f(y) or [DGcr(het)/D Gcr(hom)] varies from 0 to 1.0, as illustrated in Fig. 8. It is important to note that the value of rcr for the curvature of the critical nucleus does not change in the classical analysis of heterogeneous nucleation, but the reduction in the activation barrier shown in Fig. 8 has a significant influence on the nucleation rate. This feature is also described in Fig. 8, in which the spherical cap size as measured by the ratio (h/r) is plotted as a function of y. The dimension h is the height above the substrate. A closer examination of the trend of (h/r) in Fig. 8 reveals that there is a problem





2  DL 2prcr ð1
cos yÞ 2 2 a a 
 
Gcr ðhomÞ fðyÞ Ca exp
kT

Ihet ¼

(Eq 20)

or for typical metals:   16ps3LS Tf2 Vm2 fðyÞ Ihet ’ 1030 exp
3k 
Hf2  T 
T 2

(Eq 21)

Rapid Solidification Over the past 25 years, developments in rapid solidification have expanded the understanding of a number of central solidification microstructural characteristics, such as supersaturated solid solutions, cells, and dendrites, associated with growth kinetics and have expanded the product phase selection to include nonequilibrium structures such as metastable phases and metallic glass (Ref 12). Initially, the analysis of rapid solidification was viewed in terms of a high cooling rate; however, the understanding of the synthesis of metastable

280 / Principles of Solidification

Fig. 8

The variation in shape factor f(y) or [DGcr(het)/DGcr(hom)] and spherical cap size, h/r, as a function of the contact angle, y

Fig. 9

Comparison between heterogeneous nucleation (A) and homogeneous nucleation (B) in terms of the relative transformation kinetics below the melting point. The reduced temperature Tr = T/Tf and time t ∝ l
1. Tg, glass transition temperature

phases and metallic glass requires a consideration of nucleation behavior under conditions of high undercooling. Indeed, the development of bulk metallic glasses, where melts of centimeter size can vitrify during slow cooling (Ref 18), demonstrates the importance of melt undercooling in rapid solidification behavior. Under usual conditions, the steady-state nucleation rate given by Eq 17 or 21 is constant. During rapid melt quenching and especially at high undercooling near Tg, atomic transport may not be adequate to maintain cluster equilibrium. Under conditions of sluggish

atomic motion, the initial cluster distribution can be different from the steady-state form, and the time required to reach steady state represents an effective delay time, t (Ref 8, 9). Essentially, the delay time is the period for cluster growth over the size range covered by (DGcr
kT) in Fig. 5 and can be estimated from (bZ2)
1, with b = nSL Scr. During this transient period, the time-dependent nucleation rate I(t) increases continuously and can be described approximately by I(t) = I exp[
t/t] (Ref 19). When transient nucleation is significant, crystallization is delayed to facilitate glass formation. For a number ofmaterials, the delay time and the cooling rate, T , are inversely pro portional, so that T  1 K and may be used to estimate the relative importance of transient effects on nucleation (Ref 20). Current experience indicates that non-steady-state effects are most important for glass formation and during devitrification upon heating.

Inoculation Practice There is a wide commercial application of cast alloy treatments that modify the initial solidification characteristics to provide a means for effective control of grain size and morphology. A number of methods are available to control grain size, including electromagnetic or mechanical stirring, ultrasonic cavitation, and

directed fluid flow. However, inoculation appears to give the most consistent results and is the most economical approach. Inoculation involves the introduction of nucleating agents to a melt, either externally in the form of a fine dispersion or through internal means by phase or chemical reactions that result in the formation of a solid reaction product. Currently, grain-refining agents are usually fabricated in rod form, for example, as an aluminum-base master alloy containing insoluble particles of borides or carbides. The rod is introduced into the molten metal as it flows into a casting unit to facilitate rapid dissolution of the aluminum matrix and dispersion of the particles. Grain refinement effectiveness is often evaluated by the standard TP-1 test procedure (Ref 21). An effective inoculation treatment typically yields an equiaxed grain morphology with minimum grain sizes in the 100 to 150 mm range. In commercial practice, inoculants are used as additions to numerous molten alloys to yield fine grain sizes in castings. In cast iron, ferrosilicon is added to nucleate graphite; zirconium or carbon can be added to magnesium alloys; iron, cobalt, or zirconium is added to copper-base alloys; phosphorus is introduced to aluminumsilicon alloys to refine the silicon phase size; arsenic or tellurium can be added to lead alloys; and titanium is added to zinc-base systems. Other additives, such as sodium in aluminumsilicon alloys, are used to modify the growth morphology. In addition, there is a large body of experience on the grain refinement of aluminum-base alloys by aluminum-titanium, aluminum-titanium-boron, and aluminumtitanium-carbon melt additions. There is widespread experience with inoculation, and several models have been developed for the process, primarily (but not exclusively) in connection with aluminum alloy melts. It is not possible to discuss all of this work in this article; instead, the key features of inoculation are treated in relation to specific features of nucleation during solidification. To illustrate the principal features of inoculation and the physical mechanisms that have been developed to account for the effectiveness of inoculating agents, some of the experience with aluminum alloys is used as a basis for discussion. The basic features of inoculation in other alloy systems follow the pattern that appears to apply to aluminum alloys. The basic requirements for an efficient nucleant can be assessed from consideration of nucleation theory (Ref 22–24). To promote the formation of crystals on a nucleant, the interface between the nucleant and the liquid should be of higher energy than that between the nucleant and the crystal solid. A means of maximizing this condition is to provide a nucleant crystal relationship that is associated with a good crystallographic fit between the respective crystal lattices. In fact, the potency of a given catalyst is believed to be increased for decreasing values of relative lattice disregistry, with the degree of undercooling decreasing with a

Nucleation Kinetics and Grain Refinement / 281 lowering of lattice disregistry. In addition to the basic requirement of lattice compatibility, the melt should tend to wet the surface of the nucleant, and for a uniform refining, the solid nucleant should remain widely dispersed in the liquid. As an ancillary condition, the onset of nucleation should not be followed by a rapid and an isotropic crystal growth, because this will permit the full effect of the potential nucleants to be realized. In alloys, the growth restriction can be achieved as a consequence of solute segregation during freezing, which allows for nucleants throughout the melt to become effective and favors an equiaxed fine grain structure (Ref 24, 25). Aside from these general conditions for an efficient nucleant, it is important to note that for the typical undercooling values of less than 5  C (9  F) observed for grain refinement, the associated contact angle is well below 30 , and thus, the conventional heterogeneous nucleation model represented by Eq 20 cannot be applied for predicting nucleant performance. However, several advances in experimental observations and modeling efforts have yielded a new perspective on grain refinement, and, for the first time, a degree of predictive capability has emerged to provide a valuable basis for the optimization of grain refinement operations. Within the general requirements for effective nucleants, there appear to be two general classes of compounds that are effective in aluminum-base alloys. Similar compound particle types apply to other common casting alloys. The first group includes Al3Ti, Al3Zr, and Al7Cr and can be considered to be associated with a peritectic reaction. The second group comprises compounds such as TiC, TiB2, and AlB2, which are added either intentionally or which result from a chemical reaction in the melt with residual impurities. It is useful to note that the undercooling required to activate solidification is usually less than approximately 5  C (9  F). Despite the rather extensive discussion concerning the mechanism of action and the activity of these proposed nucleants, the strongest evidence for the operation of a compound nucleant is derived from the observed orientation relationships. Table 1 lists disregistry calculations along two directions, a and b, that are normal to each other and are located in the plane in which nucleation is considered to initiate. For the most part, the crystallographic relationships that have been observed involve planes of relatively close packing between the nucleant and aluminum, with disregistries of usually less than approximately 10%. The issue of specific nucleant identity is not insignificant, considering that in commercial melts the rather broad spectrum of background catalysts may often result in solidification at undercoolings of approximately 5  C (9  F). Indeed, only approximately 1 to 2% of all potential nucleant particles in a master alloy addition result in the formation of grains. Furthermore, there is some indication that these proposed nucleants may

not operate singly but rather in association with each other in the melt. Among the nucleant-forming elements listed in Table 1, titanium and, in particular, titanium associated with boron in the master alloy form with a typical composition (Al-5Ti-1B or Al-3Ti-1B) is the most effective grain refinement addition in use for aluminum, but other compositions, such as Al-3Ti-0.15C, are receiving increasing use. In the commercial application of grain-refining agents involving titanium and boron, it is well known that an optimum contact time (that is, the holding time necessary after the addition of refining agents to obtain maximum grain refinement) can vary considerably among different master alloy compositions. As shown in Fig. 10, this feature is generally observed to occur where the optimum contact time interval (period A-B) for minimum grain size is followed by contact time intervals that yield large grain sizes during the interval B-C, which is called a fading process. It is possible to recover from inoculant fading if it is due to nucleant particle settling by stirring the melt, as indicated by the interval C-D.

Grain Refinement Models The operation and underlying mechanism of grain refinement has been under active study for more than 50 years. During this period, numerous models have been proposed to account for the grain initiation and development process. For the most part, the early models served to rationalize existing observations at the time of their introduction but offered little predictive capability. This situation has been changed by the advances in experimental approaches and computational modeling methods so that there is currently a growing predictive capability that can be used for the design of effective grain-refining agents and the optimization of the grain refinement process. The available coverage in this article does not allow for a comprehensive review of all models, but two of the classical models are discussed briefly to provide a perspective for two of the contemporary models. It is generally viewed that Al3Ti crystals are mainly responsible for grain refinement at

alloy compositions above 0.15% Ti, that is, within the peritectic range. For compositions outside the peritectic range, where the Al3Ti phase is not stable under equilibrium conditions, there are basically two interpretations of the grain refinement effect: the carbide-boride view and the peritectic reaction theory. The carbide-boride model suggests that a compound such as AlB2, TiB2, or TiC is responsible for the nucleation of aluminum crystals, because the Al3Ti particles that are introduced into the master alloy are expected to dissolve rapidly (Ref 28, 29). The borides are hexagonal structures, and the carbide is a cubic structure with a relatively close correspondence (Table 1) with the aluminum lattice. Although it is considered that a boride or a carbide is the site of nucleation, the Al3Ti particles do play a role in the overall grain refinement process. The dissolution of Al3Ti is required, for example, to provide a constitutional deterrent (that is, excess titanium) to rapid crystal growth. Within this viewpoint, the establishment of an optimum contact time can be related to the initial dissolution of Al3Ti and to the possible conditioning of the nucleant surface. The subsequent fading reaction is related to the agglomeration and settling of nucleant particles (Ref 29).

Fig. 10

The influence of holding time after the introduction of a grain-refiner master alloy on the degree of grain refinement. The period A-B represents the contact time, the period B-C refers to a fading process, and the period C-D refers to a recovery process.

Table 1 Observed characteristics of potential nucleant compounds Lattice disregistry, % Compound

Crystallographic relationships

Al3Ti

(221)Alk(001)Al3Ti; [010]Alk[113]Al3Ti (001)Alk(001)Al3Ti; [100]Alk[100]Al3Ti (011)Alk(001)Al3Ti; [011]Alk[110]Al3Ti ð111ÞAlk(110)Al3Ti (111)Alk(001)AlB2 None observed (111)Alk(001)TiB2 (001)Alk(001)TiC

Al3(Ti,B)

AlB2 TiB2 TiC

a direction

b direction

0


0.25

5.2

Undercooling (DT)(a) 

Ref



C

F

Ref

25

...

...

25

5.2

5,25

3–5

5–9

25


0.6

5.2

25

0

0

25


5.1
5.9 ...
5.9 6.5


14.3
5.0 ...
5.8 6.5

26 26 26 25 27

... 0.5–1.0 0–0.3

... 0.9–1.8 0–0.5

... 25 26

0

0

28

(a) For peritectic compounds, DT is referenced to Tp; for other compounds, DT is referenced to Tf.

282 / Principles of Solidification Although there is evidence for the stability of boride or carbide particles in aluminum melts and reports of observations of these particles in ingots, other observations have indicated a number of problems in attributing nucleation solely to the action of the boride or carbide particles. For example, a consistent orientation relationship is observed between Al3Ti and aluminum. For TiB2 and aluminum, the reported results differ on detection of a consistent orientation relationship, as noted in Table 1. In addition, boride particles have often been observed to be located at grain boundaries rather than within grain centers, which suggests limited effectiveness (Ref 26). Peritectic Reaction Theory. There is an alternate viewpoint of the nucleation behavior during grain refinement that was originally developed on the basis of a peritectic reaction, but this viewpoint has subsequently evolved into a model with more general scope (Ref 30–35). Although many types of particles are believed to be active in the nucleation of aluminum crystals at undercoolings less than approximately 5  C (9  F), the presence of Al3Ti is believed to offer a more active catalyst requiring little or no undercooling. Thus, Al3Ti crystals, when present, will predominate in the grain refinement nucleation. For peritectic alloys such as the one illustrated by the solid lines in Fig. 4, it is clear that Al3Ti can form upon cooling below the liquidus and can promote the formation of aluminum crystals at the peritectic temperature, Tp. Even for compositions in the peritectic range, a properitectic formation of aluminum has been observed at temperatures above Tp that appears to be associated with a metastable reaction (Ref 36–39). At alloy compositions below the peritectic range, a grain refinement effect is expected to be reduced by the loss of Al3Ti particles, which are added externally in master alloy form. Indeed, independent Al3Ti particles in liquid aluminum that is not saturated with titanium dissolve within several minutes. On the other hand, boride particles, that is, TiB2 or (Al,Ti)B2, are observed to be reasonably stable in melts and offer a fine particulate dispersion. To account for the beneficial effect of incorporating boron into the master alloy compositions and also the observation of the interaction between TiB2 and Al3Ti, it appears likely that TiB2 particles may provide an effective substrate for the nucleation of Al3Ti for compositions spanning the peritectic reaction. During cooling of the melt or holding (that is, the contact time period), the TiB2 particles will react slowly with liquid aluminum to form an (Al,Ti)B2 solution and to liberate excess titanium locally to provide for the formation of a sheath of Al3Ti as a coating on the (Al,Ti)B2 particles (Ref 5). This view is supported by observations on the behavior of boride particles following direct injection into an aluminum alloy melt (Ref 40) and the proposed hypernucleation model that deals with preferential adsorption (Ref 41). However, other work

(Ref 39) indicates that the interactions between titanium, boron, and aluminum are complex and there are thermodynamic issues that do not favor Al3Ti development from the melt (Ref 42). Overall, the applicability of the peritectic reaction model to the hypoperitectic composition range is unresolved and appears to be complex (perhaps involving preferential adsorption) compared to the case for the peritectic composition range. Free-Growth Model. In recognition of the difficulties in applying the conventional heterogeneous nucleation model to the conditions of low undercooling and low contact angle, Greer (Ref 43–49) has adapted a model proposed by Turnbull (Ref 50) to treat the evolution of aluminum crystals during grain refinement. The long-standing challenge of observing the early stage of grain refinement that is obscured in bulk castings was overcome by the use of glass-forming aluminum alloys containing a dispersion of grain-refining particles. With this approach, the initial stage of grain refinement could be captured for detailed analysis, as shown in Fig. 11. The basis for the model is derived from transmission electron microscopy observations that aluminum crystals develop initially on the (0001) planes of TiB2 hexagonal prism particles (Ref 45, 51). Moreover, there is evidence that between the aluminum crystal and TiB2, there is some Al3Ti with an orientation relationship between the individual phases, such that f0001gTiB2 kf112gAl3 Ti kf111gAl . The duplex nature of the particles suggests that TiB2 may not be the actual grain initiation site.

Fig. 12

The initial event leading to the aluminum crystal is not specified completely (i.e., it could be conventional nucleation or adsorption on the particles) (Ref 52), but the key tenet of the model is that for a platelet (modeled as a disc of diameter d) following lateral growth to cover the TiB2, further growth must be outward, as illustrated in Fig. 12. The solid aluminum takes

Fig. 11

Bright-field transmission electron micrograph showing a-aluminum crystallites formed on a TiB2 particle embedded in an amorphous Al88Y8Ni5Co2 (at.%) matrix and annealed. The labeled crystallite sits on a (0001) ledge on the ð1010Þ prismatic face of the hexagonal platelet particle that is coated with a thin layer of Al3Ti. Source: Ref 45

The bold curve indicates the undercooling necessary for grain initiation. The free-growth undercooling is calculated; the nucleation undercooling is schematic only. Inset (i) shows the classical spherical cap model for heterogeneous nucleation. Inset (ii) shows a cap of a-aluminum growing on an inoculant particle through the critical hemispherical condition.

Nucleation Kinetics and Grain Refinement / 283


Tfg ¼ 4sLS =
Sf d

(Eq 22)

As the undercooling continues to increase, free growth can commence on particles of successively smaller size. This model represents a deterministic description of aluminum grain development that is fundamentally different from a stochastic nucleation description. In the implementation of this model, the cooling path is considered as a sequence of isothermal steps. For each step, the number of grain initiation events is determined as well as the latent heat release from growth of existing grains. The balance between the latent heat release and the external heat extraction yields the cooling rate and the temperature for the next calculation step. Once recalescence begins due to the release of the latent heat, the number of grains per unit volume is established, since there are no further grain initiation events. This approach is similar to the original model of Maxwell and Hellawell (Ref 25), except that they assumed conventional heterogeneous nucleation and a uniform nucleant particle size. At the low undercooling values that are typical for grain refinement, the growth of an aluminum crystal of radius r can be approximated by a spherical growth model to yield a growth rate v: v ¼ dr=dt ¼ ðDL
Ts Þ=ðrQÞ

(Eq 23)

where DTs is the solutal undercooling, and Q is a growth restriction parameter defined by: Q ¼ mL ðk
1ÞCo

(Eq 24)

where Co is the alloy solute content. Sometimes, the growth restriction is related to the constitutional undercooling parameter, P = Q/k. A listing of the relevant parameter values for the main solutes of interest in aluminum grain refinement is given in Table 2. From the values given in Table 2, it is evident that titanium is the most effective solute in terms of restricting growth of aluminum grains. In the free-growth model, the measured nucleant particle size distribution is incorporated in the numerical analysis. For TiB2 particles in an Al-Ti-B grain refiner, a log-normal distribution was shown to be appropriate (Ref 45). In a standard grain-refining test based on the TP-1 procedure, the predictions of the free-growth model for grain size as a function of grain-refiner level, alloy solute content, and cooling rate are in good agreement with the observations. For example, in Fig. 13, the

Table 2

Relevant parameters for the main solutes used in aluminum grain refinement

Solute element

Ti Ta V Hf Mo Zr Nb Cr Si Ni Mg Fe Cu Mn

k

7–8 2.5 4.0 2.4 2.5 2.5 1.5 2.0 0.12 0.007 0.49 0.02 0.17 0.62

mL, K/wt%

mL(k
1) = Q/Co

mL(k
1)/k = P/Co

Maximun concentration, wt%

Reaction type

26–33 70 10.0 8.0 5.0 4.5 13.3 3.5
6.8
3.3
5.7
3.0
3.4
1.2

156–236 105 30 11.2 7.5 6.8 6.6 3.5 5.9 3.3 2.9 2.9 2.8 0.46

22–29.5 42 7.5 4.7 3.0 2.7 4.4 1.8 49.8 4714 5.9 145 16.5 0.74

0.15 0.10 0.1 0.5 0.1 0.11 0.15 0.4 12.6 6 37 1.8 33.2 2

Peritectic Peritectic Peritectic Peritectic Peritectic Peritectic Peritectic Peritectic Eutectic Eutectic Eutectic Eutectic Eutectic Eutectic

1013 Number of grains, m–3

the form of a growing spherical cap. At a given undercooling, growth ceases when the radius of curvature reaches rcr for the undercooling, as given by Eq 12. As the undercooling increases, rcr decreases and the crystal can grow accordingly. When rcr has decreased to equal d/2, the solid forms a hemisphere, and the solid can now grow freely to develop into a grain. The critical undercooling, DTfg, for the onset of free growth is given by:

Predicted 1012

Maxwell & Hellawell

1011

(A)

(B)

1010 1010

Measured 1014 1012 1013 1011 Number of particles, m–3

1015

Fig. 13

The number of grains per unit volume as a function of the number of refiner particles per unit volume, showing a general trend to lower efficiency at higher addition level. Data from grain diameters measured in TP-1 tests (closed circles) are compared with predictions of the free-growth model (open circles). Source: Ref 44. The predictions are qualitatively different from those of Maxwell and Hellawell (Ref 25) and are a much better fit to the data.

Grain diameter (mean linear intercept) for commercially pure aluminum inoculated with A1-5Ti-lB at various levels. The grain diameters measured in TP-1 tests (diamonds) are compared with free-growth model predictions based on an isothermal melt (circles) or on directional solidification (squares). Source: Ref 45

number of aluminum grains as a function of the number of nucleant particles is shown and demonstrates an agreement between the freegrowth model analysis and the measurements. Similarly, in Fig. 14, the observed grain size dependence on refiner addition level shows a similar trend with the free-growth model prediction, except at small levels. The free-growth model has also been applied to the grain refinement of magnesium-base alloys by SiC particles and again has revealed the capability to predict the resultant grain size (Ref 53). The agreement between the free-growth model and grain refinement measurements is a strong support for the model, since the input parameters are determined independently and are not adjustable. The apparent success of the free-growth model that is essentially an athermal treatment of grain initiation rather that the usual stochastic nucleation process provides a valuable means to treat grain development at low undercooling, where conventional nucleation does not apply. In fact, the behavior treated by the free-growth model is consistent with some of the modeling approaches used in simulating

the development of solidification microstructure (Ref 54–56). At the same time, the freegrowth model requires further development to treat large casting volumes, where significant liquid flow can contribute to grain initiation from dendrite fragments, and particle agglomeration and settling are important. Similarly, solute effects such as those involved in poisoning are treated only in terms of their effect on Q. However, in multicomponent alloys, strong chemical interactions can influence particle surfaces as well as growth restriction. Constitutional Undercooling Model. Along with the development of the free-growth model analysis of grain refinement, an alternate viewpoint has been proposed by Easton and St. John that incorporates the chemical effects of solute content on restricting grain growth and the structural characteristics of grain refinement as reflected by nucleant potency (Ref 57– 65). In this model, the analysis of constitutional undercooling, DTc, plays a major role in the evolution of the as-cast grain size. It is assumed that following the initial thermal undercooling that is generated by the mold chill, constitutional undercooling is the dominant form of

Fig. 14

284 / Principles of Solidification undercooling. As an additional assumption, it is considered that nucleation initiates when DTc = DTn, the nucleation undercooling, but again, the grain initiation mechanism on grain-refining particles is not specified. At this point, the value of the fraction solid, fs, corresponds to the final grain size, since further growth is arrested. Due to the shallow thermal gradients in the melt, a progression of nucleation events develops from the mold wall toward the casting center. Separate nucleation events in this progression initiate as the previously nucleated grains develop sufficient DTc to reach DTn, which is a measure of nucleant potency. To satisfy this condition, a small amount of growth is necessary and is evaluated by expressing the constitutional undercooling as:

Tc ¼ ml c0 1


1 ½1
fs p

 (Eq 25)

where p = 1
k. The initial rate of development of constitutional undercooling is then related to the growth restriction factor as: d
Tc ¼ ml c0 ðk
1Þ ¼ Q dfs

(Eq 26)

By rearranging Eq 25, an expression for the relative grain size (RGS) is obtained in terms of solute content and particle potency as:
RGS ¼ fs;n ¼ 1


ml c0 ml c0

Tn

1=p (Eq 27)

where fs,n is the amount of growth required to develop sufficient constitutional undercooling to reach DTn and initiate grain nucleation. The application of Eq 27 to reveal the relationship between grain size and the combined effects of nucleant potency and solute content is illustrated in Fig. 15. It is evident that nucleant potency has a strong effect on grain size for all solute levels, as expected, while solute content is a significant factor at low levels and has little effect at high levels. As a further development in this analysis, the constitutional undercooling model for grain refinement was related to the semiempirical expression: D ¼ a þ b=Q

(Eq 28)

where D is the grain size, and a and b are constants. Since Q is related to the rate of development of constitutional undercooling by Eq 26, the grain size was taken as proportional to (DTn/Q), so that b = (b0 DTnm/Q), where m may be between 1 and 2 to reflect either plane front (m = 1) or dendritic (m = 2) growth. Moreover, it is apparent that D is dependent on the separation between active particles. By assuming that the grain density is proportional to the nucleant particle density, the parameter a was evaluated as [rf ]
1/3, where r is the nucleant particle density, and f is the fraction of particles

Fig. 15

Relative grain size versus solute content for a range of nucleant particle potencies, DTn as evaluated from Eq 27. The solute used is titanium added to pure aluminum.

that are active. From these considerations, Eq 28 becomes: a0 b0
Tn ffiffiffiffiffiffi þ D¼p 3 Q rf

(Eq 29)

By replotting results such as those in Fig. 15 in the form indicated by Eq 28 or 29, some useful insight is obtained on the separate factors controlling grain size. For example, in Fig. 16(a), the slope b (or b0 ) is related to the nucleant potency. A steeper slope corresponds to a lower potency. The intercept a (or a0 ) represents the number of nucleant particles that are active in grain initiation at infinite Q values, as shown in Fig. 16(b). One example is given in Fig. 17, where the measured grain size for several aluminum alloys is plotted against (1/Q) for different TiB2 addition levels. The zero level represents the background nucleant effect. With all TiB2 levels shown, the potency has increased to a constant amount, with further grain refinement related to the increased number of particles. The relationships shown in Fig. 16 have been applied to a wide variety of aluminum and magnesium alloys and have also been used to interpret other effects, such as the influence of melt superheating and poisoning solutes on grain size. As with the free-growth model, the constitutional undercooling model requires further development to account for fluid flow effects. Both the free-growth model and the constitutional undercooling model require input information on the particle sizes and number density, but the constitutional undercooling model also requires grain refinement results in order to evaluate the parameters in Eq 28 and 29. Once the model is calibrated, it can provide useful design guidance.

Fig. 16

(a) Effect of changing the nucleant particle potency on the gradient b with a constant number of particles. Particles that produce a slope b00 are the most potent particles. (b) Effect of adding more particles at constant potency on the value of the intercept a

Nucleation Kinetics and Grain Refinement / 285 1200

1000

Grain size, µm

800

0TiB2 0.005TiB2

600

0.01TiB2 0.02TiB2

400

0.05TiB2 200

0 0

0.1

0.2

0.3

0.4

1/Q, 1/K

Fig. 17

Measured grain size versus the reciprocal of the growth restriction factor, Q, for 0, 0.005, 0.01, 0.02, and 0.05% TiB2 additions in several aluminum alloys

Fig. 18

Micrograph of an Al-6Ti master alloy illustrating the range of size and shape of Al3Ti particles

Fig. 19

The Al3Ti structure and the arrangement of the (011) planes. Three successive (011) planes identified by dashed lines contain only aluminum atoms and offer a low disregistry to the atom arrangement on (012) planes of aluminum.

Master Alloy Processing Another important aspect of the grain refinement reaction that has not received much consideration relates to the influence of preparation conditions during master alloy processing and structural differences on the effectiveness of nucleant particles (Ref 5, 33). This feature has taken on new significance with the observation that Al3Ti particles may appear with different morphologies (Fig. 18) and may exhibit several possible crystallographic orientation relationships with aluminum crystals. The degree of refinement is observed to be dependent on the Al3Ti crystal morphology. Essentially, three types of Al3Ti crystals are observed to be present: a flake type exposing (001) planes to the melt that forms at high preparation temperatures (>760  C, or 1400  F), a blocky form that tends to form at low temperatures, and a petallike morphology. There is

some indication that each particle morphology can differ in nucleating potency. This has been examined in droplet studies where individual particles can be isolated into separate droplets (Ref 37, 38). The blocklike crystals expose (011) planes to the melt, as indicated in Fig. 19, for a [(011) Al3Tik (012) Al] relationship with aluminum. For flake-type crystals, the crystallographic axes of Al3Ti and aluminum are closely parallel, that is, (001)Al3Tik(001)Al. Although the latter relationship also gives a relatively small disregistry, the (011) Al3Ti surface as indicated in Fig. 19 is occupied by three successive layers of aluminum atoms, which can be viewed as somewhat distorted (012) aluminum lattice planes. The formation of aluminum crystals on such a surface merely requires a growth (that is, epitaxy) of this existing distorted lattice at little or no undercooling below the peritectic temperature Tp. The (001) Al3Ti surface, on

the other hand, contains titanium atoms so that a simple growth of the existing surface is not sufficient to generate an aluminum crystal, and a lower nucleating potency may be expected and is observed. In addition, the observation of different morphologies and twinning for Al3Ti crystals may account for the variety of orientation relationships that have been observed between Al3Ti and aluminum. Other particles, such as TiB2, TiC, TiN, ZrB2, and TaB2, have been examined, but the undercooling required to initiate solidification was minimized when Al3Ti particles were present. Usually, the grain-refining particles are added externally to the melt, but there is recent work that has demonstrated an in situ nucleant process that can operate on a commercial scale (Ref 66). From grain-refining tests with in situ nucleant particles, it was observed that the ascast grain size did not depend on the nucleant particle size, which is not consistent with the predictions of the free-growth model. Since the particle surface in contact with the melt is the critical point in grain initiation, it is possible that ex situ particles have size-dependent surface characteristics that are absent for in situ particles. Thermal Analysis Techniques. Grain-refining characteristics can also be conveniently assessed through thermal analysis techniques. In examining cooling curves after master alloy dilution, three main groups of characteristic curves are often observed, as shown in Fig. 20. The characteristic curve A in Fig. 20 is detected for a high degree of grain refinement and is related to a large number of aluminum crystals that form on surviving Al3Ti particles

286 / Principles of Solidification

Fig. 20

Representative types of cooling curves for an aluminum melt treated with different master alloys to yield varying levels of grain refinement. A, high degree of grain refinement coincides with Al3Ti crystals present in melt; B, poor grain refinement due to few Al3Ti crystals in melt; C, intermediate case in which there are a sufficient number of Al3Ti crystals but an insufficient local titanium concentration to promote grain refinement. The characteristic peritectic hump for all three curves lies between the growth temperature, TG, and the equilibrium liquidus temperature, TL. Source: Ref 31

at Tp. With continued cooling upon reaching the growth temperature, TG, the aluminum crystals can grow, consuming the liquid and raising the temperature to the equilibrium liquidus temperature, TL. This reaction creates a peritectic hump, which is an indication that Al3Ti crystals are present in the melt. The type A cooling curve in Fig. 20 was detected with blocky Al3Ti crystals following a short holding time but also with flake-type Al3Ti after a long contact time of 1 h or more. With a cooling curve of the type B configuration in Fig. 20, relatively few Al3Ti crystals remain; therefore, poor grain refinement occurs. A cooling curve of the type C configuration in Fig. 20 represents an intermediate case in which there are a sufficient number of Al3Ti crystals, but the local titanium concentration is not sufficient to allow for growth at high temperatures, a condition that yields poor grain refinement. Although the results of recent work represent a significant improvement in the description of grain refinement as well as in the improved characterization and predictive modeling of grain refinement agents and their effectiveness, the overall understanding of the grain refinement of aluminum or other metallic melts by compound particles remains incomplete. The success of the contemporary models reveals that while nucleant particle size is an important factor in grain refinement effectiveness, constitutional factors are also significant. Other chemical effects such as adsorption remain to be considered as well as the influence of fluid flow. The issue of the actual grain initiation mechanism is also open, and the resolution is likely to be related to the details of the nucleant

surface structure. However, the new models seem to provide the necessary foundation for further significant progress in the development of a real understanding of grain refinement.

REFERENCES 1. J.C. Baker and J.W. Cahn, in Solidification, American Society for Metals, 1971, p 23 2. J.H. Perepezko and W.J. Boettinger, Proc. Mater. Res. Soc., Vol 19, 1983, p 223 3. D.R. Gaskell, Introduction to the Thermodynamics of Material, 4th Edition, Taylon and Francis, 2003 4. J. Cisse, H.W. Kerr, and G.F. Bolling, Metall. Trans., Vol 5, 1974, p 633 5. L. Arnberg, L. Backerud, and H. Klang, Met. Technol., Vol 9, 1982, p 1, 7, 14 6. M.C. Flemings, Solidification Processing, McGraw-Hill, 1974 7. W. Kurz and D.J. Fisher, Fundamentals of Solidification, 3rd ed., Trans Tech, 1992 8. H.I. Aaronson and J.K. Lee, in Lectures on the Theory of Phase Transformations, 2nd ed., H.I. Aaronson, Ed., The Minerals, Metals and Materials Society, 1999, p 165 9. D.T. Wu, Solid State Phys., Vol 50, 1997, p 37 10. J.H. Hollomon and D. Turnbull, Prog. Met. Phys., Vol 4, 1953, p 333 11. J.H. Perepezko, Mater. Sci. Eng. A, Vol 65, 1984, p 125; Vol 178, 1994, p 105 12. W.J. Boettinger and J.H. Perepezko, in Rapidly Solidified Alloys: Processes, Structures, Properties, Applications, H.H. Liebermann, Ed., Marcel Dekker, 1993, p 17

13. D. Turnbull, J. Appl. Phys., Vol 21, 1950, p 1022 14. F. Spaepen, Solid State Phys., Vol 47, 1994, p 1 15. J.J. Hoyt and M. Asta, Phys. Rev. B, Vol 65, 2002, p 214106 16. D.Y. Sun, M. Asta, J.J. Hoyt, M.I. Mendelev, and D.J. Srolovitz, Phys. Rev. B, Vol 69, 2004, p 020102 17. R.L. Davidchack, J.R. Morris, and B.B. Laird, J. Chem. Phys., Vol 125, 2006, p 094710 18. A. Inoue, Acta Mater., Vol 48, 2000, p 279 19. K.F. Kelton, A.L. Greer, and C.V. Thompson, J. Chem. Phys., Vol 79, 1983, p 6261 20. K.F. Kelton and A.L. Greer, J. Non-Cryst. Solids, Vol 79, 1986, p 295 21. “Standard Test Procedure for Aluminum Alloy Grain Refiners: TP-1,” The Aluminum Association, 1987 22. D. Turnbull and B. Vonnegut, Ind. Eng. Chem., Vol 44, 1952, p 1292 23. P.B. Crosley, A.W. Douglas, and L.F. Mondolfo, The Solidification of Metals, The Iron and Steel Institute, 1968, p 10 24. A. Hellawell, Solidification and Casting of Metals, The Metals Society, 1979, p 161 25. I. Maxwell and A. Hellawell, Acta Metall., Vol 23, 1975, p 895 26. I.G. Davies, J.M. Dennis, and A. Hellawell, Metall. Trans., Vol 1, 1970, p 275 27. J. Cisse, G.F. Bolling, and H.W. Kerr, J. Cryst. Growth, Vol B-14, 1972, p 777 28. A. Cibula, J. Inst. Met., Vol 76, 1949– 1950, p 321; Vol 80, 1951–1952, p 11 29. G.P. Jones and J. Pearson, Metall. Trans. B, Vol 7, 1976, p 223 30. F.A. Crossley and L.F. Mondolfo, Trans. AIME, Vol 141, 1951, p 1143 31. J.H. Perepezko and S.E. LeBeau, in Aluminum Transformation—Technology and Applications, American Society for Metals, 1982, p 309 32. J.A. Marcantonio and L.F. Mondolfo, J. Inst. Met., Vol 98, 1970, p 23 33. M.M. Guzowski, G.K. Sigworth, and D.A. Senter, Metall. Trans. A, Vol 18, 1987, p 603 34. L.F. Mondolfo, in Grain Refinement in Castings and Welds, G.J. Abbaschian and S.A. David, Ed., The Metallurgical Society, 1983, p 3 35. D.A. Granger, “Practical Aspects of Grain Refining Aluminum Alloy Melts,” Laboratory Report 11-1985-01, Aluminum Company of America, 1985 36. E. Nes and H. Billdal, Acta Metall., Vol 25, 1977, p 1031 37. M.K. Hoffmeyer and J.H. Perepezko, Scr. Metall., Vol 23, 1989, p 315 38. M.K. Hoffmeyer and J.H. Perepezko, Light Metals 1989, P.G. Campbell, Ed., TMS, 1989, p 913 39. M. Johnsson, L. Backerud, and G.K. Sigworth, Metall. Trans. A, Vol 24, 1993, p 481 40. P.S. Mohanty and J.E. Gruzleski, Acta Metall. Mater., Vol 43, 1995, p 2001

Nucleation Kinetics and Grain Refinement / 287 41. G.P. Jones, in Solidification Processing 1987, Institute of Metals, London, 1988, p 496 42. G.K. Sigworth, Scr. Mater., Vol 34, 1996, p 919 43. A.M. Bunn, P.V. Evans, D.J. Bristow, and A.L. Greer, in Light Metals 1998, B. Welch, Ed., The Minerals, Metals & Materials Society, 1998, p 963 44. A.L. Greer, A.M. Bunn, A. Tronche, P.V. Evans, and D.J. Bristow, Acta Mater., Vol 48, 2000, p 2823 45. A.L. Greer, Philos. Trans. R. Soc. (London) A, Vol 361, 2003, p 479 46. T.E. Quested and A.L. Greer, Acta Mater., Vol 52, 2004, p 3859 47. T.E. Quested and A.L. Greer, Acta Mater., Vol 53, 2005, p 2683, 4643 48. A.L. Greer and T.E. Quested, Philos. Mag., Vol 86, 2006, p 3665 49. S.A. Reavley and A.L. Greer, Philos. Mag., Vol 88, 2008, p 561

50. D. Turnbull, Acta Metall., Vol 1, 1953, p 8 51. P. Schumacher, A.L. Greer, J. Worth, P.V. Evans, M.A. Kearns, P. Fisher, and A.H. Green, Mater. Sci. Technol., Vol 14, 1998, p 394 52. W.T. Kim and B. Cantor, Acta Metall. Mater., Vol 42, 1994, p 3115 53. R. Gu¨nther, Ch. Hartig, and R. Bormann, Acta Mater., Vol 54, 2006, p 5591 54. W. Oldfield, Trans. ASM, Vol 59, 1966, p 945 55. J.D. Hunt, Mater. Sci. Eng., Vol 65, 1984, p 75 56. M. Rappaz and Ch.-A. Gandin, Acta Metall. Mater., Vol 41, 1993, p 345 57. M. Easton and D. St. John, Metall. Mater. Trans. A, Vol 30, 1999, p 1613 58. Y.C. Lee, A.K. Dahle, and D.H. St. John, Metall. Mater. Trans. A, Vol 31, 2000, p 2895 59. M.A. Easton and D.H. St. John, Acta Mater., Vol 49, 2001, p 1867

60. M.A. Easton and D.H. St. John, in Light Metals 2001, J.I. Anjier, Ed., The Minerals, Metals & Materials Society, 2001, p 927 61. M. Qian, D.H. St. John, and M.T. Frost, Scr. Mater., Vol 50, 2004, p 1115 62. D.H. St. John, M. Qian, M.A. Easton, P. Cao, and Z. Hildebrand, Metall. Mater. Trans. A, Vol 36, 2005, p 1669 63. M. Easton and D.H. St. John, Metall. Mater. Trans. A, Vol 36, 2005, p 1911 64. L. Lu, A.K. Dahle, and D.H. St. John, Scr. Mater., Vol 54, 2006, p 2197 65. P. Cao, M. Qian, and D.H. St. John, Scr. Mater., Vol 56, 2007, p 633 66. J.A. Megy, D.A. Granger, G.K. Sigworth, and C.R. Durst, in Light Metals 2001, J.I. Anjier, Ed., The Minerals, Metals & Materials Society, 2001, p 943

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 288-292 DOI: 10.1361/asmhba0005208

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Transport Phenomena During Solidification Jonathan A. Dantzig, University of Illinois at Urbana-Champaign

Fundamentals Some of the most important tools that materials scientists use regularly are equilibrium phase diagrams. These diagrams, derived from thermodynamics, define the phases present and their relative amounts in equilibrium as a function of overall alloy composition, C, and temperature, T. However, there is no information in the diagram about how a system evolves in space, x, and time, t. Transport phenomena is the generic name used to describe the dynamical and spatial aspects of the system behavior. There are more specific names. The study of energy transport, embodied in the temperature field T(x,t), is called heat transfer. The transport of solute, given by C(x,t), is called mass transfer. In fluid systems, it is also necessary to consider the velocity field v(x,t), the study of which is called momentum transport or, more commonly, fluid dynamics. Clearly, these subjects are much deeper and broader than the scope of a Handbook article. Therefore, the focus here is on an introduction to the topic, all given in the context of solidification. One begins with the balance equations for mass, energy, and solute and the necessary boundary conditions for solving problems of interest in casting and solidification. The phenomena cover a vast range of length and time scales—from atomic dimensions up to macroscopic casting size, and from nanoseconds for interface attachment kinetics to hours for casting solidification. It is essential to understand how to determine which phenomena are most important at the particular length and time scale of interest for the problem at hand. Methods for doing so are described in the section “Scaling” in this article. Finally, several examples are given of the application of transport phenomena in solidification, focusing in particular on microstructure formation.

Balance Equations The derivation of the general forms of the balance equations is given in Ref 1 and 2. Instead, a shorter form is presented here that begins with

the general balance equations and then specializes them using common models for materials behavior appropriate for metal systems. Since the subject of interest is solidification, it is necessary to examine more closely how these equations are applied when there is a moving solidification front. This should become more clear as each balance equation is developed. Mass Balance. The mass balance for a fluid of density r moving at velocity v is given by:



@r þ r ðrvÞ ¼ 0 @t

(Eq 1)

This is written in the so-called conservative form, which is quite useful if one decides to integrate over a control volume. Doing so, and choosing a fixed control volume V with surface S, yields: ð ð @ rdV þ rv ndS ¼ 0 @t V



S

(Eq 2)

where the divergence theorem has been applied to convert the second term to a surface integral. The surface normal vector is n. The interpretation of Eq 2 is clear: The time rate of change of total mass in the volume, represented by the first term, is balanced by the total mass leaving the volume through the surface. Now consider the control volume shown Fig. 1, which encloses a portion of the solid-liquid interface. The interface, solid, and liquid are moving at velocities v*, vs, and vL, respectively, and the interface normal vector is n*. (The subscripts “s” and “L” are used throughout to refer to the solid and liquid, respectively, and the superscript “*” indicates quantities on the liquid-solid interface.) Now evaluate Eq 2 as the thickness of the control volume goes to zero. The first term vanishes, and the second term then consists of four parts:





rs ðvs n  v n Þ



where b is called the solidification shrinkage. Equation 4 states that there is a net flow of liquid into the interface in order to balance the density difference between the liquid and solid phases. This flow is the underlying cause of porosity and hot tearing in castings. It is less important for the other balances. Therefore, the shrinkage flow is omitted from the development of the momentum, energy, and solute balances that follow. Momentum Balance. The balance of linear momentum is the continuum form of Newton’s second law of motion, that is, that the sum of the forces on a body is balanced by the time rate of change of linear momentum. In conservative form:



where s is the stress tensor, and b is a body force such as gravity. The dyadic product vv is a tensor with components vivj. Integrating over a control volume as for the mass balance shows more clearly the relation to Newton’s second law: ð ð @ rvdV þ rvv ndS @t v S ð ð ¼ f ndS þ rbdV



S







(Eq 6)

v

where f = s n is the external traction force applied on the surface of the control volume. It is often convenient to separate the stress into vL n∗ v∗

(Eq 3) vs

The physical meaning of Eq 3 is made more clear by assuming that the solid is not moving (vs = 0) and by rearranging terms to obtain:



(Eq 5)

Liquid



þ rL ðv n  vL n Þ ¼ 0

vL



@rv þ r ðrvvÞ ¼ r s þ rb @t



r  rL   n ¼  s v n ¼ bv n rL

(Eq 4)

Solid Interface

Fig. 1

Schematic of a control volume including the liquid-solid interface

Transport Phenomena During Solidification / 289 hydrostatic pressure, p, and the “extra” stress tensor, t, leading to the form s = pI + t. Following a procedure similar to the one followed to develop the interface form of the mass balance leads to balance equations for the normal and tangential components of the force at the interface: ðpL  ps Þ  ðts  tL Þ : ðn n Þ ¼ 2g k

(Eq 7)



ðts  tL Þ : ðn t Þ ¼ rsurf g t ¼ 0

(Eq 8)

where g is the surface energy of the liquid-solid interface, and k is the mean curvature. The notation “:” implies a double-dot product, and rsurf indicates the gradient within the surface. This term captures the so-called Marangoni effect, where gradients in temperature or composition along the surface create tangential forces that can drive a flow in the liquid. These forces are very important for welding, where there is a free surface, but less so for liquids and solids. Equation 7 provides the well-known relation that the surface tension is balanced by a pressure difference across the curved interface. Energy Balance. The energy balance is one of the most important balances for solidification problems. The most convenient thermodynamic variables are temperature, T, and pressure, p. The energy balance is written in terms of the enthalpy density, H, as:

 

@rH þ r ðrHvÞ @t ¼ r ðkrT Þ  pr v þ t : D þ Q_



(Eq 9)

where D is the rate of deformation tensor, 2D = rv + (rv)T, and t:D is the work done by the shear stress, also called the viscous dissipation. For most solidification problems, the pressure effects, viscous dissipation, and internal heat generation, Q_ , usually can be neglected, leaving the form:





@rH þ r ðrHvÞ ¼ r ðKrT Þ @t



(Eq 13)

Often, the ratio Ks/KL  1. Solute Balance. The solute balance is developed in a similar way to the energy balance. For the sake of brevity, the treatment here is restricted to binary alloys (A and B) with no chemical reactions. In analogy with Fourier’s law for conduction, Fick’s law for diffusion is adopted, which gives the diffusive flux of species B as proportional to the composition gradient, J = Dr(rC). The composition is given as a mass fraction, that is, grams B divided by total grams, the units of the diffusivity D are m2/s, and thus, the flux has units grams B per unit area per unit time. Note that there are numerous subtleties associated with whether the mass appears inside or outside the gradient operator when the density is not constant, but these issues are beyond the scope of this article. The solute balance is given by:





@rC þ r ðrCvÞ ¼ r ðDrrCÞ @t

(Eq 14)

and the interfacial balance for solute is also analogous to the interfacial balance for energy:





ðrL CL  rs Cs Þv n ¼ Ds rðrs Cs Þ n



DL rðrL CL Þ n (Eq 15)

As opposed to heat conduction, where Ks  KL, usually Ds  DL, and the first term on the right side of Eq 15 is often neglected. The left side is usually modeled by assuming local thermodynamic equilibrium at the interface, and therefore the compositions CS and CL are given by the equilibrium phase diagram. In that case, the segregation coefficient, k, is introduced, such that rsCs = krLCL. Introducing both of these simplifications in Eq 15 gives:





rL CL ð1  kÞv n ¼ DL rðrL CL Þ n

(Eq 16)

Scaling

ð ð ð @ rhdV þ rhv ndS ¼ KrT ndS @t V S S ð ¼  q ndS







(Eq 11)

where Fourier’s law for the conductive heat flux, q = KrT, has been applied. The interfacial balance, neglecting the shrinkage flow, is:





rs H ¼ Ks rTs n  KL rTL n

(Eq 10)

Integrating over the control volume shows that the time rate of change of enthalpy in the control volume is balanced by that advected and conducted through the surface:

S

Inserting DH gives the Stefan condition:



rL HL  rs Hs ¼ Ks rTs n  KL rTL n

(Eq 12)

The left side of Eq 12 is the difference in enthalpy between the liquid and solid phases, the latent heat of fusion, DH; the right side is the net heat conduction away from the interface.

The governing equations, while necessary, are only the beginning. The geometry, boundary and initial conditions, and material properties also must be specified. When all of these data are gathered, one can, in principle, evolve the appropriate equations in time and find a solution to the problem at hand. This is usually done numerically, perhaps using one of the many available commercial codes. However, that is usually much more work than is either necessary or even possible, because the range of length and time scales involved with solidification processes is vast—from atomic-scale kinetic processes that take place over picoseconds to macroscopic heat flow at the size of the casting, extending over as much as an hour for large parts. Practical simulations can cover at most 3 to 4 orders of magnitude in both space and time, and this means that some phenomena must be omitted from the problem. But which ones should be kept and which ones

neglected (or are better to include via an analytical model)? Clearly, the important scales should be included, but how can one decide which those are? That is the subject of this section. The decision about which length and time scales are important is done by scaling the governing equations. This provides a systematic means for ensuring that the important terms are kept while the negligible terms are discarded. First, the procedure is described, then a simple example is given. Scaling begins by choosing characteristic scales for variables such as length, temperature, and so on that usually come from the problem geometry and boundary conditions. Some characteristic values are unknown and are defined by the governing equations. The characteristic values are used to define dimensionless variables that are order one. For example, suppose that the temperature, T, is known (e.g., from initial and boundary conditions) to lie in the interval T1  T  T0. A dimensionless temperature is defined, y = (T  T1)/(T0  T1) E [0,1]. Unknown characteristics are just given a symbol and carried along. When the dimensionless variables are defined, they are substituted into the governing equations. Each term in the equation then has a (dimensional) coefficient multiplying a scaled (dimensionless) term. The equation is then divided by the coefficient of a term that is expected to be important. This requires some engineering judgment, but it is expected that something about the problem at hand is known! The unknown characteristics are then evaluated by setting the value of the coefficient where they appear to 1. Now that all of the terms in the equation are order 1 or less, terms whose coefficients are small compared to 1 can be safely neglected and the remaining problem solved. A concrete example makes the scaling process seem much less mysterious. Consider a thin, circular rod of length L and radius R  L, initially at temperature T0, quenched into a large bath at temperature T1. Some simplifying assumptions are made in order to focus on the scaling process. Assume that the bath temperature stays at T1, unaffected by the quench, and that the heat transfer from the surface of the rod to the bath can be characterized by a constant heat transfer coefficient, h. All of the material properties will be assumed to be constant, and the temperature distribution is axisymmetric. The problem is solved in polar coordinates. Finally, the rod is solid (v = 0), and there is no internal heat generation (Q_ ¼ 0). The energy equation for the rod (0  r  R;  L/2  z  L/2) is then:     @T 1@ @T @2T ¼a r þ 2 @t r @r @r @z

(Eq 17)

and the initial and boundary conditions are: T ¼ T0 K

@T ¼ hðT  T0 Þ @r

t¼0

(Eq 18)

r¼R

(Eq 19)

290 / Principles of Solidification @T þK ¼ hðT  T0 Þ  @z

z ¼ L=2

(Eq 20)

where a = K/rcp is the thermal diffusivity. Now the problem specification is used to define the following dimensionless variables: r~ ¼

r ; R

z¼

z þ L=2 ; L

y¼

T  T1 T 0  T1

(Eq 21)

Time must also be scaled, but the characteristic value tc is not yet known. Therefore, t~ ¼ t=tc is defined, and guidance is awaited. The scaled variables are now inserted into Eq 17, and the entire equation is divided by a(T0  T1)/R2, the coefficient of the term associated with conduction in the radial direction. The result is:     R2 @y 1@ @y R2 @ 2 y ¼ r~ þ 2 2 atc @ t~ L @ r~ @~ r @~ r

(Eq 22)

Setting the dimensionless group on the left side to 1 gives the characteristic time tc = R2/a, and the ratio at/R2 is called the Fourier number. It can be seen that the geometry of the problem, where R  L, indicates that axial conduction, associated with the term @ 2y/@z2, is negligible in comparison to radial conduction. This makes the problem effectively one-dimensional, a great simplification. The initial condition is simply y = 1 at t~ ¼ 0. The scaled form of the boundary conditions becomes: @y hR ¼ y @~ r K

r~ ¼ 1

(Eq 23)

@y hR L y þ ¼  @ K R

z ¼ 0;1

(Eq 24)



The dimensionless parameter hR/K is called the Biot number, Bi. Notice that information about the problem geometry is retained in the aspect ratio L/R, which now appears in the boundary condition. To review concepts of the scaling: The characteristic time for conduction is R2/a, conduction in the axial conduction can be neglected with respect to radial conduction for R/L, and the important parameter that determines the rate of heat loss to the environment is the Biot number, hR/K. The remaining one-dimensional problem can be solved, for example, using Fourier series

methods (Ref 3). The simplification to a onedimensional problem came at a price, however. Since the derivatives with respect to z were eliminated from the differential equation, the end conditions in Eq 24 cannot be satisfied. It is expected that there will be a small region near the boundary, called a boundary layer, where the solution is two-dimensional. In this region, the axial and radial conduction terms must be comparable, and scaling should make it obvious that this layer extends a few R from each end. Figure 2 shows temperature solutions for a problem where Bi = 10 and L/R = 20, at several values of t~. Notice that the temperature is nearly uniform at T1 at t~ = 0.7—this shows the significance of the characteristic time.

Transport and Microstructure Using the fundamentals presented in the preceding sections, solidification problems now can be discussed. Perhaps the most important thing to understand is that for most materials, Ds  DL  aL  as, and these large differences produce a separation of characteristic time and length scales. Because the thermal diffusivity is much larger than the chemical diffusivities, the solidification rate is controlled by heat flow, and chemical diffusion tries to follow. This eventually leads to the formation of microstructure, such as that shown in Fig. 3 (Ref 4). Several aspects of the solidification process are now considered as well as how transport phenomena are expressed in the development of microstructure.

Planar Front Growth As discussed in other articles in this Handbook, the two processing variables that are used to correlate microstructures are the temperature gradient in the liquid at the interface, GL, and

the solidification front speed, vn*. For this reason, a common method for either assessing microstructural features experimentally or actually controlling them in practice is directional solidification. In a simplified version of the experiment, a thin layer of alloy is encapsulated between microscope slides that are then placed atop controlled hot and cold blocks. The blocks are maintained at predetermined temperatures and at a fixed distance, in order to establish a desired temperature gradient, GL. The slide assembly is then pulled at constant speed vn* through the gradient. The experiment is often done with a transparent organic alloy, and the solidification morphology is observed using a microscope. (Recently, similar experiments have been performed in metallic systems using synchrotron radiation to observe the microstructure (Ref 5). Next, a model of the system is built with the goal of finding out how the important parameters interact to produce the observed microstructural features shown in Fig. 3. It is observed that if the pulling speed is slow enough, the interface remains flat. In this case, after a brief transient at the beginning of the experiment, the process reaches a quasi-steady state where the microstructure is constant in time. It is convenient to analyze the process in a frame that is fixed on the interface (z = 0) and to assume that in this frame the transport is independent of time. The material moves through the interface at velocity vn*. It is assumed that the material properties are all constant (thus there is no shrinkage flow) and, for convenience, that the thermal diffusivities of the solid and liquid are equal. Finally, variations in all fields perpendicular to the pulling direction are ignored, making the problem one-dimensional. Note that all these assumptions are not essential to the analysis, and either everything could be included or the assumptions could be justified through a scaling analysis similar to that given for the quenched rod. For the sake of brevity, they are simply taken

0.8 t = 0.05

0.6 0.4

t = 0.25

0.2

t = 0.70

Fig. 2 Temperature distributions through the midplane of a quenched rod for several values of the dimensionless time

Fig. 3

Micrographs from directional solidification of a succinonitrile-acetone alloy. Left: initial planar interface with small perturbations; right: developing dendrites, with primary and secondary arms indicated. Source: Ref 4

Transport Phenomena During Solidification / 291 as given. The transport equations for heat and solute now reduce to: vn

@T @2T ¼a 2 @z @z

vn

@CL @ 2 CL ¼ DL @z @z2

vn

@Cs @ Cs ¼ Ds @z @z2

1
2

0z z0

(Eq 25) (Eq 26) (Eq 27)

Looking at the micrograph in Fig. 3, it seems that the system chooses a length scale l1, the primary dendrite arm spacing. These equations are partially scaled by defining z = z/l1, with the result:    v l1 @T @2T  n ¼ 2 a @z @z    vn l1 @CL @ 2 CL  ¼ DL @z @z2    v l1 @CS @ 2 CS ¼  n DS @z @z2

(Eq 28) (Eq 29) (Eq 30)

Each equation has as a prefactor a variation of a dimensionless group called the Pe´clet number, corresponding to the ratio of advective to diffusive transport of heat or solute, depending on the equation. Actual values can be taken from the experiments for the material properties, vn* and l1, to evaluate the Pe´clet numbers: vn l1  102 ; a

vn l1  1; DL

vn l1  104 Ds

(Eq 31)

Substituting these results into Eq 28 to 30 allows terms to be dropped that have coefficients that are much less than 1. It is also convenient to then redimensionalize the equations so that boundary conditions can be applied. The result is: @2T 0 @z2 @CL @ 2 CL ¼ DL @z @x2 @CS 0  @x

vn

(Eq 32) (Eq 33) (Eq 34)

The solution to Eq 32 is a simple linear profile. Applying the boundary conditions T = Tcold at z = zcold and T = Thot at z = zhot gives: T ¼ Tcold þ

Thot  Tcold ðz  zcold Þ zhot  zcold

(Eq 35)

The solution clearly shows how the apparatus selects the temperature gradient. The result that the temperature field is linear and undisturbed by the moving interface is sometimes called the frozen temperature approximation, but it is really a result of scaling. The solution for the composition field in the liquid can be found after applying the boundary

λ2

condition given in Eq 16 at z = 0 and setting CL = C0 for z ! 1: CL ¼ C0 þ C0

  1  k vn z=DL e k

(Eq 36)

Thus, there is a solute boundary layer ahead of the interface of thickness roughly 3DL/vn*. The solution to Eq 33 is just Cs is constant. Examination of Eq 36 shows that CL = C0/k at z = 0, and this permits evaluation of the constant, so that: Cs ¼ C0

L Z

X

(Eq 37)

This example, although rather simple, indicates how the transport phenomena naturally separate into different length scales due to the disparity in the rate of transport of energy and solute. This is very important for understanding the origin of microstructure during solidification. It is beyond the scope of this article to pursue the analysis much further, but it is important to understand that there is one other important aspect. On considering the stability of the planar front growth morphology, it is found that for small values of the ratio GL/vn*, the interface becomes unstable to perturbations and breaks down first into cells and then into dendrites. In the stability analysis, a length scale associated with the surface energy of the liquid-solid interface enters the problem through a material property called the capillary length, d0 = Gk/mLC0(1  k), where G is the Gibbs-Thomson coefficient, and mL is the slope of the liquidus line. The characteristic length selected by the dendritic arms can be correlated to the geometric mean of the two length scales, that is: l

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi d0 ðDL =vn Þ

Microsegregation After the initially flat interface becomes unstable, it evolves into the dendritic pattern shown in the micrograph in Fig. 3. The structure is characterized by the primary (l1) and secondary (l2) dendrite arm spacings. In the previous analysis of transport, l1 is chosen as the length scale. At a finer scale, such as in the interdendritic region between secondary arms, l2 would be the appropriate length scale. Figure 4 illustrates the geometry. The fundamental concepts of scaling can be applied to this new problem. As illustrated, the array of secondary arms is approximately periodic, and a control volume can be chosen as shown, with dimensions l2 L and the expectation that l2  L. This time, there is no velocity, because the control volume is fixed in the dendrite array. Writing the governing equations for energy and solute within this volume yields:  2  @T @ T @T þ ¼a @t @x2 @z

(Eq 38)

Fig. 4

Schematic drawing of the local geometry of secondary dendrite arms

 2  @CL @ CL @CL ¼ DL þ @t @x2 @z  2  @Cs @ Cs @Cs ¼ Ds þ 2 @t @x @z

(Eq 39) (Eq 40)

Next, some scaled variables are defined: x z ; z¼ ; l2 =2 L T  Tsol y¼ Tliq  Tsol

x¼

t¼

t ; tf

(Eq 41)

where tf is the local freezing time, that is, the time it takes to go from the liquidus temperature, Tliq, to the solidus, Tsol. Substituting these scaled variables into the governing equations and invoking the assumption that l2  L gives: @y 4atf @ 2 y ¼ 2 @t l2 @x2

(Eq 42)

@CL 4DL tf @ 2 CL ¼ @t l22 @x2

(Eq 43)

@Cs 4Ds tf @ 2 Cs ¼ @t l22 @x2

(Eq 44)

Metallurgists often use measured correlations between l2 and tf, which show that for aluminum alloys (Ref 6):   1=3 l2  105 m=s1=3 tf

(Eq 45)

For these same alloys, a  106 m2/s, DL  1010 m2/s, and Ds  1014 m2/s. Combining these data with Eq 45, one finds that for solidification times between approximately 10 s and 1 day, Eq 42 to 44 reduce to: @2y 0 @x2

(Eq 46)

@ 2 CL 0 @x2

(Eq 47)

@Cs 0 @t

(Eq 48)

The dendritic array is periodic, so the derivatives @y/@x and @CL/@x must be zero at both

292 / Principles of Solidification x = 0 and x = 1. From this, it is concluded that both the temperature and the composition in the liquid are constant, and that the composition in the solid does not change with time. However, the composition at the liquid-solid interface must adjust to maintain local equilibrium. Applying the solute balance at the interface from Eq 16 gives a relation between the composition of the solid, Cs, and the solid fraction, fs, known as the Scheil equation: Cs ¼ kC0 ð1  fs Þk1

(Eq 49)

(This is probably not completely obvious from what has been presented here—see Ref 1 or 7 for a more thorough derivation.) Equation 49 can be recast in terms of temperature using the phase diagram, with the result:  fs ¼ 1 

T  Tf Tliq  Tf

needed to homogenize the structure is simply l22 =Ds . A conservative approach would be to homogenize the structure for two to three times this value. This implies that the microstructure, in the form of l2, is known. From where does that come? It is correlated with the local solidification time, as in Eq 45, but how does one obtain tf? This comes from solving the transport equation for heat at the macroscopic scale, for example, using a commercial casting solidification analysis package. Note that there is also a connection between the microsegregation and heat flow in the code, because the user usually must specify the enthalpy as a function of temperature, which is obtained from Eq 50 from the definition: ðT HðT Þ ¼

1=ðk1Þ (Eq 50)

This equation is very important, because it provides a very good estimate for the microsegregation in cast alloys.

Homogenization The microsegregation described in the previous section is inevitable in normal casting practices. This is often undesirable, and many castings are heat treated to reduce the segregation. A detailed analysis of the problem could be done, but it really is not necessary to do so if one wishes only to estimate the proper homogenization time. It is known that the segregation is more or less periodic on the length scale of the secondary dendrite arm spacing. Based on scaling, then, the estimated time

cp dT þ Hð1  fs Þ

(Eq 51)

298

The integral begins at 298 K (77  F), where enthalpy is taken to be zero by convention.

Summary The governing equations for transport based on balance of mass, momentum, energy, and solute were written. Equations for these balances were developed for a moving solidification front. It was found that the differences in density, enthalpy, and composition between the liquid and solid are very important for the understanding of solidification processes. Considerable effort was spent to understand how material properties and geometry can be analyzed in the context of the governing equations, a process called scaling.

Several example problems were considered to show how the hierarchy of time and length scales associated with transport leads to important features of cast microstructures. An analysis of planar front growth revealed the important length scale associated with diffusion DL/vn*. On the scale of secondary dendrite arms, the limited transport of solute in the solid, combined with rapid transport in the liquid, leads to the Scheil equation for microsegregation. Finally, the principles of transport and scaling are applied to homogenization processes to ameliorate the undesirable segregation pattern. REFERENCES 1. J.A. Dantzig and C.L. Tucker III, Modeling in Materials Processing, Cambridge University Press, New York, 2001 2. R.B. Bird, W.E. Stewart, and E.N. Lightfoot, Transport Phenomena, 2nd ed., Wiley, New York, 2002 3. H.S. Carslaw and J.C. Jaeger, Conduction of Heat in Solids, 2nd ed., Oxford University Press, London, 1959 4. R. Trivedi and K. Somboonsuk, Constrained Dendritic Growth and Spacing, Mater. Sci. Eng., Vol 65, 1984, p 65–74 5. R. Mathiesen and L. Arnberg, X-Ray Monitoring of Solidification Phenomena in Al-Cu Alloys, Mater. Sci. Forum, Vol 508, 2006, p 69–75 6. T.F. Bower, H.D. Brody, and M.C. Flemings, Measurements of Solute Redistribution in Dendritic Solidification, Trans. AIME, Vol 236, 1966, p 624 7. W.F. Kurz and D.J. Fisher, Fundamentals of Solidification, Trans-Tech, Aedermannsdorf, 1984

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 293-298 DOI: 10.1361/asmhba0005209

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Plane Front Solidification Wilfried Kurz, Ecole Polytechnique Fe´de´rale de Lausanne, Switzerland

PLANE FRONT SOLIDIFICATION (PFS) is not a commonly used process since it is generally difficult to obtain the required conditions. It is only in cases where a crystal has to be highly homogeneous across the section that the corresponding processes are developed. One impressive example of PFS is the industrial production of large silicon single crystals, used mainly as substrates for integrated circuits. Other examples of PFS can be found in more basic work of solidification phenomena; the morphological stability of plane fronts at both small and large growth rates has generated substantial interest. This part is fundamental for the theoretical understanding of most solidification phenomena and the reason for examining the topic here. In this article, the PFS of a single phase is treated without taking convection into account. The sections in this article discuss the solute buildup at the solid-liquid interface forming transients and steady state, the morphological stability/instability and perturbation theory, and rapid solidification effects, including solute trapping and oscillatory instabilities. Finally, a microstructural selection map is presented that gives an overview of interface stability as a function of composition for a given alloy.

Plane Front Solidification Technology Silicon single-crystal growth is a technical application of PFS. The component that forms the substrate of a multitude of integrated

Fig. 1

Element distribution as a function of volume fraction solid, fs, in plane front Czochralski processing of silicon single crystals. Source: Ref 1

circuits with minute electronic elements must have reliable properties on a scale smaller than the smallest elements of the circuit. Silicon substrates must also have a high degree of chemical purity and a high degree of crystalline perfection. This can only be obtained by PFS in either the Czochralski (CZ) or the float zone (FZ) melting process. For PFS of silicon, very low growth rates, on the order of several millimeters per minute, are required. Figure 1 shows the distribution of some elements in silicon for CZ growth (Ref 1). The doping elements used, such as boron or

Table 1 Symbol

Cl Co Cs D Di Ds d fs G Gc J K k m P T* Tf Tl To Ts V Va Vc Vo y z z0 DCo DTo dC di dK dT e G f l o xc

Definitions and units of symbols Definition

Liquid composition Alloy composition Solid composition Diffusion coefficient in liquid Diffusion coefficient across interface Diffusion coefficient in solid Diameter Solid fraction Temperature gradient Concentration gradient Solute flux Curvature ðK1 1 þ K2 1 Þ Distribution coefficient (Cs/Cl) Liquidus slope Pe´clet number Interface temperature Melting temperature Liquidus temperature Temperature of equal free energy (k(V) = 1) Solidus temperature Interface velocity Velocity of absolute stability Velocity of constitutional undercooling Collision rate of atoms Coordinate parallel to interface Coordinate perpendicular to interface Coordinate of specimen Solidus-liquidus interval at Ts Solidus-liquidus interval (Tl  Ts) Solute diffusion length (D/V) Interface diffusion length (Di/V) Capillary length (G/DTo) Thermal length (DTo/G) Amplitude of perturbation Gibbs-Thomson parameter Driving force for instability (G  mGc) Wavelength Wave number (2p/l) Stability function

Units

% % % 2 m /s 2 m /s 2 m /s m ... K/m %/m % m/s 1 m ... K/% ... K K K K K m/s m/s m/s m/s m m m % K m m m m m Km K/m m 1 m ...

phosphorus, have distribution coefficients (k) close to unity, while many other elements have much smaller coefficients. (See Table 1 for the definitions and units of symbols used in this article.) Due to the different volumes of liquid and solid in the CZ and FZ processes, the composition profile of the single crystal produced is also different. In CZ, the originally large liquid volume produces a crystal where the concentration of the first half corresponds closely to kCo, leading in its second half to an important final transient (Fig. 1). In the FZ process, the liquid volume is small, and saturation of this liquid, at least for the elements with k values close to unity, is rapid. After a short distance, the composition profile reaches the mean concentration of the crystal, and the final transient is small. In the early years of silicon single-crystal growth, the FZ process was widely used; however, today (2008) large-diameter crystals are generally produced by the CZ process. The size of industrial single crystals has dramatically increased over the last 55 years (Fig. 2), and progress in crystal growth techniques has made substrates for integrated circuits progressively less expensive.

Plane Front Solidification Science An understanding of most solidification phenomena began essentially in the 1950s with the work on constitutional undercooling (CU).

Fig. 2

Plane front solidification processing. Evolution of the size of silicon single crystals over the years. Source: Ref 1

294 / Principles of Solidification Chalmers and his group initially produced bicrystals in order to study grain boundaries. Using the Bridgman method for PFS, they found various patterns at the decanted solidification front. These observations led to the most successful research on morphological instability: cell and dendrite growth (see the interesting article on the history of CU by Tiller in Ref 2). It was discovered by theory and experiment that a plane front is stable only if the growth rate is sufficiently small and the temperature gradient across the interface is positive and sufficiently large. The paper on CU by Tiller et al. in 1953 (Ref 3) was a major starting point for a half-century of solidification science (Ref 4). A decade later, Mullins and Sekerka (Ref 5, Ref 6) published their important papers on a rigorous linear stability analysis. They showed that, taking into account the interface energy, the wavelength of the morphological instabilities could be determined. Furthermore, a new limit of stability, absolute stability, was predicted for high growth rates. From perturbation theory, Langer and Mu¨ller-Krumbhaar (Ref 7) proposed marginal stability in 1978 as the operating point of dendrites (i.e., the selection of the tip radius for a given dendrite velocity), which substantially improved theoretical predictions and initiated much new work on dendrite theory.

at the composition kCo (Fig. 3(b)). The difference between the liquid and solid composition at the moving solid-liquid interface forms the source of the solute, J1, which dissipates into the liquid by diffusion, J2. Due to slow diffusion in the liquid and much slower diffusion in the solid, solute piles up at the solidification front. Neglecting the small flux into the solid, the fluxes at the interface may be written as: J1 ¼ V Cl ð1  kÞ J2 ¼ DGc

At low growth rates, interface stability requires a positive temperature gradient; that is, the temperature must increase from the solid into the melt. Such a situation is typical for directional solidification. Here, the heat flux is unidirectional to produce planar isotherms and interface shapes. For example, in the Bridgman process, a crucible that contains the alloy is moved through a heater and cooler, both being separated by a buffer that thermally isolates the central zone. For a theoretical analysis of the phenomena that are produced in such a process, the situation is often simplified, and the thermal transients at the beginning and end of solidification are not taken into account. It is assumed that the temperature gradient and the growth rate are instantaneously established and that they stay constant throughout the specimen of constant cross section.

(Eq 1)

J2 ¼ V ðCl  Co Þ

1k k

(Eq 5)

and the liquid composition profile is: Cl ¼ Co þ Co expðVz=DÞ

(Eq 6)

(Eq 2)

where z is the distance into the melt from the solid- liquid interface, perpendicular to that interface. This concentration field is essential to understanding constitutional undercooling.

(Eq 3)

Final Transient

the flux into the melt becomes:

Initially, C*l is close to Co, and the sink term is smaller than the source term, that is, J2 < J1. Therefore, the concentration peak at the interface increases with distance and approaches the steady-state value Co/k (Fig. 3b). After a certain distance along the crucible, given by: z0 ¼ nD=V k

Transients and Steady State

Co ¼ Co

where C*l is the liquid composition at the solid-liquid interface. With the concentration gradient at the interface: Gc ¼ ðCo  Cl ÞV =D

by the surface (A), is equal to the mass in the boundary layer (B) (Fig. 3(b)). In the steady state, the liquid composition far from the interface is equal to the solid composition, Co, and the interface grows at the solidus temperature of the alloy, Ts. In the steady state, the solute rejection at the interface is given by:

(Eq 4)

steady state is closely approached, that is, J2 ! J1. As a good approximation, n = 4. From Eq 4, it can be seen that for small k- values, the initial transient becomes very long, and steady state may never be reached. This, for example, is true for aluminum in silicon or salt in ice.

Steady State Mass balance considerations require that the missing mass in the solid, given by the difference between the mean alloy composition, Co, and the solid composition, C(z0 ) represented

When the boundary layer interacts with the end of the specimen, the dissipation of solute becomes limited due to the decreasing volume of liquid. Therefore, when k < 1, both the concentration of the liquid at the interface, C*l, and, correspondingly, of the solid, C*s = kC*l, increase. (In the case of k > 1 the final transient leads to a solid with a decreasing concentration.) The final transient becomes measurable when the interface reaches a multiple of the boundary layer thickness, D/V. It has been evaluated in terms of a series by Smith et al. (Ref 8). Unlike the initial transient, the final transient does not depend on k. Due to the very slow diffusion of substitutional elements in the solid, the final composition may reach very large values—at the limit, 100% of solute. The final concentration can be estimated by the Brody-Flemings equation modified by a function of the Fourier number (Ref 9, 10). For D ! 1 and Ds ! 0, the final transient is already present at the very beginning of

Initial Transient The phase diagram in Fig. 3(a) shows that the solubility of a solute in a solvent is different in the liquid and the solid phases. This reflects the thermodynamic requirement of equal chemical potentials. In most alloys, the melting point of the solvent is reduced by the solute and, according to the generally accepted definition, the distribution coefficient, k (= Cs/Cl), is smaller than unity. In this case, assuming that there is no nucleation barrier, the solid starts to grow

Fig. 3

Steady-state solute distribution. (a) Phase diagram and (b) concentration field in a directional solidification experiment

Plane Front Solidification / 295 solidification, and the concentration variation with solid fraction, fs, follows the GulliverScheil equation: Cs ¼ ð1  fs Þk1 kCo

(Eq 7)

The form of the solute distribution in this case resembles but, due to geometrical differences, is not exactly the same as the curves of Fig. 1.

Constitutional Undercooling Constitutional undercooling is a general phenomenon where heat diffusion is much more rapid than solute diffusion, thereby tending to produce steep concentration gradients (small concentration boundary layers) and low-temperature gradients (extensive temperature boundary layers). The concentration field ahead of the moving interface can be related through the liquidus slope, m = dTl/dC, to a corresponding liquidus temperature field. When this field is superimposed on the temperature field due to the heat flux (neglecting a very small undercooling from atom attachment kinetics), an undercooled liquid region is obtained under certain conditions. The concentration gradient for the steadystate concentration field (Eq 6) at the interface z = 0 is: (Eq 8)

This concentration gradient gives rise to a liquidus temperature gradient: dTl =dz ¼ ðdTl =dCÞðdC=dzÞ ¼ mGc

G ¼ mGc

(Eq 9)

(Eq 10)

or G=V ¼ To =D

Morphological Stability

Gc ¼ Co ðV =DÞ

Constitutional undercooling occurs when the heat flux gradient at the interface, G, is smaller than the liquidus temperature gradient. This case is shown in Fig. 4. The limiting condition of CU is obtained with the equality of the two gradients (Ref 3):

(Eq 11)

with DTo = mDCo. The left side of Eq 11 represents the process variables, and the right side represents the alloy properties. Stability corresponds to G/V > DTo/D. Equation 11 can also be understood as the equality of two lengths: the diffusion length, dC (= D/V), and the thermal length, dT (= DTo/G) (Ref 11).

correspond to this sequence of events. Eventually, the waves develop into cells and dendrites. Clearly, in order to form such an undulated interface structure, the system requires energy for creating the curvature that must come from interface undercooling. Not every structure is possible, and the strength of the CU defines the possible curvature (l-1). The details of the Mullins-Sekerka analysis can be found in the original papers (Ref 5, 6) or in Appendix 7 of Ref 10. Here, only the simplest possible interface stability analysis is developed, which, however, contains most of the essential physics (Chapter 3 of Ref 10). The starting point of a linear stability analysis is an infinitesimally small perturbation, that is, a wave of much smaller amplitude, e, than wavelength, l(e/l << 1). The interface temperature, T*, may be written as: T  ¼ Tf þ mC1  K

Perturbation Analysis The simple criterion of CU depends only on transport phenomena. It does predict the possibility of morphological interface instability. However, since the interface energy is not considered, it does not say anything about the interface morphology, such as the wavelength, l, of a perturbation in the case of instability. The capillarity effect is taken into account in the perturbation theory, which was developed for the first time by Mullins and Sekerka (Ref 5, 6) for solidification. From the inclined interface of succinonitrile under the microscope (Fig. 5) (Ref 12), it can be seen that at the beginning of instability of a plane front, a wavy small-amplitude pattern develops across the solid-liquid interface. This happens first at the grain-boundary grooves, where solute piles up more strongly than at the plane front, then at the subgrain-boundary grooves and finally at the planar interface. The figure shows that the amplitudes

(Eq 12)

with the product of the Gibbs-Thomson parameter, G, and curvature, K, equal to the capillarity undercooling. With a sinusoidal deformation of the interface: z ¼ e sinðoyÞ

(Eq 13)

the temperature difference between the extremes of the perturbation (tip and depression) is given by: Tt  Td ¼ mðCt  Cd Þ  ðKt  Kd Þ

(Eq 14)

and: Kt ¼ Kd ¼ 4p2 e=l2

(Eq 15)

Due to the infinitesimal small amplitude of the perturbation and the much larger diffusion distances in metals, the temperature and concentration differences between the tips and the depressions can be obtained to a good approximation from the gradients existing at the unperturbed planar solid-liquid interface (small thermal and solutal Pe´clet number approximation, Pt < Pc < 1). The temperature and concentration differences are simply given by: Tt  Td ¼ 2eG

(Eq 16)

Ct  Cd ¼ 2eGc

(Eq 17)

This leads to a relationship for the perturbation wavelength of a marginally stable interface: li ¼ 2pð=fÞ0:5

Fig. 5 Fig. 4

Constitutional undercooling. (a) Interface concentration field with boundary layer dc. (b) Temperature fields at the solid-liquid interface

Image of a solid-liquid interface at the beginning of instability. For clarity, the interface of succinonitrile in a directional solidification experiment is inclined to the optical axis of the microscope. Grain boundaries and subgrain boundaries can be seen. The mean wavelength of the perturbations is 5 mm (200 min.). Source: Ref 12

(Eq 18)

where f = (mGc  G) is the driving force for unstable growth. Equation 18 shows that the critical (marginally stable) perturbation wavelength, where neither the tip nor the depression will grow with respect to the mean interface location, depends on the ratio of the GibbsThomson parameter (resisting deformation)

296 / Principles of Solidification and the positive difference between the liquidus temperature gradient, mGc, and the temperature gradient, G (driving force of deformation). From the full Mullins-Sekerka analysis, a relationship is obtained between the amplitude growth and the wavelength, which is graphically shown in Fig. 6. The unstable modes of the perturbation lead to a range of wavelengths with li, the lower bond of possible instabilities. The wavelength range, li, is used in the Ivantsov-marginal stability (I-MS) dendrite theory (Ref 13) for calculating the tip radius. If it is assumed, for the sake of simplicity, that the thermal properties of liquid and solid are the same and that the temperature gradient across the interface is constant, that is, no latent heat is produced, then for the marginally stable state, the Mullins-Sekerka analysis gives:

o2  G þ mGc xc ¼ 0

(Eq 19)

The stability function, xc, has a value of unity for solute Pe´clet numbers up to P = 1 and drops to zero at higher P-values (Ref 14). It can be easily seen that, when xc equals unity, Eq 19 corresponds to Eq 18, and, in addition, when the interface energy is set to zero, the limit of CU is obtained (Eq 10).

Rapid Solidification Effects Absolute Stability When the interface velocity becomes large, the stabilizing capillarity effect dominates the

destabilizing solute diffusion effect. This phenomenon leads to a restabilization of the plane front when the growth velocity exceeds a critical value given by: Va ¼

DTo k

(Eq 20)

This phenomenon of absolute stability can be understood by comparing the variation of the controlling length scales with velocity: The diffusion distance decreases with V1, and the characteristic length of the perturbation decreases with V1/2. At high velocity, the diffusion distance becomes much smaller than the perturbation wavelength, and diffusion becomes essentially one-dimensional (localized). This phenomenon can also be qualitatively understood as the equality of two length scales: the diffusion length, dc (= D/V), and the capillarity length, dK (= G/DTo). According to this theory, morphological stability is therefore observed when the interface velocity is either below Vc (= DG/DTo) or above Va (= DDTo/kG).

Solute Trapping Up to now, it has been assumed that there is local equilibrium at the interface; that is, the compositions at the interface of solid and liquid correspond to the equilibrium phase diagram. This is a good assumption so long as the diffusion processes across the interface are much more rapid than the interface displacement. If, however, the interface velocity becomes large, the local equilibrium assumption is no longer applicable, and the distribution coefficient tends to unity. The relationship between k and V for a dilute solution can be written as (Ref 15):

Fig. 6

Amplitude growth (de/dt

 1/e) as a function of the perturbation wavelength for three velocities (G = 10

4

kðV Þ ¼ K/m)

k þ ðdi V =Di Þ 1 þ ðdi V =Di Þ

(Eq 21)

where the parameter in parentheses represents the interface Pe´clet number (ratio of interface width, di, to diffusion distance across the interface, Di/V). A decreasing difference between the liquid and solid compositions (solute trapping) appears in the phase diagram where both liquidus and solidus lines converge to the To line (Ref 16–18), as shown in Fig. 7(a). When T1 = Ts = To, the distribution coefficient k = 1. In this limiting case, the Gibbs free energies of both phases are equal, but the chemical potentials are not. This can be expressed through the following equations for the interface temperature, T*, and the nonequilibrium liquidus slope, m(V) (Ref 18): T  ¼ Tf þ mðV ÞCl þ

Fig. 7

Nonequilibrium solidification. (a) Equilibrium (cm) and nonequilibrium phase diagram mv. The interrupted line represents To, which corresponds to equal free energies, i.e., to k(V) = 1. (b) Interface temperature functions for plane front and cells/dendrites

mðV Þ ¼ m

m V 1  k Vo

1  kðV Þ þ kðV Þ lnðkðV Þ=kÞ 1k

(Eq 22)

(Eq 23)

Plane Front Solidification / 297 The last term of Eq 22 introduces the atom attachment kinetic undercooling, which depends on the collision rate of the atoms, Vo. In most alloys, absolute stability corresponds to the regime of local equilibrium loss, which reduces the limit, Va. A detailed analysis of this problem leads essentially to Eq 20, with DTo and k as velocity-dependent parameters (Ref 19, 20): Va ¼

DTo ðV Þ kðV Þ

(Eq 24)

where DTo(V) = Com(V)[k(V)1]/k(V). In Fig. 7(b), the interface temperatures of the various growth forms—plane front, cells, and dendrites—are shown as functions of V for a constant G (>0). The lower (partially interrupted) curve represents the plane front temperature, and the upper curve the cell/dendrite temperature. At two velocities, Vc and Va, the cell/dendrite tips have the same growth temperature as the plane front. The increase in the plane front temperature at growth velocities near to Va is due to solute trapping (k(V)), while the decrease of the plane front temperature at very high growth velocities is due to the increasing effect of atom attachment kinetics (Ref 21).

Oscillatory Instability Once the growth velocity exceeds absolute stability, the nonequilibrium conditions make the plane front unstable with time; that is, the planar interface develops oscillations. Oscillatory instabilities have been predicted by Corriel and Sekerka (Ref 22) and Davis and co-workers (Ref 19, 23, 24). Karma and Sarkissian (Ref 25) analyzed this phenomenon numerically and showed that the oscillatory instability induces large-amplitude relaxation oscillations at a critical velocity. This phenomenon can be qualitatively understood as follows: When, beyond Va, the isotherm displacement rate reaches the plane front branch with its positive T(V) slope, (Fig. 7(b)) the interface accelerates as k approaches unity, and less solute is rejected. When the accelerating plane front reaches the maximum in the T(V) curve, the solidification velocity must decrease in order to conform with the thermal field until it reaches the interface temperature minimum, where it may develop a cellular morphology. Either a banded, plane front structure, or bands of alternating plane front and cells is obtained (Ref 26, 27). Figure 8(a) shows the regimes of existence of the various instability phenomena according to the model of Huntley and Davis in a Co-V diagram (Ref 24). Below the concentration minimum at 5  103 at.% Fe, a plane front is stable at all growth velocities. At this point, CU and absolute stability have the same effect. The hatched region at the branch of absolute stability corresponds to the experimental observation of banded structures (Ref 26). The

schematic representation of Fig. 7(b) is shown in quantitative form for an Al-1at.%Fe alloy in Fig. 8(b) (Ref 29). This diagram has been calculated using the plane front and IMS dendrite model (Ref 13). For the parameters used in the figure, plane front growth is unstable with respect to cellular and dendritic growth between Vc = 5  105 and Va = 7  101 m/s. Beyond VM = 4 m/s, atom attachment kinetics dominate, and the plane front is stable in every respect. From this, it is evident that the results of both theories, stability theory (Ref 24) (Fig. 8a) and interface response (Ref 29, 30) (Fig. 8b), and of experiments are consistent and complementary.

microstructure formation phenomena and is therefore fundamental to solidification theory. Interface stability theory allows the limits of microstructure formation to be predicted. At low rates, it is the ratio G/V that controls the limit of CU, while at high rates, the limit of absolute stability depends on capillarity. Solute trapping reduces the limit of absolute stability. The wavelength of the marginally stable state, li, is useful for modeling dendritic growth with the I-MS theory. In the high V transition region from cellular-dendritic to absolutely stable plane front growth, banded microstructures appear that are indicative of oscillatory interface instability.

Solidification Microstructure Selection Maps

REFERENCES

Figure 8(a) represents a microstructural selection map for binary aluminum-iron alloys. It shows two V-shaped regimes of instability. The low-V branch of the continuous curve corresponds to constitutional undercooling, while the high-V branch of the continuous curve represents absolute morphological stability. At low-V (above the solid curve), morphological instability is predicted, and at high velocities (above the dashed curve), oscillatory instabilities develop. The V-range of instability increases with solute concentration. To the left and right of the curves, PFS is observed, and below the concentration minimum, PFS occurs for all interface velocities. Similar diagrams for various alloys, silvercopper, silicon-tin, aluminum-copper, and aluminum-tin, can be found in Ref 19 and 24. Figure 9 is a microstructural selection map for iron-nickel alloys (Ref 28, 30). If the coordinates of the axes are interchanged, Fig. 9 and Fig. 8 are similar diagrams for different alloys. In alloys such as iron-nickel, where two solid phases compete in the range of the peritectic composition, the microstructures that may form are more complicated. At concentrations below 3.8 at.% Ni, the d-phase (ferrite) is stable, while at high nickel concentrations, the g-phase (austenite) replaces the d-phase. Because kNi g > kNi d , the V-range of unstable g plane front growth is much narrower than that of d. In the low-V regime (high G/V values) at 4 at. %, two-phase growth is observed, either in the form of d/g bands or of peritectic coupled growth (Ref 31).

1. W. Zulehner, Mater. Sci. Eng. B, Vol 73, 2000, p 7–15 2. W.A. Tiller, Mater. Sci. Eng., Vol 65, 1984, p 3 3. W.A. Tiller, K.A. Jackson, J.W. Rutter, and B. Chalmers, Acta Metall., Vol 1, 1953, p 428

Fig. 8

Summary Plane front growth can be observed at both low and high velocities. It is a phenomenon that is rarely observed and seldom used in practice, except in single-crystal growth and in directional solidification experiments. However, the understanding of PFS is basic to most

Stability diagrams for an Al-1at.%Fe alloy as a function of growth rate, V, (G = 5  106 K/m). Source: Ref 29. (a) Composition dependence of morphological and oscillatory instabilities showing complete morphological stability below 5  103 at.%. Source: Ref 23. (b) Interface temperature (response) for plane front, Tp(V ), and dendritic front, Td(V ), compare with Fig. 7(b). Source: Ref 29. The characteristic interface velocities are Vc, limit of constitutional undercooling; Vm, velocity of minimum interface temperature; Va, limit of absolute stability; and VM, velocity of maximum interface temperature.

298 / Principles of Solidification

Fig. 9

Microstructure selection map (V- Co diagram) of iron-nickel alloys for G = 105 K/m showing regions of plane front solidification of cellular/dendritic structures of delta and of gamma, respectively. Source: Ref 28

4. W. Kurz, Mater. Sci. Forum, Vol 508, 2006, p 313–324 5. W.W. Mullins and R.F. Sekerka, JAP, Vol 34, 1963, p 323 6. W.W. Mullins and R.F. Sekerka, JAP, Vol 35, 1964, p 444 7. J.S. Langer and H. Mu¨ller-Krumbhaar, Acta Mater., Vol 26, 1978, p 1681 8. V.G. Smith, W.A. Tiller, and J.W. Rutter, Can. J. Phys., Vol 33, 1955, p 723

9. H.D. Brody and M.C. Flemings, Trans. Metall. Soc. AIME, Vol 236, 1966, p 615 10. W. Kurz and D.J. Fisher, Fundamentals of Solidification, 4th ed., Trans Tech, 1998 11. R. Trivedi and W. Kurz, Acta Metall. Mater., Vol 42, 1994, p 15–23 12. D.J. Fisher and W. Kurz, unpublished work, EPFL, 1977 13. W. Kurz, B. Giovanola, and R. Trivedi, Acta Metall., Vol 34, 1986, p 823

14. R. Trivedi and W. Kurz, Acta Metall., Vol 34, 1986, p 1663–1670 15. M.F. Aziz, JAP, Vol 53, 1982, p 1158 16. J.C. Baker and J.W. Cahn, in Solidification, American Society for Metals, 1971, p 23 17. W.J. Boettinger, S.R. Coriell and R.F. Sekerka, Mater. Sci. Eng., Vol 65, 1984, p 27 18. W.J. Boettinger and S.R. Coriell, in Science and Technology of the Undercooled Melt, P.R. Sahm, H. Jones, and C.M. Adam, Ed., Martinus Nijhoff, 1986, p 81 19. D.A. Huntley and S.H. Davis, Acta Metall. Mater., Vol 41, 1993, p 2025–2043 20. A. Ludwig and W. Kurz, Acta Mater., Vol 44, 1996, p 3643–3654 21. R. Trivedi and W. Kurz, Int. Mater. Rev., Vol 39, 1994, p 49–74 22. S.R. Corriel and R.F. Sekerka, J. Cryst. Growth, Vol 61, 1983, p 499 23. D.A. Huntley and S.H. Davis, Phys. Rev. B., Vol 53, 1996, p 3132 24. S.H. Davis, Chap. 6, Theory of Solidification, Cambridge Univ. Press, 2001 25. A. Karma and A. Sarkissian, Phys. Rev. E, Vol 47, 1993, p 513 26. M. Gremaud, M. Carrard, and W. Kurz, Acta Metall. Mater., Vol 39, 1991, p 1431 27. M. Carrard, M. Gremaud, M. Zimmermann, and W. Kurz, Acta Metall. Mater., Vol 40, 1992, p 983–996 28. M. Vandyoussefi, H.W. Kerr, and W. Kurz, in Solidification Processing 1997, J. Beech and H. Jones, Ed., University of Sheffield, 1997, p 564 29. W. Kurz and S. Dobler, Interfaces for the 21st century, M.K. Smith, M.J. Miksis, G.B. Mc Fadden, G.P. Neitzel, D.R. Canright, Ed, Imperial College Press, 2002 203–212 30. W. Kurz, Adv. Eng. Mater., Vol 3, 2001, p 443–452 31. S. Dobler, T.S. Lo, M. Plapp, A. Karma, and W. Kurz, Acta Mater., Vol 52, 2004, p 2795–2808

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 299-306 DOI: 10.1361/asmhba0005210

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Non-Plane Front Solidification Rohit Trivedi and Erin Sunseri, Ames Laboratory, USDOE and Iowa State University

DURING THE SOLIDIFICATION of alloys, a nonplanar interface is most commonly present under conditions of casting and welding, playing a crucial role in determining the properties of the solidified material. The most common nonplanar interfaces form cellular or dendritic microstructures, which lead to microsegregation and possible second-phase formation that control the properties and reliability of the solidified product. Since microstructures form a key link between properties and processing conditions, it is important to develop basic ideas on how processing conditions influence microstructures so that one can predict, modify, and control the microstructure of the material by designing appropriate composition and processing conditions.

Columnar and Equiaxed Microstructures The thermal field in a casting is very important in determining the microstructure of the cast alloy. Two distinctly different conditions may exist in a mold that can give rise to equiaxed or columnar crystals. First, consider the case of equiaxed crystals that form in an undercooled melt. As the solid nucleates and grows, latent heat and solute are rejected into the liquid so that negative temperature and concentration gradients are present in the liquid at the solid/liquid interface. As discussed by Kurz in the previous article, “Plane Front Solidification,” in this Volume, the stability of the interface at low velocity is controlled by constitutional undercooling and interface energy. Since the thermal and solute gradients in the liquid are negative, constitutional supercooling will always be present to cause the interface to become unstable. The interface energy will initially stabilize the spherical nucleus, but the effect of interface energy becomes small as the sphere radius becomes approximately an order of magnitude larger than the critical radius of nucleation. Consequently, instabilities will develop during the early stage of the growth. In most cubic metals, instabilities occur in six directions. These instabilities grow and form side branches, so that six dendrite branches form from a nucleus

and give an equiaxed shape to the solid, as shown in Fig. 1. Equiaxed structures generally form at the center of a casting, where a substantial part of the liquid becomes constitutionally undercooled. The nucleation density controls the final grain size, and the nucleation density is often increased by the use of inoculants. Columnar crystals form when the temperature gradients in the liquid and the solid are positive, such that the latent heat generated at the interface is dissipated through the solid. Such a temperature field gives rise to directional solidification and results in the columnar zone in a casting. The conditions for the stability of a planar front, discussed in the previous article, “Plane Front Solidification,” by Kurz, show that a planar front will be stable when V < Vc or V > Va, where, for clarity, the thermal gradient and composition are taken to be constant. These limiting velocities are given by Vc = G D/DT0 and Va = DDT0/kG, where D, k, and G are the diffusion coefficient in the liquid, the solute distribution coefficient, and the Gibbs-Thomson parameter, respectively, and DT0 = mC0(k
1)/k, in which m is the slope of the liquidus at the solidus temperature. Thus, within the velocity range Vc < V < Va, planar interfaces become unstable and reorganize into a periodic structure consisting of shallow or deep cells or dendrites (Ref 1), as shown in Fig. 2. If the velocity is increased further, dendrites transform to cells and then to a planar interface at the velocity Va.

Fig. 1

Equiaxed dendrites undercooled liquid

that

form

from

an

Microstructure and Processing Conditions The formation of planar, cellular, or dendritic microstructure for a single-phase growth is governed by the processing conditions of G and V and the composition of the alloy. The effect of processing conditions (G,V) on microstructure is illustrated in Fig. 3, in which regimes of planar, cellular, and dendritic microstructures are

Fig. 2

Columnar structures that form under directional solidification conditions. (a) Shallow cellular. (b) Deep cellular. (c) Dendritic interfaces

300 / Principles of Solidification

Fig. 3

Microstructure-processing diagram showing the regimes of planar, cellular, and dendritic microstructures as a function of the Gibbs-Thomson parameter (G) and the velocity (V ) that are present in different processing technologies

Fig. 4 shown for the directional solidification of Al4.0wt%Cu. Typical ranges of G and V, as well as the cooling rate ðT_ ¼ GV Þ, present in different solidification processing techniques are also shown in the figure. A planar front growth occurs at very low velocity, as in single-crystal casting, or at very high velocity, as in laser processing and atomization of very fine droplets. A cellular microstructure forms at low velocities when a planar front becomes unstable. It also forms at very high velocities before the planar interface becomes stable. Under normal casting conditions, dendritic structures are commonly observed (gray area) in most alloys used in casting. The scale of microstructure becomes finer as one goes from single-crystal growth to normal casting to continuous casting to strip casting to laser welding. The processing conditions (V and G) are critical not only in the selection of nonplanar microstructures, but they also govern microstructural scales that are related to properties.

Characteristic Lengths of Microstructures To relate properties of a material to the microstructure, it is first necessary to define microstructure in terms of some characteristic parameters that can be quantitatively measured and can be related to properties. For equiaxed dendritic structure, the important characteristic lengths of the microstructure are the dendrite tip radius, R, the secondary arm spacing, l2, and the average distance between the grains. For directional solidification, important lengths are the tip radius, R, the primary spacing, l, the secondary spacing, l2, and the length of the dendrite in the two-phase region, or the extent of the mushy zone, ‘. The formation of a nonplanar interface causes solute segregation patterns, since the solid grows with a smaller composition than the alloy composition. The rejected solute

(a) Array of cells showing variation in the shape and local spacing. (b) Rescaling of two cells with different local spacing gives a unique shape.

accumulates in the intercellular/interdendritic region for alloy systems with phase diagrams in which the solute segregation coefficient is less than 1 (decreasing liquidus temperature with concentration). From a practical viewpoint, the interarm spacings characterize the solute segregation pattern, and often, a secondphase distribution in the interdendritic region results, which also has a strong influence on the properties of the material. Theoretical models of microstructure formation in a casting are thus directed toward quantitatively understanding how the processing conditions, alloy composition, and thermophysical properties control the magnitudes of these spacings and of the tip temperature or tip composition. First, the basic model of cellular and dendrite growth is described, and then the conditions for the columnar-to-equiaxed transition are discussed.

Cellular Microstructures When growth conditions exceed the critical condition for planar interface stability, the thermal fluctuations cause perturbation in the shape of the interface. These perturbations grow with time and reorganize into a periodic array of cells. For velocities just above the critical velocity, shallow cells form in which the length of the cell is small and is of the same order of magnitude as the cell spacing (Fig. 2a). The tip region of the cell is broader, and the cell has a larger tip radius. As the velocity is increased, the shallow cells transform to deep cells in which the length of the cell is much larger than the cell spacing, as shown in Fig. 2 (b). As the velocity is increased, the tip region of the deep cell assumes a nearly parabolic shape, which is similar to the dendrite tip shape

so that the term cellular dendrite is often used to characterize this structure (Ref 2, 3).

Scaling Behavior of Cellular Microstructures The steady-state shape of the cell is rather complicated, since it is influenced by the solute as well as thermal field interactions between the neighboring cells. A detailed numerical model has been developed by Hunt and Lu (Ref 2) to characterize the shape, cell tip undercooling, and cell spacing as functions of G, V, and composition. A simple analytical expression for the cell tip undercooling or tip temperature (Tt), which is a good approximation in the limit of small k, has been developed by Bower, Brody, and Flemings (Ref 4) as: Tt ¼ TL
ðGD=V Þ

(Eq 1)

where TL is the liquidus temperature of the alloy. The shape and the local cellular spacing are found not to be unique, but a narrow range of stable shapes and spacing has been observed for different cells in an array, as seen in Fig. 4(a). For spacing below this range, cell elimination occurs, while for spacing larger than this range, tip-splitting occurs to create a new cell. Although different cells in an array have different shapes and different local spacing, it is shown that the shapes are self-similar (Ref 5), and they scale with the local spacing; that is, all cell shapes in an array superpose when they are rescaled to the local spacing, as shown in Fig. 4(b) for two cells having minimum and maximum spacing value in the array. This scaling behavior implies that l/R is constant for all cells in an array. Although the shape and spacing vary for different cells, the position of the cell tips, or the tip temperature, is nearly constant.

Non-Plane Front Solidification / 301 Microsegregation The formation of cellular structures gives rise to solute segregation in the solid. The tip of the cell is at a higher temperature or lower composition for systems with k < 1, so that the solute is rejected in the intercellular region that will give rise to higher composition in the solid at the base of the cell. A periodic concentration profile will be present in the solid with a periodicity that is equal to the primary spacing. Since the liquid region between the primary cells is very small compared to the diffusion distance in the liquid, the concentration is nearly uniform in the lateral direction, that is, normal to the growth direction, so that under negligible diffusion in the solid, the solute profile in the solid is given by the Scheil-Gulliver equation (Ref 4, 6): Cs ¼ kC0 ½1
fs k
1

(Eq 2)

where fs is the fraction of solid. When the diffusion in the solid is not negligible, the aforementioned equation is modified as: 0

k
1 1
2a0 k

Cs ¼ kC0 ½1
fs ð1
2a kÞ

(Eq 3)

0

in which the parameter a is given empirically in terms of the dimensionless parameter a = Dstf/L2 as: a0 ¼ a½1
expð
1=aÞ
0:5 expð
1/2aÞ

(Eq 4)

Ds is the diffusion coefficient in the solid, and L is the distance between the cells. The modified equation ensures that the Scheil-Gulliver equation is obtained when Ds = 0, and lever rule is obtained when Ds = 1. For single-phase solidification in which the diffusion in the solid is important, microsegregation is characterized by the ratio of the maximum solid composition (at the cell base) to the minimum solid composition (at the cell tip), and this segregation ratio (SR) can be quantitatively related to properties. If ‘ is the length of the cell, GM the average temperature gradient in the two-phase region, and DT 0 is the cell tip undercooling with respect to the melting point isotherm of the solvent, then: SR ¼ 1 þ



‘ GM T 0

(Eq 5)

As the cellular structure becomes cellular dendritic or dendritic, ‘ increases sharply and DT 0 decreases under normal solidification conditions. Thus, the segregation ratio will increase significantly upon the formation of cellular dendrites or dendrites. The cellular spacing, l, is important since it characterizes the periodicity of the segregation pattern in the solid. Thus, if homogeneous composition is desired, the time required for the homogenization is related to the spacing as:



t ¼ 0:47 l2 =Ds

(Eq 6)

where t is the time required to homogenize to 1% of the original composition difference, and Ds is the diffusion coefficient in the solid. One of the important applications of the microsegregation equation is to predict the possible formation of different phases in the intercellular region. When the phase diagram shows a limited solubility in the solid, a new phase can form in the intercellular region when the local composition of the solid, predicted by the modified Scheil-Gulliver equation, reaches its solubility limit. In this case, the remaining liquid may form a new phase, such as a eutectic or peritectic phase, or a compound, depending on the phase diagram. As the composition in the liquid increases further, it is possible to form additional phases depending on the solubility limit for the new phase given by the phase diagram. The formation of a peritectic phase in the intercellular region of the primary phase in the tin-cadmium system is shown in Fig. 5. When the second phase forms, the properties of the material are related to the fraction of the second phases.

High-Velocity Cells The transition from dendrite to cell also occurs at high velocity, and these cells are generally referred to as high-velocity cells. Since the cellular structure is very fine at high velocities, it is sometimes referred to as microcellular structure. In most alloy systems, this transition occurs at very high velocity, which is near the absolute velocity for planar front solidification. Microstructural observations in very fine, atomized aluminum-silicon powders have characterized the transition from dendrites to cells at high velocity (Ref 7). The dendrite-to-cell transition also depends on the freezing range of the alloy, so that it can occur at relatively lower velocities in some systems. For example, experimental studies by Lee et al. (Ref 8) on Fe-18.0%Cr-5%Al have shown that the dendrite-to-cell transition occurs at approximately 0.050 mm/s (0.002 in./s), whereas the absolute velocity for planar front growth is approximately 50.0 mm/s (2 in./s). There are several similarities and differences between the low- and high-velocity cells: (1) The interface temperature in the low-velocity cellular regime increases with an increase in velocity, whereas it decreases with an increase in velocity in the high-velocity regime. (2) The primary spacing is found to decrease with velocity in both the regimes, and the spacing variation is found to be continuous between the two regimes. (3) The cells at low velocity grow in the direction of heat flow, while the growth of the high-velocity cells occurs along preferred crystallographic directions, which are the same as those for the dendrites.

Dendrite Growth Dendritic structures are classic examples of branching patterns in which the branched

Fig. 5

Cellular primary phase with the peritectic phase forming in the intercellular region

structure evolves with primary, secondary, tertiary, and eventually higher-order branches. Such a branched structure forms because it can effectively allow rapid dissipation of the latent heat of fusion as well as any solute redistribution required for growth due to the difference in compositions in the solid and the liquid. Dendritic structures most commonly form during the solidification of metallic systems. Three important aspects of dendritic structures that are examined here are dendrite growth direction, dendrite tip characteristics such as tip composition and tip temperature, and microsegregation that is controlled by the secondary arm spacing.

Dendrite Growth Direction Dendrites have been observed to grow in specific crystallographic direction. This crystallographic effect not only can form texture during the columnar growth, but it is also critical for the steady-state growth of dendrites and in the transition from chill to columnar crystals in a casting. In cubic crystals such as facecentered cubic and body-centered cubic structures, dendrites have been observed to grow in <100> directions. Chalmers (Ref 9) discussed that the dendrite growth direction will be along the axis of a pyramid whose sides are the most closely packed planes with which a pyramid

302 / Principles of Solidification can be formed. However, other growth directions have also been observed, particularly in aluminum-zine and aluminum-magnesium alloys, where the growth direction changes from <100> to <110> as the solute content is increased (Ref 10, 11). A more rigorous criterion for the dendrite growth direction has now been developed (Ref 12), showing that the growth direction at low velocities is controlled by the anisotropy in interfacial energy, while at very high velocities it is selected by the anisotropy in interface kinetics. For clarity, first consider a two-dimensional case in which the interface energy, g, in {100} planes, for materials with an underlying cubic symmetry, can be expressed as a function of the orientation as: gðyÞ ¼ g0 ð1 þ e4 cos 4yÞ

(Eq 7)

The anisotropy is characterized by only one parameter, e4, and g0 is the mean value of g. The anisotropy parameter, e4, can be measured experimentally from the equilibrium shape of a droplet in the solid; for the Al-4.0wt%Cu alloy, e4 = 0.0097 (Ref 13). The capillary undercooling for anisotropic interface energy is proportional
to the stiffness,  S, which is defined as S ¼ g þ @ 2 g=@n21 . The growth of perturbations will be preferred at the orientations where the stiffness is at a minimum. In this case, the polar plot of 1/S with orientation will show protrusions at the orientations that have the minimum stiffness, as illustrated in Fig. 6 for a cubic system with preferred <001> growth directions. The aforementioned model assumes fourfold symmetry and (100) dendrite growth in cubic crystals. In order to predict orientations with other symmetries, a three-dimensional model must be considered. In three dimensions, two anisotropy parameters, e1 and e2, are required. From phase-field calculations, Haxhimali et al. (Ref 12) have shown that the term with a positive value of e1 favors a (100) orientation, whereas the n term with a negative value of e2 favors a (110) orientation. Combinations of these values allow for various crystallographic growth directions. These predictions have been used to explain different orientations that are observed in the aluminum-zinc and aluminummagnesium alloys (Ref 10, 11). The effect of anisotropy in interface energy on the dendrite tip radius selection and on the dendrite tip undercooling is presented in the next section.

complex casting (Ref 14). (2) Relationships for the dendrite tip composition and the dendrite tip temperature with composition and processing conditions are predicted. The dendrite tip composition is important in determining microsegregation patterns, and the dendrite tip temperature is critical in predicting the selection of dendritic structures over other (such as a eutectic) morphologies. (3) The concentration field ahead of the dendrite is required to evaluate the region of constitutional supercooling that is important in the determination of the columnar-to-equiaxed dendrites. Since the dendritic pattern evolves to effectively dissipate heat or solute, dendrite growth models have been based on the solutions of the thermal and mass transport equations. For most metals, the Schmidt number (the ratio of thermal to solute diffusivity) is large, so that the formation of columnar and equiaxed dendrite in alloys is largely dictated by the solute transport process and the interfacial free energy. There are two important aspects of the dendrite growth model based on the transport processes that must be addressed. First, the steady-state dendrite growth problem is a freeboundary problem in which the shape of the interface is not known a priori, but it must be determined in a self-consistent manner such that this shape is preserved during growth. Second, if only thermal and mass transport processes were important for the formation of dendrites, the dendrite tip will keep splitting to form finer dendrites or nondendritic structures to obtain more efficient heat and solute dissipation into the liquid. Thus, the thermal and solute diffusion processes are not sufficient to describe the steady-state dendrite growth, and a stabilizing effect such as that of interfacial free energy must be included in the dendrite growth model. A steady-state growth model was first developed by Ivantsov (Ref 15) under the constraint of constant interface temperature and

composition. He showed that a parabolic dendrite satisfies the steady-state shape-preserving constraint in that the entire interface will not change shape during growth. However, the solution gave a relationship between the growth rate, tip radius, and tip composition (or the tip undercooling). Thus, for a known V, the model predicts an infinite number of solutions that relate tip undercooling to the tip radius. Experimental studies, however, show that a unique tip radius is selected under given growth conditions so that another criterion is required for the selection of a specific value of the dendrite tip radius. Furthermore, if only diffusion is considered, the dendrite tip will be unstable, since it can split and dissipate solute more effectively. The stabilizing effect of interface energy must be considered to obtain the stable tip radius value. However, the presence of isotropic interface energy effect will cause the composition to vary along the interface, and it is shown that no steady-state solution of dendrite is possible (Ref 16). However, a steady-state solution is recovered if anisotropy in interface energy is introduced in the model. This stability condition introduces a dimensionless parameter, s*, that relates the dendrite tip radius to velocity. The parameter s* is referred to as the dendrite tip selection parameter and is a function of the anisotropy parameter, e4. A functional relationship between s* and e4 has been obtained numerically by Karma and Rappel (Ref 16). The Ivantsov solution, along with the stability condition, predicts the dendrite tip temperature, dendrite tip composition, and dendrite tip radius as a function of velocity and alloy composition (Ref 17). Although the interface energy and its anisotropy are critical in the stable, steady-state growth of dendrites, the solutions of the transport equations are not influenced by these effects. Thus, the interface energy effect on tip undercooling is negligible at low velocities, and the dendrite tip undercooling is close to the solutal undercooling at the tip. In addition, the dendrite tip radius and dendrite tip temperature do not depend on the thermal gradient. The variation in dendrite tip composition and dendrite tip temperature with velocity has been calculated for Al-3.0wt%Cu by using the Ivantsov model with the dendrite tip stability condition; the results are shown in Fig. 7. For casting conditions where the velocity (or the Peclet number, VR/2D) is small, the

Dendrite Growth Model Dendrite growth models have been developed to correlate the effect of processing variables, such as velocity and thermal gradient, on dendrite microstructures. The aim of the model is to predict the following important aspects of dendrite growth. (1) The relationship between the velocity and dendrite tip undercooling is critical in correlating microstructure with macroscopic heat flow profiles in a

Fig. 6

Schematic illustration of the polar plot of the inverse of stiffness showing bumps at easy growth directions

Fig. 7

Temperature and composition at the dendrite tip as a function of velocity in Al-3.0wt%Cu

Non-Plane Front Solidification / 303 dendrite growth model is often simplified by approximating the Ivantsov solution to give a much simpler relationship for the dendrite tip undercooling as (Ref 17):



Tt ffi K V

by the final secondary arm spacing near the base of the dendrite, which is given by (Ref 20, 21): 1=3

n

l2 ¼ const: tf

where K and n are constants whose values depend on the range of Peclet number that is relevant to the problem. The value of n is of the order of 0.3 to 0.5 for small Peclet number conditions. Note that this form of relationship is valid for columnar dendrites and also for equiaxed crystal if the ratio of thermal to solute diffusivity is large.

Microsegregation and Secondary Arm Spacing The growth of dendritic structures is associated with solute segregation since the solid rejects solute during growth (for k < 1), which accumulates in the surrounding liquid. In columnar growth, the secondary arm spacing controls the microsegregation profile in the solidified alloy. In equiaxed crystals, the solute accumulation in the liquid will also result in solute segregation at the boundaries of neighboring grains, so that the grain size is important. The primary dendrite arm spacing, l, becomes finer with increase in velocity or thermal gradient in the liquid, and a general relationship is given by (Ref 18, Ref 19): h pffiffiffi i1=2 l ¼ 4 2D=s ½V GR
1=2

(Eq 10)

(Eq 8)

(Eq 9)

The variation in primary spacing with velocity for cellular and dendritic structure is shown in Fig. 8. There is an increase in primary spacing as the cells transform to dendrites. Recently, Lee et al. (Ref 8) have shown that a decrease in primary spacing is observed at the velocity where dendrites transform to high-velocity cells. For dendritic microstructures, secondary arm spacing is important since it governs the solute segregation pattern. The secondary arms originate close to the dendrite tip, and their initial spacing scales with the dendrite tip radius. However, the secondary spacing coarsens with time in the mushy zone. Initial coarsening occurs by the competition in the growth process among secondary arms. However, once the diffusion fields of the tips of the secondary branches interact with those of the neighboring dendrite, the growth of the secondary arms is reduced, and a coarsening process to reduce interfacial energy begins. The final secondary arm spacing near the dendrite base is significantly larger than that near the dendrite tip. This final secondary arm spacing controls the microsegregation profile in the solidified alloy. This microsegregation pattern is analogous to that discussed for the cellular structure, except that the periodicity of segregation is controlled

where tf is the local solidification time. For directional solidification, the local solidification time is inversely proportional to the cooling rate, which is a product of G and V, so that secondary arm spacing is generally related to the cooling rate, and this has been verified experimentally by Bower et al. (Ref 4) in the aluminum-copper system. The secondary spacing is the smallest length scale over which solute segregation occurs in dendritic structures, and it is correlated to the mechanical properties of single-phase alloys. In reality, these alloys are often not single phase but frequently contain small amounts of second phases, such as eutectic, precipitates, pores, and so on, that form due to microsegregation and control the properties.

Columnar-to-Equiaxed Transition Most practical metallurgical alloys solidify with a dendritic structure in a mold, and both columnar and equiaxed dendrites may be present in a casting (Fig. 9). As the liquid metal is poured into the mold, solid nuclei appear at the mold wall that give rise to a chilled zone. Some of these crystals are then favorably oriented for growth under directional heat flow conditions, and the selection of the grains occurs due to the anisotropy property of the interface energy. Initially, the chill crystals are oriented randomly. However, those crystals that are oriented with the preferred crystallographic direction in the heat flow direction will be selected. This is because the stiffness of the advancing front in the preferred growth direction is smaller, and thus, the capillary undercooling is smaller, so that they will grow ahead of the neighboring grains and crowd them out. These favorably oriented grains produce a columnar zone in a casting and exhibit [100] texture in most cubic metals. For alloy castings, an equiaxed zone can form ahead of the columnar crystal. This transition from the columnar to equiaxed zone is now examined.

Fig. 8

Effect of velocity on cellular, cellular dendrite, and primary dendrite spacings in the pivalic acid-ethanol system. Source: Ref 6

The basic concepts for the columnar-toequiaxed transition (CET) were first presented by Hunt (Ref 22). The transition from columnar to equiaxed grains occurs when enough large equiaxed grains form and grow ahead of the columnar front, hindering its movement. The size of the transition zone depends on the amount of competition between the columnar and equiaxed grains. To form an equiaxed zone, there are two requirements: the presence of nuclei ahead of the columnar front, and sufficiently fast growth of these nuclei to effectively block the columnar front. If either of these conditions are not met, then columnar grains will still predominate. One of the prerequisites for the formation and growth of nuclei is the presence of a constitutionally supercooled region ahead of the columnar front. Since the stable dendrite tip growth requires a balance between the constitutional supercooling and curvature effects, a constitutionally undercooled zone is always present ahead of the dendrite tip. Equiaxed growth can occur if the required nucleation undercooling is smaller than the undercooling present ahead of the columnar dendrites, as shown in Fig. 10. The length of the supercooled region, LT, ahead of the columnar front, shown in Fig. 10, is given by LT = DTt/G. This characteristic length provides the distance over which equiaxed nuclei can grow for the limiting case of zero nucleation undercooling. The second important parameter is the distance between the nuclei, LN, which forms the equiaxed zone. Once the nucleation undercooling is reached, nuclei begin to form, and the nucleation rate increases rapidly as the undercooling increases slightly. Thus, it is reasonable to assume that all effective nucleation sites are activated close to the nucleation undercooling. If N0 is the number of nuclei per unit volume, then
1=3 LN ¼ N0 . Once the nucleation undercooling is reached, the nuclei will form and grow as they approach

Fig. 9

Grain structures of an ingot casting showing the regions of chill, columnar, and equiaxed grains

304 / Principles of Solidification

Fig. 11

Fig. 10

Formation of equiaxed crystals ahead of the columnar front, and the length of the constitutionally supercooled liquid ahead of the columnar front

the columnar front. The increase in the radius of the grain with time (or undercooling) is obtained from Eq 8 for the undercooling changing from DTN to DTt. From the volume of the equiaxed grains at the columnar front and the number of nuclei per unit volume, the fraction of the equiaxed grains present at the columnar front can be determined. Hunt (Ref 22) proposed that equiaxed crystals will block the columnar front when the fraction of the equiaxed grains at the columnar front is 0.49. Since the nucleation sites are arranged randomly, a correction for the impingement of the equiaxed grains can be made by using the extended volume fraction concept, which requires that the extended volume fraction of the equiaxed grains be larger than 0.66 for the transition from columnar to equiaxed grains. Numerical calculations of CET can be obtained by calculating the columnar dendrite tip undercooling by using a more rigorous dendrite growth model (Ref 16). First, a simple picture of the condition for the CET is obtained, and then the results from the rigorous model are shown. The CET will occur if the time required for the nuclei to obtain required volume fraction to block the columnar front is less than the time available to the nuclei in the constitutionally supercooled zone. These times can be correlated with the characteristic distances. For a limiting case of very small nucleation undercooling, that is, DTN << DTi, CET occurs when the distance between the nuclei is less than the length of the constitutionally supercooled zone, that is, LN < LT. Substituting the

Numerical calculations of the columnar-to-equiaxed transition in Al-3.0wt%Cu alloy

values of these lengths, the CET condition 1=3 becomes G < N0 Tt . Thus, to promote CET, one needs a low thermal gradient, large number of nuclei, and a large dendrite tip undercooling that is obtained at higher velocity (Fig. 7). By using the general dendrite growth model to calculate the tip undercooling, Ga¨umann et al. (Ref 23) have numerically calculated the CET condition, shown on a G-V plot for different nucleation undercooling values, as shown in Fig. 11. Note that anisotropy in interface energy also influences CET, since a larger anisotropy parameter will sharpen the tip radius and reduce the tip undercooling at a given velocity. This will stabilize the columnar region, and the CET curve in Fig. 11 will move upward as the anisotropy parameter is increased. One of the ways to enhance the equiaxed zone is to increase the density of nuclei by the addition of effective grain refiners or inoculants. In casting and welding, the nuclei often originate from the detachment of the secondary arms close to the columnar dendrite tip region due to the fluctuating thermal field caused by convection. These fragments grow at the slightest undercooling; therefore, they are highly effective in producing CET. Fragmentation is often induced by stirring the liquid in the mold to obtain castings with equiaxed grains (Ref 24).

Nonplanar Eutectic Growth Eutectic alloys are important for casting because of their good fluidity caused by the small freezing range and low melting temperatures. The fine-scale composite microstructure also gives them superior mechanical properties. Thus, theoretical models have been developed to correlate eutectic spacing with velocity (Ref 25, 26), and very fine spacing of the order of 10 nm has been observed in aluminum-copper alloys at a velocity of 0.2 m/s (0.7 ft/s)

(Ref 27). A eutectic alloy in a binary system grows with a planar interface. However, it is observed that at higher velocities, a planar two-phase interface becomes unstable, and two-phase cellular and dendritic microstructures can form (Ref 28, 29). These nonplanar eutectic fronts in a model transparent system of carbon tetrabromide and hexachlorethane, observed by Han (Ref 29), are shown in Fig. 12, which shows the formation of planarto-cellular-to-dendritic two-phase eutectic as the velocity is increased. Two types of eutectic cells are observed: one in which the anisotropy effects are small (Fig. 12b), and the other (Fig. 12c), in which the anisotropy effects are significant. The eutectic cell is generally referred to as the colony formation. Detailed analytical (Ref 30) and numerical (Ref 31) models have been developed to examine the long wavelength instability of a planar eutectic front in binary alloys. They have clearly shown that a large-scale morphological instability of a two-phase planar eutectic front does not occur in binary alloys. The instability of a planar two-phase interface is attributed to the presence of a small amount of impurity (Ref 32). As in the case of a morphological instability of a planar single phase, the instability of a planar eutectic interface is due to the rejection of the third component by both of the solid phases that build up a boundary layer of the ternary impurity. When the solute buildup increases to the point where the concentration gradient of the third element produces a constitutionally supercooled region, the planar eutectic interface becomes unstable and forms two-phase cells, which then transform to two-phase dendrites as the velocity is increased. Since most binary alloys contain some impurity, a nonplanar eutectic will form when the velocity exceeds some critical value. Thus, colony formation is common in many rapidly solidified eutectics.

Non-Plane Front Solidification / 305 further increase in velocity, the eutectic spacing in the colony decreases. Furthermore, the spacing between the colonies is found to decrease with an increase in velocity, and the colony spacing is found to be approximately an order of magnitude larger than the eutectic spacing. However, as the planar interface becomes unstable, the interface temperature, the eutectic spacing at the center of the colonies, and the colony spacing are all found to increase initially, reach a maximum, and then decrease as the cellular structure transforms into a two-phase dendritic structure. Experimental results are analyzed by using appropriate theoretical models on cellular, dendritic, and eutectic growth.

Summary

Fig. 12

Typical microstructures observed in the carbon tetrabromide-hexachlorethane system of eutectic composition as a function of velocity. (a) At low velocity, below 1.0 mm/s, a macroscopically planar eutectic interface is observed. (b) Cellular two-phase interface forms when the velocity is increased. (c) Another type of cellular morphology forms depending on the orientation and the anisotropy in interface energy. (d) At higher velocities, a dendritic two-phase eutectic is observed. Source:Ref 29

G mc C0 ðkc
1Þ  V kc D

Fig. 13

(a) Variation in colony tip temperature with velocity. (b) Variation in the lamellar spacing at the center of the colony with velocity. An increase in spacing is observed when the planar interface becomes cellular eutectic.

The model for the onset of instability has been based on the single-phase instability, which predicts that the eutectic interface will be unstable when:

(Eq 11)

where C0 is the concentration of the impurity, and mc is the slope of the liquidus with the impurity content. The solute distribution coefficient is taken to be the weighted average value from both the solid phases, that is, kc = b la kac + lb kbc c/l, in which l is the eutectic spacing, and la and lb are the width of the a- and b-lamella. The distribution coefficients of the solute for the a- and b-phases are kac and kbc, respectively. The variation of the interface temperature with velocity is predicted by using the analogy with the model for the single-phase cell and dendrite (Ref 33). For a planar eutectic interface, the interface temperature decreases with velocity until a critical velocity is reached where the interface becomes unstable. The interface temperature increases as 1/V in the cellular colony regime, and then decreases when the cellular colony transforms to a dendritic colony. Quantitative measurements on the variation of interface temperature, eutectic spacing, and colony spacing with velocity were carried out by Han (Ref 29) over the two-phase planar, cellular, and dendritic structures. The results are shown in Fig. 13(a). The eutectic spacing varies within a colony, and it increases near the boundary of the colony due to the curvature of the front, so that one characterizes the spacing at the center of the colony. The variation of eutectic spacing with velocity is shown in Fig. 13(b). It decreases with an increase in velocity in the planar front region, but then it increases discontinuously as the planar front becomes nonplanar. With

Nonplanar microstructures form most frequently during the solidification of alloys, and they play a crucial role in governing the properties of the solidified material. The formation of cellular and dendritic structures in one- and two-phase structures is presented with emphasis on the effect of processing conditions and composition on the selection of microstructure and microstructure scales. A general microstructure map is shown that correlates processing conditions of G and V, or the cooling rate, in different solidification technologies. The basic ideas and the processing conditions required to control the columnar and equiaxed microstructures are also emphasized. ACKNOWLEDGMENTS This work was carried out at Ames Laboratory, which is operated by Iowa State University for the Office of Basic Energy Science, Division of Materials Science, U.S. Department of Energy under Contract No. W7405-Eng-82. REFERENCES 1. B. Billia and R. Trivedi, Pattern Formation in Crystal Growth, Chap. 13, Handbook of Crystal Growth, Vol 1, D.T.J. Hurle, Ed., Pergamon Press, 1993, p 899 2. J.D. Hunt and S.-Z. Lu, Numerical Modeling of Cellular/Dendritic Array Growth: Spacing and Structure Predictions, Metall. Mater. Trans. A, Vol 27, 1996, p 611 3. R. Trivedi, Y.X. Shen, and S. Liu, Cell to Dendrite Transition in Directional Solidification of Alloys, Mater. Trans. A, Vol 34, 2003, p 395 4. T.F. Bower, H.D. Brody, and M.C. Flemings, Measurements of Solute Redistribution in Dendritic Solidification, Trans. TMS-AIME, Vol 236, 1966, p 624 5. R. Trivedi, Y.X. Shen, and S. Liu, A Unique Correlation among Cellular Microstructural Length Scales in Directionally Solidified Binary Systems, Proc. Int. Conf. on Advanced Materials and Processing,

306 / Principles of Solidification

6.

7.

8.

9. 10. 11. 12. 13.

14. 15.

N. Chakraborty and U. Chatterjee, Ed., Tata McGraw-Hill, New Delhi, 2001, p 42 T.W. Clyne and W. Kurz, Solute Redistribution during Solidification with Rapid Solid State Diffusion, Metall. Trans. A, Vol 12, 1981, p 965 R. Trivedi, F. Jin, and I.E. Anderson, Dynamical Evolution of Microstructure in Finely Atomized Droplets of Al-Si Alloy, Acta Mater., Vol 51, 2003, p 289 J.H. Lee, H.C. Kim, C.Y. Jo, S.K. Kim, J.H. Shin, S. Liu, and R. Trivedi, Microstructure Evolution in Directionally Solidified Fe-18Cr Stainless Steels, Mater. Sci. Eng. A, Vol 413–414, 2005, p 30 B. Chalmers, Principles of Solidification, John Wiley and Sons, New York, 1964 F. Gonzales and M. Rappaz, Dendrite Growth Direction in Aluminum Zinc Alloys, Metall. Mater. Trans. A, Vol 37, 2006, p 2797 E. Sunseri, M.S. thesis, Iowa State University, 2008 T. Haxhimali, A. Karma, F. Gonzales, and M. Rappaz, Orientation Selection in Dendritic Evolution, Nature Mater., Vol 5, 2006 S. Liu, S.R. Napolitano, and R. Trivedi, Measurement of Anisotropy of CrystalMelt Interfacial Energy for Binary Al-Cu Alloy, Acta Mater., Vol 49, 2001, p 42, 71 M. Rappaz, Modeling of Microstructure Formation in Solidification Processes, Int. Mater. Rev., Vol 34, 1989, p 93 G.P. Ivantsov, Temperature Field Around Spherical, Cylindrical and Needleshaped

16.

17. 18. 19. 20.

21.

22. 23.

24.

Crystals Which Grow in Supercooled Melt, Dokl. Akad. Nauk SSSR, Vol 58, 1947, p 567 A. Karma and W.J. Rappel, Quantitative Phase-Field Modelling of Dendritic Growth in Two and Three Dimensions, Phys. Rev. E, Vol 57, 1998, p 4323 R. Trivedi and W. Kurz, Dendritic Growth, Int. Mater. Rev., Vol 34, 1990, p 93 J.D. Hunt, Primary Dendrite Spacing, Solidification and Casting of Metals, Book 192, The Metals Society, London, 1979, p 3 R. Trivedi, Interdendritic Spacing, Part II: A Comparison of Theory and Experiments, Metall. Trans. A, Vol 15, 1984, p 977 T.Z. Kattamis and M.C. Flemings, Dendrite Morphology, Microsegregation, and Homogenization of Low Alloy Steel, Trans. TMS-AIME, Vol 233, 1965, p 992 U. Feurer and R. Wunderlin, Einfluss der Zusammensetzung und der ErstarrungsBedingungen auf die Dendritenmorphologie Binarer Al-Legierungen, Fachbericht DGM, Oberursel, Germany, 1977 J.D. Hunt, Steady State Columnar and Equiaxed Growth of Dendrites and Eutectic, Mater. Sci. Eng., Vol 65, 1984, p 75 M. Ga¨umann, R. Trivedi, and W. Kurz, Nucleation Ahead of the Advancing Interface in Directional Solidification, Mater. Sci. Eng. A, Vol 226–228, 1997, p 763 S. Kobayashi, H. Tomono, and K. Toda, Ame´lioration de la Qualite´ des Brames de Coule´e Continue Graˆce au Brassage E´lectromagne´tique par Courant de Conduction,

25. 26.

27.

28. 29.

30. 31. 32. 33.

Rev. Me´tall., Cah. Int. Tech., Vol 80, 1983, p 887 K.A. Jackson and J.D. Hunt, Lamellar and Rod Eutectic Growth, Trans. TMS-AIME, Vol 236, 1966, p 1129 R. Trivedi and W. Kurz, Microstructure Selection in Eutectic Alloy Systems, Solidification Processing of Eutectic Alloys, D.M. Stefanescu, G.J. Abbaschian, and R. J. Bayuzik, Ed., Metallurgical Society of AIME, 1988, p 3–34 M. Zimmerman, M. Carrad, and W. Kurz, Rapid Solidification of Al-Cu Eutectic Alloy by Laser Remelting, Acta Metall., Vol 37, 1989, p 3305 H.W. Weart and D.J. Mack, Eutectic Solidification Structures, Trans. Metall. Soc. AIME, Vol 212, 1958, p 664 S.H. Han, “Stability of a Eutectic Interface during Directional Solidification,” Ph.D. thesis, Iowa State University, Ames, IA, 1995 V. Datye and J.S. Langer, Stability of Thin Film Lamellar Eutectic Growth, Phys. Rev. B, Vol 24, 1981, p 4155 A. Karma and A. Sarkissian, Morphological Instabilities of Lamellar Eutectics, Metall. Trans. A, Vol 27, 1996, p 635 W.A. Tiller, Liquid Metals and Solidification, American Society of Metals, 1958 D.G. McCartney, J.D. Hunt, and R.M. Jordan, The Structures Expected in a Simple Ternary Eutectic System, Part 1: Theory, Metall. Trans. A, Vol 11, 1980, p 1243

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 307-311 DOI: 10.1361/asmhba0005211

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Eutectic Solidification—An Introduction Sumanth Shankar, McMaster University

METALLIC EUTECTIC SYSTEMS have been used for several centuries to cast engineering components. The use of eutectic systems predates the understanding of the solidification of such systems. Systems wherein two pure metals can be mixed to specifically yield a low-melting alloy have enabled ease of casting near-net shape components of high strength and performance. Rapid developments in our understanding of the solidification of such systems over the second half of the 20th century have enabled improvements in the processing of eutectic metallic systems. Presently, commercially important metallic alloy systems used for casting engineering components are predominantly eutectic systems. Aluminumsilicon, aluminum-copper, and iron-carbon systems are typical examples.

Binary Eutectic Phase Diagram Figure 1 shows a schematic of a typical phase diagram of a binary eutectic system between two components, A and B. In Fig. 1, there is no solubility of B in A or vice versa. This type of eutectic system is typical of nonmetallic systems such as salts and organic alloys. The LiBr-LiF and NaF-RbF systems are typical examples of such systems. However, metallic eutectic systems seldom exhibit the behavior shown by Fig. 1. Figure 2 shows a schematic of a typical phase diagram of a binary metallic eutectic system. In Fig. 2, there is a solid solubility of component B in A (designated as a-phase) and a solid solubility of component A in B (designated

as b-phase). Although A and B have high melting points individually, the alloy composition denoted by point “E” in Fig. 2 has a considerably lower melting point than components A or B. Alloys with composition close to point “E” will have a short freezing range and a low melting point. Lead-tin, aluminum-silicon, aluminum-copper, iron-carbon (Fe-Fe3C), Silvergermanium, bismuth-lead, and Zn-Mg2Zn11 are typical examples of binary metallic eutectic systems exhibiting the behavior shown in Fig. 2. In Fig. 2, the lines CE and CF are the liquidus and solidus lines for the alloys, respectively, which solidify to form solid-solution a. Similarly, lines DE and DG are the liquidus and solidus lines for the alloys, respectively, which solidify to form the solid-solution b. Intermediately, consider the liquid of composition CE, which will cool until temperature TE, at which point the liquid will be a part of two liquidus lines, CE and DE, and be in equilibrium with two solid solutions of compositions Ca and Cb. Both of these solid solutions will form at temperature TE from the liquid composition CE. The combination of these solid solutions is known as the eutectic. The eutectic is a mixture of two solid solutions wherein the two solid phases intimately coexist. However, the distribution of these two solid phases in the eutectic structure will vary vastly among various binary metallic eutectic systems and will solely depend on the nature of the components that form the system. To further understand the various structures that evolve in a binary eutectic system during solidification, it is useful

to examine the various classifications of eutectic structures and relate them to principles of solidification.

Eutectic Structures For the solids with Ca and Cb to evolve from the liquid at temperature TE in Fig. 2, there must be a nucleation event and subsequent growth of the solid phases to form the final solidified eutectic structure. From studies of over 60 binary metallic eutectic systems containing solid solutions a and b (Ref 1–4), researchers have shown that if solid-solution a can be nucleated by b, then solid-solution b cannot be nucleated by a. For example, in the system shown in Fig. 2, assume that a is nucleated by b and b cannot be nucleated by a. For an alloy with a composition of B greater than CE (hypereutectic alloys), the eutectic b-phase will nucleate or grow epitaxially on the primary b-phase at the eutectic temperature, TE. Subsequently, the eutectic a-phase will nucleate on the b-phase with minimal undercooling to initiate the growth of the eutectic structure. However, in an alloy with a composition of B less than CE (hypoeutectic), the eutectic a-phase will initially nucleate or grow epitaxially on the primary a-phase; however, the eutectic b-phase cannot nucleate on the a-phase and hence will require a nucleating agent in the form of an impurity phase in the alloy or a substantially large undercooling to nucleate.

Classification of Eutectic Structures The most widely accepted classification of the eutectic structures was first proposed by Sheil (Ref 5, 6) and was later modified and elaborated by Jackson et al. (Ref 7–9). Jackson et al. (Ref 7–9) introduced a parameter, a, to explain the entropy of melting: a¼


Fig. 1

Schematic of a typical simple binary eutectic system with no solubility of component A in B or vice versa

Fig. 2

Schematic of phase diagram of a typical metallic binary eutectic system



SF R

(Eq 1)

where (DSF/R) is the entropy of melting in dimensionless units, and x is a crystallographic factor that is less than and almost unity. Solids with high entropy of melting will have a > 2, and these solids grow with crystalline facets (atomically

308 / Principles of Solidification smooth interface). Solids with low entropy of melting will have a < 2, and these grow isotropically with no facets (atomically rough interface). Eutectic structures can be broadly classified into predictable morphologies based on the value of a for the two solid solutions, respectively. Further, Scheil (Ref 5, 6) proposed that the volume fraction of solute phase in the eutectic structure will also influence the morphology of the eutectic structure. Hence, the morphology of the eutectic structure in a binary metallic system can be broadly classified as shown in Fig. 3. In Fig. 3, the regular or normal eutectic structure, such as the rodlike and the lamellar, will form when the two phases have low entropies of fusion (a < 2) (nonfaceted/ nonfaceted-type growth). The rodlike structure typically occurs when the volume fraction, VF, of the minor phase is less than approximately 30% (Fig. 3a), and the lamellar structure typically occurs when VF of the minor phase is greater than approximately 30% (Fig. 3b). The irregular or anomalous eutectic structure forms when one of the solid phases has high entropy of fusion (a > 2) (nonfaceted/faceted-type growth). In the nonfaceted/faceted-type growth, when VF < 30%, the faceted rodlike or spherical eutectic phases will typically occur (Fig. 3c), and when VF > 30%, the faceted branched (irregular) lamellar or the acicular structure will typically occur (Fig. 3d). Metallic systems very rarely exhibit a eutectic system wherein the two solids have high entropy of fusion (faceted/ faceted-type growth). These are typically found in nonmetallic systems. Unlike regular eutectic structures that are either rodlike or lamellar, the anomalous eutectic structures have a wide variety of morphology, depending on the system and the solidification conditions. Croker et al. (Ref 10) further refined the classification of the eutectic structures, as shown in Fig. 4. In Fig. 4, 5.5 cal mol1 K1 (23 J mol1 K1) was evaluated as the critical value to differentiate between normal eutectic structures, such as lamellar or rodlike







(two nonfaceted solids), and anomalous eutectic structures (nonfaceted/faceted). Moreover, Fig. 4 was drawn for various eutectic structures solidified at a growth rate of approximately 5  104 cm s1 (2  104 in s1). Figure 5 shows a comprehensive map of the location of the eutectic structures of various binary metallic alloy systems solidified at a growth rate of 5  104 cm s1 (2  104 in s1). In all of these alloys, the first eutectic solid forms by heterogeneous nucleation, and the second solid nucleates contiguously and heterogeneously but may not necessarily nucleate on the first solid phase. In Fig. 5, regions marked 1 and 2 are termed normal or regular eutectic structures and constitute structures formed with two solids growing without crystalline facets. At high volume fraction (VF) of over 32%, a lamellar eutectic structure dominates, as shown in region 1; however, at low VF (<32%), the rodlike eutectic structure dominates. Regions 3 to 6 represent anomalous eutectic structures obtained by one of the solids having a faceted structure and the other a nonfaceted structure. Region 3 occurs at VF  10%, wherein the









structure is that of broken lamellar. As the VF increases, the broken lamellar structure transitions to an anomalous irregular flake structure (region 4). As VF increases to greater than approximately 20%, the irregular structure transitions into a complex regular structure (region 5) that consists of large regions of regular eutectic structure that are cellular in nature. In between the complex cellular regular structure, the anomalous irregular structure exists. As VF increases to higher values, the VF of the irregular structure diminishes. Further increase in the value of VF leads to region 6, wherein the morphology of the eutectic structure is termed quasi-regular. Typically, fibrous morphology coupled with platelike morphology, with varying proportions of each, is characteristic of the quasi-regular structure. The solid phase with the high entropy value forms the matrix of this eutectic structure. The structure in this region appears to be more regular than the anomalous structures with lower values of VF because of the reduced tendency of overgrowth exhibited by the higher VF of the faceted phase. Eutectic structures in region 6 are very sensitive to



Fig. 3

Schematic of the four broad categories of regular eutectic structures. (a) Regular rodlike (volume fraction, VF, <30%; entropy of fusion, a, <2). (b) Regular lamellar (VF > 30% and a < 2). (c) Faceted rods or spherical (VF < 30% and a > 2). (d) Faceted branched lamellar or acicular (VF > 30% and a > 2). The growth direction in all of these schematics is perpendicular to the plane of the paper.

Fig. 4

General classification of eutectic structures as a function of the entropy of fusion (DSF) and the volume fraction (VF). There are six regions of classification based on the morphology of the eutectics. These are typical microstructures corresponding to the six regions: 1, Pb-AuPb2 eutectic; 2, Cd-Pb eutectic; 3, Ag-Bi eutectic; 4, CdSbCd eutectic; 5, Bi-Pb2Bi eutectic; and 6, Bi-Cd eutectic. The eutectic growth directions in all of these microstructures are perpendicular to the plane of the paper, and the growth rate is approximately 5  104 cm s1 (2  104 in s1). Source: Ref 10





Eutectic Solidification—An Introduction / 309

Fig. 5

Classification of eutectic structures into six distinct regions as a function of entropy of fusion (DSF) and volume fraction (VF). The structures obtained were for a growth rate of 5  104 cm s1 (2  104 in s1). Structures left of the vertical line at 5.5 cal mol1 K1 (23 J mol1 K1) are formed by nonfaceted/nonfaceted-type growth, and those to the right of this line are formed by nonfaceted/faceted-type growth. Region 1, predominantly regular lamellar (□); 2, regular rodlike (); 3, broken lamellar and a few fibrous (▲); 4, irregular morphologies (); 5, complex regular structures with contours showing the volume fraction regular cells ( ); and 6, quasi-regular structures ($). Source: Ref 10













▪

310 / Principles of Solidification growth conditions. Ledeburite (white iron eutectic of the Fe3C-Fe system) is a typical example of the quasi-regular structure. The classification of the eutectic structures mentioned in this system is intended to present a general and broad overview of the types of morphologies of various systems, depending on the entropy of melting and VF of the two solid phases. It must be noted that the variations in factors such as nucleation, solidification rate, levels of impurities, surface energies, and crystallographic relationships can significantly influence and alter the morphology of eutectic phases in a given binary alloy system. Nucleation plays a significant role in defining the morphology of eutectic structures. In nearly all cases of binary eutectic reactions, nucleation of the eutectic phases occurs by a heterogeneous mode, since the homogeneous mode is seldom viable in freezing alloys. Hence, the inoculant impurity levels, distribution in melt, and their potency are critical in defining the undercooling required for nucleation. The potency of nucleation by an impurity inoculant is mainly controlled by the surface energies (Ref 4). If the energy of the nucleus/liquid surface is higher than the energy of the nucleated phase/liquid surface, heterogeneous nucleation takes place. Further, if the second solid eutectic phase easily nucleates on the first phase, the resultant structure is typically that of a normal lamellar or rodlike eutectic. Alternatively, if the two eutectic solid phases nucleate independently on separate inoculant phases, the resultant structure is typically anomalous.

subsequent solidification of a- and b-phases together. The same mechanism occurs in alloy 3, wherein the a-phase separates out of the supersaturated liquid to freeze around the primary b-phase (forming a halo) to initiate simultaneous evolution of a- and b-phases from the eutectic liquid. In alloy 2, the b-phase cannot nucleate until point “B2” is reached, wherein the liquid is at a temperature in which the a-phase can nucleate. Subsequently, the evolution of b and a renders the liquid to eutectic composition for the simultaneous evolution of both the eutectic phases. In alloy 2, formation of a halo is unnecessary. The size of the halo is proportional to the distances ED. Large halos are observed in alloys (Fig. 7a) requiring high undercooling for nucleation and have a small slope of the metastable liquidus line; insignificant halos are observed in alloys (Fig. 7b) requiring small undercooling for nucleation and have a large slope for the metastable liquidus line (Ref 3). This discussion explaining the observed morphologies of eutectic structures in metallic binary systems is a general overview based on

thermodynamic considerations alone. There are several researchers (Ref 9, 11–15) who have successfully predicted the growth and structure of regular lamellar or rodlike eutectics, wherein the entropy of melting is less than 5.5 cal mol1 K1 (23 J mol1 K1) (Fig. 5). Constitutive relationships among growth rate, n, eutectic spacing, l, and degree of undercooling, DT, have been developed. The general relationships among them are:





 2 nl ¼ Constant





(Eq 2)

and

T pffiffiffi ¼ Constant n

(Eq 3)

Various theories have been proposed to explain these relationships, and several rigorous mathematical and phenomenological models have been developed to explain the various features of the regular lamellar and rodlike eutectic structures (Ref 9, 13–16). However, the

Formation of Halos In most eutectic structures, the minor solid phase typically solidifies around the primary solid phase, forming a halo. Halos are observed in eutectic microstructures of most alloy systems. In some, they are visually prominent, and in others they are insignificant. Sundquist et al. (Ref 3) explained the formation of these halos based on nucleation of the phases in a solidifying eutectic alloy. Figure 6 shows a schematic of a typical binary eutectic equilibrium phase diagram with the metastable line extensions of the two liquidus lines. Halo formation can be understood by examining the solidification of the three alloys marked 1, 2, and 3. In alloy 1, the nucleation of the primary phase, a, takes place at point “B1”, after sufficient undercooling from point “A1”. Immediately, the liquid composition moves to point “L1”, and further cooling eventually renders the liquid at point “E”. Nucleation of b-phase requires an undercooling as well and takes place at point “D1”. During cooling between points “E” and “D1,” additional growth of the a-phase occurs along the metastable liquidus line, EC8. The b-phase can be nucleated by a or some other impurity in the melt. The liquid at point “D1” is supersaturated with b-phase, and the excess b has to solidify from the liquid onto the primary a-phase (forming a halo of b on primary a) to render the remaining liquid back to the eutectic composition for

Fig. 6

Schematic of eutectic phase diagram showing the solidification sequence of alloys 1, 2, and 3. Points “A1,” “A2,” and “A3” are equilibrium freezing points for alloys 1, 2, and 3, respectively. Point “E” is the eutectic point. Point “B” is the actual temperature of initiation of solidification of primary a-phase. Line EC is the metastable liquidus line. Point “D” is the nucleation point of the second solid phase. Line DE is the excess of the secondary phase. The subscripts in all of the points denote the respective alloys 1, 2, and 3.

Fig. 7

Typical halos found in eutectic structures. (a) Ag-36 wt% Cu showing primary copper (black) with silver halo. (b) Al-35 wt% Cu showing eutectic radiating from primary CuAl2. Source: Ref 3

Eutectic Solidification—An Introduction / 311 predictability of the structure of eutectic growth produced by a pair of nonfaceted/faceted solid solutions is still an unsolved problem. There have been many theories proposed to explain the anomalous eutectic structures, but it has now come to be widely accepted that the influence of extraneous factors such as solidification rates, impurity levels, and crystallographic factors of solid phases play a significant role in determining these structures. It is best that these structures be explained on a case-by-case basis for each binary eutectic alloy and processing condition. Two of the most significant alloy systems for commercial engineering applications are the aluminum-silicon and the iron-carbon (Fe-Fe3C) systems. A detailed review of the evolution of eutectic structures in these systems is presented in subsequent articles in this Volume.

REFERENCES 1. E. Scheil and R. Zimmermann, Z. Metallkd., Vol 48, 1957, p 509 2. B.E. Sundquist and L.F. Moldolfo, Trans. Metall. Soc. AIME, Vol 221, 1961, p 157 3. B.E. Sundquist, R. Bruscato, and L.F. Moldolfo, J. Inst. Met., Vol 91, 1962–1963, p 204 4. L.F. Moldolfo, J. Aust. Inst. Met., Vol 10, 1965, p 169 5. E. Scheil, Z. Metallkd., Vol 37, 1946, p 1 6. E. Scheil, Z. Metallkd., Vol 45, 1954, p 298 7. K.A. Jackson, Liquid Metals and Solidification, American Society for Metals, 1958, p 174 8. K.A. Jackson, Growth and Perfection of Crystals, R.H. Doremus, D. Turnbull, and B.W. Roberts, Ed., John Wiley and Sons, New York, 1958, p 319

9. J.D. Hunt and K.A. Jackson, Trans. Metall. Soc. AIME, Vol 236, June 1966, p 843 10. M.N. Croker, R.S. Fidler, and R.W. Smith, Proc. R. Soc. (London) A, Math. Phys. Sci., Vol 335 (No. 1600, Oct 2, 1973, p 15 11. C. Zener, Trans. AIME, Vol 167, 1946, p 550 12. W.A. Tiller, Liquid Metals and Solidification, American Society for Metals, 1958 13. M.C. Flemings, Solidification Processing, McGraw-Hill, 1974 14. D.A. Porter and K.E. Easterling, Phase Transformation of Metals and Alloys, 2nd ed., CRC Press, Boca Raton, FL, 2004 15. W. Kurz and J.D. Fischer, Fundamentals of Solidification, Trans Tech Publications, Clausthol-Zellerfield, Germany, 1984 16. B. Chalmers, Principles of Solidification, John Wiley and Sons, New York, 1964

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 312-316 DOI: 10.1361/asmhba0005212

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Solidification of Eutectic Alloys— Aluminum-Silicon M.M. Makhlouf, Worcester Polytechnic Institute

Figure 1 shows the aluminum-silicon equilibrium phase diagram (Ref 5). It is a simple binary eutectic with limited solubility of aluminum in silicon and limited solubility of silicon in aluminum. The solubility of silicon in aluminum reaches a maximum 1.5 at.% at the eutectic temperature, and the solubility of silicon increases with temperature to 0.016 at.% at 1190
C (2175
F). There is only one invariant reaction in this diagram, namely Eq 1, where L is the liquid phase, a is predominantly aluminum, and z is predominantly silicon. It is now widely accepted that this eutectic reaction occurs at 577
C (1070
F) and at a silicon concentration of 12.6 wt%: L!þ

(Eq 1)

It is interesting to note that in 1926, Gwyer and Phillips (Ref 6) determined the composition and temperature of the eutectic reaction in the aluminum-silicon binary system to be 11.7 wt% Si and 577
C (1070
F), respectively. Although the eutectic temperature is in accord with the currently accepted value, the eutectic þ 0.1 composition was later changed to 12.2  at.% Si (Ref 7–9). The initial error in establishing the eutectic point is due to the fact that the temperature of formation of the primary constituents (aluminum and silicon), as well as the temperature of the aluminum-silicon eutectic plateau, is cooling-rate dependent. Primary silicon undercools more than primary aluminum;

0

10

20

30

40

Equilibrium Metastable extensions T0 curves

1400

1200

Depression of the eutectic temperature with increased cooling rate may also be explained on the basis of the coupled region effect (Ref 12). Coupled regions represent fields within the phase diagram where the two phases of the eutectic structure are organized in the solid in such a way that diffusion in the liquid occurs effectively at a duplex solid/liquid front (Ref 8, 12–15). Regions of coupled growth are shown schematically in Fig. 3 for both a symmetric and an asymmetric hypothetical phase diagram. In the aluminum-silicon system, silicon is a nonmetal with directed covalent bonds; therefore, it tends to grow anisotropically into faceted crystals. Hence, it requires more undercooling for its growth than the isotropic

Silicon, wt% 60 50

70

80

90

100 2500

L 2000

1000 1500

800

600 1000

Temperature, ⬚F

The Aluminum-Silicon Equilibrium Phase Diagram

hence, the eutectic structure forms at 10 to 12
C (18 to 22
F) below the eutectic temperature without appreciable recalescence (Ref 10). Consequently, at higher cooling rates, the system behaves as though the eutectic point is shifted to higher silicon contents, and the eutectic temperature is depressed. Figure 2 illustrates the effect of this apparent shift in the eutectic point on the microstructure of a typical aluminum-silicon alloy. Figure 2(a) depicts the microstructure of a eutectic aluminum-silicon alloy that is slowly cooled and shows no primary aluminum; on the other hand, Fig. 2(b) depicts the microstructure of the same alloy cooled at a significantly faster rate and shows primary aluminum dendrites and finer silicon particles.

Temperature, ⬚C

THE ALUMINUM-SILICON EUTECTIC REACTION has been the subject of intense interest ever since Pacz introduced aluminumsilicon alloys to the United States in 1921 (Ref 1). In the years that followed, there was an explosion of ideas concerning this technologically important transformation, and much of the current understanding of the reaction may be traced back to the work of Mondolfo (Ref 2) and to the more recent work of Lu and Hellawell(Ref 3, 4).

400 500 200

(Si)

0

0

-200 0 Al

Fig. 1

10

20

30

40

50 Silicon, at.%

60

70

80

90

100 Si

Equilibrium phase diagram for the aluminum-silicon system showing metastable extensions of the liquidus and solidus lines. Source: Ref 5

Solidification of Eutectic Alloys—Aluminum-Silicon / 313

Fig. 2

(a)

Fig. 3

Al-12.5wt%Si alloy (a) slowly cooled and (b) chill cast. Source: Ref 11

(b) Coupled zones in (a) symmetrical and (b) asymmetrical phase diagrams

aluminum phase. Consequently, the coupled region in the aluminum-silicon system is asymmetric.

Classification of the AluminumSilicon Eutectic Structure Typical eutectic structures of binary alloys form by the simultaneous growth of the two phases from the liquid; therefore, they may exhibit a variety of microstructures that can be classified according to two criteria:  Lamellar versus fibrous morphology of the

individual phases: In general, when the two eutectic phases are present in approximately equal volume fractions, eutectics of binary alloys exhibit a lamellar structure. However, if one phase is present in a small volume fraction, this phase tends to be fibrous. As a rule of thumb, the eutectic structure obtained will tend to be fibrous when the volume fraction of the minor phase is less than 0.25; otherwise, it will tend to be lamellar (Ref 16).  Regular versus irregular eutectic structure: If both phases in the eutectic structure are nonfaceted, the eutectic will exhibit a regular morphology. In this case, the microstructure is made up of either lamellae or fibers that have a high degree of regularity and periodicity. However, if one phase is faceted, the eutectic structure is often irregular.

The typical aluminum-silicon eutectic is closer to a lamellar structure than to a fibrous one, even though the volume fraction of silicon in the aluminum-silicon binary is less than 0.25. (The volume fraction of silicon in the aluminum-silicon eutectic is 0.143.) This structure is usually attributed to the strong anisotropy of growth of silicon and to the relatively low interfacial energy between silicon and aluminum (Ref 16). Moreover, the aluminum-silicon eutectic has always been designated an irregular eutectic; in fact, it is the most widely used example of the irregular eutectic structure.

Microstructure of the AluminumSilicon Eutectic The aluminum-silicon eutectic can exhibit either of two morphologies: an unmodified morphology and a modified morphology. The unmodified morphology, such as that shown in Fig. 2(a), is characterized by coarse and flaky silicon particles and is usually observed in slowly cooled foundry alloys (typically at cooling rates between 1 and 10
C/min, or 2 and 18
F/min) when no chemical modifiers are added. The silicon particles in an unmodified alloy seem to be disconnected when viewed at a low magnification without deep chemical etching. Consequently, it was earlier believed that each of the coarse silicon particles is an isolated crystal. Later, Gurtler (Ref 17) observed that radial silicon particles grew across the primary aluminum dendrites, as shown in Fig. 4, and concluded that such an occurrence is not possible unless the silicon particles extend in three dimensions. Crosley and Mondolfo (Ref 18) suggested that the needlelike silicon particles observed in unmodified alloys must be flakes or sheets. Now, with the availability of electron microscopy, it is confirmed that silicon in unmodified aluminum-silicon eutectics has a flakelike microstructure. At relatively fast cooling rates, such as those encountered in chill casting, the aluminum-silicon eutectic is much finer, and silicon assumes a fibrous morphology. Day and Hellawell (Ref 19) classified the morphology of eutectic silicon in directionally

Fig. 4

Continuity of eutectic silicon flakes around a primary aluminum dendrite. Source: Ref 18

solidified, unmodified eutectic aluminum-silicon alloys as a function of the temperature gradient (G), the growth rate (R), and the alloy composition, as shown in Fig. 5. The region marked “A” in Fig. 5 is characterized by massive silicon particles growing by long-range diffusion and by a solidification interface that is essentially planar and isothermal. Some of the silicon crystals in the eutectic are interconnected, and the individual silicon crystals are irregular, banded, and tend to contain numerous {111} growth twins. In the region marked “B,” silicon assumes a variety of morphologies depending on the temperature gradient (G). At higher temperature gradients, close to those of the region marked “A,” silicon appears as fibers with a round cross section, but as the temperature gradient decreases, silicon forms lateral branches with a characteristic fourfold symmetry (Ref 7, 21, 22). In the region marked “C,” the silicon phase displays a flakelike, interconnected morphology. This is the general class of irregular eutectic observed in chemically unmodified alloys. At high growth rates (R), the structure develops as shown in regions “C” and “D.” Figure 6 shows the microstructure of a chemically modified aluminum-silicon alloy. When observed at low magnification, the silicon particles in these alloys appear as isolated, more-orless spherical crystals, as shown in Fig. 6(a). Figure 6(b), which is a higher-magnification scanning electron microscope image of a deep-etched sample, reveals the fibrous morphology of this silicon phase. It should be noted, however, that in spite of the apparent similarities in the morphology of the silicon phase in the chemically modified alloy and the chill-cast alloy, the growth mechanism of silicon is different in the two cases.

Theories of Solidification of the Aluminum-Silicon Eutectic The flakelike, interconnected silicon phase characteristic of typical unmodified aluminumsilicon eutectics is believed to grow by twinning via a mechanism known as twin plane re-entrant edge (TPRE). The TPRE mechanism

314 / Principles of Solidification

A

G, K/mm

100

D

10

B

C

1 B+C

1

10

100

R, µm/s

(a)

(b)

(c)

(d)

(e)

Fig. 5

Plot of temperature gradient (G) versus growth rate (R) along with the microstructures in the different regions. (a) Region A, (b) Region B (R = 0.3 m/s, or 1 ft/s, and G = 20
C/mm, or 915
F/in.). (c) Region B (R = 0.3 m/s, or 1 ft/s, and G = 3
C/mm, or 135
F/in.). (d) Region C. (e) Region D. Source: Ref 19, 20

Fig. 6

Microstructure of an aluminum-silicon alloy modified with strontium. (a) At low magnification. (b) At higher magnification and deeply etched. Source: Ref 20

was first introduced by Hamilton and Seidensticker (Ref 23) to explain the growth of germanium dendrites and was later extended to the growth of silicon. The equilibrium habit of silicon is an octahedron bound by eight {111} planes. A twin crystal is half of the equilibrium crystal reflected across the remaining solid along the twin composition plane. Consequently, the outline of the twin silicon crystal consists of six edges of the intersection of pairs of {111} planes, as shown in Fig. 7. The

external angles between these bounding planes are 141
and 219
. The bounding planes that make a 141
external angle form a re-entrant corner, while those planes that make a 219
external angle form a ridge. Because of the more favorable bonding for an atom joining a re-entrant corner than one joining a ridge, re-entrant corners are more favorable nucleation sites than ridges. Thus, the presence of a re-entrant corner leads to rapid growth along the [211] direction. This rapid growth on the

re-entrant corner stops when a trigonal solid that is bounded entirely by ridges is formed, as shown in Fig. 7(b). However, if the crystal has two twins instead of one, as shown in Fig. 7(c), it will have six re-entrant corners located along the h211i direction. In this case, growth on the re-entrant corners generates more reentrant corners, as shown in Fig. 7(d), and the newly generated re-entrant corners relieve the blockage of nucleation sites that is caused by the formation of ridges. Figure 7(e) shows a solid with several steps that are growing simultaneously. Although Billings and Holmes (Ref 24) experimentally verified that the TPRE mechanism is responsible for the growth of germanium dendrites and observed that all germanium dendrites invariably contain two or more twins and never a single twin, it has not been shown that the TPRE mechanism is the mechanism responsible for growth of silicon dendrites or the flakelike silicon. Moreover, Kitamura et al. (Ref 25) questioned the adoption of the TPRE mechanism to explain growth in real crystals, because it was originally developed for perfect crystals. Kitamura et al. argue that screw dislocations, which are prevalent in real crystals, should be accounted for when developing a growth mechanism for faceted crystals such as silicon. Sunagawa and Yasuda (Ref 26) proved by means of x-ray topography that growth in quartz crystals occurs at the re-entrant corner of a twin junction due to the concentration of screw dislocations. Extension of the Sunagawa and Yasuda hypothesis to eutectic flakelike silicon crystals would invariably cast doubt that the TPRE is the operating mechanism in these crystals. The contemporary understanding of the formation of the aluminum-silicon eutectic (Ref 27, 28) has its origins in the nucleation rather than the growth aspects of the eutectic reaction and is based on the fact that aluminum-silicon alloys invariably contain iron. For example, alloys that are prepared from very pure materials, that is, 99.999% purity aluminum and 99.999% purity silicon, could have up to 50 ppm Fe (Ref 29–32). Although this level of iron is normally considered a trace-level impurity of little consequence, it has been established that it plays a significant role in the solidification of the aluminum-silicon eutectic (Ref 27). It is maintained that iron, except when present in exceedingly low amounts (0.0015%), results in the formation of an iron-containing b-(Al, Si,Fe) phase that plays an important role in the nucleation of the aluminum-silicon eutectic. Therefore, it is more appropriate to think of this system when discussing the eutectic reaction as essentially an Al-Si-Fe ternary system with the eutectic phases being Aleut + Sieut + b-(Al,Si, Fe), rather than a binary aluminum-silicon system. In hypoeutectic alloys of this system, nucleation of the eutectic phases proceeds as illustrated schematically in Fig. 8. During solidification, the primary aluminum phase forms as dendrites at the liquidus temperature of the

Solidification of Eutectic Alloys—Aluminum-Silicon / 315

219⬚ 219⬚ ––

[211] 219⬚ (b) Twin planes

II 219⬚ 141⬚

(a)

I ––

––

[211]

I

[211]

––

[211] 141⬚

141⬚

219⬚

Twin planes

Twin planes

(c)

(d)

Fig. 8

Sequence of events during nucleation of the eutectic phases in aluminum-silicon hypoeutectic alloys. Source: Ref 27

Twin planes (e)

Fig. 7

Schematic illustrating the twin plane re-entrant edge mechanism. (a) Crystal with a single twin. (b) Closure of twins due to ridge formation. (c) Crystal with two twins. (d) Creation of extra re-entrant corners I and II. (e) Propagation of crystal due to re-entrant corners. Source: Ref 23

10 µm (a)

Fig. 9

50 µm (b)

Microstructure of an aluminum-silicon alloy. (a) Unmodified. (b) Modified with strontium. Source: Ref 33

alloy. This is followed by the evolution of a secondary b-(Al,Si,Fe) phase at some temperature between the liquidus temperature and the eutectic temperature of the alloy, depending on the concentration of iron in the alloy. At the eutectic temperature, and at an undercooling of 0.4 to 0.8
C (0.7 to 1.4
F), eutectic silicon (Sieut) nucleates on the secondary b-(Al,Si,Fe) phase in the solute field ahead of the growing aluminum dendrites. Once nucleated, the eutectic silicon grows as flakes into the eutectic liquid. The liquid surrounding the eutectic silicon flakes becomes enriched with aluminum as it is being depleted of silicon; consequently, eutectic aluminum (Aleut) nucleates and grows on the edges and tips of the eutectic silicon flakes. Finally, the aluminum dendrites stop growing as they impinge on the growing eutectic aluminum grains.

Theories of Chemical Modification of the Aluminum-Silicon Eutectic Chemical modification refers to the morphological transformation that occurs in the eutectic phases of aluminum-silicon alloys upon addition of trace quantities of certain elements, such as strontium or sodium, and converts the otherwise platelike (flake) silicon phase, shown in Fig. 9(a), to the fine, fibrous (corallike) structure of Fig. 9 (b). This morphological transformation significantly enhances the mechanical properties and overall performance of components that are cast from these alloys (Ref 32). Consequently, much of the fundamental research in the aluminum-silicon alloy system since Pacz introduced it into the United States in 1921 (Ref 1) has been directed toward understanding the mechanism

behind the modification of the morphology of the eutectic silicon phase by trace element additions (Ref 33, 34). However, and in spite of many hypotheses proposed to explain the modification of the eutectic structure in aluminum-silicon alloys (Ref 34), the genesis of this technologically important morphological transformation remains uncertain. This is due mainly to the lack of conclusive evidence provided by experimentation in support of the proposed hypotheses. The main assumption underlying many popular hypotheses on the mechanism of chemical modification of the silicon phase morphology is that the eutectic silicon phase nucleates on the primary aluminum dendrites during solidification of the hypoeutectic alloys, and that the modifying element inhibits the growth of the eutectic silicon phase, thus transforming its morphology from platelike to fibrous (Ref 18, 35, 36). In 1987, Lu and Hellawell (Ref 3) introduced the concept of impurity-induced twinning to explain the mechanism by which chemical modifiers such as sodium and strontium affect the morphology of eutectic silicon. Soon after, this became the accepted eutectic modification theory, in spite of its inability to explain several observed phenomena associated with chemical modification of the aluminum-silicon eutectic. According to the impurity-induced twinning theory, chemical modifiers are impurities that poison already growing atomic silicon layers by becoming adsorbed onto surface steps and kinks, thus preventing the attachment of silicon atoms to the crystal. Furthermore, the adsorbed impurity atoms induce twinning by altering the stacking sequence of atomic layers as the newly added layers seek to grow around the adsorbed impurity atom. Lu and Hellawell assumed a face-centered cubic structure and calculated the ratio of impurity atom radius (ri) to matrix atom radius (r) that is required for impurity-induced twinning. They found it to be ri/r = 1.6457. Analyses for sodium in modified aluminumsilicon alloys using Auger electron spectroscopy (Ref 3) and electron microprobe measurements (Ref 37) show that sodium segregates in the

316 / Principles of Solidification silicon fibers and in the aluminum matrix, which seems to validate the impurity-induced twinning theory. However, careful examination shows that the impurity-induced twinning theory cannot explain many observed phenomena that are associated with chemical modification of the aluminum-silicon eutectic. Particularly, it cannot explain the relatively large undercooling during solidification that is observed with the evolution of the eutectic phases when modifying elements are present. It also cannot explain the occurrence of eutectic modification without chemical additives (i.e., eutectic modification due to an increased superheat and/or a relatively fast cooling rate). It is interesting to note that sodium, which has less than the ideal radius ratio for eutectic modification as predicted by the impurity-induced twinning theory, is a better modifier than ytterbium and calcium, which are very close to the ideal radius ratio. Moreover, lithium, which is much smaller than the ideal radius ratio, modifies the eutectic silicon morphology when added in large concentrations to aluminum-silicon alloys (Ref 38). Because of the inability of the impurity-induced twinning theory to account for these facts, research efforts that aim to develop a better understanding of chemical modification and to rationalize these well-established facts into a consistent theory are prevalent. The most promising of these efforts are those that maintain that the morphology of the aluminum-silicon eutectic is decided during its nucleation in the liquid phase (Ref 28). REFERENCES 1. A. Pacz, U.S. Patent 1,387,900, 1921 2. L. Mondolfo, J. Aust. Inst. Met., Vol 10, 1965, p 169

3. S. Lu and A. Hellawell, Metall. Trans. A, Vol 18, 1987, p 1721 4. S. Lu, and A. Hellawell, J. Cryst. Growth, Vol 73, 1985, p 316 5. J.L. Murray and A.J. McAlister, Bull. Alloy Phase Diagrams, Vol 5, 1984, p 74 6. A.G.C. Gwyer and H.W.L. Phillips, J. Inst. Met., Vol 36, 1926, p 283 7. J.A. Bell and W.C. Winegard, J. Inst. Met., Vol 93, 1965, p 318 8. H.W. Kerr, J.A. Bell, and W.C. Winegard, J. Aust. Inst. Met., Vol 10, 1965, p 64 9. R.A. Meussner, U.S. Naval Research Laboratory Report, NRL 5331, U.S. Naval Research Laboratory, 1959 10. M.L.V. Gayler, J. Inst. Met., Vol 39, 1927, p 157 11. A. Hellawell, Prog. Mater. Sci., Vol 15, 1970, p 1–78 12. G. Tamman and A.A. Botschwar, Z. Anorg. Chem., Vol 157, 1926, p 26 13. G. Chadwick, Prog. Mater. Sci., Vol 12, 1963, p 208 14. V.D.L. Davies and J.M. West, J. Inst. Met., Vol 92, 1963–1964, p 208 15. E. Scheil, Z. Metallkd., Vol 37, 1946, p 1 16. W. Kurz and D.J. Fisher, Fundamentals of Solidification, Trans. Tech. Publications, 1988 17. G. Gurtler, Z. Metallkd., Vol 44, 1953, p 503 18. P.B. Crosley and L.F. Mondolfo, Mod. Cast., Vol 46, 1966, p 89 19. M.G. Day and A. Hellawell, Proc. R. Soc. (London) A, Vol 305, 1968, p 473 20. H.V. Guthy, M.S. thesis, Worcester Polytechnic Institute, 2002 21. F.N. Rhines and W.F.B. Timpe, Z. Metallkd., Vol 48, 1957, p 109

22. Z. Obinata and N. Komatsu, Sci. Rep. Tohoku Univ. A, Vol 9, 1957, p 107 23. D.R. Hamilton and R.G. Seidensticker, J. Appl. Phys., Vol 31, 1960, p 1165 24. E. Billings and P.J. Holmes, Acta Cryst., Vol 8, 1955, p 353 25. M. Kitamura, S. Hosoya, and I. Sunagawa, J. Cryst. Growth, Vol 47, 1979, p 93 26. I. Sunagawa and T. Yasuda, J. Cryst. Growth, Vol 65, 1983, p 43 27. S. Shankar, Y.W. Riddle, and M.M. Makhlouf, Acta Mater., Vol 52, 2004, p 4447 28. M.M. Makhlouf, S. Shankar, and Y.W. Riddle, AFS Trans. Paper 05-088(02), p 2005 29. C.R. Ho and B. Cantor, Acta Mater., Vol 43, 1995, p 3231 30. D.L. Zhang and B. Cantor, Metall. Trans. A, Vol 24, 1993, p 1195 31. C.R. Ho and B. Cantor, J. Mater. Sci. Eng. A, Vol 173, 1993, p 173 32. B. Cantor, J. Mater. Sci. Eng., Vol 151, 1997, p 226 33. M.M. Makhlouf and H.V. Guthy, J. Light Met., Vol 1, 2001, p 199 34. F. Yilmaz and R.J. Eliott, Mater. Sci., Vol 24, 1989, p 2065 35. A.K. Dahle, K. Nogita, J.W. Zindel, S.D. McDonald, and L.M. Hogan, Metall. Mater. Trans. A, Vol 32, 2001, p 949 36. G. Heiberg, K. Nogita, A.K. Dahle, and L. Arnberg, Acta Mater., Vol 50, 2002, p 2537 37. O.A. Atasoy, Z. Metallkd. Vol 78, 1987, p 177 38. L. Clapham, Ph.D. thesis, Queens University, 1987

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 317-329 DOI: 10.1361/asmhba0005213

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Solidification of Eutectic Alloys— Cast Iron* Andreas Bu¨hrig-Polackzek and Daniel dos Santos, Gießerei-Institut

Structure of Liquid Iron-Carbon Alloys Observations from x-ray, neutron diffraction, and sound velocity measurements on liquid binary iron-carbon alloys at temperatures approximately 20  C (35  F) above the liquidus indicate that for up to 1.8% C, the distance between nearest iron neighbors, rI, as well as the number of nearest neighbors, NI, in the first coordination sphere increases (Fig. 1). Above 1.8% C, the distance remains constant, while the number of nearest neighbors continues to grow. Above 3.5% and up to 5.5% C, both the distance and the number of neighbors remain constant. Above 3.5% C, short-range order regions rich in carbon exist in the melt. This means that the melt becomes more dense with the addition of carbon. A maximum packing

0.27 rI

11

10

0.26 NI

9

8

0.25 0

1

2 Carbon, wt%

3

4

Fig. 1

Variation of distance between nearest neighbors (rI) and the number of nearest neighbors (NI) as a function of the percentage of carbon in the ironcarbon alloy. Source: Ref 1

10

Nucleation of Eutectic in Cast Iron Structure of Graphite in the AusteniteGraphite Eutectic. The graphite phase is a faceted crystal bounded by low index planes. For graphite crystallizing from an iron-carbon melt, the normally observed bounding planes are (0001) and f1010g, as shown in Fig. 3(a). The crystallographic structure of graphite and the possible growth directions, A and C, are shown in Fig. 3(b). Because unstable growth occurs on

Distance between nearest neighbours (rl), nm

Depending on composition, cooling rate, and liquid treatment, it is also possible to produce a mixed white-gray eutectic called mottled structure. The two basic types of eutectic are very different, with mechanical properties such as strength, ductility, and hardness varying over very large intervals as a function of the type and the amount of eutectic formed. To understand the mechanism of the solidification of cast iron, it is necessary first to discuss the structure of liquid iron-carbon alloys.

12

10

9

rI ↑, NI ↑

8

rI const. NI ↑

rI, NI const.

8

7

7 Short range ordered regions

6 5 4 0

9

6 5

0.5

1.0 1.5 2.0 2.5 3.0 3.5 Carbon concentration, wt%

4.0

Fig. 2

4 4.5

Viscosity of iron-carbon alloys, cP

stable diagram, Fe-Fe3C, the white eutectic or austenitic (g), iron carbide (Fe3C) forms.
If solidification follows the stable diagram iron-graphite (a significant amount of silicon is required for this to occur), the gray eutectic, austenite-graphite (Gr), results.

f1010g planes, the edges of the platelike graphite crystals are not well defined. Graphite growing out of liquid iron-carbon alloys has a layer-type structure, with strong covalent bonds (4.19  105 to 5  105 J/mol, or 3.09  105 to 3.70  105 ft  lbf/mol) between atoms in the same layer. There is a trielectronic bond of each atom with its neighbors, while the fourth electron is common for the layer giving the metallic Number of nearest neighbours (Nl), atoms


If solidification occurs according to the meta-

density is reached at 3% C, and it remains constant at higher carbon concentrations. The excess carbon forms carbon-rich regions (nonhomogeneities) in the melt. Viscosity measurements (Fig. 2) show a correlation between viscosity and percentage of carbon. This correlation can be further explained in terms of increased viscosity as the interatomic distance becomes smaller. Liquid iron-carbon alloys with low carbon content (<3.5% C, that is, steels and cast irons poor in carbon) are microscopically homogeneous. Liquid iron-carbon alloys with high carbon (>3.5% C, that is, cast irons rich in carbon) are colloidally dispersed systems with microgroups of carbon in liquid solution. The nature of these microgroups is not clear. It is hypothesized that they are either Fe3C clusters (Ref 3, 4) or Cn clusters (Ref 1, 5). Because the nucleation energy for Fe3C is smaller than that for graphite, it is thermodynamically possible for the carbon-rich regions to exist as Fe3C clusters. Other investigators consider the Cn clusters to be stable (Ref 3). Their size is supposed to be in the range of 1 to 20 mm (40 to 800 min.), and it increases with the carbon equivalent, lower silicon content, and lower holding time and temperature. It is to be expected that the carbon-rich configurations existing in molten iron-carbon alloys are in dynamic equilibrium and that they diffuse within the melt.

Viscosity of iron-carbon alloys, Pa · s ⫻ 10−3

CAST IRON is a binary iron-carbon or a multicomponent Fe-C-X alloy that is rich in carbon and exhibits a considerable amount of eutectic in the solid state. Two such possible eutectics may result, as follows:

Viscosity of iron-carbon alloys as a function of carbon concentration at a temperature 20 above liquidus. Source: Ref 2

*This article has been updated and revised from D.M. Stefanescu, Solidification of Eutectic Alloys: Cast Iron, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 168–181.

318 / Principles of Solidification properties of graphite. Weak molecular forces exist between layers (4.19  103 to 8.37  103 J/mol, or 3.09  103 to 6.17  103 ft  lbf/mol). The prism plane is a high-energy plane at which impurities absorb preferentially. Strength and hardness are higher in the C direction of the graphite crystal. Complete destruction of the graphite structure occurs only at approximately 4000  C (7230  F). This explains the presence of some graphite aggregates in molten iron even at temperatures considerably higher than the liquidus temperature. Depending on the chemical composition and the temperature gradient/growth rate ratio, G/R, and/or cooling rate, G  R, a variety of graphite shapes can solidify as part of the austenitegraphite eutectic. Basically, they are as follows: Flake (actually plate) graphite (FG) Compacted/vermicular graphite (CG) Coral graphite Spheroidal (nodular) graphite (SG)

Crystalline structure of graphite. (a) Crystal of graphite bounded by (0001) and ð1010Þ type planes; the hexagonal arrangement of the atoms within the (0001) plane is shown relative to the bounding ð1010Þ faces. (b) Hexagonal structure of graphite showing the unit cell (heavy lines). Source: Ref 6

Fig. 3

Schematic of graphite types occurring in the austenite-graphite eutectic. (a) Flake graphite. (b) Compacted/ vermicular graphite. (c) Coral graphite. (d) Spheroidal graphite

2.0

212

752

932 108

50

90

40

72

30

54

20

36

10

18

0

0

Fig. 5

100

200 300 Superheating, ⬚C

18

Undercooling (ΔT ), ⬚F 36 54

72

90 50

Undercooling by rapid freezing

400

0 500

Relationship between superheating and maximum undercooling in flake graphite cast iron. Source: Ref 7

1.6 Eutectic cell count per mm

32 60

Superheating, ⬚F 392 572

0

Undercooling by superheating

40

1.2

30

0.8

20

0.4

10

0 0

Fig. 6

10

20 30 Undercooling (ΔT ), ⬚C

40

0 50

Influence of undercooling on eutectic cell count in flake graphite cast iron. Source: Ref 8

Eutectic cell count per in.

Fig. 4

Undercooling (ΔT ), ⬚F

A schematic of these graphite types, showing the traces of the f1010g planes, is given in Fig. 4. Nucleation of the Austenite-Flake Graphite Eutectic. Experimental evidence indicates that various types of nuclei become effective at various temperatures. Indeed, if only one type of nuclei would be active in a cast iron melt, an undercooling-superheating curve will show a single arrest for the temperature region over which the nuclei become effective. However, as shown in Fig. 5, a number of steps are observed in the relationship, suggesting that various foreign nuclei become effective as superheating is increased. Increasing the superheating apparently destroys the effective nuclei. Consequently, the number of eutectic grains (eutectic cells) decreases as the superheating increases, and as a result undercooling increases. However, when undercooling, DT, is increased at constant superheat by increasing the cooling rate or the growth rate, R, the eutectic cell count will increase (Fig. 6). Mathematically, this is shown by r* / 1/DT, where r* is the critical size radius. As DT increases, r* decreases, yielding an increasing number of critical size nuclei as well as more eutectic cells. Homogeneous nucleation is improbable in cast iron because typical undercoolings are much smaller than required by the theory of homogeneous nucleation (1 to 10  C, or 2 to 20  F, as opposed to 230  C, or 415  F). Nevertheless, some Cn clusters (as discussed earlier in this section) or undissolved graphite may act as heterogeneous nuclei for graphite formation. When two separate melt charges of similar composition containing white iron and gray iron are compared, the resulting structure provides support for the idea that some type of residual graphite serves as graphite nuclei during solidification (Ref 9). Further, additions of graphite to the melt enhance nucleation because the eutectic cell count is increased while the chilling tendency decreases (Ref 10).

Undercooling (ΔT ), ⬚C







Solidification of Eutectic Alloys—Cast Iron / 319 A rather wide variety of compounds have been claimed to serve as nuclei for FG cast iron, including oxides (for example, silicon dioxide) or silicates, sulfides, nitrides (boron nitride), carbides (for example, Al4C3), and intermetallic compounds (Ref 9). In addition, a number of metals, such as sodium, potassium, calcium, strontium, barium, yttrium, and the lanthanides, act as inoculants in FG iron and can therefore be considered to play a significant role in the heterogeneous nucleation of the austenite-FG eutectic (Ref 9, 11–13). High silicon concentrations in the melt can also contribute to the heterogeneous nucleation of graphite. The heterogeneous nucleation of FG iron occurs mainly on silicon dioxide particles (Ref 14). This assertion is based on experimental results as well as on the following theoretical considerations. If the oxygen dissolved in the iron is in equilibrium with carbon in an iron-acid lining-slag-atmosphere system, Eq 1 can be written: ðSiO2 Þ þ 2½C Ð ½Si þ 2fCOg

metals that have been proved to act as nuclei are as follows: Group I

Group II

Group III

NaHC2 KHC2

CaC2 SrC2 BaC2

YC2 LaC2

These carbides have a number of properties that support the hypothesis of their role as nuclei:
Due to the strong ionic bond between carbon

and the metal, they will not dissolve in the

melt but will precipitate as insoluble particles.
They are regions of concentration in the melt as compared with the bulk composition of the liquid and can serve as heterogeneous nuclei.
They have carbon planes in their structures that can serve as a basis for the epitaxial growth of graphite from the melt (Fig. 7). A schematic of the mechanism of inoculation by metals forming saltlike carbides is shown in Fig. 8. A metal from the I, II, or III group in the periodic table is added to the melt and reacts with carbon to form a saltlike carbide, MzC.

(Eq 1)

The equilibrium temperature for Eq 1 can be calculated as: log

½Si ½C2

¼

27; 486 þ 15:47 T

(Eq 2)

where T is the temperature in degrees K. Below the equilibrium temperature, solid silica is formed by precipitation deoxidation; above the equilibrium temperature, silica is decomposed. Silica serves as nuclei for graphite growth. The formation of silica particles in turn requires nucleation, which can be promoted by inoculation. Therefore, the inoculation of FG iron can be described as a deoxidation process with the following requirements:

Fig. 7

Epitaxial growth of graphite on a CaC2 crystal. Source: Ref 11

Fig. 8

Schematic of the inoculation of cast iron by metals to form salt-like carbides. Source: Ref 11


Enough oxygen must be available in the

melt.


There must be the addition of elements with

high affinity to oxygen, such as calcium, aluminum, and barium, to form nuclei for silicon dioxide precipitation (heterogeneous catalysis).
Inoculation temperature must be a maximum of 50  C (90  F) above the equilibrium temperature to avoid dissolution of the silicon dioxide particles.
The cooling rate after inoculation must be low enough to allow sufficient time for silicon dioxide formation, but high enough to prevent coalescence and segregation of silicon dioxide. A different theory of heterogeneous nucleation in FG iron has also been proposed (Ref 11). Based on the study of the influence of the inoculation of pure iron-carbon and Fe-C-Si alloys with pure metals, it was concluded that the addition of elements capable of forming saltlike carbides increases the number of eutectic cells and therefore the number of nuclei in the cast iron. Saltlike carbides are carbides containing the ion C2 2 . Typical saltlike carbides of

320 / Principles of Solidification Graphite then grows heterogeneously from the melt on the carbon planes of the carbide. Nucleation of the Austenite-Spheroidal Graphite Eutectic. Because most of the inoculants used in FG iron treatment are also effective for SG irons, it is reasonable to assume that similar mechanisms and substrates are active in both types of irons. Extensive transmission electron microscopy and scanning electron microscopy studies have been done to identify the composition of graphite nuclei in SG irons, with a rather wide variety of results. Numerous types of compounds have been found in the middle of graphite spheroids; therefore, it was hypothesized that they could act as nuclei. Some of these compounds are as follows: 3MgO2SiO22H2O (cristobalite) (Ref 15), xMgOyAl2O3zSiO2, xMgOySiO2, xMgOySiO2zMgS (Ref 16), MgS (Ref 17), Te + Mn + S (Ref 18), and lanthanide sulfides (Ref 19). A rather extensive work on what appeared to be substrates for spheroidal graphite in chilled ductile iron concluded that the substrates are duplex sulfide-oxide inclusions with a diameter of 1 mm (40 min.). The core is probably made of calcium-magnesium or Ca-Mg-Sr sulfides, while the outer shell is made of complex MgAl-Si-Ti oxides with a spinel structure (Ref 20). The orientation relationships were established as follows. For the nucleus core/nucleus shell: ð110Þsulfide k ð111Þoxide ½1010sulfide k ð211Þoxide For the nucleus shell/graphite: ð111Þoxide k ð0001Þ graphite ½110oxide k ½1010graphite The x-ray diffraction data showed that the first few graphite layers, adjacent to the (111) oxide, ˚, had a dilated lattice (0.264 nm, or 2.64 A ˚ ). It was suginstead of 0.246 nm, or 2.46 A gested that the spacing within the graphite layers decreases away from the oxide until unconstrained spacing is reached. Dislocations were frequently observed in the matrix, and it was suggested that these were generated to relieve some of the elastic strain in the graphite layers adjacent to the oxide. These observations allow for a theory of nucleation of spheroidal graphite similar to the catalytic theory of nucleation of flake graphite on silicon dioxide, as discussed previously. It can be assumed that the nucleation process begins with the formation of complex sulfides that serve as nuclei for complex oxides, which in turn serve as nuclei for spheroidal graphite. Recent experimentations (Ref 21) suggest that even adding small amounts of sulfur (less than 0.0059%) to hypereutectic postinoculated ductile irons results in higher nodule counts than in comparable melts without any later sulfur additions. This strengthens the theory that sulfur plays an important role in the formation of substrates for the graphite. Structure of Iron Carbide in the AusteniteIron Carbide Eutectic. The iron carbide

(cementite) consists of an orthorhombic unit cell with 12 iron atoms and 4 carbon atoms per cell and therefore has a carbon content of 6.7 wt%. Its density is 7.6 g/cm3. Nucleation of the Austenite-Iron Carbide Eutectic. Very little information is available on the nature of the nuclei of the white eutectic. Nevertheless, it is accepted that the nucleation of Fe3C occurs at lower undercooling than that of graphite (Ref 8). This is supported by the slopes of the solubility lines in the phase diagram, which shows that the supersaturation of cementite increases faster than that of graphite, resulting in an increased probability for the nucleation of cementite as the temperature is lowered. Cooling curve data also show that the solidification of white iron begins at a lower temperature than that of gray iron. There is some evidence that silicon dioxide and aluminum oxide can serve as substrates for the growth of the Fe-Fe3C eutectic and that the nature of the substrate can influence the morphology of the eutectic (Ref 22).

Nucleation of Primary Phases in Cast Iron Depending on the carbon equivalent (or saturation degree), the primary phase can be either austenite for hypoeutectic cast iron or graphite for hypereutectic cast iron. Primary Austenite. The formation and growth characteristics of austenite dendrites have received considerably less attention from investigators than eutectic cell count and morphology simply because dendrites are not readily discernible in the structure. Nevertheless, it has been demonstrated that inoculants that are effective in increasing the cell count (such as calcium-titanium, strontium, and 75% FeSi) have little effect on the solidification pattern of primary austenite dendrites (Ref 23). A number of elements, such as titanium, vanadium, and aluminum, were proved to increase the number of dendrites and their length when compared with the base iron. For titanium and vanadium, this was attributed to the formation of carbides, nitrides, and carbonitrides, which then act as substrates for austenite solidification. The influence of aluminum was not explained. Other elements, such as cerium and boron, although being instrumental in increasing the number and length of dendrites, were not considered to be nucleants for austenite. It was suggested that these elements restrict the growth of the eutectic cells, which subsequently results in larger undercooling for eutectic solidification. This means that more time (a larger temperature interval) is available for the nucleation and growth of austenite (Ref 23). Primary Graphite. It is reasonable to assume that nuclei that are active in the solidification of eutectic graphite also serve as nuclei for hypereutectic primary graphite. Nevertheless, it must be noted that inoculation does not seem to increase

significantly the number of eutectic cells in hypereutectic FG iron. On the other hand, inoculation is quite effective in hypereutectic SG iron.

Growth of Eutectic in Cast Iron Coupled Zone in Cast Iron The degree and type of eutectic growth that occurs in cast iron can be determined by using tools such as growth rate curves to locate coupled zone regions, isothermal time-temperature diagrams to gage susceptibility to carbide formation, and growth rate-composition plots to ascertain parameters that affect both directional and multidirectional solidification. In cast irons, the coupled growth region of the eutectic is asymmetric. It is possible to construct a theoretical coupled zone for gray iron from the condition of equal growth rate of the austenite (g) and graphite (Gr) phases (Ref 24). First, one must consider the growth rate curves for g and Gr in the iron-carbon system (Fig. 9). For FG iron, the growth rate of austenite, Rg, and that of graphite along the ½1010 direction, RGr ½1010, intersect; therefore, a coupled zone

Fig. 9

Suggested growth rate curves for austenite (g) and graphite (Gr) in the Fe-C-Si alloys. (a) Austenite-flake graphite eutectic. (b) Austenitespheroidal graphite eutectic. (c) Different graphite growth rates in lamellar growth. RGr ½1010 determines the lamellar habits of graphite flakes, while RGr [0001] gives the rate of thickening for the flake. Source: Ref 6

Solidification of Eutectic Alloys—Cast Iron / 321 Gr γ

ΔTGr⬘ ΔTγ⬘

γ

Gr

ΔTγ/Gr ΔTGr⬙

L

2

E ΔTγ/Gr ΔTγ ⬙

Growth rate, R

1

P Halo (shell)

ΔTγ /Gr

ΔTγ⬘

Undercooling, ΔT

Graphite leading phase

Q

ΔTGr⬘ ΔTGr⬙

R

ΔTγ ⬙

(a)

RGr⬙ = R g (Co line)

(b)

Fig. 10

Constructions of coupled growth line in Fe-C-Si alloys. (a) Austenite (g) and graphite (Gr) growth curves. (b) Construction of the equal growth rate curve Rg = RGr. Source: Ref 24

20 400 10 5 Au + Eu1 2

Temperature, ⬚C

Velocity (v), μm/s

Eu1

Gr + Eu1

Velocity (v), μin./s

4 ⫻ 103

D

L+A

1976

1080 L + A + Gr + C

1940

1060 L+A

N

1040

1904

F

1020 1000 Z 980 0.1

40

2012

I 0.5

1

2

1

1868 1832

L+A+C

K

Temperature, ⬚F

L 1100

100 R = 7K/mm

2048

1120

Gray eutectic

50

R

B

L + A + Gr

White eutectic Gray eutectic + γ- dendrite

500 200

1140

4 ⫻ 104

103

5 10 20 Time, s

50 100

(a) 0.5

1140 0.2

B

O

R

1120 1100

2120

1160

1155

1145

2102

Gr + Eu1

Au + Eu1

Ce + Eu2

Au + Eu2

2012 N

D

Liq 1080

1976 1940

1060 Liq + A 1040

1904

L Liq + A + Gr + C

2084

1140

1020

Temperature, ⬚F

Temperature, ⬚C

Eu1 1150

2048 Liq + A + Gr Temperature, ⬚F

(a)

3.75 4.0 4.25 4.5 4.75 Concentration (C∞), wt% carbon

4 5.0

Temperature, ⬚C

0.1 3.5

1868 F

1000 980 0.1

I

Z 0.5

1

K 2

1832 Liq + A + C

5 10 20

50 100

Time, s (b)

1135

1130 3.75

Eu2

4.0 4.25 4.5 4.75 Concentration (C∞), wt% carbon

Fig. 12 2066 5.0 5.25

Time-temperature transformation diagrams for isothermal solidification of eutectic cast iron. (a) Flake graphite cast iron. (b) Spheroidal graphite cast iron. L, liquid; A, austenite; C, cementite; Gr, graphite. Source: Ref 8

(b)

Fig. 11

and are equal (DTg/Gr); at points 1 and 2 on the curves, the growth rates are the same, but the undercooling differs. Figure 10(b) shows the phase diagram in the vicinity of the eutectic point, with DTg0 and DTg00 drawn for growth conditions represented by the points of Fig. 10 (a). Point “Q” in Fig. 10(b) corresponds to DTg/Gr, where Rg = RGr. At the left of the RGr = Rg line, primary g dendrites will form, while at the right of this line, graphite will be the leading phase. Coupled eutectic growth can only occur for combinations of undercoolings and compositions on the dotted line RGr = Rg. The transition from a fully eutectic to a eutectic plus dendrite structure in pure iron-graphite alloys of eutectic composition has been determined, and the g-iron and graphite-iron eutectic boundaries have been calculated (Fig. 11).

Coupled zone of directionally grown ironcarbon eutectic. (a) Growth rate-composition diagram showing calculated curves and experimental points. (b) Temperature-composition diagram with superimposed phase diagrams, stable (Eu1) and metastable (Eu2). Au, austenite dendrites; Gr, primary graphite plates; Ce, primary cementite plates. Source: Ref 25

can be constructed. For SG iron, where the predominant growth direction is along [0001], the two rates, Rg and RGr[0001], do not intersect, which means that coupled growth is impossible. Figure 10(a) shows the point at which the graphite and g growth curves for gray iron cross

Isothermal Solidification Solidification studies are usually performed athermally because of the high rate of the liquid-solid transformation. Nevertheless, useful information can be extracted from the isothermal time-temperature-transformation diagrams shown in Fig. 12. It is apparent that SG iron is more susceptible to carbide formation than FG iron. Graphite precipitates earlier in SG iron than in FG iron at all undercoolings, although the time interval for complete gray solidification is smaller in FG irons.

Growth in Directional Solidification It is quite obvious from the previous discussion that undercooling and composition must be carefully selected to achieve coupled growth of the eutectic in cast iron. The g-FG eutectic is a coupled irregular eutectic of the faceted (Gr)/nonfaceted (g) type. The g-SG eutectic is a divorced eutectic. The basic parameters affecting the morphology of the eutectic are the G/R ratio and composition.

322 / Principles of Solidification As shown in Fig. 13, it is possible to achieve a variety of structures in cast iron when varying G/R and/or the level of impurities (for example, cerium). The gray eutectic can solidify with a planar interface. The theoretical relationship l2R = constant (Jackson and Hunt), where l is the spacing, is not obeyed in the growth of the g-FG eutectic. Different experimental results are compared with the theoretical behavior in Fig. 14. The departure of experimental values from the theoretical line is a clear indication of irregular rather than regular cooperative growth in FG cast iron. For G/R ratios selected to produce a planar interface during the directional solidification of FG iron, it has been measured that the growth constant in the R = m  DT2 equation is m = 8.7  108 m  s1  K2 (Ref 25). In the case of SG iron, even for hypereutectic irons, the graphite spheroids do not grow in independent austenite envelopes but rather are associated with austenite dendrites (Fig. 15). The eutectic austenite is dendritic and cannot be distinguished from the primary austenite. Spheroidal graphite precipitates directly from the melt, becomes enveloped in an austenite shell that is very soon incorporated into the dendrites, and then grows together with the dendrites (Ref 27, 28). Directional solidification experiments (Ref 29) demonstrated that nucleation of SG in eutectic and hypereutectic irons occurs over the entire eutectic solidification temperature interval.

Growth in Multidirectional Solidification Flake (Lamellar) Graphite Eutectic. The austenite-FG eutectic solidifies with the formation of eutectic colonies (cells) that are more or less spherical in shape. It is generally thought that each eutectic cell is the product of a nucleation event. The eutectic cell is made of interconnected graphite plates surrounded by austenite. The degree of ramification of graphite within the cell depends on undercooling, with higher undercooling resulting in more graphite branching (Fig. 16). The leading phase during the eutectic growth is the graphite. Graphite spacing is determined by the same parameters as for regular eutectics, with branching occurring as a response to interface instability. In turn, interface instability is determined by localized changes in composition, convection currents, crystallographic orientation different from the heat extraction direction, and a change in temperature gradient. Figure 17 shows a scanning electron microscopy (SEM) micrograph of graphite eutectic cell and prior austenitic dendrite structures in gray cast iron. The variations in graphite structures have been classified, together with the length of the flakes, by standards that have been used for many years. Flake graphite in gray cast iron can be designated as:
Type

A: uniform orientation

distribution,

random


Type B: rosette grouping, random orientation
Type C: superimposed flake sizes, random

orientation


Type D: interdendritic segregation, random

orientation


Type E: interdendritic segregation, preferred

orientation The formation of the eutectic flake graphite (types A, B, C, and D) is greatly influenced by the amount by which the iron melt cools below the equilibrium temperature for the austenite-graphite eutectic before appreciable solidification occurs. Gray iron with type A graphite undergoes only small amounts of undercooling, whereas that with type D graphite undercools significantly below this equilibrium temperature. The undercooling that occurs with type B graphite is intermediate between the two, producing fine graphite flakes, like type D, in the center of the eutectic cells or rosette and a coarser type, like type A, at the outer cell boundaries. Type E graphite occurs in strongly hypoeutectic gray irons with carbon equivalents well below 4.3%. Figure 18(a) shows an SEM image of flake graphite and respective three-dimensional reconstruction (Fig. 18b) using focused ion beam (FIB) nanotomography. In the represented volume, it is possible to see that almost all graphite particles are connected, and in the cropped-out magnified regions (Fig. 18c) some inclusions can be found. A chemical analysis (energy-dispersive x-ray) of these inclusions determined them to be MnS precipitates (Ref 31). Calculations using data from continuous cooling experiments given in Ref 32 allowed estimation of the growth constant m in the range of 7.25  108 to 9.5  108 m  s1  K2, which is of the same magnitude as the values given previously for the case of directional solidification. The solidification of off-eutectic FG iron will start with the nucleation and growth of a primary phase, which is austenite dendrites for hypoeutectic iron or primary graphite plates for hypereutectic iron. It is not unreasonable to assume that the primary phases may play a role in the nucleation of the eutectic, with further growth of the eutectic occurring, as experimental evidence shows, on the primary phases. Spheroidal Graphite Eutectic. Growth of the austenite-SG eutectic is more complicated and less understood than that of the g-FG eutectic, although a good number of theories have been proposed. As discussed previously in this section, the g-SG eutectic is a divorced eutectic. It has been rather widely accepted that the growth of this eutectic begins with nucleation and the growth of graphite in the liquid, followed by early encapsulation of these graphite spheroids in austenite shells (envelopes). A schematic of the process is shown in Fig. 19. Graphite nucleation and growth deplete the melt of carbon in the vicinity of the graphite; this creates conditions for austenite nucleation

and growth around the graphite spheroid. Once the austenite shell is formed, further growth of graphite can occur only by solid diffusion of carbon from the liquid through the austenite. Calculations of the diffusion-controlled growth of graphite through the austenite shell were originally made based on Zener’s growth equation for an isolated spherical particle in a matrix of low supersaturation (Ref 33). Equation 3 was derived: RGr ¼

VmGr g rg X g=L  X g=Gr D Vmg c rGr ðrg  rGr Þ X Gr  X g=Gr

(Eq 3)

where RGr is rate of growth of graphite; VmGI and Vmg are the molar volumes of graphite and austenite, respectively; Dgc is the diffusivity of carbon in austenite; rGr and rg are the radii of graphite and austenite, respectively; Xg/L and Xg/Gr are the molar fractions of austenite at the austenite/liquid and austenite/graphite boundaries, respectively; and XGr is the molar fraction of carbon in graphite. It has also been shown that rg = 2.4 rGr. A slightly different approach, based on a steady-state diffusion model of carbon through the austenite shell (Ref 34), also allowed derivations for RGr and rg. However, recent research has shown that the solidification mechanism of SG iron is more complicated and that austenite dendrites play a significant role in eutectic solidification (Ref 27, 28). The eutectic austenite is dendritic and can scarcely be distinguished from primary austenite dendrites. The sequence of solidification is as follows:
At the eutectic temperature, austenite den







drites and graphite spheroids nucleate independently in the liquid. Limited growth of spheroidal graphite occurs in contact with the liquid. Flotation or convection then determines the collision of spheroidal graphite with the austenite dendrites. Graphite encapsulation in austenite can occur before or immediately after the contact between graphite and austenite dendrites. Further growth of graphite occurs by carbon diffusion through the austenite shell.

This sequence is illustrated schematically in Fig. 20. Graphite shape in SG iron is not perfectly spheroidal. Depending on cooling rate and chemical composition, significant deviations from the true spheroidal shape can be obtained (various nodularities) (Fig. 21). It remains now to explain the reasons for the change of the habits of graphite from lamellar to spheroidal. Many theories have been proposed (see, for example, review works in Ref 6 and 9), but because of space limitations, only a few are discussed in this section. Many theories capitalize on the observation that the graphite/liquid surface energy is higher

Solidification of Eutectic Alloys—Cast Iron / 323 0.4

Spacing (λ), mm

0.2

C

0.1 0.06 0.04 D 0.02

0.01 10−4

B A

10−3 Growth rate (R ), mm · s−1

0.01

l-R relationships in flake graphite cast iron. A, Lakeland l = 3.8  105  R0.5 cm; B, Nieswaag and Zuithoff l = 0.56  105  R0.78 cm (0.004% S); C, Nieswaag and Zuithoff l = 7.1  105  R0.57 cm (>0.02% S); D, Jackson and Hunt l = 1.15  105  R0.5 cm (theoretical). Source: Ref 6

Fig. 14

Fig. 13

Schematic showing the influence of composition (% cerium), G, and R over the eutectic morphology of Fe-C-Si alloys. SG, spheroidal graphite; CG, compacted graphite; FG, flake graphite. Source: Ref 26

Fig. 15

Microstructure of the solid/liquid interface for a directionally solidified hypereutectic cerium-treated spheroidal graphite cast iron. Source: Ref 27

in SG iron than in FG iron. These theories explain SG formation by either simply implying that a sphere will have less free surface energy than a lamella with the same volume above a certain critical interface energy (Ref 35) or by suggesting that the high interface energy will curve the growing crystal in order to decrease

the energy/volume ratio, resulting in spheroidal rather than lamellar graphite (Ref 36). The defect growth of graphite theory explains SG formation based on the possible growth mechanisms of faceted crystals such as graphite (Ref 37). Three growth mechanisms are considered: two-dimensional nucleation

(Fig. 22a), step of a defect (twisted) boundary (Fig. 22b), and screw dislocation. The first two mechanisms are governed by exponential laws and apply to the ð1010Þ surface, but the third is governed by a parabolic law and applies to the (0001) surface of the graphite crystal.

324 / Principles of Solidification

Fig. 16

Schematic of solidification of flake graphite. (a) Typical eutectic colonies (cells). (b) Growth sequence for a eutectic colony. g, austenite; Gr, graphite

Fig. 17

Fig. 18

Fig. 19

SEM micrograph showing graphite, eutectic cell, and prior dendrite structure in gray cast iron. Original magnification: 200. Source: Ref 30, courtesy of Gary F. Ruff

(a) SEM micrograph of flake graphite, deep etched. (b) Reconstructed three-dimensional structure. (c) Cropped-out magnified region with inclusions and flake graphite. Source: Ref 31

Isothermal growth of a graphite spheroid within an austenite shell and growth of the shell with a smooth interface. (a) Growth of spheroidal graphite in contact with melt. (b) Envelopment by austenite. (c) Growth of spheroidal graphite within the austenite shell. Source: Ref 24

Solidification of Eutectic Alloys—Cast Iron / 325 When weak, reactive impurities such as sulfur are present in the melt, a contaminated environment occurs. These elements change the edge energy of steps, resulting in a relative position change of the growth rates involved, as shown in Fig. 23(a). The curve for growth on the step of a defect boundary, Rstep, is at a lower undercooling than those for growth by two-dimensional nucleation, R2D, or by screw dislocation, Rscrew. In a pure environment such as an Fe-C-Si alloy with no sulfur contamination, the growth rate curves are displaced to higher undercooling (Fig. 23b). In a melt of sufficient purity, or when increasing cooling rate, the higher degree of undercooling may allow growth with Rscrew so that graphite spheroids can form. This has been achieved experimentally for pure nickelcarbon alloys by increasing the cooling rate of the melt, or for ultrapure iron-carbon alloys by cooling slowly in a vacuum. In an environment with reactive impurities (for example, magnesium), the impurity will react with the surface, and the growth at a step of a twist boundary will be neutralized. Only the curves for R2D and Rscrew are left, and they are displaced to greater undercoolings (Fig. 23c). Another theory relates graphite shape in cast iron with undercooling (kinetic plus constitutional) during solidification (Ref 6). As shown in Fig. 24, each graphite form has its own temperature for growth, which is achieved by a specific cooling rate and composition. Steps on surfaces can change graphite morphology from plate to rod. With an increase in undercooling, pyramidal instabilities will occur on the faces of the graphite crystal. At undercoolings of 29 to 35  C (50 to 65  F), instabilities occur on the ð1011Þ faces of the pyramid, and it is suggested that graphite spheroids form at these undercoolings. Finally, at large undercoolings of 40  C (70  F), the growth form noted is a pyramidal one, bounded by ð1011Þ faces. These pyramidal crystals are part of the series of imperfect forms observed particularly in thick-wall SG iron castings. Imperfect SG shapes, such as those in Fig. 21(b) and (c), are due to incorrect kinetic undercooling and an imbalance of impurities influencing constitutional undercooling. An insufficient amount of impurities influencing kinetic undercooling (for example, magnesium) will favor the formation of intermediate graphite shapes, such as compacted/vermicular, while an insufficient amount of impurities influencing constitutional undercooling (for example, lead) will perturb the surface of the spheroid and promote the formation of protuberances. Surface Adsorption Theory. This somewhat older theory of graphite growth postulates that the change from lamellar to spheroidal graphite occurs because of the change in the ratio between growth on the ð1010Þ face and growth on the (0001) face of graphite (Ref 38). For

equilibrium conditions, the Gibbs-Curie-Wulf law states that the crystalline phase with the higher interface energy has a slow rate of growth in the normal direction. Bravais’s rule stipulates that the growth rate in the direction normal to a plane is inversely proportional to the density of atoms located on the plane. Accordingly, it follows that under equilibrium conditions, the crystallographic plane with the highest density of atoms has the lowest interface energy and the minimum growth rate in a direction perpendicular to the plane. Nevertheless, under the nonequilibrium conditions prevailing during the solidification of cast iron, kinetic considerations become important. Assuming growth by two-dimensional nucleation, the highest rate of growth will be experienced by the face with the higher density of atoms, where the probability for nucleation is higher. Therefore, in a pure environment, the highest growth rate will be in the (0001) direction of the graphite crystal (Fig. 25), resulting in the formation of unbranched single crystals (coral graphite). In a contaminated environment, surface-active elements such as sulfur or oxygen are absorbed on the high-energy plane ð1010Þ, which has fewer satisfied bonds. Subsequently, the ð1010Þ plane face achieves a lower surface energy than the (0001) face, and growth becomes predominant in the ð1010Þ direction, resulting in lamellar (plate) graphite. Finally, the reactive impurities (such as magnesium, cerium, and lanthanum) in an environment scavenge the melt of surface-active elements (sulfur, diatomic oxygen, lead, antimony, titanium, and so on), after which they also block growth on the ð1010Þ prism face. A polycrystalline SG results. Compacted/Vermicular Graphite Eutectic. The sequence of growth of compacted/vermicular graphite during the eutectic transformation is shown schematically in Fig. 26, based on experimental data from Ref 39 on rapidly quenched samples from successive stages during the solidification process. It can be seen that at the beginning, graphite precipitates as spheroids, which then degenerate during growth and subsequently develop into compacted graphite. Compacted graphite develops as interconnected segments within an austenitic matrix. Typical compacted graphite is shown in Fig. 27(a). The reconstructed three-

(a) 10 μm

(b) 10 μm

(c) 1 μm

Fig. 21

Fig. 20

Schematic illustrating the progression of growth in austenite-spheroidal graphite eutectic

SEM micrographs of deep-etched spheroidal graphite samples showing a fractured graphite spheroid (a). Nodularity decreases from (a) through (c).

326 / Principles of Solidification dimensional form (Fig. 27b) was obtained using FIB nanotomography (Ref 31). It was proposed that the growth of compacted graphite occurs by the twin/tilt of boundaries (Ref 40). The initiation of a twin/tilt is related to the unstable growth of the ð1010Þ interface, which can be induced by the presence of some reactive elements. The formation of a ð1012Þ twin/tilt plane is shown in Fig. 28. It was further hypothesized that when insufficient reactive element is present, the tilt orientation of twin boundaries can alternate, resulting in typical compacted graphite (Fig. 29a). On the other hand, when there is enough impurity in the melt, the tilt orientation is singular, and spheroidal graphite may occur (Fig. 29b). Based on this growth mechanism, compacted graphite can be considered to grow either by transition from flake graphite (Fig. 30a) or by degeneration of spheroidal graphite (Fig. 30b). The degree of interconnection of the compacted graphite within the eutectic cell is also apparent from Fig. 30. Based on cooling curve configuration, it is reasonable to assume that the growth rate of compacted graphite is similar to that of spheroidal graphite, which is not surprising because the growth direction is supposedly in the [0001] direction of the graphite crystal. Austenite-Iron Carbide Eutectic. Growth of the austenite-iron carbide (ledeburite) eutectic begins with the development of a cementite plate on which an austenite dendrite nucleates and grows (Fig. 31). This destabilizes the Fe3C, which then grows through the austenite. As a result, two types of eutectic structure develop: a lamellar eutectic with Fe3C as the

leading phase in the edgewise direction, a, and a rodlike eutectic in the sidewise direction, c. Cooling rate has a significant influence on the morphology of the g-Fe3C eutectic. At moderate undercoolings, ledeburite structure is expected. High cooling rates, as obtained in quenching experiments, produce a degenerated eutectic structure dominated by Fe3C plates. A coarse mixture of Fe3C and g fills the spaces

Suggested DT-R correlation for ð1010Þ and (0001) crystal faces of graphite growing in various environments. (a) Contaminated environment (for example, sulfur-containing Fe-C-Si alloy). (b) Pure environment (for example, Fe-C-Si alloy). (c) Environment with reactive impurities (for example, magnesium-containing Fe-C-Si alloy). Three growth mechanisms are discussed: A, on the step of the defect boundary (Rstep); B, two-dimensional nucleation (R2D); C, screw dislocation (Rscrew). Source: Ref 37

between the Fe3C plates. This structure obviously does not result from cooperative growth (Ref 8). A platelike carbide structure associated with equiaxed eutectic grains can be obtained by increasing the undercooling by superheating, by decreasing the silicon content, or by adding chromium or magnesium. The gray or white solidification mode of cast iron apparently depends on the relative nucleation probability and the growth rates of the graphite and Fe3C phases. In turn, this will be a function of the cooling rate and chemistry of the alloy. As shown in Fig. 32, only the graphite eutectic can nucleate and grow between the eutectic temperature for the gray eutectic (1153  C, or 2107  F) and that for the white eutectic (1148  C, or 2098  F). Below 1148  C (2098  F), both eutectics can occur. The growth rate of the g-Fe3C eutectic rapidly exceeds that of the g-Gr eutectic, and at a temperature of approximately 1140  C (2085  F), the gray-towhite transition occurs (Ref 6, 43).

Fig. 23

Growth of graphite in the ð1010Þ direction. (a) Growth by two-dimensional nucleation on ð1010Þ faces illustrates that steps on (0001) surfaces will advance only as far as bounding crystal edges. (b) Growth from step to twist boundary illustrates that the (1010) faces grow by nucleation of planes at the step. Source: Ref 37

Fig. 22

Fig. 24

Correlation among the different types of instability observed in graphite growth and growth morphologies with increasing undercooling, DT. (a) DT = 4  C (7  F). (b) DT = 9  C (16  F). (c) DT = 30  C (54  F). (d) DT = 40  C (72  F). Source: Ref 6

Solidification of Eutectic Alloys—Cast Iron / 327

Cooling Curve Analysis Cooling curves have been extensively used in the experimental study and production control of cast irons. The history of solidification can be traced on the cooling curve. A detailed description of the principles involved in this analysis and its application in cast iron technology can be found in other articles in this Volume. REFERENCES 1. S. Steeb and U. Maier, Structure of Molten Fe-C Alloys by Means of X-Ray and Neutron Wide Angle Diffraction as well as Sound Velocity Measurements, The Metallurgy of Cast Iron, B. Lux et al., Ed., Georgi Publishing, 1975, p 1 2. W. Krieger and H. Trenkler, Arch. Eisenhu¨ttenwes., Vol 42 (No. 3), 1971, p 175 3. A.A. Vertman et al., Liteinoe Proizvod., No. 10, 1964 4. J. Luo, Q. Zhai, and P. Zhau, Microstructures of Liquid Pure Iron and Fe-C Alloy near the Melting Point, Acta Metall. Sin. (Chinese Lett.), Vol 39, 2003, p 5–8 5. L.S. Darken, “Equilibria in Liquid Iron with Carbon and Silicon,” Technical Publication 1163, American Institute of Mining, Metallurgical, and Petroleum Engineers, 1940, p 1 6. I. Minkoff, The Physical Metallurgy of Cast Iron, John Wiley & Sons, 1983 7. W. Patterson and D. Ammann, Solidification of Flake Iron-Graphite Eutectic in Gray Iron, Giesserei, Jan 1959, p 1247

8. H.D. Merchant, Solidification of Cast Iron—A Review of Literature, Recent Research on Cast Iron, H.D. Merchant, Ed., Gordon and Breach, 1968, p 1 9. J.F. Wallace, Effects of Minor Elements on the Structure of Cast Iron, Trans. AFS, Vol 83, 1975, p 363 10. S.L. Liu, C.R. Loper, Jr., and S. Shirvani, Inoculation of Gray Cast Irons with Carbonaceous Materials, Trans. AFS, Vol 93, 1985, p 501 11. B. Lux, Nucleation of Graphite in Fe-C-Si Alloys, Recent Research on Cast Iron, H.D. Merchant, Ed., Gordon and Breach, 1968, p 241 12. J.V. Dawson and S. Maitra, Recent Research on the Inoculation of Cast Iron, Br. Foundryman, April 1967, p 117 13. D.M. Stefanescu, Inoculation of Gray Iron with Sodium and Barium, Giesserei-Prax., No. 24, 1972, p 429 14. W. Weiss, The Importance of Deoxidation in the Crystallization of Cast Iron, The Metallurgy of Cast Iron, B. Lux et al., Ed., Georgi Publishing, 1975, p 69 15. M.B. Zeedijk, Identification of the Nuclei in Graphite Spheroids by Electron Microscopy, J. Iron Steel Inst., July 1965, p 737 16. P. Poyet and J. Ponchon, Contributions to the Study of Graphite Nucleation in Cast Iron for Ingot Molds, Fonderia, No. 277, 1969, p 183 17. J.C. Mercier et al., Inclusions in Graphite Spheroids, Fonderia, No. 277, 1969, p 191 18. H. Nieswaag and A.J. Zuithoff, “The Occurrence of Nodules in Cast Iron Containing Small Amounts of Tellurium,” Paper

19. 20.

21.

22. 23.

24.

25.

26.

27. 28.

29.

Fig. 25

Schematic of the change in the growth rate of graphite due to the absorption of foreign atoms in spheroidal graphite eutectic. Three variations of an Fe-C-Si cast iron are as follows. (a) With nodularizer added as reactive impurity environment. (b) Pure environment. (c) Contaminated environment in which surface-active elements such as oxygen and sulfur are absorbed into system. For (a) and (b), density in the basal plane, VB, is greater than the density in the prism face, VP, and either branch polycrystalline or unbranched single crystals result. For (c), VB < VP. Sulfur adsorption makes prism faces the most densely packed, and graphite flakes are subsequently formed.

presented at the 37th International Foundry Congress (Brighton, England), Sept 1970 R.J. Warrick, Spheroidal Graphite Nuclei in Rare Earth and Magnesium Inoculated Irons, Trans. AFS, Vol 76, 1968, p 722 M.M. Jacobs, T.J. Law, D.A. Melford, and M.J. Stowell, Basic Processes Controlling the Nucleation of Graphite Nodules in Chill Cast Iron, Met. Technol., Vol 1, Part II, No. 1974, p 490 O.M. Suarez, R.D. Kendrick, and C.R. Loper, Jr., Late Sulfur Additions to Postinoculated High-CE Ductile Iron Melts, Trans. Am. Foundrymens Soc., Vol 108, 2000, p 63–69 J. Rickard and I.C.H. Hughes, Eutectic Structure in White Cast Iron, BCIRA J., Vol 9 (No. 1), 1961, p 11 G.F. Ruff and J.F. Wallace, Control of Graphite Structure and Its Effect on Mechanical Properties of Gray Iron, Trans. AFS, Vol 84, 1976, p 705 B. Lux, F. Mollard, and I. Minkoff, On the Formation of Envelopes Around Graphite in Cast Iron, The Metallurgy of Cast Iron, B. Lux et al., Ed., Georgi Publishing, 1975, p 371 H. Jones and W. Kurz, Growth Temperatures and the Limits of Coupled Growth in Unidirectional Solidification of Fe-C Eutectic Alloys, Metall. Trans. A, Vol 11, 1980, p 1265 D.M. Stefanescu, P.A. Curreri, and M.R. Fiske, Microstructural Variations Induced by Gravity Level During Directional Solidification of Near-Eutectic IronCarbon Type Alloys, Metall. Trans. A, Vol 17, 1986, p 1121 S.K. Biswal, D.K. Bandyopadhyay, and D.M. Stefanescu, unpublished research A. Rickert and S. Engler, Solidification Morphology of Cast Irons, The Physical Metallurgy of Cast Iron, Vol 34, H. Fredriksson and M. Hillert, Ed., Proceedings of the Materials Research Society, North Holland, 1985, p 165 D.K. Banerjee and D.M. Stefanescu, Structural Transitions and Solidification Kinetics of SG Cast Iron during Directional Solidification Experiments, Trans. Am.

Fig. 26

Schematic of the sequence of development of compacted/vermicular graphite: (a) small spheroids; (b) and (c), some spheroids have tails; (d) compacted graphite plus spheroidal graphite; and (e) compacted graphite

328 / Principles of Solidification

Fig. 29

Change of direction of the tilt of the boundary due to the amount of reactive impurity. (a) Insufficient spheroidization. (b) Sufficient spheroidization. Source: Ref 40

(a)

(b)

Fig. 27

(a) SEM micrograph of deep-etched compacted graphite iron sample. (b) Reconstructed three-dimensional form. Source: Ref 31

Fig. 28

Diagram of sequences (a) through (c) involved in the formation of a twin/tilt plane for graphite growing in the [1010] direction. Source: Ref 40

Foundrymens Soc., Vol 99, 1991, p 747– 759 30. D.M. Stefanescu, Cast Iron, Casting, Vol 15, ASM Handbook, ASM International, 1988 31. A. Velichko, C. Holzapfel, and F. Mu¨cklich, 3-D Characterization of Graphite Morphologies in Cast Iron, Adv. Eng. Mater., Vol 9 (No.1–2), 2007, p 39–45 32. H. Fredriksson and S.E. Wetterfall, A Study of Transition from Undercooled to Flake Graphite in Cast Iron, The

Metallurgy of Cast Iron, B. Lux et al., Ed., Georgi Publishing, 1975, p 227 33. S.E. Wetterfall, H. Fredriksson, and M. Hillert, Solidification Process of Nodular Cast Iron, J. Iron Steel Inst., May 1972, p 323 34. K.C. Su, I. Ohnaka, I. Yamauchi, and T. Fukusako, Computer Simulation of Solidification of Nodular Cast Iron, The Physical Metallurgy of Cast Iron, Vol 34, H. Fredriksson and M. Hillert, Ed., Proceedings of the Materials Research Society, North Holland, 1985, p 181

35. H. Geilenberg, A Critical Review of the Crystallization of Graphite from Metallic Solutions After the “Surface Tension Theory,” Recent Research on Cast Iron, H.D. Merchant, Ed., Gordon and Breach, 1968, p 195 36. J.P. Sadocha and J.E. Gruzleski, The Mechanism of Graphite Spheroid Formation in Pure Fe-C-Si Alloys, The Metallurgy of Cast Iron, B. Lux et al., Ed., Georgi Publishing, 1975, p 443 37. I. Minkoff and B. Lux, Graphite Growth from Metallic Solution, The Metallurgy of Cast Iron, B. Lux et al., Ed., Georgi Publishing, 1975, p 473 38. K. Herfurth, Investigations into the Influence of Various Additions on the Surface Tension of Liquid Cast Iron with the Aim of Finding Relationships Between the Surface Tension and the Occurrence of Various Forms of Graphite, Freib. Forschungsh., Vol 105, 1965, p 267 39. E.N. Pan, K. Ogi, and C.R. Loper, Jr., Analysis of the Solidification Process of Compacted/Vermicular Graphite Cast Iron, Trans. AFS, Vol 90, 1982, p 509 40. P. Zhu, R. Sha, and Y. Li, Effect of Twin/ Tilt on the Growth of Graphite, The Physical Metallurgy of Cast Iron, Vol 34, H. Fredriksson and M. Hillert, Ed., Proceedings of the Materials Research Society, North Holland, 1985, p 3 41. X. Den, P. Zhu, and Q. Liu, Structure and Formation of Vermicular Graphite, The Physical Metallurgy of Cast Iron, Vol 34, H. Fredriksson and M. Hillert, Ed., Proceedings of the Materials Research Society, North Holland, 1985, p 141 42. M. Hillert and H. Steinhauser, The Structure of White Iron Eutectic, Jernkontorets Ann., Vol 144, 1960, p 520 43. M. Hillert and V.V. Subba Rao, “The Solidification of Metals,” Publication 110, The Iron and Steel Institute, 1968

Solidification of Eutectic Alloys—Cast Iron / 329

Fig. 30

Transition from (a) flake to compacted graphite and from (b) spheroidal to compacted graphite based on the twin/tilt of boundaries growth mechanism. Source: Ref 41

1170

1153 ⬚C

Spontaneous change from white to gray mode of solidification

B

1150

2120

2102

1148 ⬚C 2084

1140 Spontaneous change from gray to white mode of solidification

A

1130 0.01

Temperature, ⬚F

Temperature, ⬚C

1160

2066 10−3

10−4

10−5

10−6

Growth rate (R ), cm · s−1

Fig. 32

Gray-to-white eutectic transition as a function of undercooling and growth rate. A, white structure; B, gray structure. Source: Ref 43

Fig. 31

Schematic illustrating growth of ledeburite (austenite-iron carbide) eutectic. (a) Lamellar eutectic with cementite as the leading phase in the

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 330-337 DOI: 10.1361/asmhba0005214

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Peritectic Solidification THE TERM PERITECTIC, in the science of heterogeneous equilibria, may be used to define all reactions in which two or more phases (gas, liquid, solid) react at a defined temperature, Tp, to form a new phase that is stable below Tp. Usually, the term peritectic refers to reactions in which a liquid phase reacts with at least one solid phase to form one new solid phase. This reaction can be written as a + liquid ! b. Furthermore, the term peritectoid denotes the special case of an equilibrium phase in which two or more solid phases (which are stable above the temperature Tp) react at Tp to form a new solid phase. This reaction can be written as a + b ! g. The phases formed during a peritectic or peritectoid reaction are a solid solution of one of the components, an allotropic phase of one of the components, or an intermetallic base. Schematics of peritectic phase diagrams are shown in Fig. 1. Many interesting alloys undergo these types of reactions, for example, iron-carbon- and iron-nickel-base alloys as well as copper-tin and copper-zinc alloys. Controlled peritectic reactions and transformations are rarely used to optimize the microstructure of engineered materials. Usually, peritectic structures are avoided because of their detrimental effect on material properties. However, an understanding of peritectic solidification structure development is necessary in order to concisely interpret microstructures. Also, there may be some technical importance attributed to the possibility of producing metallic alloys with aligned fibers or plates (Ref 2–5) and of achieving grain refinement by means of peritectic decay of primary crystals (Ref 6, 7).

Phase Diagrams Phase diagrams are very instructive when describing peritectic phase transformations. Figure 2 shows a phase diagram with a peritectic reaction. This diagram shows that, under equilibrium conditions, all alloys to the left of I will solidify to a-crystals. Similarly, all alloys to the right of III will solidify to b-crystals. Alloys between II and III first solidify to a-crystals and then transform to stable b-crystals. Alloys between I and II also solidify to a-crystals, but they are partially transformed to b-crystals later. The volume fraction of each phase will be given by the lever rule if the alloy solidifies under equilibrium conditions (the lever rule is defined in the article “Thermodynamics and Phase Diagrams” in this Volume). In practice, the lever rule usually will not give the volume fraction of the different phases, because the nucleation and growth kinetics, determined by the diffusion rate, in the solid phases determine the time required to reach equilibrium.

Peritectic Solidification Peritectic structures form by means of at least three mechanisms:  Peritectic reaction, where all three phases (a, b,

 Direct precipitation of b from the melt, when

enough undercooling below the peritectic temperature (Tp) occurs. Direct precipitation of b from the melt can occur when a peritectic reaction or peritectic transformation is sluggish, as is often the case. Peritectic reactions can proceed only as long as a-phase and liquid are in contact. The b-solid nucleates at the a/liquid interface, readily forming a layer that isolates a from the liquid. This mechanism occurs through short-range diffusion, as shown in Fig. 3(a). In contrast, peritectic transformations occur through a long-range mechanism, where A atoms and B atoms migrate through the a-layer to form solid b at the a/b and the b/liquid interfaces, respectively (Fig. 3b). These distinctions between peritectic reaction and peritectic transformation were introduced by Kerr (Ref 8). The mechanism by which b has formed is not apparent from the final microstructure. In any case, all three of the mechanisms require some undercooling, because the driving force is zero at the peritectic temperature. The time dependence is pronounced for the peritectic transformation. Therefore, the amount of b formed will depend on the cooling rate or on the holding time if isothermal conditions are established. Localized nucleation of b and shape changes of b by solution reprecipitation, which is driven

and liquid) are in contact with each other

 Peritectic transformation, where the liquid

and the a-solid solution are isolated by the b-phase. The transformation takes place by long-range diffusion through the secondary b-phase.

Typical peritectic phase diagrams. (a) Peritectic reaction a + liquid ! b and peritectoid reaction a + b ! g. (b) Peritectic formation of intermetallic phases from a high-melting intermetallic. (c) Peritectic cascade between high- and low-melting components. Source: Ref 1

Fig. 1

Fig. 2

Phase diagram with a peritectic reaction

Peritectic Solidification / 331

Mechanisms of peritectic reaction and transformation. (a) Lateral growth of a b-layer along the a/liquid interface during peritectic reaction by liquid diffusion. (b) Thickening of a b-layer by solid-state diffusion during peritectic transformation. The solid arrows indicate growth direction of b; dashed arrows show the diffusion direction of the atomic species. Source: Ref 8

Fig. 3

Fig. 5

Start of the peritectic reaction in a directionally solidified Cu-20Sn alloy. Primary a-dendrites (white) are covered by peritectically formed b-layer (gray) shortly after the temperature reaches Tp. Matrix (dark) is a mixture of tin-rich phases. Mechanically polished, etched in HNO3. Original magnification: 40. Source: Ref 12

Types of peritectic systems. (a) Type A system where the b/a + b solvus and the b-solidus have slopes of the same sign. (b) Type B system where the slopes have opposite signs. (c) Type C system where the b-phase has a limited composition. Source: Ref 1

Fig. 4

by surface energy through diffusion in the liquid, also influence the microstructure of peritectic alloys. The tendency for final solidification microstructures to be far from equilibrium is another factor limiting their widespread commercial use. St. John and Hogan (Ref 9) ascertained that the degree of completion may be approximated based on certain characteristics of the phase diagram. From their work on the aluminumtitanium system and the work of Titchener and Spittle (Ref 10), three classes of peritectic phase diagram (Fig. 4) have been identified based on the shape of the peritectic (b) solid-solution region (Ref 11). Additional information is available in the article “Invariant Transformation Structures” in Metallography and Microstructures, Volume 9 of ASM Handbook.

Peritectic Reaction In the classical description of peritectic reactions, heterogeneous nucleation of b occurs at the a/liquid interface at the peritectic equilibrium temperature, Tp. If nucleation is limited to a few locations, and lateral growth of the b-nuclei does not readily occur, no continuous layer of the peritectic phase is formed. Two types of the peritectic reactions can occur:  Nucleation and growth of the b-crystals in

the liquid without contact with the a-crystals

 Nucleation and growth of the b-crystals in

contact with the primary a-phase

In the first case, the secondary phase is nucleated in the liquid and does not contact the primary phase because of surface tension conditions. Following nucleation, the secondary phase grows freely in the liquid. At the same time, the primary phase will dissolve. The secondary phase will not develop a morphology similar to a precipitated primary phase. This type of peritectic reaction has been observed for the reaction g + L ! b in the aluminummanganese system (Ref 12). The second phase tends to grow around the primary phase at increasing cooling rates. Similar reactions have been observed in nickel-zinc (Ref 13) and aluminum-uranium (Ref 14) systems. In the second type of reaction, which is the most common, nucleation of the secondary b-phase occurs at the interface between the primary a-phase and the liquid. A lateral growth of the b-phase around the a-phase then takes place. In an ideal peritectic reaction, undercooling is rather low (up to a few degrees Kelvin), and a plateau is observed in the cooling curve, as in the aluminum-titanium system (Ref 8). Envelopes of the peritectic phase around the primary phase form by direct reaction in some systems—copper-tin and silver-tin, for example— through interrupted directional solidification experiments (Ref 12). Figures 5 and 6 depict microstructures of Cu-20Sn and Cu-70Sn alloys

Fig. 6

Start of the peritectic reaction in a directionally solidified Cu-70Sn alloy. The primary e-phase (dark) is covered by the peritectically formed Z-layer (white), which thickens with increasing undercooling below Tp. The matrix is the Sn-Z eutectic. Mechanically polished, etched in HNO3. Original magnification: 100. Source: Ref 12

that demonstrate the onset of the peritectic reactions a + liquid ! b and e + liquid ! Z, respectively. Figure 7 shows, at a higher magnification, the homogeneous thickness of the blayer around the a-dendrites. The copper-tin phase diagram is shown in Fig. 8. Remelting is normally much faster than solidification, and there are no forces hindering dissolution. Therefore, dissolution will occur at the same rate as precipitation of the secondary phase. Thus, the peritectic reaction can proceed very rapidly by liquid diffusion over a very short distance in the lateral direction, as shown

332 / Principles of Solidification

Fig. 7

Start of the peritectic transformation in the same directionally solidified Cu-20Sn alloy shown in Fig. 5, but at higher magnification. Note the homogeneous thickness of the b-layers (gray) around the primary a (white). The matrix (dark) is a mixture of tinrich phases. Mechanically polished, etched in HNO3. Original magnification: 160. Source: Ref 12

Fig. 8 in Fig. 3(a). The thickness of the layers has been calculated with fairly good agreement to experimental results for copper-tin and silvertin alloys on the basis of maximum growth rate or minimum undercooling from the laws derived for solidification at low undercoolings (Ref 12). The thickness depends to some extent on the cooling rate but more strongly on the interfacial energies, s, with s (liquid/b) + s (a/b)  s (liquid/a) as the determining factor. Nucleation of the secondary phase on the primary phase is sometimes favored. When this occurs, a number of crystals will form around the primary phase, as illustrated in Fig. 9 (Ref 13).

Peritectic Transformation and Direct Precipitation The precipitation of b directly from the liquid and the solid depends on the shape of the phase diagram and the cooling rate. The diffusion process through the b-layer depends on the diffusion rate, the shape of the phase diagram, and the cooling rate. The thickness of the b-layer will normally increase during subsequent cooling. There are three reasons for this:  Diffusion through the b-layer  Precipitation of b directly from the liquid  Precipitation of b directly from the a-phase

The thickness of the b-phase envelope surrounding the a-phase is determined by the peritectic reaction followed by an increase in thickness due to a precipitation directly from the liquid. The rate of the peritectic transformation is influenced by the diffusion rate and the extension of the b-phase region in the phase

Copper-tin phase diagram. Source: Ref 15

diagram. If the diffusion rate is small, the peritectic transformation will be negligible compared to the peritectic reaction. During continuous cooling, this diffusion-controlled growth is affected by precipitation from the liquid and from the primary a or by dissolution of b, according to the slopes of the solubility limits in the phase diagram. Under simplifying conditions, the growth rates have been calculated numerically and found to be in reasonable agreement with the experimental findings in the copper-tin and silver-tin systems (Ref 12). The kinetics of peritectic transformations can more easily be studied under isothermal conditions. Then, b is formed exclusively by diffusion of the two atomic species in the b-layer and at the a/b and b/liquid interfaces. From calculations described in Ref 16 and 17, one can assume that the growth of the b-layer is controlled by the diffusion through it at a temperature just below the peritectic temperature. The diffusion process and the concentration profile and its relation to the phase diagram are illustrated in Fig. 10. A mass balance gives the following: a ! b : Db ¼

ðxb=L  xb=a Þ ‘b

d‘b=a ðxb=a  xa=b Þ dt



(Eq 1)

and L ! b : Db ¼

b=L

d‘ dt

ðxb=L  xb=a Þ ‘b

 ðxL=b  xb=L Þ

(Eq 2)

Fig. 9

Microstructure in a Cd-10Cu sample that has passed a peritectic reaction. The primary Cu5Cd8 crystals are white, the dark matrix is cadmium, and the peritectically formed CuCd3 is gray.

where ‘b is the thickness of the b-phase, t is the time, and Db is the diffusion coefficient in the b-phase. All other terms are concentrations that are defined in Fig. 10: d‘b d‘b=a d‘b=L Db ¼ þ ¼ b ðxb=L  xb=a Þ dt dt ‘ dt  1 1 þ L=b b=a a=b b=a ðx  x Þ ðx  x Þ





(Eq 3)

Peritectic Solidification / 333 dissolved alloying elements, the diffusion rates are much higher than for substitutionally dissolved elements in face-centered cubic metals. In this case, the diffusion process has a much larger influence on the precipitation process.

Iron-Carbon System A peritectic transformation occurs in the important iron-carbon system The iron-carbon phase diagram (Fig. 11) is very similar to the diagram illustrated in Fig. 10(a). A detailed numerical calculation of the transformation for one alloy has been performed (Ref 18). However, a simpler analytical model is used in this discussion (Ref 19). The model is the same as the isothermal model described earlier but is used with the assumption that the boundary conditions change during cooling. Equation 3 can therefore be used. In Eq 3, d‘b/dt can be expanded to (d‘b/dT)(dT/dt), where dT/dt is the cooling rate of the sample. Using the phase diagram, the concentrations can be transferred to a temperature. If the slope of the lines in the phase diagrams and the cooling rate are both assumed to be constant, Eq 3 can be simplified in the following way: ‘b d‘b ¼ A

Fig. 10

The peritectic transformation in a system with a high diffusion rate in the b-phase. (a) The phase diagram. (b) Concentration distribution. The dashed lines in the concentration profile are for a system with a low diffusion rate in the b-phase where the volume fraction of b increases with decreasing temperature. See Eq 6 and corresponding text. See also Fig. 13(b).

Integrating Eq 3 results in: ð‘b Þ2 ¼Db ðxb=L  xb=a Þ   1 1 þ ðxb=a  xa=b Þ ðxL=b  xb=L Þ



(Eq 4)

Equations 3 and 4 show that the growth rate increases with increasing undercooling. For example, at the peritectic temperature, the expression (xb/L  xb/a) is 0 and increases with increasing undercooling. Equations 3 and 4 also show that the growth rate is dependent on the diffusion coefficient. For substitutionally dissolved alloying elements in face-centered cubic metals, the diffusion coefficient near the melting point is of the order of  1013 m2/s. In such a case, the growth rate will be very low, and the time for the peritectic transformation will be unrealistically large. In a normal casting process, the reaction rate will be so low that the amount of b-phase formed by the peritectic reaction will be negligible in comparison with the precipitation of b from the liquid. For bodycentered cubic metals and for interstitially

dt

 dT  dT

(Eq 5)

where A is a constant. By integrating Eq 5 under the assumption of a constant cooling, one can obtain a relation that can be used to calculate the temperature interval under which the peritectic transformation takes place. This has been done in Ref 19, and the result of those calculations is shown in Fig. 12. Figure 12 shows that the peritectic reaction in iron-carbon alloys is very rapid and is finished at a maximum of 6 to 10 K below the peritectic temperature. The shape of the curve is a result of the change in the volume fraction of ferrite with the carbon content. The curve is in agreement with the results of other experiments (Ref 18, 20). As mentioned previously, the effect of the phase diagram must also be considered (Fig. 4). This was not taken into consideration when deriving Eq 3 and in the previous discussion for iron-carbon alloys. The effect of the phase diagram is illustrated in Fig. 10(a) and 13(a). Figures 10(a) and 13(a) show two different types of phase diagrams, in which the slopes of the a/b regions are different. In the first case (Fig. 10a), the b-layer will increase in thickness both at the interface a/b and at the interface b/L. In the second case (Fig. 13a), the b-layer will increase only at the side against the liquid and will decrease at the side against the a-phase. These two cases are somewhat more difficult to treat theoretically than the isothermal example or the case involving a high diffusion rate. The concentration profiles for the transformation are given in Fig. 10(b), as shown by the dashed lines, and Fig. 13(b). Equation 3 will now be changed in the following way:

 b dx dy y ¼ ‘a=b  b Db dx þ L=b x  xb=L dy y ¼ ‘b=L  b Da dx þ a=b x  xb=a dy y ¼ ‘a=b

d‘b Db ¼ dt ðxa=b  xb=a Þ

þ



‘L dxL xL=b  xb=L

(Eq 6)

where the first and second terms on the right side describe the increase in thickness due to diffusion into the b-phase and a-phase, respectively, from the boundary a/b. The third term describes the increase in thickness due to diffusion into the b-phase from the boundary b/L. The last term is the increase in the b-phase due to the changing of the composition in the liquid dxL with a decrease in the temperature dT according to the phase diagram. Solving Eq 6 is a difficult task that can require the use of numerical methods. The equation can be simplified by assuming that the diffusion distance corresponds to the thickness of the b-phase and that there are no concentration gradients in the a-phase. Figure 14 compares simulated and experimental determinations of the thickness of the secondary phase layer for a Cu-20Sn alloy (Ref 12), showing close agreement between the experiments and the theory. Equation 6 has been solved in Ref 19 by assuming that diffusion into the b-phase could be described in the same manner as in Ref 21 for back diffusion during segregation at a primary precipitation. The model was used to calculate the concentration distribution of austenite for iron-nickel alloys, as shown in Fig. 15. These calculations are in agreement with experiments reported in Ref 22 and 23.

Primary Metastable Precipitation of Beta In many systems, the secondary b-phase has been observed to form as a primary phase for alloys with a composition left of point III in Fig. 2. This is especially true for iron-base alloys (Ref 22, 23). The formation of a metastable b-phase directly from the liquid under different cooling rates can be demonstrated by examining the binary iron-nickel phase diagram. In Fig. 16, the metastable extensions of the equilibrium between liquid and ferrite and between liquid and austenite are represented by dashed lines. The melting point of pure iron as austenite is 11  C (20  F) below the melting point of pure iron as ferrite. It can be seen that there is a difference in the partition coefficient between ferrite and liquid on one hand and austenite and liquid on the other. The growth rates were calculated for needles of ferrite and austenite. The alloy composition chosen is to the left of the peritectic point, and the calculations of the growth rates were carried out for different undercoolings.

334 / Principles of Solidification

Fig. 11

Carbon-iron phase diagram. Source: Ref 15

Fig. 12

Temperature range of peritectic reaction in iron-carbon alloys as a function of carbon content and the solidification rate. The temperature gradient, G, is 6000 K/m.

Only ferrite can form at low undercoolings. Austenite can also form if the temperature is chosen just below the extension of its liquidus

line, but ferrite grows faster and should therefore dominate in the solidification process. However, below a critical line, austenite will have the highest growth rate and may dominate, although the temperature is still above the peritectic temperature. This kinetic advantage of austenite is due to its partition coefficient being smaller than that for ferrite. The critical line is indicated by a dashed line in Fig. 16. The partition coefficient of an element can be closer to unity for ferrite than for austenite in ternary alloys, and ferrite will then have the kinetic advantage and will be favored by a high cooling rate. It has also been reported that an aligned structure can be formed in peritectic systems (Ref 2). However, it has been argued that this type of structure is formed when a- and b-primary phases are growing at the same rate (Ref 12). In addition, the composition of the liquid must be chosen so the sum of the volume fraction of the two solid phases will be unity.

Peritectic Cascades The theoretical analysis shows that the rate of the peritectic transformation is influenced by the diffusion rate and the extension of the b-phase region in the phase diagram. If the diffusion rate is small, the peritectic transformation will be negligible compared to the peritectic reaction. The thickness of the b-phase envelope surrounding the a-phase is determined by the peritectic reaction followed by an increase in thickness due to a precipitation directly from the liquid. In many systems, one peritectic reaction at a high temperature is followed by one or more peritectic reactions at lower temperatures. If the diffusion rate is low in the initially formed peritectic layer, a second peritectic layer can be formed when the second peritectic temperature is reached. This type of series of peritectic reactions is referred to as a cascade (Ref 13). A characteristic microstructure formed by cascading peritectic reactions in a Sn-17Co alloy is shown in Fig. 17.

Peritectic Solidification / 335

Fig. 15

Nickel distribution after peritectic reaction in a steel containing 4 wt% Ni. The temperature gradient was 60 K/cm. Calculations were made at different solidification rates. The dotted line shows the nickel distribution at the start of the peritectic reaction. d is primary ferrite, g is austenite. Source: Ref 22

Fig.

13

The peritectic transformation during continuous cooling in a system with a low diffusion rate and where the volume fraction of b increases with decreasing temperature. (a) The phase diagram. (b) Concentration distribution

Fig. 17

Microstructure of a Sn-17Co alloy that was cooled to 1000  C (1830  F) and held 2 h, then cooled to 225  C (435  F) and held 22 h. The primary g-phase has completely transformed and dissolved into relatively small CoSn crystals that form the dark centers surrounded peritectically by CoSn2 layers (gray), followed by a peritectoid envelope of the CoSn3 phase. The temperatures of peritectic formation of CoSn and CoSn2 are approximately 910 and 540  C (1670 and 1005  F), respectively; CoSn3 forms peritectoidally at approximately 235  C (455  F). Scanning electron micrograph (backscattered electron image). Original magnification: 200

Fig. 16

The binary iron-nickel system. The broken line shows the undercooling at which ferrite, d, and austenite, g, grow at the same rate. Source: Ref 24

Peritectic Transformations in Multicomponent Systems

Fig. 14

Thickness of the secondary phase layer as a function of temperature below the peritectic temperature in the copper-tin system. The solidification rate was 100 mm/min (4 in./min). The diffusion units are given in cm2/s. The volume fraction is defined as ‘b/l, where l corresponds to the dendrite arm space.

Alloys often consist of more than two alloying elements. Investigations of iron-base alloys have shown that peritectic reactions are very common in stainless steels (Ref 22, 25). The peritectic reaction in these alloys gives the same type of nickel distribution as shown in Fig. 15. In stainless steels, the peritectic reaction will become a eutectic reaction if the chromium content is increased to 20% or more. This transition is also influenced by the molybdenum content, as shown in Fig. 18.

Fig. 18

The transition from a peritectic to a eutectic reaction as a function of chromium and molybdenum content in a stainless steel containing 11.9% Ni

Both chromium and nickel are substitutionally dissolved elements. Iron-base alloys often consist of carbon with some other elements. Carbon is interstitially dissolved and has a very high diffusion rate. The other alloying elements are primarily substitutionally dissolved with very low diffusion rates. This gives rise to

336 / Principles of Solidification

Fig. 19

Three stages of a peritectic reaction in a unidirectionally solidified high-speed steel. (a) First-stage structure. Dark gray is austenite, white is ferrite. The mottled structure is quenched liquid. (b) Subsequent peritectic transformation of (a). (c) Further peritectic transformation of (a) and (b). Dark gray in the middle of the white ferrite is newly formed liquid. Source: Ref 26

transformations that are determined by the movement of the substitutional elements, and carbon is distributed according to equilibrium conditions. As a result, a normal peritectic transformation does not occur. To fulfill the criterion that carbon should follow the equilibrium conditions, liquid must be formed at the border between ferrite and austenite. This reaction is illustrated in Fig. 19. This type of reaction has been both experimentally and theoretically analyzed in Ref 26. ACKNOWLEDGMENTS This article has been adapted from H. Fredriksson, “Solidification of Peritectics,” Casting, Vol 15, Metals Handbook, 9th ed., ASM International, 1988, p 125–129, and H.E. Exner et al., “Peritectic and Peritectoid Structures,” Metallography and Microstructures, Vol 9, ASM Handbook, ASM International, 2004, p 156–164. REFERENCES 1. H.E. Exner and G. Petzow, Peritectic Structures, Metallography and Microstructures, Vol 9, ASM Handbook, American Society for Metals, 1985, p 675–680 2. W.J. Boettinger, The Structure of Directionally Solidified Two-Phase Sn-Cd Peritectic Alloys, Metall. Trans. A, Vol 5, 1974, p 2023–2031 3. H.D. Brody and S.A. David, Controlled Solidification of Peritectic Alloys, Solidification and Castings of Metals, The Metals Society, London, 1979, p 144–151 4. A.P. Tichener and J.A. Spille, The Microstructure of Directionally Solidified Alloys That Undergo a Peritectic Transformation, Acta Metall., Vol 23, 1975, p 497–502 5. A. Ostrowski and E.W. Langer, Unidirectional Solidification of Peritectic Alloy,

6. 7.

8.

9. 10. 11. 12. 13. 14. 15. 16.

17. 18.

Solidification and Castings of Metals, The Metals Society, London, 1979, p 139–143 J.A. Sartell and D.J. Mack, The Mechanism of Peritectic Reactions, J. Inst. Met., Vol 93, 1964, p 19–24 I. Maxwell and A. Hellawell, An Analysis of the Peritectic Reaction with Particular Reference to Al-Ti Alloys, Acta Metall., Vol 23, 1975, p 901–909 H.W. Kerr, J. Cisse, and G.F. Bolling, On Equilibrium and Non-Equilibrium Peritectic Transformation, Acta Metall., Vol 22, 1974, p 677 D.H. St. John and L.M. Hogan, Acta Metall., Vol 35, 1987, p 171–174 A.P. Titchener and J.A. Spittle, Met. Sci., Vol 8, 1974, p 112–116 J.W. Rutter et al., Mater. Sci. Technol., Vol 14, 1998, p 182–186 H. Fredriksson and T. Nyle´n, Mechanism of Peritectic Reactions and Transformations, Met. Sci., Vol 16, 1982, p 283–294 G. Petzow and H.E. Exner, Zur Kenntnis Peritektischer Umwandlungen, Radex Rundsch., 3/4, 1967, p 534–539 S. Uchida, “Systematik and Kinetik Peritektischer Umwandlund,” Ph.D. thesis, Technical University, 1980 Alloy Phase Diagrams, Vol 3, ASM Handbook, ASM International, 1992, p 2.110 M. Hillert, Eutectic and Peritectic Solidification, Solidification and Casting of Metals, The Metals Society, London, 1979, p 81–87 D.H. St. John and L.M. Hogan, The Peritectic Transformation, Acta Metall., Vol 25, 1977, p 77–81 Y.K. Chuang, D. Reinisch, and K. Schwerdtfeger, Kinetics of the Diffusion Controlled Peritectic Reaction During Solidification of Iron-Carbon Alloys, Metall. Trans. A, Vol 6, 1975, p 235–238

19. H. Fredriksson and J. Stjerndahl, Solidification of Iron-Base Alloys, Met. Sci., Vol 16, 1982, p 575–585 20. J. Stjerndahl, “The Solidification Process of Iron Base Alloys,” thesis, Department of Casting of Metals, Royal Institute of Technology, 1978 21. H.D. Brody and M.C. Flemings, Solute Redistribution in Dendritic Solidification, Trans. Met. Soc. AIME, Vol 236, 1966, p 615–624 22. H. Fredriksson, The Solidification Sequence in an 18-8 Stainless Steel, Metall. Trans., Vol 3, 1972, p 2989–2997 23. H. Fredriksson and J. Stjerndahl, On the Formation of a Liquid Phase During Cooling of Steel, Metall. Trans. B, Vol 6, 1975, p 661 24. H. Fredriksson, Segregation Phenomena in Iron-Base Alloys, Scand. J. Metall., Vol 5, 1976, p 27–32 25. H. Fredriksson, The Mechanism of the Peritectic Reaction in Iron-Base Alloys, Met. Sci., Vol 11, 1976, p 77–86 26. H. Fredriksson, Transition from Peritectic to Eutectic Reaction in Iron-Base Alloys, Solidification and Casting of Metals, Publication 192, The Metals Society, London 1977, p 131–136 SELECTED REFERENCES  I.V. Belyaev, M. Kursa, and Y. Drapala,

Directed Crystallization of Peritectic Transformed Alloys, METAL 2007: 16th International Metallurgical and Materials Conference, 2007  C. Bernhard and G. Xia, Influence of Alloying Elements on the Thermal Contraction of Peritectic Steels during Initial Solidification, Ironmaking Steelmaking, Vol 33 (No. 1), 2006, p 52–56

Peritectic Solidification / 337  B.K. Dhindaw et al., Characterization of the





 









Peritectic Reaction in Medium-Alloy Steel Through Microsegregation and Heat-ofTransformation Studies, Metall. Mater. Trans. A, Vol 35 (No. 9), 2004, p 2869–2879 B.K. Dhindaw and H. Fredriksson, Characterisation of the Peritectic Reaction in Medium Alloy Steel Through Directional Solidification and Micro-Segregation Studies, Trans. Indian Inst. Met., Vol 56 (No. 5), 2003, p A5 R.J. Dippenaar, High-Temperature Phase Transitions and Mechanical Properties in Steel of Peritectic Composition, Mater. Sci. Forum, Vol 539–543, 2007, p 4243–4248 S. Dobler et al., Peritectic Coupled Growth, Acta Mater., Vol 52 (No. 9), 2004, p 2795– 2808 A. Karma and M. Plapp, New Insights into the Morphological Stability of Eutectic and Peritectic Coupled Growth, JOM, Vol 56 (No. 4), 2004, p 28–32 E. Kozeschnik, W. Rindler, and B. Buchmayr, Scheil-Gulliver Simulation with Partial Redistribution of Fast Diffusers and Simultaneous Solid-Solid Phase Transformations, Int. J. Mater. Res. (Z. Metallkd.,) Vol 98 (No. 9), 2007, p 826–831 Y. Liu et al., Primary Cellular/Dendritic Spacing Selection in Rapidly Solidified Peritectic Alloy, Mater. Lett., Vol 59 (No. 23), 2005, p 2915–2919 T.A. Lograsso, B.C. Fuh, and R. Trivedi, Phase Selection during Directional Solidification of Peritectic Alloys, Metall. Mater. Trans. A, Vol 36 (No. 5), 2005, p 1287–1300 D. Ma et al., On Secondary Dendrite Arm Coarsening in Peritectic Solidification,









 





Mater. Sci. Eng. A, Vol 390 (No. 1–2), 2005, p 52–62 N.J. McDonald and S. Seetharaman, “Peritectic Transition in the Hypoperitectic IronNickel System,” Ferrous Alloys, April 27–30, 2003 (Indianapolis, IN) N.J. McDonald and S. Sridhar, Peritectic Reaction and Solidification in Iron-Nickel Alloys, Metall. Mater. Trans. A, Vol 34 (No. 9), 2003, p 1931–1940 N. McDonald and S. Sridhar, Observations and Analysis of the Peritectic Reaction for Fe-Co Alloys, Z. Metallkd., Vol 96 (No. 3), 2005, p 304–310 J. Miettinen, Thermodynamic-Kinetic Model for the Simulation of Solidification in Binary Copper Alloys and Calculation of Thermophysical Properties, Comput. Mater. Sci., Vol 36 (No. 4), 2006, p 367–380 J.H. Perepezko, Nucleation Controlled Phase Selection during Solidification, Mater. Sci. Eng. A, Vol 413–414, 2005, p 389–397 D. Phelan, M. Reid, and R. Dippenaar, Experimental and Modelling Studies into High Temperature Phase Transformations, Comput. Mater. Sci., Vol 34 (No. 3), 2005, p 282–289 D. Phelan, M. Reid, and R. Dippenaar, Kinetics of the Peritectic Phase Transformation: In-Situ Measurements and Phase Field Modeling, Metall. Mater. Trans. A, Vol 37 (No. 3) 2006, p 985–994 R. Pierer and C. Bernhard, High Temperature Behavior during Solidification of Peritectic Steels Under Continuous Casting Conditions, MS&T ’06: Materials Science and Technology 2006 Conference and Exhibition, 2007

 S. Raghavan, M.J.M. Krane, and D. Johnson,

 













“Different Rule Sets for Cellular Automata Modeling of Peritectic Dendritic Growth,” Modeling of Casting, Welding and Advanced Solidification Processes—X, May 25–30, 2003 (Destin, FL) L. Snugovsky et al., Phase Equilibria in SnRich Corner of Cu-Ni-Sn System, Mater. Sci. Technol., Vol 22 (No. 8), 2006, p 899–902 Y. Su et al., Microstructure Evolution of Ti-Al Peritectic System during the Initiated Stage of Directional Solidification, Mater. Sci. Forum, Vol 546–549, 2007, p 1447–1450 Y. Su et al., Microstructure Selection during the Directionally Peritectic Solidification of Ti-Al Binary System, Intermetallics, Vol 13 (No. 3–4), 2005, p 267–274 Y. Su et al., Numerical Model for Nucleation of Peritectic Alloy during Unidirectional Solidification, J. Mater. Sci., Vol 42 (No. 14), 2007, p 5360–5368 M. Sumida, Growth Analysis of Banded Microstructure Formed during Directional Solidification of Peritectic Alloys, Mater. Trans., Vol 45 (No. 2), 2004, p 388–391 R. Trivedi, The Role of Heterogeneous Nucleation on Microstructure Evolution in Peritectic Systems, Scr. Mater., Vol 53 (No. 1), 2005, p 47–52 R. Trivedi and J.H. Shin, Modelling of Microstructure Evolution in Peritectic Systems, Mater. Sci. Eng. A, Vol 413–414, 2005, p 288–295 M.F. Zhu and C.P. Hong, Modeling of Microstructure Evolution in Eutectic and Peritectic Solidification, Modeling of Casting, 2003

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 338-347 DOI: 10.1361/asmhba0005215

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Microsegregation Harold D. Brody, University of Connecticut

MICROSEGREGATION is the nonuniform distribution of alloying elements within a volume characteristic of the solidification microstructure, that is, chemical nonuniformity that can be observed and measured on the scale of the dendrite spacing and/or the grain size. The cored dendritic structure of a cast Ti-15%Mo alloy is shown, for example, in Fig. 1. Microsegregation results when the partitioning and redistribution of solute between liquid and solid phases that accompany the solidification of alloys does not reach equilibrium, and uniform chemical potential is not attained within all phases of the as-cast alloy. Diffusion, typically diffusion of solute in the primary solid solution, is the rate-limiting step in the redistribution of solute during solidification and postsolidification cooling and is the dominant factor in determining the extent of microsegregation in as-cast alloys. In contrast, solute redistribution processes that do not preserve the local solute concentration result in macrosegregation. Bulk convective flow, typically of solute-enriched (or solute-depleted) liquid in the mushy zone, is the dominant factor in determining the extent of macrosegregation in as-cast alloys. The segregation ratio, S, the local maximum solute content divided by the local minimum, and the fraction of nonequilibrium constituents, f nsq, are two, among several, measures of the extent of microsegregation. Because solute redistribution processes continuously alter the apparent microstructure during solidification and postsolidification cooling, interrupted solidification and in situ imaging are used to follow the kinetics of processes that lead to microsegregation. Microsegregation and other manifestations of nonequilibrium solidification limit castability, further processability, and the final properties of the alloy. Microsegregation and its effects can be eliminated or reduced by homogenization and solution heat treatments, which, in the case of wrought alloys, are combined with mechanical deformation processes to reduce the scale of the solidification microstructure.

Solidification Models In general, alloys freeze over a temperature range, and the solute is continually redistributed,

or partitioned, among solid and liquid phases in an attempt to maintain chemical equilibrium. Microsegregation results when overall equilibrium is not attained. For detailed modeling of solute redistribution and microsegregation, see the article “Modeling of Microsegregation and Macrosegregation” in this Volume. Herein, the process and material parameters that control microsegregation are discussed in relation to the manifestations of microsegregation in simple and then increasingly complex alloy systems. First, the two extremes of solute redistribution are discussed, equilibrium solidification and nonequilibrium Gulliver-Scheil solidification, for which solid redistribution of solute within the primary solid phase is the distinguishing parameter. C The partition ratio, ki ¼ CLiSi (or distribution coefficient), describes the difference in compo sition for solute i of the liquid, CLi , and solid,  CSi , phases that would be at equilibrium at a temperature within the freezing range as given by the equilibrium tie line. (See the article “Thermodynamics and Phase Diagrams” in this Volume for phase diagram fundamentals.) For ki < 1, the liquid is enriched in solute i during solidification. For ki > 1, the liquid is depleted in solute i during solidification. The further ki is from unity, the greater the potential for microsegregation. If the liquidus and solidus for a terminal solid solution are straight, the partition ratio is a constant.

can be used to compute the progress of solidification: fS ¼

CL  CO T  TL  CL  CS ðT  TM Þð1  kÞ

(Eq 1)

where fS is the fraction solid, TM is the equilibrium melting temperature, TL is the liquidus temperature for the alloy, and the latter approximate ratio assumes straight liquidus and solidus phase boundaries. Equilibrium is one extreme of solidification behavior (Ref 2–4).

Equilibrium Solidification For solidification to proceed through the freezing range, the equilibrium lever law requires that there must be (1) negligible undercooling before the nucleation of the solid phases; (2) negligible undercooling at the solid-liquid interface due to curvature, interface kinetics, or a solute boundary layer (local equilibrium at the solid-liquid interface); (3) sufficient mass transport in the remaining liquid phase for the composition of the liquid to remain uniform as it enriches (or depletes) in solute; (4) constant local average solute concentration, that is, negligible macrosegregation; and (5) sufficient mass transport in the forming solid phase(s) to maintain uniform chemical potential. The equilibrium lever rule

Fig. 1

As-cast microstructure showing dendritic coring in Ti-15%Mo alloy, and electron microprobe trace showing variation of molybdenum content across dendrite structure. Source: Ref 1

Microsegregation / 339 Nonequilibrium Solidification, The Gulliver-Scheil Relation In most commercial casting and welding practices, and for many alloys, equilibrium freezing would not be attained. The freezing range would be extended. The primary dendritic phase would be cored. Nonequilibrium phases will be formed. The nonequilibrium lever rule proposed by Gulliver and Scheil describes another extreme of solute redistribution during solidification. Conditions 1 through 4 listed previously are assumed by the Gulliver-Scheil relation, which may be used to compute the progress of solidification for condition 5, negligible redistribution of solute within solid phases by diffusion or other processes: fS ¼ 1 

 1=1k   CO TL  TM 1=1k  1  (Eq 2) CL T  TM

The composition of the primary phase forming at the solid-liquid interface is given by another form of the Gulliver-Scheil relation: CS ¼ ð1  fS Þk1

(Eq 3)

The composition of the remaining liquid phase changes continuously and uniformly from the initial melt composition, CO, and the composition of the primary solid phase changes from kCO. Solidification is not complete until an invariant reaction, such as the melting point or a eutectic, is reached. The Gulliver-Scheil conditions lead to the maximum extent of microsegregation (Ref 2–7).

Nonequilibrium Solidification with Back Diffusion The actual solidification path and the extent of microsegregation for a cast alloy will be somewhere between the two extremes. The conditions 1 through 4 represent the solidification of most alloys over a wide range of welding and casting processes. The kinetics of solute redistribution in the solid phase(s), condition 5, to the first approximation, depend on solute diffusion within the local characteristic volume. The mass transport Fourier number, a, is a useful predictor of the extent of microsegregation: a¼

DS tdiff DSðTL Þ yf 
das 2 L2diff

(Eq 4)

2

where the characteristic time for diffusion, tdiff, is approximated by the local solidification time, yf, and the characteristic diffusion distance, Ldiff, is approximated by one-half the final dendrite arm spacing (das) for the local region of the casting under consideration (Ref 1–4, 8–22). For a highly grain-refined casting, onehalf the grain size could be used to approximate Ldiff (Ref 20). The solid diffusion coefficient, DS, changes exponentially with temperature

during solidification and during postsolidification cooling. For simplicity, the diffusivity around the liquidus temperature, DSðTL Þ , can be selected to estimate a. The transition from the extremes of Gulliver-Scheil to equilibrium behavior occurs as 0.01 < a < 100. For low-melting-temperature alloys, microsegregation approaches the Gulliver-Scheil condition. For high-melting-temperature alloys, and especially for interstitial solute elements, solute redistribution tends to equilibrium. The measurement and kinetics of microsegregation are discussed for the binary isomorphous system, titanium-molybdenum; for binary eutectic systems, aluminum-copper and aluminum-silicon; for a binary peritectic system, copper-Zinc; for a multicomponent eutectic system, Al-Si-Cu-Mg; and, finally, for systems with both eutectic and peritectic reactions, Fe-C-Cr and nickel-base superalloy.

Binary Isomorphous Systems, k > 1: Titanium-Molybdenum For many titanium-base alloys, as-cast microsegregation is close to the minimum— approaching equilibrium—and the observed segregation ratios are close to unity. In those titanium-base alloy systems that freeze over a narrow temperature range close to or above the melting point, the partition ratios are close to 1, and solute diffusivities in b-titanium are relatively high. Solute redistribution is extensive in b-titanium dendrites during solidification. Microsegregation present at the end of freezing is reduced by diffusion as the alloy cools in the mold from the high solidus temperatures. The relatively high diffusivities of alloying elements in b-titanium are countered by relatively large as-cast dendrite arm spacing (when compared, for example, to aluminum alloys with local solidification times), presumably the result of relatively high solute diffusivities in the liquid. Solute redistribution is less extensive and microsegregation more noticeable when alloying elements that lower the freezing range, such as iron, are added to titanium (Ref 1, 21, 22). Titanium-molybdenum is an isomorphous system. The liquidus and solidus boundaries

slope upward from the melting temperature of titanium to the melting point of molybdenum. For the compositions 5 < wt% Mo < 30%, k  1.4. As titanium-molybdenum alloys freeze, the liquid is depleted of molybdenum and enriched in titanium. The equilibrium freezing ranges, DTf, for these binary alloys (Table 1) are relatively narrow. The liquidus temperature, TL, for a Ti-5%Mo alloy would be approximately 1700  C (3100  F) and for a Ti-30%Mo alloy would be approximately 1870  C (3400  F) (Table 1) (Ref 1). At equilibrium, titanium-molybdenum alloys would freeze as uniform single-phase b-titanium. According to the Gulliver-Scheil assumptions, all the alloys would be cored b-titanium with G-S , at the maximum molybdenum content, Cmax the dendrite centers and pure titanium in G -S the interdendritic boundaries ðCmin ¼ 0%MoÞ. The effective solidus temperature, according to the Gulliver-Scheil conditions, would be the melting point of pure titanium for all titanium-molybdenum alloys. Figure 1 shows the microstructure of an as-cast Ti-15%Mo alloy polished and etched to reveal the coring. Above the micrograph is a plot of the molybdenum content measured by an electron microprobe line scan. As expected, the maximum molybdenum content is at the dendrite centers, and the minimum molybdenum content is at the interdendritic meas boundaries. Table 1 shows the values of Cmax meas and Cmin measured after cooling to room temperature for Al-5, 10, 15, and 30% Mo alloys arc melted and drop cast into copper chill molds. Table 1 includes the segregation ratio computed from the listed data and a residual segregation index (SI): SI ¼

meas meas Cmax  Cmin G -S G S Cmax  Lmin

(Eq 5)

where the superscript “meas” indicates a measured composition, and the superscript “G-S” indicates a composition computed from the Gulliver-Scheil relation. Both measures of the extent of segregation in titanium-molybdenum alloys are compared to the estimated value of a, the mass transport Fourier number, which is based on measured solidification times, and

Table 1 Solidification and microsegregation data for isomorphous titanium-molybdenum alloy castings DTf

TL Mo, wt%



C



F



C



F

-S CG max , wt%

5

1701 3094

10 18

7

10

1734 3153

19 34

14

15

1767 3213

29 52

21

30

1866 3391

57 103

52

Source: Ref 1

Cmeas max , wt%

Cmeas min , wt%

S, Cmax Cmin

meas meas SI, Cmax  Cmin G-S -S CG max  Cmin

uf, s

das, mm

5.4 5.3 11.5 11.1 17.4 16.8 35.6 34.6

4.0 4.2 8.5 8.8 12 12.5 22.8 24.6

1.35 1.26 1.36 1.27 1.45 1.34 1.56 1.41

0.20 0.16 0.21 0.17 0.26 0.20 0.30 0.24

0.9 29 1 31 1.1 33 7.6 131

28 130 33 140 26 128 32 128

a,

4Ds uf das2

0.06 0.09 0.06 0.09 0.03 0.04 0.03 0.035

340 / Principles of Solidification dendrite arm spacing and an estimate of DS at the liquidus temperature of the alloy. The qualitative trend is excellent. Both indices show a decrease in extent of microsegregation with increasing values of a. The trend is slightly stronger with the segregation index (Ref 1).

Binary Eutectic Systems, k < 1: Hypoeutectic Aluminum-Copper Alloys

the composition of the solid changes continuously to be equal to the interface composition (indicated by the dotted curve in Fig. 3), and the final distribution of solute (between the equilibrium solidus and equilibrium solvus temperatures) is uniformly CO, the same as the initial melt composition. For the Gulliver-Scheil conditions, there is no redistribution of solute behind the solid-liquid interface, and the

composition that forms in the solid at the interface remains unchanged as the region cools to the eutectic temperature, where solidification is complete, and continues unchanged as the region cools to the ambient temperature. Nonequilibrium Solidification with Back Diffusion—Directional Solidification. A mass transport Fourier number of a = 0.5 represents well the measured behavior for a

660 1200

1150

Eq

620

600 1100 a = 0.5

Temperature, ⬚F

Temperature, ⬚C

640

580 G-S 1050 560

540 0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1000 1.00

Fraction solid

Fig. 2

Solidification curves (temperature vs. weight fraction solid) for three cases of Al-4.5%Cu alloy: equilibrium (Eq: a ! 1), Gulliver-Scheil (G-S; a ! 0), and directional solidification with a = 0.5

6.00

5.00

Cu concentration, wt%

Two convenient ways to represent the extent of microsegregation are the solidification curve, the progress of solidification with temperature in the freezing range, and the solute distribution curve, the fraction of solid within the characteristic volume enclosed by a range of isoconcentration surfaces. These curves are illustrated in Fig. 2 and 3 for a model binary eutectic system, aluminum-rich aluminum-copper alloys, in which much of the initial research on alloy solidification has been reported (Ref 3, 8–11). The melting temperature of aluminum is 660.2  C (1220.4  F); the eutectic temperature is 547.8  C (1018.8  F) for the invariant reaction: liquid (33.2% Cu) ! a-aluminum (5.65% Cu) + y-phase (CuAl2); and the a-phase solvus recedes by approximately 1% Cu per 50  C (90  F) below the eutectic. Over the freezing range for hypoeutectic aluminum-copper alloys, 0.14 < k < 0.17. In practice, an Al-4.5%Cu alloy would not solidify completely until it cools to its eutectic temperature. The eutectic microconstituent, a-aluminum + y-CuAl2, would form in the last regions to freeze locally. The primary solidification phase would be cored—a gradient in copper content will be apparent on polished and etched surfaces. Solidification Curve. The progress of solidification with temperature in the freezing range, the solidification curve, for an Al-4.5%Cu alloy is compared for the two extremes in Fig. 2. For equilibrium freezing, the alloy is 100% solid at the equilibrium solidus. For the Gulliver-Scheil conditions, the Al-4.5%Cu alloy is 88% solid (by weight) at the equilibrium solidus temperature and 91% solid at the eutectic temperature. The liquid remaining at the eutectic temperature transforms to the eutectic microconstituent, 9% by weight. The freezing range is extended (and the incipient melting point is lowered) by 25  C (45  F). The solidification curve is shallow near the liquidus temperature and becomes progressively steeper as it approaches the eutectic temperature. Solute Distribution Curve. The solute distribution curve (solute profile) is a plot of the solute content versus the fraction of solid within a characteristic volume that would be enclosed by the isoconcentration surface at the composition indicated on the ordinate. Figure 3 compares, for the two extremes, the copper concentration in the solid phase(s) at the solid/ liquid interface as a function of the cumulative fraction of solid. For equilibrium freezing,

Eq-final

4.00

G-S

3.00

a = 0.5

2.00

Eq-intf

1.00

0.00 0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

Fraction solid

Fig. 3

Solute distribution curve (weight percent copper vs. weight fraction solid contained) for three cases of Al-4.5% Cu alloy: equilibrium (Eq: a ! 1), Gulliver-Scheil (G-S; a ! 0), and directional solidification with a = 0.5. For equilibrium, the final as-cast solute distribution (Eq-final) and the solid composition at each stage of solidification (Eq-intf) are shown.

Microsegregation / 341 directionally solidified experimental casting of Al-4.5%Cu alloy (Ref 9,10). The solidification curve (Fig. 2) and as-cast copper distribution (Fig. 3) indicate that the freezing range still extends from the equilibrium liquidus to the eutectic temperature, where approximately 8 wt% of eutectic microconstituent will form. The minimum copper content is approximately 1.35% Cu, and the copper profile increases slowly across the first 40% of the primary phase. The maximum copper content in the primary phase is still 5.65%. As a result of back diffusion, the average copper content of the primary a-phase is 2% as compared to 1.65% for the Gulliver-Scheil conditions. Most of the copper segregates to the last region to freeze locally, the eutectic microconstituent, which has an average composition of 33.2% Cu. The segregation ratio, based on the primary phase, nsq will be S  5:65=1:35 ¼ 4:2 and fsut  8% for a = 0.5, as opposed to the extremes: nsq S  5:65=0:69 ¼ 8:2 and fsut  9% for Gullinsq ¼ 0% ver-Scheil conditions and S = 1 and fsut for equilibrium. Note that in all cases, the rate of solidification with temperature is very high near the liquidus temperature; that is, the initial slope of the solidification curve, @T =@fS , is shallow. For the Al-4.5%Cu alloy, 50% of the solid forms within the first 10 of the freezing range. Near the end of primary phase formation, the solidification curve is steep; the fraction solid changes slowly as the remaining liquid and now coherent solid coexist over the majority of the temperature range of freezing. For nonequilibrium solidification, the steep portion of the solidification curve leads to the isothermal transformation of the eutectic liquid to the eutectic mixture at the eutectic temperature. Concomitantly, the initial slope of the solute distribution curve, @C=@fS , is shallow, becoming progressively steeper as the eutectic composition is approached. Influence of Local Cooling Rate. The mass transport Fourier number, a, and the extent of back diffusion depend on the local solidification time and the reciprocal of the square of the characteristic length. The as-cast dendrite arm spacing has been found to depend on the local solidification time to a power, n, between 0.3 and 0.5. Using n = 0.39, the exponent found 1=5 for aluminum-copper alloys is a / yf . The extent of microsegregation decreases, but very slowly, as the local solidification time increases, and cooling rate decreases. Table 2 illustrates

Table 2

the decrease in segregation ratio and fraction eutectic as the local solidification time varies from a local solidification time and dendrite spacing characteristic of a weld (a = 0.1), to permanent mold and sand castings, to extremely low cooling rates. Over this wide range of processing conditions, the variation in microsegregation is minor. Slow cooling and long solidification times are not effective in reducing microsegregation. Conversely, short solidification times, to obtain a fine microstructure, followed by slow cooling in the mold or by solution heat treatment are more effective in removing microsegregation. Dendrite Coarsening. Dendrite (and grain) coarsening that involves remelting can be another source for solute redistribution in solid phases during solidification. The apparent dendrite arm spacing increases during the time solid and liquid phases coexist. Several mechanisms for dendrite coarsening have been proposed and observed (Ref 15, 16, 18, 23–26). Those coarsening mechanisms that involve remelting of solid phases redistribute the solute. As an example, for Al-4.5%Cu consider that an existing thin dendrite arm with high curvature melts, and a neighboring thicker dendrite arm with lower curvature thickens (as a result of diffusion of copper through the liquid from the thicker to the thinner arm). The solid being deposited on the thickening arm would have a higher copper content than the disappearing dendrite arm (that formed at higher temperatures). Consequently, the average copper content of the primary phase will increase, and the fraction of nonequilibrium eutectic will be reduced. For Al-4.5%Cu alloys, the extent of solute redistribution by solid diffusion or by coarsening is low during the initial stages of freezing. The copper content builds up slowly from kCO (note Fig. 3). The gradient for diffusion is small, as would be the difference in copper content between melting and solidifying dendrite arms. During the later stages of primary phase solidification, the copper gradient is higher. Solute redistribution by solid diffusion becomes appreciable. Coarsening involving melting of existing, cored dendrite arms would reduce microsegregation appreciably. On the other hand, studies of dendrite coarsening mechanisms indicate remelting mechanisms are more likely in the early stages of solidification (Ref 26). Certainly, in low-melting systems (e.g., tin and lead alloys), diffusion in the solid is

Variation in microsegregation with change in solidification time for Al-4. 5%Cu

Process

Weld Chill Permanent mold Sand mold Directional

a

0.1 0.15 0.2 0.3 0.5

uf, s

1.2 7.4 27 172 1756

das, mm

Cmin, wt% Cu

S

feut, wt%

8 16 27 56 138

0.93 1.00 1.05 1.15 1.29

6.1 5.7 5.4 4.9 4.4

8.7 8.5 8.3 8.0 7.5

negligible, and deviation from Gulliver-Scheil behavior is due to dendrite coarsening. For higher-melting alloy systems, higher solute diffusivities and the mass transport Fourier numbers indicate solute redistribution is predominantly by back diffusion. Influence of Microsegregation and Partitioning on Density and Solidification Shrinkage. As the compositions of the solid and liquid phases change as a result of partitioning, the relative densities of the phases can change significantly. The alloy Al-4.5%Cu provides an example where substantial density changes due to solute partitioning strongly influence castability. As an Al-4.5%Cu alloy cools through the freezing range, the copper content of the liquid increases from 4.5 to 33.2%, while the average composition of the primary solid phase only increases from 0.69 to 2% Cu. The first solid to form is denser than the initial liquid. After only a few percent of solidification, the copper-enriched liquid is denser than the primary solid a-phase. Equiaxed dendrites and dendrite fragments float in the copper-enriched liquid. Solidification expansion accompanies a large fraction of the formation of the primary a-phase. Importantly, though, the final solidification of eutectic liquid at 33.2% Cu to the eutectic mixture (a + y) is accompanied by a 10% decrease in specific volume. A large fraction of the overall solidification shrinkage of the Al-4.5%Cu alloy occurs at the last-stage of solidification. This tendency for late-stage solidification shrinkage makes Al-4.5%Cu alloys susceptible to microshrinkage porosity and to hot tearing. Interaction of Microsegregation and Macrosegregation. Macrosegregation is the difference in the local average solute content  from the initial alloy after solidification, C, composition, CO. Macrosegregation is evident in the as-cast microstructure (and in properties) by the way it changes microsegregation. Using Al-4.5%Cu alloy as an example, the main manifestation of macrosegregation in the as-cast microstructure will be the amount of nonequilibrium eutectic constituent (Ref 27). For positive macrosegregation, C > CO , there will be slightly less primary phase formation and noticeably more nonequilibrium eutectic. The increased eutectic will likely be accompanied by increased microporosity. Standard solution treatment will leave more undissolved y-phase. Without corrective treatment, regions of Al-4.5%Cu alloys with positive macrosegregation can be expected to have reduced ductility. For negative macrosegregation, C < Co , there will be slightly more primary phase formation and noticeably less nonequilibrium eutectic. Regions of Al-4.5%Cu alloys with negative macrosegregation can be expected to have reduced strength. Microsegregation influences the relative density of solid and liquid phases, the driving force for interdendritic fluid flow, and the extent of macrosegregation. Interdendritic flow of solute

342 / Principles of Solidification (Figure 1 is an example for k > 1.) Then, a second, perpendicular trace, crossing at the first minimum, can be made to check for a lower minimum; another perpendicular trace can be made until the extreme minimum in that plane of polish is located (Ref 10). Quantitative metallography can be used to measure the volume fraction of secondary phases. That measurement is converted to a weight fraction, and the difference between the measured value and the equilibrium value is reported as f nsq. For example, y-phase is a nonequilibrium constituent in Al-4.5%Cu. All measured y-phase can be reported as fnsq , or the measured value can be divided by the mass fraction of y-phase in the eutectic microconstineq (Ref 8). tuent to report fsut An array of electron probe point counts can be used to statistically determine solute distribution curves (such as the copper profile in Fig. 3) (Ref 32–36). Composition is measured across a rectangular grid of n  m = P points, taking care that the spacing between points is not a multiple of the dendrite arm spacing. For a binary alloy with k < 1, such as Al4.5%Cu, the measured points can be sorted and indexed (j) in order of ascending solute content. Measured solute content, Cij, is plotted versus fS ¼ j=P , the ratio of the index of the sorted point, j, to the total number of points, P. The reproducibility and accuracy of this statistical analysis can be tested by comparing the average composition of the measured points to CO and by separating the points at random into two groups, sorting, plotting the two groups, and comparing the solute profiles obtained from one-half of the points and the profile using all the points. If agreement is not satisfactory, more points should be added to the grid. Uncertainty in the composition measurements around

Cmin and Cmax may distort the shape of the solute profile. Also, noting that the measured volume may contain more than one phase, composition measurements that fall outside of the stability range of the phases can be prorated. Generally, secondary phases will have a fixed composition, such as y-phase in hypoeutectic aluminum-copper alloys. Figure 4 presents, as an example, solute (zinc) distribution curves measured for the binary alloy Cu-33%Zn (Ref 33). Figure 5 is an example of solute (copper and silicon) distribution curves for multicomponent and multiphase Al-Si-Cu-Mg, where the solute concentrations in the primary a-phase dendrites were computed by assuming the silicon-phase and y-phase have fixed compositions (Ref 36). Fig. 4 and Fig. 5 are discussed in later sections. Observation and Measurement of Kinetic Processes Leading to Microsegregation. The kinetics of solute redistribution processes can be examined more effectively by interrupted solidification and/or in situ solidification. Because of dendrite (or grain) coarsening, diffusion in the solid, and bulk flow of solid and liquid, the solidification morphology and solute distribution in the solid phases change continuously during solidification and can continue to change during postsolidification cooling. Measurements and images made after continuous cooling through the solidification range and to room temperature do not provide unambiguous information on the dynamic evolution of microstructure and solute distribution through the solidification range. Interrupted solidification refers to a variety of techniques to discontinuously increase the cooling rate of an alloy that has cooled only part way through the solidification range. In principal, the microstructure present before

35 34.5 34 Zn concentration, wt%

enriched (or depleted liquid) in the mushy zone that results from the requirement to feed solidification shrinkage and from thermal and solutal convection due to density gradients is the dominant factor in determining the extent of macrosegregation. For an Al-4.5%Cu alloy, the density of the liquid in the mushy zone increases as the local temperature decreases, the fraction of solid increases, and the liquid enriches in copper. The densest liquid is at the eutectic isotherm. Copper-enriched interdendritic liquid tends to settle, and copper-depleted solid grains and fragments tend to float. In contrast, for the aluminum-silicon alloys discussed subsequently, the density of the liquid in the mushy zone decreases as the liquid enriches in silicon. Solutal convection will be the opposite as for aluminum-copper alloys. In both cases, horizontal temperature gradients will give rise to solutal convection and the potential for macrosegregation. One method to minimize the extent of macrosegregation is to balance the alloy composition to minimize density changes as the solute is partitioned between solid and liquid phases during solidification. Measurement of As-Cast Microsegregation. Qualitatively, the extent and distribution of microsegregation will be evident in cut and polished surfaces. For optical and scanning electron microscope images, an etchant may be used to bring out the microstructure. Coring in the primary phase will be revealed by a gradation in color or texture after etching. The likely positions of relative composition maxima and minima can be inferred from the microstructures. Differential heat treatment can be used to emphasize composition differences. As an example, consider hypoeutectic aluminumcopper alloys. The alloy can be solutionized at a temperature below the equilibrium solvus and then quenched. The dendrite cores will appear single phase (a), and the regions above the solvus composition for the solutionizing temperature will precipitate y-phase. The boundary between single and two-phase regions will mark an isoconcentration surface. Changing the solutionizing temperature will indicate the position of a range of isoconcentration surfaces. Serial sectioning can be used to determine the shape of isoconcentration surfaces and as-cast dendrite morphology in three dimensions (Ref 28–30). Differential heat treatment is an especially useful technique to bring out microsegregation patterns in systems that undergo polymorphic transformations that mask microsegregation, for example, steels and titanium alloys (Ref 21, 31). The segregation ratio, S, the segregation index, SI, and the fraction of nonequilibrium microconstituent, f nsq, have already been mentioned as quantitative indicators of the extent of as-cast microsegregation. Electron microprobe traces can be used for quantitative analysis of Cmin and Cmax. For example, an electron beam trace (line scan) can be made across a dendrite arm, crossing an area where a minimum in solute concentration is expected.

Columnar

33.5

Equiaxed

33 32.5 32 31.5 31 30.5 30

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Cumulative fraction of measured points

Fig. 4

Measured solute distribution curves for Cu-33%Zn alloy, for two directionally solidified ingots with the same thermal history but different grain structures. Adapted from Ref 33

Microsegregation / 343 Measured Cu, Si distribution of alloy 1 (as-cast, 0.10 ⬚C/s or 32 ⬚F/s) 4 % Cumin = 0.8 Cu and Si in a-phase, wt%

3.5

gprimary = 78% gbinary = 12%

3

geutectic = 10% ga = 92.7%

2.5

Cu

gSi = 5.6%

2

Fe

gq = 1.7%

1.5 Si 1 0.5

Mg

0 0

Fig. 5

0.2

0.6 0.4 Cumulative fraction of measured points

0.8

1

Silicon and copper solute distribution curves measured by electron microprobe grid in a five-component aluminum alloy, Al-6.6%Si-3.4%Cu-0.2%Mg-0.5%Fe. Source: Ref 36

Fig. 6

Al-7%Si-3.5%Cu-0.3%Mg-0.5%Fe alloy slowly cooled to two temperatures in the solidification range and then quenched. (a) Aluminum-rich dendrites and enriched liquid are present at 571  C (1060  F). (b) Al5FeSi intermetallic plates start to form at 564  C (1047  F). Colonies of aluminum and silicon eutectic will form at a slightly lower temperature, followed by Q-phase and y-phase. Source: Ref 37

solidification is interrupted will be coarse and easily contrasted from the fine microstructure that will form in regions that were liquid immediately before the quench. In one variation of interrupted solidification, several small crucibles containing molten alloy of the same composition are continuously cooled at a controlled, relatively slow rate; each crucible is extracted from the furnace at a different temperature (and fraction solid) within the freezing range, and the partly solidified alloy is quenched. When the samples are cut, polished, and etched, the regions that were solid before the quench will have a relatively coarse microstructure and be in contrast with the

regions that were liquid at the time the samples were quenched (Ref 10). The general morphology of the solid at the time and temperature of the quench, the phases present, and the composition profile within the preexisting solid will be apparent and measureable. For example, the microstructure of a multicomponent aluminum alloy slowly cooled and then quenched from a temperature within the freezing range is presented in Fig. 6 (which is discussed in a later section) (Ref 10, 37). Care must be exercised in analyzing the microstructure in interrupted solidification samples. There will be some delay after the quench before the solid growing at the preexisting

interface breaks down into fine dendrites. The fraction solid at the temperature of the quench will be overestimated. Further, the rapid change in solidification rate will lead to inverse macrosegregation in the fine structure at the preexisting solid-liquid interface and give the false appearance of a solute boundary layer. On the other hand, measurements of the solute content in the coarse solid phase will reliably represent the solute concentration in the solid at the time of the quench. For example, electron beam microprobe measurements of the minimum solute content in the a-aluminum dendrites of Al-4.5%Cu alloy samples quenched at different stages of solidification increase from kCO  0.65 to 1.35% Cu. Thermal analyses and theoretical estimates of the solute buildup at dendrite tips indicate undercooling less than 1  C (1.8  F), which would increase the copper content of the initial solid by only 0.05%. The reduction in as-cast microsegregation from the GulliverScheil extreme can be attributed primarily to back diffusion (Ref 9, 10). In situ solidification refers to techniques to image and measure solidification morphology and solute distribution in alloys as they evolve during solidification. Microradiography makes use of solute partitioning to provide contrast and make optically opaque alloys sensitive to dynamic processes. The microsegregation in thin sections of as-cast alloys, such as Al-4.5%Cu, is readily apparent in radiographic images wherein the wavelength of incident x-rays is selected to maximize the difference in aluminum and copper absorption coefficients. If thin sections are solidified in front of an x-ray beam and sequential images captured, the liquid and solid phases will be distinguished as the liquid progressively is enriched in copper, the more absorbing element (Ref 26, 38–41). Unlike imaging of as-cast samples, for dynamic images, exposure time must be short to obtain a “stop-action snapshot.” Wavelength, coherency, and intensity of the incident beam must be optimized to obtain images with good contrast and brief exposure times. Figure 7 is an example of an in situ image of equiaxed dendritic grains in a thin section (0.1 mm, or 0.004 in.) of an Al-30%Cu alloy solidifying under a shallow temperature gradient (Ref 39). In successive images, these dendrites are seen to float on the copper-enriched liquid, rotate, and impinge, hence isolating liquid pools at what will become grain-boundary junctions. Figure 8, by contrast, is an in situ image of columnar dendrites of Sn-13%Bi solidifying under a steep temperature gradient (Ref 42). The composition of the interdendritic liquid changes from CO = 13% Bi at the dendrite tips (top) to CE = 52% Bi at the dendrite roots (bottom). For this low-temperature alloy, the secondary dendrite arms appear to separate the alloy into isolated characteristic volumes where solidification proceeds in accordance with the Gulliver-Scheil model.

344 / Principles of Solidification

Fig. 7

Synchrotron microradiograph of Al-25%Cu alloy solidifying under shallow temperature gradient. Equiaxed aluminum-rich dendrites (light) are floating in copper-enriched liquid (dark). Source: Ref 39

Fig.

8

Microradiograph of Sn-13%Bi alloy directionally solidifying in a steep temperature gradient. Liquid (dark) at top is at the initial melt composition (13% Bi), and remaining liquid at bottom is at eutectic composition (51% Bi). Composition of the columnar dendrites varies from 3.5 to 21% Bi. Source Ref 42

Thermal analysis of aluminum-copper alloys indicates the freezing range is consistent with nonequilibrium solidification, and the cooling curve resembles the solidification curves in Fig. 2. Conversely, the solidification curve, if known, can be used to compute the rate of heat evolution through the freezing range. Just as the solidification curves for Gulliver-Scheil and nonequilibrium freezing with back diffusion are very similar, the enthalpyversus-temperature curves for the two solidification paths will not differ appreciably.

Binary Eutectic Systems: Hypoeutectic Aluminum-Silicon Alloys Microsegregation is discussed in terms of the behavior of another model binary eutectic system, aluminum-rich aluminum-silicon alloys.

The eutectic temperature is 577  C (1071  F) for the invariant reaction: liquid (12.2% Si) ! a-aluminum (1.65% Si, a-phase) + siliconphase (essentially pure silicon); the a-phase solvus recedes by approximately 1% Si per 50  C (90  F) below the eutectic temperature. The partition ratio for aluminum-silicon alloys is approximately the same as for aluminum-copper alloys. The diffusivity of silicon in a-aluminum is approximately 2½ times greater than the diffusivity of copper when compared at the same temperatures. Silicon is an important component of most aluminum-base casting alloys, such as Al-356, because it improves castability by increasing fluidity and resistance to hot tearing and decreasing the tendency for shrinkage microporosity. In commercial casting alloys, specified silicon content exceeds the solubility limit at the eutectic temperature, and the a-aluminum + silicon eutectic is an equilibrium microconstituent of the as-cast microstructure. Consider, for example, an Al-7%Si alloy. Approximately 50% of the a-aluminum primary phase solidifies above the eutectic temperature, and the remaining 50% of solidification is the eutectic microconstituent (secondary a-aluminum + silicon). The secondary a-aluminum forms at the solubility limit, 1.65% Si. The silicon content at the center of the dendrite arms, Cmin, increases by back diffusion from approximately 0.9% Si at the liquidus temperature to 1.15% Si at the eutectic temperature. At the eutectic temperansq  2%, and ture, S   1:65=1:15 ¼ 1:4; fsut nsq fSi  0:2%. Anomalous Microsegregation. As an Al7%Si alloy cools below the eutectic temperature, the solubility of silicon in the a-phase decreases rapidly. Within the interdendritic regions, silicon diffuses out of the a-phase and increases the volume fraction of the siliconphase particles. The direction of the silicon composition gradient reverses, and silicon diffuses away from the dendrite cores toward the interdendritic regions. The slower the postsolidification cooling rate, relative to the solidification rate, the more the a-phase will be depleted in silicon. The maximum silicon content in the a-phase will be in the center of the dendrite, and the minimum silicon content will be in the interdendritic a-phase. The silicon microsegregation profile will be the opposite of that observed in aluminum-copper alloys. This solute distribution has been called anomalous microsegregation (Ref 43).

Binary Peritectic Systems: Copper-Zinc a-Brass Alpha-brass, Cu-30–33%Zn, is an example of a hypoperitectic system in which solute redistribution and as-cast microsegregation are between the extremes of Gulliver-Scheil and equilibrium behavior. The melting temperature of copper is 1083  C (1981  F); the peritectic

temperature is 903  C (1657  F) for the invariant reaction: liquid (38% Zn) + a-brass (32.5% Zn) ! b-brass (36.9% Zn); the solubility limit for a-phase increases from 32.5% Zn at the peritectic temperature to 39% Zn at room temperature. A Cu-33%Zn alloy will begin to freeze as a-brass at the liquidus temperature with an initial composition of kCo  28.3% Zn. For equilibrium solidification, the alloy would be 91% a-brass and 9% liquid at the peritectic temperature and transform to 89% a-brass + 11% b-brass right below the peritectic temperature. As the alloy continues to cool toward room temperature, the b-brass will transform to a. For equilibrium solidification, the segregation ratio is S = 1 and fbnsq ¼ 0 in the as-cast microstructure. At the other extreme, Gulliver-Scheil solidification, the alloy would be 62% a-brass and 38% liquid at the peritectic temperature. Right below the peritectic temperature, b-brass would nucleate and surround the a-brass. The remaining liquid would transform to increasingly zinc-rich b-brass that would remain in the interdendritic region of the as-cast microstructure. For Gulliver-Scheil solidification, the segregation ratio is S  57=28:3 ¼ 2 and fbneq ¼ 38%. The diffusivity of zinc in a-brass near the Cu-33%Zn liquidus temperature is approximately 7 times the diffusivity of copper in a-aluminum around the liquidus temperature for Al-4.5%Cu. Back diffusion is prevalent in a-brass. Right above the peritectic temperature, approximately 22 wt% liquid will remain. As the alloy cools below the peritectic temperature, the b-phase will nucleate and surround the a-brass. The peritectic reaction (L + a ! b) occurs in three steps: (1) L ! b at the b/L interface, (2) zinc diffuses through the b layer from the b/L to the a/b interface, and (3) a ! b at the a/b interface. Diffusion through the b layer (step 2) is the rate-controlling step for the peritectic reaction. Eventually, at the nonequilibrium solidus, the final zinc-enriched liquid transforms to zinc-rich b-brass. For a local solidification time of 1500 s, the segregation ratio is S  44=31:3 ¼ 1:42 and fbnsq  8% (Ref 33). The influence of grain structure on microsegregation has been investigated for a-brass. Figure 4 compares for columnar and equiaxed grain structures the zinc distributions measured by point counting on 13 by 13 grids and plotting the points in ascending order of measured zinc content. The columnar grain structure was obtained by directional solidification. The equiaxed structure was obtained by adding 0.75% Fe to the liquid before directional solidfication. The equiaxed grain size is approximately the same as the secondary dendrite arm spacing in the columnar-grained casting. The locations compared in Fig. 4 have essentially the same thermal history. The difference in zinc profile for the equiaxed and columnar structures is within the precision of the measurement technique. The dependence in a-brass of the extent of microsegregation on grain structure is weak (Ref 33).

Microsegregation / 345

Multicomponent, Multiphase Eutectic Systems: Hypoeutectic Al-Si-Cu-Mg Alloys Common heat treatable aluminum casting alloys are strengthened by precipitation of the precursor phases to Mg2Si (b0 ) and CuAl2 (y0 and y00 ). Al-6.5%Si-3.5%Cu-0.5%Mg-0.5%Fe is representative of Al-319 alloys, which are used for automotive structural applications that require good castability, good heat resistance, good machinability, and tolerate low ductility. The solidification sequence is L ! a-aluminum (from liquidus  600  C, or 1110  F), L ! aaluminum + silicon-phase binary eutectic (from  566  C, or 1050  F), L ! a-aluminum + silicon-phase + Q-phase ternary eutectic (from  524  C or 975  F), L ! a-aluminum + silicon-phase + Q-phase + y-phase quaternary eutectic (at  505  C, or 940  F). Figure 6(b) is a micrograph of the microstructure of an Al-6.5%Si-3.5%Cu-0.5%Mg-0.5%Fe alloy that was slowly cooled to 564  C (1047  F) and then water quenched. The coarse a-dendrites and Al5FeSi platelets formed before the quench. The fine-structured quenched liquid is enriched in copper, silicon, magnesium, and iron. Divorced eutectic colonies of a and silicon, Qphase and y-phase will form at lower temperatures. Analyses of the interrupted solidification microstructures and of the compositions of the quenched solid provide an understanding of the kinetics of the solidification processes (Ref 36, 37, 43, 44). Note that the liquidus and eutectic temperatures of the multicomponent alloy are considerably lower than for the binary alloys, which tends to lower the diffusivity of the alloying elements in the solid phases. On the other hand, the presence of less than ½% Mg more than doubles the diffusivity of both copper and silicon in the primary a-phase. Also, the diffusion of copper in the binary eutectic a-aluminum + silicon-phase region is increased 5 to 20-fold over the diffusivity in the single-phase region. During primary phase solidification, the partition ratios for silicon, copper, and magnesium are less than 1 (Ref 44, 45). Copper redistribution during primary phase solidification approaches the Gulliver-Scheil conditions, and silicon redistribution approaches equilibrium. When the silicon nucleates, the copper content of the remaining liquid and of the forming a-aluminum continues to increase. The silicon concentration of the liquid and of the forming a-phase decreases slightly. Silicon begins to diffuse from the primary phase at the center of the dendrites into the two-phase region. After ternary and quaternary eutectic solidification, and as the alloy cools below the eutectic composition, the solubilities of copper and silicon in the a-phase decrease sharply. Copper and silicon diffuse away from the a-aluminum + silicon-phase toward the interdendritic region. The interdendritic y and silicon-phase particles act as sinks

for diffusing copper and silicon. Figure 5 presents the copper and silicon profiles measured by electron microprobe in the a-phase of a sample from a sand casting, allowed to cool in the mold. The silicon profile within the a-phase is at a maximum at the center of the dendrite and at a minimum in the interdendritic regions, that is, anomalous microsegregation. The copper profile within the a-phase is a minimum at the center of the dendrites and a maximum near, but not at, the interdendritic region. Knowledge of the as-cast microsegregation in this alloy can be used to optimize the solution heat treatment prior to quenching and aging. A brief solution treatment will return the silicon distribution in the a-phase to the level present at the eutectic temperature, over 1%. Magnesium and copper in the core of the dendrites will be at a low level. The predominant precipitate on aging will be b0 . After moderate solution treatment, magnesium will diffuse into the dendrite cores, and b00 precipitates will be prevalent. For longer solutionizing times, copper will diffuse into the dendrite cores, and y0 will contribute to strengthening. Brief solutionizing times, or rapid cooling from just below the eutectic temperature, can be used to achieve adequate hardness and strength and save energy. Extended solution treatment can be used to substantially increase strength and ductility if required (Ref 36).

Multicomponent Multiphase Steel: Fe-C-Cr Alloying elements in steel are both substitutional and interstitial solutes. Interstitial elements, such as carbon, are highly mobile, especially in ferrite, d-iron. Redistribution of carbon during solidification can be expected to attain equilibrium. Substitutional elements, such as chromium, are much less mobile, and their diffusivities depend strongly on the temperature range of solidification. In the ironchromium binary system, the freezing range is very narrow; the liquidus temperature is close to the melting temperature of iron, and k approaches 1. In both iron-carbon and ironchromium binary alloys, microsegregation will be negligible, and solidification will follow equilibrium conditions: S = 1 and fbneq ¼ 0. Fe-1.5%Cr-1%C is the nominal composition for 52100, a wear-resistant, hardenable bearing steel. The liquidus temperature is reduced by 100  C (180  F) in the ternary alloy, and the freezing range is extended. Diffusivity of chromium in the ternary alloy is reduced considerably from the binary alloy. In the ternary alloy, redistribution of carbon attains equilibrium (uniform activity but not uniform composition), and redistribution of chromium by diffusion is negligible. The dendrites are cored with respect to both carbon and chromium; the freezing range is extended to the line of twofold saturation where carbides form. Carbon diffuses

to attain uniform chemical potential. The presence of chromium alters the activity of carbon, and the nonuniform distribution of chromium causes the carbon to diffuse to a nonuniform distribution. In directionally solidified columnar grain ingot, SCr  4; in small, slowly cooled, equiaxed ingot, SCr  3; and in commercial wrought 52100 alloy, SCr  2 (Ref 31). Solidification sequence and residual microsegregation in austenitic stainless steels depend, sensitively, on composition and solidification rate. Retention of a small fraction of ferrite is considered to reduce the susceptibility of stainless steel castings and weldments to the ratio of nickel (and other austenite stabilizers) and chromium (and other ferrite stabilizers) to hot tearing and crater cracking. Figure 9 presents detailed measurements of the isoconcentration lines in an Fe-25%Cr-20%Ni alloy that illustrate the as-cast dendritic morphology (Ref 30, 46). Measurements such as these are used to select geometry for simulation models of solute redistribution during dendritic solidification (Ref 19).

Multicomponent, Multiphase Nickel-Base Superalloy Nickel-base superalloys, such as Mar-M-200, contain several alloying elements, some, such as titanium and aluminum, with k < 1 and some, such as molybdenum, with k > 1. Alloying elements such as titanium and aluminum tend to decrease the density of the liquid during

Fig. 9

Isoconcentration profile in an Fe-25%Cr-20%Ni columnar dendrite. Source: Ref 45, 46

346 / Principles of Solidification solidification, and others, such as tungsten, tend to increase the liquid density. Hafnium is added to directionally solidified columnar-grained Mar-M-200 airfoil castings to modify the ductility of transverse grain boundaries during elevated-temperature service. For hafnium, k  1, and its addition extends the freezing range and significantly increases eutectic fraction. The increased eutectic liquid amount enhances the castability of columnar-structured Mar-M-200 + hafnium by decreasing the tendency to hot tearing (in the last stages of solidification). On the other hand, the increase in the volume and density of the eutectic liquid increases the susceptibility of the hafnium-modified alloy to macrosegregation (Ref 47). REFERENCES 1. J.I. Nurminen, “Solidification Structure and Segregation in Binary Titanium Base Alloys,” Ph.D. dissertation, University of Pittsburgh, 1972. 2. B. Chalmers, Principles of Solidification, Wiley, New York, 1964 3. M.C. Flemings, Solidification Processing, McGraw-Hill Book Co., New York, 1974 4. W. Kurz and D.J. Fisher, Fundamentals of Solidification, Trans Tech Publications Ltd., 1986 5. G.H. Gulliver, The Quantitative Effect of Rapid Cooling upon the Constitution of Binary Alloys, J. Inst. Met., Vol 9, 1913, p 120–157 6. E. Scheil, Bemerkungen Zur Schictkristallbildung, Z. Metallk., Vol 34, 1942, p 70–72 7. M.E. Glicksman and R.N. Hills, Origins of Non-Equilibrium Segregation Theory, Proc. M.C. Flemings Symp. on Solidification and Materials Proc., R. Abbaschian, H.D. Brody, and A. Mortensen, Ed., TMS, 2001, p 81–85 8. A.B. Michael and M.B. Bever, Solidification of Aluminum Rich Aluminum-Copper Alloys, Trans. Met. Soc.-AIME, Vol 200, 1956, p 47 9. H.D. Brody and M.C. Flemings, Solute Redistribution in Dendritic Solidification, Trans. Met. Soc.-AIME, Vol 236, 1966, p 615–624 10. T.F. Bower, H.D. Brody, and M.C. Flemings, Measurements of Solute Redistribution in Dendritic Solidification, Trans. Met. Soc.-AIME, Vol 236, 1966, p 624–634 11. D.H. Kirkwood, Microsegregation, Mater. Sci. Eng., Vol 9 (No. 6), 1984, p 101–109 12. T.W. Clyne and W. Kurz, Solute Redistribution during Solidification with Rapid Solid State Diffusion, Metall. Mater. Trans., Vol 12 (No. 6), 1981, p 965–971 13. J.A. Warren and W.J. Boettinger, Prediction of Dendritic Growth and Microsegregation Patterns in a Binary Alloy Using the Phase-Field Method, Acta Metall. Mater., Vol 43 (No. 2), 1995, p 689–703

14. V.R. Voller and S. Sundarraj, Modeling of Microsegregation, Mater. Sci. Technol., Vol 9 (No. 6), 1993, p 474–481 15. A. Roosz, E. Halder, and H.E. Exner, Numerical Calculation of Microsegregation in Coarsened Dendritic Microstructures, Mater. Sci. Technol., Vol 2 (No. 11), 1986, p 1149–1155 16. T. Kraft, M. Rettenmayr, and A. Roosz, Undercooling Effects in Microsegregation Modeling, Scr. Mater., Vol 35 (No. 1), 1996, p 77–82 17. T. Kraft and Y.A. Chang, Predicting Microstructure and Microsegregation in Multicomponent Alloys, JOM, Vol 49 (No. 12), 1997, p 20–28 18. V.R. Voller and C. Beckermann, A Unified Model of Microsegregation and Coarsening, Metall. Mater. Trans. A, Vol 30 (No. 8), 1999, p 2183–2189 19. T. Umeda, Microsegregation, Proc. M.C. Flemings Symp. on Solidification and Materials Proc., R. Abbaschian, H.D. Brody, and A. Mortensen, Ed., TMS, 2001, p 161–170 20. V. Mathier, A. Jacot, and M. Rappaz, Coalescence of Equiaxed Grains during Solidification, Mater. Sci. Eng., Vol 12, 2004, p 479–490 21. H.D. Brody and S.A. David, Application of Solidification Theory to Titanium Alloys, The Science and Technology of Titanium, R. Jaffee and N. Promisel, Ed., Pergamon Press, Oxford, 1970, p 21–34 22. J.I. Nurminen and H.D. Brody, Dendrite Morphology and Microsegregation in Titanium Base Alloys, Titanium Science and Technology, Vol 3, R.T. Jaffee and H.M. Burte, Ed., Plenum Publishing Corp., New York, 1973, p 1893–1914 23. T.Z. Kattamis, J.C. Coughlin, and M.C. Flemings, Influence of Coarsening on Dendrite Arm Spacing of Aluminum-Copper Alloys, Trans. Met. Soc.-AIME, Vol 239, 1967, p 1504–1511 24. P.W. Voorhees, Coarsening in Binary Solid-Liquid Mixtures, Metall. Trans. A, Vol 21, 1990, p 27–37 25. T.Z. Kattamis and P.W. Voorhees, Coarsening of Solid-Liquid Mixtures: A Review, Proc. M.C. Flemings Symp. on Solidification and Materials Proc., R. Abbaschian, H.D. Brody, and A. Mortensen, Ed., TMS, 2001, p 119–170 26. B. Li, H.D. Brody, and A. Kazimirov, Real-Time Observation of Dendrite Coarsening in Sn-13%Bi Alloy by Synchrotron Microradiography, Metall. Mater. Trans. A, Vol 38 (No. 3), 2007, p 599–605 27. G.E. Nereo and M.C. Flemings, Macrosegregation, I, Trans. Met. Soc.-AIME, Vol 239, 1967, p 1449–1461 28. B.P. Bardes and M.C. Flemings, Dendrite Arm Spacing and Solidification Time in Cast Aluminum-Copper Alloy, Trans. AFS, Vol 74, 1966, p 405–412

29. J. Alkemper and P.W. Voorhees, Three Dimensional Characterization of Dendritic Microstructures, Acta Mater., Vol 49 (No. 5), 2001, p 897–902 30. I. Ohnaka, Microsegregation and Macrosegregation, Casting, ASM Handbook, Vol 15, ASM International, 1988 31. M.C. Flemings, R.V. Barone, D.R. Poirier, and H.D. Brody, Segregation in Iron Base Alloys, JISI, April 1970, p 371–381 32. P.J. Ahearn and M.C. Flemings, Dendrite Morphology of a Tin-Bismuth Alloy, Trans. Met. Soc.-AIME, Vol 239, 1967, p 1590–1593 33. J.S. Lee, “Dendrite Structure and Solute Redistribution in Copper-33%Zinc Alloy,” M.S. thesis, University of Pittsburgh, 1972 34. M.N. Gungor, A Statistically Significant Experimental Technique for Investigating Microsegregation in Cast Alloys, Metall. Trans. A, Vol 20 (No. 11), 1989, p 2529– 2533 35. M. Ganesan, D. Dye, and P.D. Lee, A Technique for Characterizing Microsegregation in Multicomponent Alloys and Its Application to Single Crystal Superalloy Castings, Metall. Mater. Trans. A, Vol 36 (No. 8), 2205, p 2194–2204 36. Y. Ma, “Integrated Study of the Influence of Casting and Heat Treatment Parameters on Microstructures of Multiphase Multicomponent Aluminum Alloys,” Ph.D. dissertation, University of Connecticut, 2008 37. C. Lin, “Sequence of Phase Appearance and Microstructure Development during the Solidification of Al-Si-Cu-Mg-Fe Alloys,” M.S. thesis, University of Connecticut, 2003 38. B. Li, H.D. Brody, and A. Kazimirov, Real Time Observation of Dendrite Coarsening in Sn-13%Bi Alloy by Synchrotron Microradiography, Phys. Rev. E, Vol 70, 2004 39. B. Li, H.D. Brody, D.R. Black, H. Burdette, and C. Rau, A Compact Design of a Temperature Gradient Furnace for Synchrotron Microradiography, Meas. Sci. Technol., Vol 17, 2006, p 1883–1887 40. B. Li, H.D. Brody, and A. Kazimirov, Synchrotron Microradiography of Thermal Gradient Zone Melting in Directional Solidification, Metall. Mater. Trans. A, Vol 37, March 2006, p 1039–1044 41. H. Nguyen Thi, G. Reinhart, A. Buffet, T. Schenk, N. Mangelinck-Noe¨la, H. Jung, N. Bergeon, B. Billia, J. Ha¨rtwig, and J. Baruchel, In Situ and Real-Time Analysis of TGZM Phenomena by Synchrotron X-Ray Radiography, J. Cryst. Growth, Vol 310 (No. 11), 2008, p 2906–2914 42. L.H. Cadoff, “Effect of Dendrite Coarsening on Solute Redistribution in Cast Alloys,” unpublished research, University of Pittsburgh, 1970 43. M. Qian, F. Yi, D. Zhang, X. Pan, H.D. Brody, and J.E. Morral, Solute Redistribution and Phase Appearance in As-Cast

Microsegregation / 347 Al-Si-Cu-Mg-Fe Alloys, Metallurgical Modeling for Aluminum Alloys, M. Tiryakioglu and L.A. Lalli, Ed., ASM International, 2003, p 69–78 44. X.M. Pan, C. Lin, J.E. Morral, and H.D. Brody, An Assessment of Thermodynamic Data for the Liquid Phase in the Al-Rich Corner of the Al-Cu-Si System and Its Application to the Solidification of 319

Alloy, J. Phase Equilibria Diffusion, Vol 26 (No. 3), June 2005, p 225–233 45. D. Zhang, J.E. Morral, and H.D. Brody, Measurements for Cu and Si Diffusivities in Al-Si-Cu Alloys by Diffusion Couples, Mater. Sci. Eng. A, 2007, p 217–221 46. T. Umeda, J. Matsuyama, H. Murayama, and M. Sugiyama, Tetsu-to-Hagane´ (J. Iron Steel Inst. Jpn.), Vol 63, 1977, p 441

47. R. Sellamuthu, H.D. Brody, and A.F. Giamei, Effect of Fluid Flow and Hafnium Content on Macrosegregation in the Directional Solidification of Nickel Base Superalloys, Metall. Trans. B, Vol 17, 1986, p 347–356

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 348-352 DOI: 10.1361/asmhba0005216

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Macrosegregation Christoph Beckermann, University of Iowa

MACROSEGREGATION refers to spatial variations in composition that occur in metal alloy castings and range in scale from several millimeters to centimeters or even meters. These compositional variations negatively impact the subsequent processing behavior and properties of cast materials and can lead to rejection of cast components or processed products. It is present in virtually all casting processes, including continuous, ingot, and shape casting of steel, cast iron, aluminum and copper alloys, casting of single-crystal superalloys, semisolid casting, and growth of semiconductor crystals. Macrosegregation is especially important in large castings and ingots. Because of the low diffusivity of the solutes in the solid state and the large distances involved, macrosegregation cannot be mitigated through processing of the casting after solidification is complete. The cause of macrosegregation is relative movement or flow of segregated liquid and solid during solidification. Most alloy elements have a lower solubility in the solid than in the liquid phase, as shown by the partial equilibrium phase diagram in Fig. 1. During solidification, the solutes are therefore rejected into the liquid phase, leading to a continual enrichment

k=

C0

Tm

of the liquid and lower solute concentrations in the primary solid. This segregation occurs on the scale of the microstructure that is forming, which often consists of dendrites having arm spacings of the order of 10 to 100 mm. It is therefore termed microsegregation and results in a nonuniform, cored solute distribution in the dendrite arms. Consider now a small volume element that contains several dendrite arms and the liquid between them, that is, an element inside the liquid-solid (mushy) zone. In the absence of any solute transport into or out of the volume element, the average composition of the mixture inside of the volume element will remain constant at the nominal alloy composition, C0. However, if liquid or solid having a solute concentration different from the one of the liquid or solid inside of the volume element flows into the volume element, the average composition of the mixture in the volume element will change away from the nominal alloy composition. Since such flow generally occurs over large distances, the casting becomes macroscopically segregated. Positive (negative) macrosegregation refers to compositions above (below) the nominal alloy composition. All macrosegregation must average out to zero over the entire casting. There are numerous causes of liquid flow and solid movement in casting processes:
Flow that feeds the solidification shrinkage

CS CL




Temperature

m CL

CS

Liquidus


Solidus

A

C0

B

Alloying element B, mass%

Fig. 1

Schematic of a partial equilibrium phase diagram of a binary alloy. See text for an explanation of parameters.







and the contractions of the liquid and solid during cooling Buoyancy-induced flows due to thermal and solutal gradients in the liquid. The thermal and solutal buoyancy forces can either aid or oppose each other, depending on the direction of the thermal gradient and whether the rejected solutes cause an increase or a decrease in the density of the liquid. Flows due to capillary forces at liquid-gas interfaces Residual flows from pouring Flows induced by gas bubbles Forced flows due to applied electromagnetic fields, stirring, rotation, vibration, and so on Movement of small (equiaxed) grains or solid fragments that have heterogeneously nucleated in the melt, separated from a mold wall or free surface, or melted off dendrites.

The solid can either float or settle, depending on its density relative to the liquid.
Deformation of the solid network in the mushy zone due to thermal and shrinkage stresses, head pressures, or external forces on the solid shell Efforts to prevent macrosegregation are aimed at controlling liquid flow and movement of solid. Examples include adjustments to the alloy composition or thermal gradients to induce a stable density stratification in the liquid; application of nozzles, baffles, porous materials, or electromagnetic fields to redistribute the flow; controlled deformation such as soft reduction during continuous casting of steel to squeeze highly enriched liquid away from the centerline of slabs; changing the cooling rate or the alloy composition to decrease the dendrite arm spacing and hence increase the resistance to flow through the solid network; and increasing the prevalence of equiaxed grains in a casting, for example, by inoculation or stirring, in order to obtain a more uniform and dense solid distribution.

Macrosegregation Induced by Flow of the Liquid Liquid flow through the mushy zone, where the developing solid network can be assumed to be rigid and fixed, is the most common cause of macrosegregation. Such macrosegregation can be best understood by considering the local solute redistribution equation (LSRE) derived by Flemings and coworkers (Ref 1–3). This equation is based on a solute balance on a small-volume element inside the mushy zone, as illustrated in Fig. 2. As in a standard Scheil-type analysis (see other articles in this section of the Volume), local solute diffusion in the solid phase is neglected, and the liquid inside the volume element is assumed to be well mixed and in equilibrium. Then, the solute concentration in the liquid, CL, is given as a function of temperature by the liquidus line of an equilibrium phase diagram (Fig. 1) according to CL = (T  Tm)/m (where T is the temperature, Tm is the melting point of the pure solvent, and m is the slope of the liquidus line).

Macrosegregation / 349 Liquid flow

velocity in the direction normal to the isotherms, vT is the isotherm velocity, and: b¼

Solid

rS  rL rS

is the solidification shrinkage, where rS and rL are the densities of the solid and liquid, respectively. In one dimension, mass conservation yields that the liquid velocity needed to feed the solidification shrinkage is given by:

Liquid

  r b un jshrink ¼ vT 1  S ¼ vT rL ð1  bÞ

Fig. 2

Schematic of liquid flowing through a small volume inside the mushy zone. The gray ovals represent cuts through dendrite arms.

Mushy zone

Temperature, T

Liquid solute concentration, CL

Isotherms

Isotherm Normal liquid velocity, un velocity, nT

Schematic of liquid flowing through a mushy zone

Distance from chill, in. 0

0.5

1.0

1.5

2.0

2.5

3.0

– Copper concentration, (C/mass)%

4.6 Measured

(Eq 3)

Note that shrinkage-driven flow is in the direction opposite to the isotherm velocity, that is, in the direction of decreasing temperature toward regions of higher solid fraction. Only if the alloy expands upon solidification, b < 0, and unjshrink has the same sign as vT. Figure 3 illustrates the meaning of some of the quantities appearing in the previous equations. Since the liquid solute concentration is uniform along isotherms, liquid flowing in or out of the volume element in the direction of the isotherms has the same solute concentration as the liquid inside of the volume element. Therefore, liquid flow parallel to the isotherms in the mushy zone does not induce macrosegregation. Defining a flow factor, x, as:   un x ¼ ð1  bÞ 1  vT

Liquid flow

Fig. 3

(Eq 2)

(Eq 4)

Equation 1 can be integrated to yield a modified Scheil equation: CL ¼ ð1  fS Þðk1Þ=x C0

(Eq 5)

where x and k are assumed to be constant. No macrosegregation occurs if the flow factor is equal to unity and the standard Scheil equation is recovered. There are two obvious cases for which x = 1 (no macrosegregation):
No shrinkage (b = 0) and no liquid flow nor-

4.4

mal to the isotherms (un = 0)

Simulated


No liquid flow other than the one required

to feed the solidification shrinkage in one dimension: un = vT b/(1  b) or un = un jshrink

4.2 Co = 4.4% 4.0

3.8

0

25

50

75

Distance from chill, mm

Fig. 4

Inverse segregation in an A1-4.1 Cu ingot with unidirectional solidification

Accounting for advection of solute by the liquid flow in or out of the volume element, and for the different densities of the solid and liquid phases, the following LSRE results:   dfS ð1  fS Þ ð1  bÞ un ¼ 1 dCL vT CL ð1  kÞ

(Eq 1)

where fS is the solid volume fraction, k is the partition coefficient, un is the liquid flow

For x 6¼ 1, macrosegregation occurs, since the liquid concentration, CL, varies differently from that predicted by the standard Scheil equation. Based on Eq 5, three different macrosegregation mechanisms and outcomes can be identified (assuming, for the purpose of this discussion, that k < 1): 1. Negative macrosegregation: x > 1 or un/vT <  b/(1  b), or equivalently, un < un jshrink. At the same solid fraction, CL is lower than for x = 1, indicating that negative macrosegregation will result. For the case of b  0, this condition is met if the liquid flows in the direction of decreasing temperature toward regions of higher solid fraction (un < 0) and with a speed (absolute value of velocity) that is greater than jun jshrinkj.

2. Positive macrosegregation: 0 < x < 1 or 1 > un/vT >  b/(1  b), or equivalently, vT > un > un jshrink. At the same solid fraction, CL is higher than for x = 1, indicating that positive macrosegregation will result. For the limiting case of b = 0, this condition is met if the liquid flows in the direction of increasing temperature toward regions of lower solid fraction (un > 0) but with a velocity that is lower than the isotherm velocity (un < vT). For b > 0, this condition can be met even if the liquid flows in the direction of decreasing temperature toward regions of higher solid fraction (un < 0), since unjshrink is negative. 3. Remelting: x < 0 or un > vT. This condition is met if the liquid flows in the direction of increasing temperature toward regions of lower solid fraction and with a velocity that is higher than the isotherm velocity. Based on Eq 1, such flow results in the solid fraction decreasing, rather than increasing, with decreasing temperature. This means that remelting occurs and that open (solid-free) channels can form inside the mushy zone. Once the casting is solidified, the macrosegregation inside of these channels is positive. The direction and magnitude of the liquid flow velocities in the mushy zone depend on numerous factors. For a given driving force, the permeability of the mush is the single most important parameter that limits the flow (Ref 4). In general, the permeability decreases with increasing solid fraction. A smaller dendrite arm spacing also reduces the permeability. The following examples illustrate in more detail the aforementioned macrosegregation mechanisms. Inverse Segregation. Near a chill face, the positive macrosegregation shown in Fig. 4 is often observed (Ref 5). This so-called inverse segregation (Ref 6) is a consequence of the second macrosegregation mechanism discussed previously. At an impenetrable chill face, the liquid velocity vanishes (un = 0). Furthermore, for a finite shrinkage, unjshrink is negative. Therefore, the condition un > unjshrink is always satisfied at the chill face, and positive macrosegregation results. If a gap is present between the mold and the casting, the solute-rich interdendritic liquid can be pushed into the gap by metallostatic pressure. This is termed exudation and results in even higher positive macrosegregation at the surface. In direct chill casting of aluminum ingots, this macrosegregation can be so severe that the exuded surface layer must later be removed by scalping the ingot. Macrosegregation due to Nonuniform Shrinkage Flow. The first macrosegregation mechanism listed previously has been demonstrated for the casting geometry shown in Fig. 5 (Ref 5). The casting is chilled from the bottom, such that the isotherm velocity is generally in an upward direction. The downward liquid flow is caused solely by shrinkage. However, due to the reduction in the cross section of

350 / Principles of Solidification

0.26

0.27 Section B–B C

0.24 0.22

Section A–A

0.20

∂T = 0 ∂n

20

B

B ∂T = 0 ∂n

Z

– Copper concentration, (C/mass)%

Simulated 0.23

Measured (0.25C–0.25Si)

0.25% C

0.21

23

0.025 mm/s ∂T = 0 ∂n

0

0.25

0.19 −15

−10

−5

0

5

10

Position (Z), mm A

A

Fig. 6 25

Macrosegregation observed along Z-direction (cross-sectional mean value) for an Fe-0.25C specimen

y ∂T =0 ∂n

y=H

–Z fs = 1

50 mm

Riser

∂T = 0 ∂n

Simulated fluid flow at 50 s after cooling and macrosegregation in an Fe-0.25C specimen

the casting near midheight, the liquid must flow at a higher downward velocity in this region. Hence, the condition un < un jshrink is satisfied in this region (since both velocities are negative), and the negative macrosegregation seen in Fig. 5 and 6 forms (Ref 5). Macrosegregation due to Buoyancy-Driven Flow. The first and second macrosegregation mechanisms can be observed simultaneously if the liquid flow pattern in the mushy zone is of a recirculating nature. This has been demonstrated for the horizontally solidified aluminum-copper casting shown in Fig. 7 (Ref 5). Here, a counterclockwise recirculating flow pattern is induced in the liquid by thermal and solutal buoyancy forces. Shrinkage-induced flow is negligibly small in this case. Figure 7 shows that the flow in the bulk liquid region penetrates into the mushy zone. In the upper part of the mushy zone, the flow is in the direction of decreasing temperature (un < 0). According to the first macrosegregation mechanism, such flow causes negative macrosegregation. Negative macrosegregation was indeed measured in the upper part of the casting, as shown in Fig. 8. Conversely, in the lower part of the mushy zone, the interdendritic liquid flows in the direction of increasing temperature

S

S+L

Cooled

Fig. 5

1 mm/s

y=0

12 mm diam Cooled 29 mm diam

0.1 mm/s

H = 100 mm

Carbon, %

0.28

250 mm

Fig. 7

∂T =0 ∂n

Simulated fluid flow at 400 s after cooling in a horizontally solidified Al-4.4Cu casting

(un > 0), back into the bulk liquid region. This causes positive macrosegregation according to the second mechanism. This was again verified by the measurements (Fig. 8). Note that there is no macrosegregation at the midheight of the casting, because there the liquid flow is primarily parallel to the isotherms. Buoyancy-driven flow is also responsible for the positive macrosegregation seen in the riser of large steel castings and ingots, as shown in Fig. 9. In steel casting, the interdendritic liquid is often lighter than the bulk liquid and flows up. This up-flow is in the direction of increasing temperature (un > 0), since the riser is the last part of the casting to solidify. Consequently, positive macrosegregation results according to the second mechanism.

Channel Segregation. The third macrosegregation mechanism leads to the formation of so-called channel segregates. One such channel segregation defect is commonly referred to as freckles, an example of which is shown in Fig. 10. Freckles sometimes occur in upward, directional solidification of single-crystal nickel-base superalloy castings. During solidification of nickel-base superalloys, a number of light elements (such as aluminum and titanium) are rejected into the liquid, and some heavy elements, such as tungsten, are preferentially incorporated into the solid, leading to strong, upward solutal buoyancy forces in the mushy zone. Despite the presence of a stabilizing thermal gradient, these buoyancy forces can trigger convection cells. As the upward liquid

Macrosegregation / 351 1.0 100 mm from chill

Fractional height, y/H

0.8 Simulated 0.6

0.4

0.2

50 mm from chill

0 4.0

Fig. 8

4.5 5.0 4.0 4.5 – Carbon concentration, (C/mass)%

5.0

Solute distribution in the Al-4.4Cu ingot shown in Fig. 7

Riser A-segregation in branched columnar dendrite zone

Positive segregation in riser

Fig. 10 V-segregation in equiaxed grain zone Negative cone Columnar dendrite zone

Fig. 9

Chill zone

Typical macrosegregation observed in steel ingots

velocities in the mush exceed the isotherm velocity (un > vT), the solid fraction decreases with decreasing temperature, according to the third macrosegregation mechanism. Eventually, this leads to open channels in the mush through which solute-rich liquid streams upward into the bulk liquid region. The numerical simulation results shown in Fig. 11 illustrate these flow patterns (Ref 7). The rodlike channels are no larger than a few millimeters in diameter and have a spacing of, at most, a few centimeters. Later, the channels are filled with dendrite fragments, which are then observed as frecklelike chains of equiaxed crystals in the otherwise single-crystal columnar structure (Fig. 10). The resulting positive macrosegregation in the channels can also be seen in Fig. 11. Channel segregates are also observed in large steel ingots. Here, they are termed A-segregates and occur in the columnar zone of the ingot (Fig. 9). Strong buoyancy forces can cause the liquid to flow upward in the mush with a velocity normal to the isotherms that is greater than the isotherm velocity (un > vT), resulting in remelting. The channels are usually inclined toward the center of the ingot and thus appear as the A-shape illustrated in Fig. 9.

(a) Freckles in a single-crystal nickel-base superalloy prototype blade. (b) Closeup of a single freckle. The freckles are approximately 2 mm (0.1 in.) in diameter.

Macrosegregation Induced by Movement of the Solid The movement or flow of solid during solidification can also induce macrosegregation in castings. Unfortunately, no simple theory, such as the LSRE discussed in the previous section, exists to explain the macrosegregation patterns resulting from solid movement (Ref 8). Therefore, only a few examples are provided to illustrate some of the complex phenomena that have been observed. The most common form of solid movement is the settling or floating up of small solid pieces formed early in the solidification process. These solid pieces may be dendrite fragments that separated from an existing solid structure or equiaxed grains that nucleated in the bulk liquid. They settle or float depending on their density relative to the liquid. The solid pieces generally have a composition different from the nominal alloy composition, and their movement to different parts of the casting thus induces macrosegregation. One example is provided by the cone of negative macrosegregation that is often observed in the bottom third of steel ingots, as illustrated in Fig. 9. This macrosegregation pattern is confined to the equiaxed zone and is caused by the settling of dendrites that are relatively poor in solute (Ref 9, 10). Buoyancy-driven flow of the liquid also induces negative macrosegregation in the bottom half of steel ingots, but the degree is not as severe as that due to settling of solid (Ref 11). Similarly, negative centerline macrosegregation in direct chill casting of aluminum ingots is sometimes

attributed to the settling of unattached solute-poor solid, although shrinkage-driven flow has been found to have a similar effect (Ref 12). On the other hand, the floating of spheroidal graphite (kish) can cause strong positive macrosegregation in the upper part of iron castings. Another example of solid-movement-induced macrosegregation is given by the V-segregates in the equiaxed region of tall steel ingots. This macrosegregation pattern is also illustrated in Fig. 9 and consists of rodlike solute-rich streaks that appear periodically along the centerline. The V-segregates arise from equiaxed crystals settling in the core and forming a loosely connected network that can easily rupture due to metallostatic head and liquid being drawn down to feed solidification shrinkage. Fissures then open up along shear planes oriented in a V-pattern and are filled with enriched liquid (Ref 9, 10). Macrosegregation can also be induced by deformations of the fully and partially solid regions during casting. One important example is centerline segregation in continuously cast steel slabs (Ref 10). Close to the bottom of the liquid pool, inadequate roll containment can cause the already solidified steel shell to bulge. The bulging draws solute-rich liquid into the center of the slab, where it freezes. This results in a rather sharp and thin line of positive macrosegregation along the center of the slab, as can be seen from the sulfur print in Fig. 12. Deformation of the solid can also be induced by thermal stresses. This is a particularly common phenomenon in shape casting, due to the restraint that the mold and/or the cores can offer. However, deformation of the solid can also

352 / Principles of Solidification ACKNOWLEDGMENT Figures 4 to 9 are Fig. 8 to 12 and 7, respectively, in I. Ohnaha, Microsegregation and Macrosegregation, Casting, Vol. 15, Metals Handbook, ASM International, 1988, p136–141 REFERENCES

Fig. 11

Simulation of freckle formation during upward directional solidification of a single-crystal nickel-base superalloy in a 5 cm (2 in.) wide by 15 cm (6 in.) tall enclosure. (a) Melt velocity vectors in the liquid region and mushy zone (longest vector corresponds to a velocity of approximately 5 mm/s, or 0.2 in./s) and solid fraction contours (in five equal intervals of 20%, starting with 0% for the uppermost contour at approximately midheight). (b) Macrosegregation pattern. The gray shades indicate the titanium concentration, with the darkest shade inside the open channels in the mush corresponding to the largest positive macrosegregation.

Fig. 12

Sulfur print showing centerline macrosegregation in a continuously cast steel slab

occur in continuous casting due to uneven cooling. The stresses are able to propagate into the high solid-fraction portion of the mushy zone, where the solid network is already coherent. If the solid network is under compression, interdendritic liquid will be forced to “drain” out of the mush. Conversely, if this part of the mush is in tension, liquid is drawn into it. In either case, complex macrosegregation patterns can result. Under sufficiently severe tension, the solid network will rupture, and the fissure may fill with solute-rich liquid. The resulting cracklike positive macrosegregation in a casting is often referred to as a healed hot tear. If the flow of the interdendritic liquid is insufficient to feed the fissure, an open hot tear will be present in the casting.

1. M.C. Flemings and G.E. Nereo, Macrosegregation: Part 1, Trans. Met. Soc. AIME, Vol 239, 1967, p 1449–1461 2. R. Mehrabian, M. Keane, and M.C. Flemings, Interdendritic Fluid Flow and Macrosegregation; Influence of Gravity, Metall. Trans., Vol 1, 1979, p 1209–1220 3. M.C. Flemings, Solidification Processing, McGraw-Hill, New York, 1974 4. D.R. Poirier, Permeability for Flow of Interdendritic Liquid in Columnar-Dendritic Alloys, Metall. Trans. B, Vol 18, 1987, p 245–255 5. I. Ohnaka and M. Matsumoto, Computer Simulation of Macrosegregation in Ingots, Tetsu-to-Hagane (J. Iron Steel Inst. Jpn.), Vol 73, 1987, p 1698–1705 6. J.S. Kirkaldy and W.V. Youdelis, Contribution to the Theory of Inverse Segregation, Trans. Met. Soc. AIME, Vol 212, 1958, p 833–840 7. M.C. Schneider, J.P. Gu, C. Beckermann, W.J. Boettinger, and U.R. Kattner, Modeling of Micro- and Macrosegregation and Freckle Formation in Single-Crystal Nickel-Base Superalloy Directional Solidification, Metall. Mater. Trans. A, Vol 28, 1997, p 1517–1531 8. C. Beckermann, Modelling of Macrosegregation: Applications and Future Needs, Int. Mater. Rev., Vol 47, 2002, p 243–261 9. M.C. Flemings, Principles of Control of Soundness and Homogeneity of Large Ingots, Scand. J. Metall., Vol 5, 1976, p 1–15 10. G. Lesoult, Macrosegregation in Steel Strands and Ingots: Characterization, Formation and Consequences, Mater. Sci. Eng. A, Vol 413–414, 2005, p 19–29 11. J.P. Gu and C. Beckermann, Simulation of Convection and Macrosegregation in a Large Steel Ingot, Metall. Mater. Trans. A, Vol 30, 1999, p 1357–1366 12. A.V. Reddy and C. Beckermann, Modeling of Macrosegregation due to Thermosolutal Convection and Contraction Driven Flow in Direct Chill Continuous Casting of an Al-Cu Round Ingot, Metall. Mater. Trans. B, Vol 28, 1997, p 479–489

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 353-356 DOI: 10.1361/asmhba0005217

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Interpretation and Use of Cooling Curves— Thermal Analysis H. Fredriksson, The Royal Institute of Technology, Sweden

THERMAL ANALYSIS is a classical method of determining phase diagrams. By melting and cooling an alloy of known composition and registering the temperature-time curves, the liquidus temperature for the respective alloy can be determined. Figure 1 shows an idealized view of the relationship between the cooling curves and the phase diagram for an alloy A-B. The starting point for solidification is given by a change in the slope of the cooling curve due to the evolution of the heat of solidification. This starting point is normally used to define the liquidus temperature. The method is described more fully in Ref 1. The idealized behavior illustrated in Fig. 1 is not seen in experimentally determined temperature-time curves. For example, Fig. 2, an experimentally determined cooling curve for an ironcarbon alloy with 3.2% C, shows that the formation of austenite and the eutectic reaction occur at a temperature well below the liquidus and the eutectic temperatures, respectively. The reason for this depends on the nucleation and growth of the crystals in the liquid. In this article, this effect is analyzed, and the use of thermal analysis in industrial processes and in research is discussed.

Quantitative Thermal Analysis Cooling curves describe a balance between the evolution of heat in the sample and the heat transport away from the sample. It is often easiest to describe this in a mathematical form. However, heat transport away from the liquid is often very complex and can be described in many different ways because of the experimental conditions. Currently, heat transport is often described by numerical methods. In the following discussion, only the expression dQ/dt is used to define the heat transport away from the experimental setup. This heat transport must be equal to the evolution of heat in the sample. Before and after the solidification process, this is balanced by a loss of heat capacity. Before the start of the solidification process, this can be written in analytical form as:



dQ dT ¼ V r Cp dt dt

(Eq 1)

where V is the volume of the sample, r is the density of the liquid, Cp is the heat capacity, and dT/dt is cooling rate of the liquid. From a cooling curve similar to that shown in Fig. 2, dT/dt is the slope of the curve before

solidification begins. By evaluating the slope as a function of time in an experiment, dQ/dt can be determined if r and Cp are known. This can later be used in the analysis of the temperaturetime curve. When the liquid cools to the liquidus temperature, crystals are nucleated and begin to grow. The heat balance from Eq 1 is now: dQ ¼ dt




Relationship between a cooling curve and phase diagram for alloy A-B

df dT



dT dt

(Eq 2)

where DH is the heat of solidification per unit volume, and df/dT is the volume fraction of solid formed at a changing temperature. The change in shape of the temperature-time curve now depends on the volume fraction of solid formed. The larger the rate of increase of volume fraction of solid, the smaller the slope (dT/dt) of the temperature-time curve. The volume fraction of solid formed is related to the number of crystals and their growth rate. Crystals are normally nucleated when the temperature of the liquid reaches the liquidus temperature as well as the eutectic temperature. In spite of this, the temperature of the liquid will decrease further, and initially the curve

Fig. 2

Fig. 1



V r Cp þ r
H

Cooling curve for an iron-carbon alloy with 3.2% C. The solidification starts with a primary precipitation of austenite, followed by a eutectic reaction.

354 / Principles of Solidification can be extrapolated to show where no crystals are growing. After a certain time, the slope of the curve begins to decrease, slowly at first and then more rapidly, forming a small plateau or even a small temperature increase near the liquidus, as seen in Fig. 2. In the early stages of crystal formation, the growing crystals are small (as is heat development) and therefore have no effect on the bulk liquid temperature. The growth rate is also low because of the low supercooling close to the liquidus temperature. The supercooling increases when the temperature decreases; therefore, the growth rate will increase as the area of the crystal increases. The result is that more heat is given off and the slope of the temperature-time curve begins to decrease. The change in the slope will begin earlier and will proceed faster. The more crystals there are in the liquid, the earlier the change in slope starts, and the faster it will proceed. When the temperature increases, the growth rate of the free crystals decreases again, but they are so large by this time that the volume solidifying per unit of time will also be large. As such, the temperature will continue to increase until the growth rate and the growth area of the crystals offset the heat extraction. The larger the area of the growing crystals, the lower the growth rate and the closer the temperature of the bulk liquid to the liquidus or eutectic temperature.

Deviation from Equilibrium The solidification process illustrated by the cooling curve in Fig. 2 begins with a primary precipitation of austenite, followed by a eutectic reaction. In this case, the liquid was cast into a sand mold. The heat extraction from the liquid can be approximated by Chvorinov’s law and follows the expression: dQ K ¼ pffiffi dt t

Fig. 3

(Eq 3)

where t is the time and K the so-called modulus constant. The value K can be determined by using Eq 1 and the slope of the cooling curve before solidification begins. The fraction of solid can be determined by using Eq 2 and 3 and the data given in Fig. 2. The results of these calculations are shown in Fig. 3(a). This type of analysis has been performed and is discussed in 2 to 4; it can be used to analyze the deviation from solidification under equilibrium conditions. The Fe-3.2C alloy discussed earlier in this article can be used as an example to illustrate this analysis. According to the iron-carbon phase diagram, the liquidus temperature for this alloy is 1268  C (2314  F). At this temperature, dendritic crystals of austenite are formed. Because carbon has a very high diffusivity in austenite, the lever rule can be used to determine the fraction of solid as a function of temperature. The lever rule can be applied to any two-phase field of a phase diagram to determine the amounts of the different phases present at a given temperature in a given alloy. A horizontal line, referred to as a tie line, represents the lever, and the alloy composition represents its fulcrum. The intersection of the tie line with the boundaries of the two-phase field fixes the compositions of the coexisting phases, and the amounts of the phases are proportional to the segments of the tie line between the alloy and the phase compositions. The rule is derived with respect to temperature, and the derivative df/dT is obtained. This derivative is inserted into Eq 2. The temperature-time curve is then easily calculated. The result is shown in Fig. 3(b) as a dashed line. The deviation from the real curve and that calculated under equilibrium conditions is greatest at the beginning of the precipitation of austenite. After the initial stage during the cooling down to the eutectic temperature, the lever rule will describe the fraction of solid formed. The eutectic reaction for the equilibrium case occurs without any undercooling. This deviation from equilibrium is interpreted as follows.

The fraction of solid phases as a function of time for the alloy with a cooling curve shown in Fig. 2. (b) Deviation of the cooling shown in Fig. 2 resulting from equilibrium solidification condition (see dashed line)

The austenite dendrites can be assumed to nucleate randomly in the liquid. They grow in leading tips in a starlike fashion. The secondary dendrite arms form behind the tips. The growth of the dendrite tips is dependent on undercooling. This undercooling is determined by the growth rate, and it increases with increasing growth rate. The growth rate of the tips is determined by the heat extraction and by the number of growing crystals. When the dendrite crystals have collided, a network of dendrite arms is formed. The solidification now proceeds by a thickening of the dendrite arms. Similarly, eutectic colonies are randomly nucleated in the liquid; they grow spherically and collide with each other. When the collision occurs, the growth area decreases, and the temperature of the liquid decreases. Theoretical models that combine growth equations, statistical laws for the collision of crystals, and heat extraction laws have been developed in recent years to describe the previous discussion (Ref 2–6). In the future, these models will probably be used to predict the structure formed during the solidification process and the resultant properties of the material.

Differential Thermal Analysis Thermal analysis has a long history of use in research. To increase accuracy, the experiments are normally carried out by so-called differential thermal analysis. Simplified Thermal Analysis. Before discussing the theoretical basis of differential thermal analysis, a simplified experimental setup is considered. This setup consists of a crucible with a melt placed inside a furnace. A thermoelement is placed in the center of the melt to measure the temperature of the sample, Ts. Another thermocouple is placed at the wall of the furnace; the purpose of this thermocouple is to regulate the temperature of the furnace, Tf, as a function of time. The furnace is assumed to cool at a constant rate in degrees Kelvin per minute. Figure 4 shows the temperature-time cycle for an aluminum-copper alloy that contains 5% Cu. The furnace has cooled during the experiment at a rate of 10 K/min (285  C/ min, or 510  F/min). Figure 4 shows that the temperature of the sample is somewhat higher than that of the furnace. This divergence increases slightly with time. A stationary condition, that is, the same cooling rate for the sample and the furnace, never occurs; the cooling rate of the sample is always somewhat lower than that of the furnace. The melt begins to solidify at the point designated tstart. Heat is emitted by the sample, and the temperature difference between the sample and the furnace increases. A hump sometimes appears in the temperature-time curve of the sample. The maximum point, or the deviation point, on the cooling curve is usually used to establish the liquidus temperature.

Interpretation and Use of Cooling Curves—Thermal Analysis / 355

Fig. 4

Thermal analysis of an Al-5Cu alloy cooled at a rate of 10 K/min (285  C/min, or 510  F/min). Source: Ref 6

the solidus temperature, tsol, during the experiment. To facilitate evaluation of the liquidus and solidus temperatures, the temperature-time curve of the sample is differentiated in most cases. In general, the derivative of the temperature-time curve for the sample shows a discontinuity when the heat of solidification ceases to develop. The solidus temperature is generally defined by this discontinuity. This sharp change becomes less pronounced in cases in which the solidification interval is large and the reaction rate is low toward the end of solidification. The reaction rate is normally evaluated by a simple mathematical analysis. This is described by:
 dTs df r Csp Vs þ
H r Vs dt dt

 

 

(Eq 4)

¼ ZðTs  Tf Þ

Fig. 5

Schematic of principal setup for differential thermal analysis equipment. Tn is the standard temperature, Ts is the sample temperature, and Tf is the furnace temperature.

The solidification rate, or the reaction rate, decreases after the initial period, and the cooling rate of the sample increases. Toward the end of solidification, or when everything has solidified, the cooling rate of the sample becomes greater than that of the furnace. This is a result of the large temperature difference that exists between the sample and the furnace when a considerable amount of solidification heat is given off. Afterward, the cooling rate of the sample approaches that of the furnace. It is often very difficult to determine the end of the solidification process and to determine

where dTs/dt is the cooling rate of the sample, CSP is the specific heat of the sample, Vs is the volume of the sample, DH is the heat of solidification, df/dt is the volume fraction of solid formed per unit time, Ts is the sample temperature, Tf is the furnace temperature, and Z is the heat transfer coefficient. The second term on the right-hand side in Eq 4 is 0 before solidification begins. The heat transfer coefficient Z can be evaluated at this time by measuring dTs/dt, Ts, and Tf. The reaction rate df/dt can be determined by using that value and by remeasuring the temperature and its first-time derivative during solidification. Differential Thermal Analysis. Figure 5 shows a schematic setup for differential thermal analysis. The sample to be analyzed and the standard sample, with a thermocouple in each, are placed in a furnace. The temperature of the furnace and the cooling rate are controlled and measured with a separate thermocouple. The standard sample and the analysis sample must have the same temperature at the beginning of the experiment. During the cooling cycle, the samples will undergo one or more phase transformations that give off heat. This emission of heat implies that, during the time of reaction, the temperature of the analysis sample differs from that of the standard sample. The temperature difference between the analysis sample and the standard sample is recorded at the same time the furnace temperature is recorded. Using the experimental setup shown in Fig. 5, one normally registers the difference in temperature between the standard sample and the analysis sample, not the difference between the sample and the furnace. Therefore, only (Ts  Tn) and Ts are known, where Tn is the temperature of the standard sample. In order to be able to calculate the reaction velocity, the term (Ts  Tf) in Eq 4 is replaced with the term (Ts  Tn) + (Tn  Tf). For the standard sample, it is true that Z (Tn  Tf) is equal to:

 

r Csp Vn

dTn dt

(Eq 5)

where dTn/dt is the cooling rate of the standard sample. Replacing this in Eq 4 gives:

 



dTs df þ ð
HÞ r Vs ¼ ZðTs  Tn Þ dt dt (Eq 6) dTn þ r CnP Vn dt

r CsP Vs





If the heat of solidification, DH, is known, the reaction rates can be calculated by measuring the temperatures and the cooling rates. If the heat transfer coefficient is known, the heat of solidification can also be determined by integrating Eq 6 over the entire solidification interval (from f = 0 to f = 1). Simplifications of Eq 6 are often found in the literature when the reaction rate and the heat of solidification are determined. It is assumed that the first term on the left-hand side is equal to the last term on the right-hand side, which leads to the following approximate relationship: df Z ðTs  Tn Þ ¼ dt ð
HÞVs

(Eq 7)

Equation 7 shows a direct proportionality between the measured temperature difference and the reaction rate. In Eq 7, Z/DHVs can be considered constant. The difference in temperature between the sample and the normal (Ts  Tn) is measured and is thus known as a function of time. The constant can be determined by integrating Eq 6 from f = 0 to f = 1. This integration is often done graphically. Equation 7 is also often used in interpreting differential thermal analysis data. However, one should keep in mind that doing so introduces many faults. This can be demonstrated by a comparison between Eq 6 and 7. For example, it is assumed that CsP , Vs, Vn, and CnP are equal. This is often not the case. To avoid this problem, the sample itself can be used as an imaginary standard sample. The temperature difference can then be determined by a simple computer analysis (Ref 6). To determine the heat of solidification, the heat transfer coefficient Z must be determined by a series of reference experiments such as that described for the simple type of experiment in Eq 4.

Determining Liquidus and Solidus Temperature Cooling curves can be used in many different ways and are often used to determine the liquidus and solidus temperatures. In such applications, the small plateau temperature shown in Fig. 2 is often used. However, as indicated earlier, this temperature does not correspond to the liquidus temperature, because of the kinetics during crystal growth. To increase the accuracy, the cooling curve is often differentiated. In this

356 / Principles of Solidification

Fig. 6

Liquidus temperature determined for an aluminum- copper alloy. Source: Ref 7

Fig. 7 Temperature-time curve for an iron-base alloy with 1.01% C, 0.25% Si, and 0.46% Mn. The dashed line represents the cooling rate dT/dt = 0. Source: Ref 8

Fig. 8

Thermal analysis used for determining the chemical composition of cast irons

case, the temperature at which the derivative changes is used as the liquidus temperature. As indicated earlier, this will not provide the correct liquidus temperature. The only way to obtain the correct liquidus temperature is to perform a series of experiments at different

cooling rates. The plateau temperature can then be plotted as a function of the cooling rate. The liquidus temperature is determined by extrapolating the line to zero cooling rates. An example of this method is shown in Fig. 6. It is more difficult to determine the solidus temperature than the liquidus temperature. Toward the end of the solidification process, the derivative of the cooling curve often shows a large change (Fig. 7). The solidus temperature is defined in general by this large change. However, the temperature evaluated in this way varies with the cooling rate much more than the liquidus temperature does. This is due to the microsegregation that occurs during solidification. This microsegregation is very much influenced by the cooling rate (Ref 9). Therefore, thermal analysis by cooling is nearly useless for determining the equilibrium solidus temperature. This temperature can be determined only by analyzing the melting process on heating curves. Thermal analysis is often used in industrial processes. One of the most common applications is determination of the chemical composition of cast iron. In this case, a linear relationship has been found between the liquidus temperature and the carbon equivalent (CE), which is defined as: CE ¼ %C þ

%Si %P þ 3 3

(Eq 8)

An example of the use of thermal analysis or determining the chemical composition of cast irons is shown in Fig. 8. Given the present knowledge of thermal analysis, it should be noted that these diagrams can be used with accuracy only if the same conditions are prevalent every time. This is due to the influence of the cooling rate and the nucleation frequency on the plateau temperature. Casting temperature, for example, can influence the results. It is well known that the superheat of the liquid influences the number of crystals. The higher the temperature, the lower the number of growing crystals. The result will be a lower liquidus temperature. However, in most cases, a foundry will use the same melting procedure consistently and can therefore determine chemical composition easily and accurately by thermal analysis, provided a curve similar to the one shown in Fig. 8 is developed for their specific conditions.

3.

4.

5.

6. 7. 8. 9.

SELECTED REFERENCES  X. Chen, H. Geng, and Y. Li, Study on the



 



REFERENCES 1. K.W. Andrews, Physical Metallurgy, Vol 1, William Clowes & Sons, 1973 2. H. Fredriksson and S.E. Wetterfall, The Metallurgy of Cast Iron, Proceedings of the

Second International Symposium on the Metallurgy of Cast Iron (Geneva), May 1974, Georgi Publishing, 1974 H. Fredriksson and I.L. Svensson, Computer Simulation of the Structure Formed During Solidification of Cast Iron, Proceedings of the Third International Symposium on the Metallurgy of Cast Iron, North Holland, 1984 D.M. Stefanescu and C. Kanetkar, Computer Modelling of the Solidification of Eutectic Alloys: The Case of Cast Iron, Proceedings of the Society for Metals and Materials Science, Computer Simulation Division, The Metallurgical Society (Toronto, Canada), 1985 H. Fredriksson and A. Olsson, Mechanism of Transition from Columnar to Equiaxed Zones in Ingots, Mater. Sci. Tech., Vol 2, 1986, p 508–516 H. Fredriksson and B. Rogberg, Met. Sci., Vol 13, 1979, p 685–690 L. Ba¨ckerud and B. Chalmers, Trans. Met. Soc. AIME, Vol 245, 1969, p 309–318 A Guide to the Solidification of Steels, Jernkontorets, Stockholm, 1977 M.C. Flemings, Solidification Processing, McGraw-Hill, 1974



Eutectic Modification Level of Al-7Si Alloy by Computer Aided Recognition of Thermal Analysis Cooling Curves, Mater. Sci. Eng. A, Vol 419 (No. 1–2), March 15, 2006, p 283–289 D. Emadi, L.V. Whiting, S. Nafisi, and R. Ghomashchi, Applications of Thermal Analysis in Quality Control of Solidification Processes, J. Therm. Anal. Calorimetry, Vol 81 (No. 1), 2005, p 235–241 J. Medved and P. Mrvar, Thermal Analysis of the Mg-Al Alloys, Mater. Sci. Forum, Vol 508, 2006, p 603–608 J. Sertucha, R. Suarez, J. Izaga, L.A. Hurtado, and J. Legazpi, Prediction of Solid State Structure Based on Eutectic and Eutectoid Transformation Parameters in Spheroidal Graphite Irons, Int. J. Cast Met. Res., Vol 19 (No. 6), Dec 2006, p 315–322 D.J. Varga L., Examination of the Eutectic Crystallization of Cast Iron by Thermal Analysis, Mater. Sci. Forum, Vol 508, 2006, p 87–92 S. Vogelgesang, G.-S. Leo, and B. Tonn, Thermal Analysis of Grain Refined Brass, Livarski Vestn., Vol 51 (No. 3), 2004, p 133–141

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 357-361 DOI: 10.1361/asmhba0005218

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

X-Ray Imaging of Solidification Processes and Microstructure Evolution Ragnvald H. Mathiesen and Lars Arnberg, Norwegian University of Science and Technology

IN THE PURSUIT for improved insight into the interplay between different transport phenomena that control alloy solidification processes, the inherent difficulties with providing direct observation of the microstructure evolution have been a serious drawback. Typically, experimental observations of metals and alloys have been limited to metallographic and electron microscopy investigations of either quenched or as-solidified material. On the other hand, in situ studies of microstructure evolution and several important process phenomena have been accessible through optical video microscopy studies with a selection of transparent, organic model systems available for primaryphase solidification, eutectics, and peritectics displaying morphologies and growth patterns analogous to those found in metals (Ref 1). Such studies have been extensive and contributed to advance theory (Ref 2–4) as well as provided benchmark data used for modeling (Ref 5, 6). Despite their impact, however, these model systems are far from complete as analogs to metals and alloys; several of their physical properties tend to differ distinctively from those of metals, for example, solute/solvent mass ratios, heat conductivities and capacities. Perhaps the most important difference concerns their Prandtl numbers, the ratio between viscous and diffusive properties, which are typically 3 to 4 orders of magnitude larger for the model systems. Therefore, solidification processes in metals may often exhibit a greater complexity than in the transparent organics, since buoyancy-driven flow is much more prominent in the former category. Accordingly, it is of significant interest to find and develop techniques that allow for direct in situ observations of real metallic alloys.

X-Ray Imaging Metal transparency and interaction with x-rays have long been recognized as obvious candidate principles from which methods for in situ monitoring of solidification processes could be developed. Yet, the limited x-ray

source brightnesses and detection efficiencies available have hampered the practical impact of x-rays as a diagnostic tool for imaging studies at the characteristic time and length scales of microstructure evolution. A general description of contrast formation by x-ray transmission in matter can be given by a wave-optical approach. Let Ei(x, y, z) be an incident monochromatic plane-wave propagating along the x-axis of a Cartesian laboratory coordinate system. The transverse cross sections of the incident wave-field, Ei (y, z), and the field transmitted through the sample, Et(y, z), are related via a complex sample transmission function, F(y,z): Et ðy; zÞ ¼ Fðy; zÞEi ðy; zÞ

(Eq 1)

with Fðy; zÞ ¼ Aðy; zÞ expðijðy; zÞÞ

(Eq 2)

The amplitude, A(y,z), and phase, j(y,z), relate to the x-ray absorption and refractive properties of the material, respectively, and can be expressed as:   ð 1 Aðy; zÞ ¼ exp  mðx; y; zÞdx 2 2p ðnðx; y; zÞ  1Þdx l



(Eq 3)

In-House Laboratory X-Ray Imaging of Solidification Processes

ð

jðy; zÞ ¼

105 or less. However, provided that phase contrast could be used, it would help to resolve phases with similar atomic compositions but adequate differences in density. Current state-of-the-art x-ray imaging detectors operate with limiting spatial resolutions similar to optical microscopes. Hence, for any phase contrast to be resolvable by the camera system, an extremely high degree of transversal coherence of the incident wave-field is required, corresponding to at least 105 l. Since a highphoton flux also must be maintained at the sample to allow for real-time studies, conditions for phase contrast can only be delivered with modern high-brilliance synchrotron radiation (Ref 7). Finally, it should be noted that the quantity measured in x-ray imaging experiments is the intensity Iðy; zÞ / Et ðy; zÞ Et ðy; zÞ ¼ jEt ðy; zÞj2 , with Et as the complex conjugate Et in Eq 1. Since the actual phase field j(y,z) is not retained in the measurements, true phase contrast imaging requires some form of mathematical reconstruction. Nevertheless, effects of phase contrast may be directly visible in I(y,z) from interference between different exit wavefield components.

(Eq 4)

where m is the linear absorption coefficient, n is the real part of the refractive index, and l is the x-ray photon wavelength. With Z as the atomic number and ra as the atomic density, m can be shown to vary roughly as raZ 4 l3, except at energies corresponding to photoelectric excitations, where it is discontinuous. Thus, absorption contrast is mainly available between phases with a distinct difference in atomic composition. The (n  1) refractive properties follow approximately a raZ l2 dependence in between the discontinuities occurring at photoelectric excitations and remain extremely weak with all materials, deviating from unity by only

The first in situ x-ray investigations of solidification were based on conventional in-house source radiography and used for studies of solute redistribution and boundary layer propagation (Ref 8, 9). The spatial resolutions available, Dr 100 mm, prevented detailed studies of solid-liquid interface morphologies, but the resolution in x-ray absorption contrast was adequate to verify proximity to Scheil-Gulliver conditions as well as to demonstrate the influence of buoyancydriven flow on the solute boundary layer by comparative measurements varying the growth direction relative to gravity (Ref 9). Later, the resolution of in-house-based radiography setups have been improved significantly by the introduction of microfocus sources. Such setups have

358 / Principles of Solidification been used to study topics like convection and microstructure evolution in gallium-indium eutectics (Ref 10, 11), hydrogen porosity formation in aluminum-copper (Ref 12), and liquid droplet formations, striations, and engulfment in aluminum-base monotectics (Ref 13, 14) reporting Dr  25 to 30 mm. In a recent real-time in-house study of metal foaming, Dr  5 mm was reported (Ref 15). As the nominal spatial resolution is improved, the temporal resolution and the source brilliance become more critical factors. Temporal resolution here refers to either: 1. the inverse of the frame-grabbing rate, that is, 1/(Dtexp + Dtr.o.), where Dtr.o. is the read-out dead-time of the camera system, when each frame is subjected to analysis; or 2. the accumulation of the former when several frames must be integrated to provide image statistics sufficient for analysis. The temporal evolution of the system under investigation will lead to blurring whenever vobj(r) Dtexp > Dr, where vobj(r) is the motion of a contrast object in local position r in the sample, relative to the x-ray camera system, and Dtexp is the exposure time. Hence, higher spatial resolutions can be obtained only by reducing the exposure time or by slowing down the dynamics and kinetics of the system. The potential for reduction of Dtexp is determined by the x-ray source brilliance, the photon-detection quantum efficiency and signal-to-noise of the detector system, and finally the transmission and contrast provided for by the sample under investigation. The latter is shown by Eq 1 to 4 to be a function of both the x-ray source and the energy in use, as well as the intrinsic x-ray properties of the different sample constituents. Decent spatial resolutions are possible to realize in laboratory studies of metal foaming and hydrogen porosity, only because the ratio between the x-ray attenuation lengths of H2 gas and metal are adequately large, 106 or more at relevant energies. The x-ray attenuation length of a medium, at a particular monochromatic x-ray photon energy, is defined as the length of propagation where photoelectric absorption by the medium has reduced the xray beam amplitude by a factor of e1, with e as the natural exponential base. In comparison, with E varying from 10 to 50 keV, the largest ratio found between x-ray attenuation lengths of any two base metals (magnesium and heavier) is 100. Consequently, high-resolution in situ studies with fully metallic samples usually require longer exposure times or improved performance of the x-ray sources and detectors. Nevertheless, in a recent study of directional solidification in gallium-indium, employing a state-of-the-art microfocus source and detector, a spatial resolution of Dr  10 mm and a temporal resolution of Dt = (Dtexp + Dtr.o.)  0.5 s was used to investigate dendritic



growth and analyze the buoyancy-driven flow of rejected solute (Ref 16).

X-Ray Imaging Studies of Solidification Processes with Synchrotron Radiation The conditions for in situ x-ray imaging are vastly improved by the eminent source brilliance and x-ray energy selectivity offered at modern synchrotron radiation (SR) sources. These facilities have opened for the use of x-ray imagingbased techniques to investigate interface morphology evolution, solute transport, and various process phenomena at spatiotemporal resolutions gradually approaching those of light optical video microscopy. In general, there are three distinct viable imaging techniques made available by SR for the real-time investigation of solidification microstructures in alloys: two-dimensional x-ray topography, two-dimensional x-ray radiography, and ultrafast three-dimensional x-ray tomography.

X-Ray Topography X-ray topography is a diffraction-based imaging technique, in principle, the x-ray variant of dark-field electron microscopy (for a review of the technique, see Ref 17). Synchrotron whitebeam x-ray transmission topography was first employed in solidification science for in situ studies of single-crystal melting-solidification processes in pure tin (Ref 18) and aluminum (Ref 19). Topography may be carried out in reflection or transmission geometry and with a nonmonochromatic (white beam) or monochromatic radiation, where the combination of a white beam and transmission geometry normally would be preferred for in situ solidification studies. Other applications include studies of planarcellular and cellular-dendritic morphology transitions during directional solidification in aluminum-copper alloys (Ref 20), including the role of strain to the onset of interface perturbations (Ref 21, 22) (Fig. 1). The main advantage of topography over the other x-ray imaging techniques is that it is unique in providing in situ information on microstructure stresses induced during solidification. Also, contrast is available at solid-liquid interfaces even for dilute alloys and pure substances, since only the solid gives rise to diffraction signals. On the other hand, in situ topography is severely limited by the output signal strength. In the x-ray energy regime below 100 keV, the elastic x-ray scattering cross sections of any element are 103 or less of their absorption cross-sectional counterparts. Typical obtainable spatiotemporal resolutions for in situ studies of metals, employing state-of-the-art live-capturing x-ray cameras and a white-beam transmission mode, are of the order Dr  15 mm and Dt  2 s.

Fig. 1

(200) reflection x-ray topograms showing local cell rotation during directional solidification of an Al-5.7wt%Ni alloy. (a) t0. (b) t0 + 360 s. Source: Ref 22

X-Ray Radiography Currently, synchrotron x-ray radiography is the technique most frequently used for in situ studies of alloy solidification microstructures and phenomena. It offers superior spatiotemporal resolutions, down to Dr  1.5 mm and Dt  0.25 s, and is reasonably flexible, provided that the alloy system under investigation contains constituent phases that can be resolved by some form of transmission contrast. With third-generation SR sources, a high degree of transversal beam coherence can be obtained, facilitating the use of x-ray phase contrast in addition to conventional absorption contrast (Ref 7). Although actual phase-contrast objects have to be reconstructed, by employing sample-camera distances within the so-called near-field of the first Fresnel region, low-order interference fringes between waves refracted in slightly different directions across phase boundaries can appear directly in the radiograms as sharp contrast features (Ref 7, 23). Thus, direct radiographic contrast enhancement can be obtained locally at interfaces and can resolve boundaries between aggregates of similar atomic compositions. Provision of additional interference contrast at phase boundaries is particularly advantageous to studies of curved fronts, because any cell or dendrite will exhibit a loss in transmission contrast at the interface due to its three-dimensional curvatures. Unlike radiation from conventional sources, synchrotron radiation provides a high-photon flux over a continuous energy band. In most cases, monochromatization is affordable, permitting selection of an x-ray energy that yields an optimum compromise between sample transmission and contrast. The first in situ SR radiography experiments in solidification science were directional solidification investigations of various tin-lead alloys to study evolution of columnar and equiaxed dendritic microstructures (Ref 24). In addition to the characteristics of the source, access to a high-resolution, fast-readout, low-noise charge-

X-Ray Imaging of Solidification Processes and Microstructure Evolution / 359 coupled device camera specially designed for time-resolved studies (Ref 25) was critical to the outcome of the experiments. Successive studies with the same setup on various aluminum-copper alloys 1. microstructure evolution during primaryphase columnar and equiaxed dendritic growth, as well as stable and destabilized eutectic growth (Ref 25); 2. destabilization of columnar dendritic fronts by local, gravity-induced settlements of rejected solute, including spatiotemporally resolved analysis of the solid-liquid interface morphology and chemical analysis of the liquid (Ref 26), shown in Fig. 2; and 3. dendrite detachment initiated by various mechanisms in different regions of the mushy zone (Ref 27), with particular emphasis on the role of solute pile-up and buoyancy-driven flow of fragments to promote mesoscale segregation, columnar front blocking, and the initial stage of columnarto-equiaxed transition (Ref 27, 28). In situ synchrotron radiography has also been used to study fragmentation (Ref 30), coarsening (Ref 31), and temperature gradient-zone melting (Ref 32) during directional solidification of tinbismuth alloys; columnar-to-equiaxed transitions and equiaxed dendritic growth in grainrefined and nonrefined aluminum-nickel (Ref 33, 34); solidification dynamics during faceted growth of AlPdMn quasi-crystals (Ref 34, 35); coarsening during semisolid processing of grain-refined aluminum-germanium (Ref 36); equiaxed dendritic growth in grain-refined aluminum-copper (Ref 29); and finally monotectic solidification in aluminum-indium (Ref 37) and aluminum-bismuth (Ref 29) demonstrating how hydrodynamics, short-range droplet-droplet interactions, and coagulation mechanisms apparent in the two-phase immiscible liquid zone can influence the monotectic reaction and perturb the primary-phase morphology.

X-Ray Tomography Ultrafast x-ray tomography is the most exclusive novelty offered with modern synchrotron facilities. Radiography and topography experiments provide two-dimensional transmission projected images of the sample. In a tomography experiment, a sequence of two-dimensional radiograms are collected while the sample is pivoted by a few hundred increments over a 180 rotation about an axis perpendicular to the propagation direction of the x-ray beam. Each radiogram provides the transmission contrast (amplitude and/or phase) corresponding to a particular sample orientation. Thus, by applying a suitable reconstruction algorithm, the three-dimensional structure of the sample can be rendered. With modern synchrotron sources, complete sets of projection data can be acquired within 10 s, providing a full threedimensional set of approximately 109 pixel

Fig. 2

X-ray radiograms showing planar eutectic and columnar dendritic growth during directional solidification parallel with gravity in an Al-30wt%Cu alloy. Left column: every fifth image from a part of the sequence representing the microstructure evolution at times t0, 2.25 and 4.5 s. The images also display the local liquid concentration, Cl(Cu), in a relative manner (as indicated by the horizontal contrast bar). Right column: examples of quantitative information extracted by image processing. (a) Liquid constitution contour map at 4.5 s, starting at Cl = 33 wt%Cu (black) and with increments of 0.5 wt%Cu. (b) Velocities for the eutectic front and columnar tips 1 to 3, enumerated from left to right, normalized against the sample pulling velocity, vsp  22.5 mm/s. (c) The time evolution of the two-dimensional solid-liquid interface projections over the full video sequence. Source: Ref 26, 29

360 / Principles of Solidification values, with an attainable three-dimensional spatial resolution of 2 mm (Ref 38). However, in order to achieve such spatial resolution in studies of evolving systems, the contrast object shapes must remain semisteady within the 2 mm resolution limit over the 10 s time required to collect a complete tomogram. Another complication is that, while topography and radiography with state-of-the-art x-ray cameras allow for direct on-line observation of the systems under study, tomography visualization is indirect because reconstruction is required. Hence, it is less flexible with respect to adjustment and fine-tuning of experimental parameters; and on-line adjustments are practically unavailable. The first demonstration of ultrafast in situ tomography in solidification science was a study of the solidification of inoculated Al-4wt%Cu alloys (Ref 39), employing a white beam to maximize the incident flux, and reporting a spatial resolution of 3 mm with a 10 s tomogram acquisition time. Within these limits, any interface velocities beyond 0.3 mm/s will cause temporal blurring, and accordingly, a slow cooling rate of 0.1 K/s was used. Image processing allowed for extraction of quantities such as three-dimensional solid-liquid interface morphologies, solid fraction versus temperatures, solid-phase shrinkage upon cooling, and copper concentration in both liquid and solid, although the latter is likely to be somewhat hampered by the fact that a white beam was used. When using a white beam, chemical analysis requires corrections for both the x-ray energy shift of the sample and the nonlinear energy

Fig. 3

response of the detector. Also, quantitative analysis had to be limited to the later stages of solidification ( fs > 0.5). In general, early stages of solidification are difficult to assess with tomography, since these tend to involve growth velocities beyond the blurring limit, and any gravityinduced relative solid-liquid flow prior to the onset of a coherent crystal network could lead to blurring by motion of the grains. More recent studies have focused on coarsening effects during partial remelting in Al-15.8wt%Cu (Ref 40) and solidification experiments in nonrefined Al-7wt%Si-10wt% Cu alloys (Ref 41), shown in Fig. 3. In these studies, cooling rates up to 3 K/s were used, while the freezing range suitable for analysis was limited between 0.18 > fs > 0.28. The lower boundary is set by the limiting contrast required for image processing, extended here as compared to Ref 39 by the heavier copper alloying. The upper limit corresponds to the point where eutectic silicon starts to grow on the aluminum dendrites, at which time further analysis is impaired since hardly any contrast can be obtained between aluminum and silicon. Coarsening phenomena have also been studied during semisolid processing of aluminum-germanium alloys, combining in situ real-time radiography and semistatic tomography (Ref 36), thus yielding information on the temporal evolution of the system by radiography and snapshots of the three-dimensional microstructure at discrete processing times by tomography. Other efforts have been made to combine different in situ x-ray imaging

Ultrafast tomography. Top row shows two-dimensional regional slices taken from the three-dimensional tomograms (size: 1.4 by 1.4 by 1.4 mm3, or 0.00009 by 0.00009 by 0.00009 in.3) at solidification times of 60, 660, and 1260 s. Bottom row shows three-dimensional reconstructions of a dendrite at solidification times of (a) 660, (b) 840, and (c) 1020 s, localized in the region of interest indicated in the 660 s slice. The grayscale in the tomograms displays the solid curvature variations; a region with prominent coarsening is highlighted. Source: Ref 41

techniques to obtain complementary information, for example, simultaneous radiography and topography (Ref 42); however, the spatiotemporal resolutions obtainable with the combined experiment are limited since the x-ray beam requirements set by topography are nonoptimal with respect to in situ radiography.

Further Reading The reader is advised to consult the original works cited for more details on the particular qualitative and quantitative results obtained in the various studies, many of which provide extensive comparisons of their results to governing theories. The original works also contain a vast amount of details on the explicit experimental conditions used for the various studies, that is, x-ray characteristics, furnace systems, detectors, and other information regarding equipment or experimental procedures. Finally, a few video demonstrations of real in situ observations of solidification microstructure evolution and various process phenomena can be accessed in the on-line version of Ref 29, distributed via the TMS website (http://www. tms.org/TMSHome.html).

REFERENCES 1. K.A. Jackson and J.D. Hunt, Transparent Compounds That Freeze Like Metals, Acta Metall., Vol 13, 1965, p 1212–1215 2. S.C. Huang and M.E. Glicksman, Fundamentals of Dendritic Solidification, Part I: Steady-State Tip Growth, Acta Metall., Vol 29, 1981, p 701–715 3. H. Esaka and W.J. Kurz, Columnar Dendrite Growth—Experiments on Tip Growth, J. Cryst. Growth, Vol 75, 1985, p 578–584 4. W. Losert, B.Q. Shi, and H.Z. Cummins, Evolution of Dendritic Patterns During Alloy Solidification: Onset of the Initial Instability, Proc. Natl. Acad. Sci. USA, Vol 95, 1998, p 431–438 5. B.T. Murray, A.A. Wheeler, and M.E. Glicksman, Simulations of Experimentally Observed Dendritic Growth-Behaviour Using a Phase-Field Method, J. Cryst. Growth, Vol 154, 1995, p 386–400 6. S. Akamatsu, M. Plapp, G. Faivre, and A. Karma, Overstability of Lamellar Eutectic Growth Below the Minimum-Undercooling Spacing, Metall. Mater. Trans. A, Vol 35, 2004, p 1815–1828 7. A. Snigirev, I. Sinigreva, V. Kohn, S. Kutznesov, and T.I. Schelekov, On the Possibilities of X-Ray Phase Contrast Microimaging by Coherent High-Energy Synchrotron Radiation, Rev. Sci. Instrum., Vol 66, 1995, p 5486–5492 8. J. Forsten and H.M. Miekkoaja, Radiographic Observations of Solidification of Metals, J. Inst. Met., Vol 95, 1967, p 143

X-Ray Imaging of Solidification Processes and Microstructure Evolution / 361 9. M.P. Stephenson and J. Beech, In-Situ Radiographic Observations of Solute Redistribution During Solidification, Proc. Int. Conf. Solidification and Casting of Metals (Sheffield, U.K.), July 1977, p 34–38 10. J.N. Koster, Visualizing Convection and Solidification in Metallic Alloys, J. Met. (JOM), March 1997, p 31–35 11. H. Yin and J.N. Koster, In-Situ Observation of Concentrational Stratification and Solid-Liquid Interface Morphology During Ga-5% In Alloy Melt Solidification, J. Cryst. Growth, Vol 205, 1999, p 590–606 12. P.D. Lee and J.D. Hunt, Hydrogen-Porosity in Directionally Solidified AluminiumCopper Alloys: In-Situ Observations, Acta Mater., Vol 45, 1997, p 4155–4169 13. P.A. Curreri and W.F. Kaukler, Real-Time X-Ray Transmission Microscopy of Solidifying Al-In Alloys, Metall. Mater. Trans. A, Vol 27, 1996, p 801–808 14. W.F. Kaukler, F. Rosenberger, and P.A. Curreri, In-Situ Studies of Precipitate Formation in Al-Pb Monotectic Solidification by XRay Transmission Microscopy, Metall. Mater. Trans. A, Vol 28, 1997, p 1705–1710 15. F.G. Moreno, M. Fromme, and J. Banhart, Real-Time X-Ray Radioscopy on Metallic Foams Using a Compact Micro-Focus Source, Adv. Eng. Mater., Vol 6, 2004, p 416–420 16. S. Boden, B. Willers, S. Eckert, and G. Gerbeth, Visualisation of the Concentration Distribution and the Flow Field in Solidifying Metallic Melts by Means of X-Ray Radioscopy, Proc. Int. Conf. Solidification and Casting of Metals (Sheffield, U.K.), July 2007, p 311–315 17. A.R. Lang, International Tables of Crystallography, Vol C, A.J.C. Wilson and E. Prince, Ed., Kluwer Academic Publishers, 1999, p 113–123 18. O. Nittono, T. Ogawa, S.K. Gong, and S. Nagakura, In-Situ Observations on Melt-Growth Process of Tin-Crystals by Means of Synchrotron X-Ray Topography, Jpn. J. Appl. Phys. 2, Lett., Vol 23, 1984, p L581–L583 19. G. Grange, C. Jourdan, A.L. Coulet, and J. Gastaldi, Observation of the MeltingSolidification Processes of an Al-Crystal by Synchrotron X-Ray Topography, J. Cryst. Growth, Vol 72, 1985, p 748–752 20. G. Grange, C. Jourdan, and J. Gastaldi, InSitu Observations by Synchrotron White Beam X-Ray Topography of the Planar, Cellular and Dendritic Growths of a Binary Alloy, J. Phys. D, Appl. Phys., Vol 26, 1993, p A98–A101 21. G. Grange, C. Jourdan, J. Gastaldi, and B. Billia, Strain Visualisation of the Onset of Morphological Instability and Defect Formation in Cellular Solidification of a Dilute Al-Cu Alloy, Acta Mater., Vol 45, 1997, p 2329–2338 22. B. Billia, N. Bergeon, H.N. Thi, H. Jamgotchian, J. Gastaldi, and G. Grange, Cumulative

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

Mechanical Moments and Microstructure Deformation Induced by Growth Shape in Columnar Solidification, Phys. Rev. Lett., Vol 93, 2004, article 126105 C. Raven, A. Snigirev, I. Snigireva, P. Spanne, A. Souvorov, and V. Kohn, Phase-Contrast Microtomography with Coherent High-Energy Synchrotron X-Rays, Appl. Phys. Lett., Vol 69, 1996, p 1826–1828 R.H. Mathiesen, L. Arnberg, F. Mo, T. Weitkamp, and A. Snigirev, Time Resolved X-Ray Imaging of Dendritic Growth in Binary Alloys, Phys. Rev. Lett., Vol 83, 1999, p 5062–5065 R.H. Mathiesen, L. Arnberg, K. Ramsskar, T. Weitkamp, C. Rau, and A. Snigirev, Time Resolved X-Ray Imaging of Aluminium Alloy Solidification Processes, Metall. Mater. Trans. B, Vol 33, 2002, p 613–623 R.H. Mathiesen and L. Arnberg, X-Radiography Observations of Columnar Dendritic Growth and Constitutional Undercooling in an Al-30%wtCu Alloy, Acta Mater., Vol 53, 2005, p 947–956 R.H. Mathiesen, L. Arnberg, P. Bleuet, and A. Somogyi, Crystal Fragmentation and Columnar-to-Equiaxed Transitions in AlCu Studied by Synchrotron X-Ray Video Microscopy, Metall. Mater. Trans. A, Vol 37, 2006, p 2515–2524 D. Ruvalcaba, R.H. Mathiesen, D.G. Eskin, L. Arnberg, and L. Katgerman, In-Situ Observations of Dendritic Fragmentation During Directional Solidification of an Aluminium Alloy, Acta Mater., Vol 55, 2007, p 4287–4292 L. Arnberg and R.H. Mathiesen, RealTime, High Resolution X-Ray Video Microscopy of Alloy Solidification, J. Met. (JOM), Aug 2007, p 20–26 H. Yasuda, I. Ohnaka, K. Kawasaki, A. Sugiyama, T. Ohmichi, J. Iwane, and K. Umetani, Direct Observation of Stray Crystal Formation in Unidirectional Solidification of Sn-Bi Alloy by X-Ray Imaging, J. Cryst. Growth, Vol 262, 2004, p 645– 652 B. Li, H.D. Brody, and A. Kazimirov, Real Time Synchrotron Microradiography of Dendrite Coarsening in Sn-13 wt pct Bi Alloy, Metall. Mater. Trans. A, Vol 38, 2007, p 599–605 B. Li, H.D. Brody, and A. Kazimirov, Synchrotron Microradiography of Temperature Gradient Zone Melting in Directional Solidification, Metall. Mater. Trans. A, Vol 37, 2006, p 1039–1044 G. Reinhart, N. Mangelinck-Noe¨l, H.N. Thi, T. Schenk, J. Gastaldi, B. Billia, P. Pino, J. Ha¨rtwig, and J. Baruchel, Investigation of Columnar-Equiaxed Transition and Equiaxed Growth of Aluminium Based Alloys by X-Ray Radiography, Mater. Sci. Eng. A, Vol 413–414, 2005, p 384–388 T. Schenk, H.N. Thi, J. Gastaldi, G. Reinhart, V. Cristiglio, N. Mangelinck-Noe¨l,

35.

36.

37.

38.

39.

40.

41.

42.

H. Klein, J. Ha¨rtwig, B. Grushko, B. Billia, and J. Baruchel, Application of Synchrotron X-Ray Imaging to the Study of Directional Solidification of Aluminium-Based Alloys, J. Cryst. Growth, Vol 275, 2005, p 201–208 H.N. Thi, J. Gastaldi, T. Schenk, G. Reinhart, N. Mangelinck-Noe¨l, V. Cristiglio, B. Billia, B. Grushko, J. Ha¨rtwig, H. Klein, and J. Baruchel, In-Situ and Real-Time Probing of Quasi-Crystal Solidification Dynamics by Synchrotron Imaging, Phys. Rev. E, Vol 74, 2006, article 031605 S. Zabler, A. Rueda, A. Rack, H. Riesemeier, P. Zaslansky, I. Manke, F. GarciaMoreno, and J. Banhart, Coarsening of Grain-Refined Semi-Solid Al–Ge32 Alloy: X-Ray Microtomography and In Situ Radiography, Acta Mater., Vol 55, 2007, p 5045–5055 H. Yasuda, T. Nagira, M. Yoshia, A. Sugiyama, I. Ohnaka, K. Kawasaki, K. Umetani, and K. Uesugi, TimeResolved X-Ray Imaging of Dendritic and Monotectic Solidification of Sn-, Al- and Cu-Based Alloys, Proc. Int. Conf. Solidification and Casting of Metals (Sheffield, U.K.), July 2007, p 306–310 M. DiMichel, J.M. Merino, D. FernandezCarreiras, T. Buslaps, V. Honkimaki, P. Falus, T. Martins, and O. Svensson, Fast Microtomography Using High Energy Synchrotron Radiation, Rev. Sci. Instrum., Vol 76, 2005, article 043702 O. Ludwig, M. DiMichel, L. Salvo, M. Sue´ry, and P. Falus, In-Situ ThreeDimensional Microstructural Investigation of Solidification of an Al-Cu Alloy by Ultrafast X-Ray Microtomography, Metall. Mater. Trans. A, Vol 36, 2005, p 1515– 1523 N. Limodin, L. Salvo, M. Sue´ry, and M. DiMichiel, In Situ Investigation by XRay Tomography of the Overall and Local Microstructural Changes Occurring During Partial Remelting of an Al-15.8 wt.% Cu Alloy, Acta Mater., Vol 55, 2007, p 3177–3191 N. Limodin, E. Boller, L. Salvo, M. Sue´ry, and M. DiMichel, In-Situ Fast X-Ray Tomography Investigation of Microstructural Changes Occurring During Partial Remelting and Solidification of Al-Cu Alloys, Proc. Int. Conf. Solidification and Casting of Metals (Sheffield, U.K.), July 2007, p 316–320 H.N. Thi, H. Jamgotchian, J. Gastaldi, J. Hartwig, T. Schenk, H. Klein, B. Billia, J. Baruchel, and Y. Dabo, Preliminary In Situ and Real-Time Study of Directional Solidification of Metallic Alloys by X-Ray Imaging Techniques, J. Phys. D, Appl. Phys., Vol 36, 2003, p A83–A86

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 362-369 DOI: 10.1361/asmhba0005220

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

The Oxide Film Defect—The Bifilm John Campbell, University of Birmingham England

MOST METALS start their lives in the liquid state. Once melted, they are usually subjected to various transfers between furnaces or ladles, involving pouring or other types of surface turbulence. These actions entrain (fold in) the surface film to create entrainment defects. These are principally:  Bubbles (that in turn create bubble trails)  Bifilms, which are doubled-over surface

films that act as cracklike defects

 Various entrained debris, traditionally known

as exogenous or external inclusions The bifilm defect is the subject of this article. It appears to be a common but serious and almost overlooked casting defect. Large populations of bifilms are usually introduced into metals at an early stage of their production. In general, their presence has been unsuspected because, even though they can have a large area, they often are only nanometers thick and not easily detected by conventional nondestructive techniques. They can be serious defects because they act as cracks in the liquid, and the cracks become frozen into the casting. The presence of cracks in suspension in liquid metal explains many otherwise inexplicable features of cast products, such as porosity, hot tearing, the morphologies of second phases, and impaired reliability of mechanical properties. The fundamental difference between such entrained defects (associated with a macroscopic unbonded interface) and defects and inclusions grown in the melt is of central significance for the failures of metals by mechanical or corrosion-type mechanisms. Analysis of bifilms provides a simple, powerful, and elegant concept to explain many features of the metallurgy of castings. Since bifilms occur mainly during turbulence involved in the pouring of liquid metals, they similarly can be avoided by avoiding turbulence. Although the effects of bifilms can be reduced (but often not eliminated) by expensive postcasting operations such as hot isostatic pressing or working, the major future potential lies in techniques for their avoidance. Some casting operations are already taking the first steps in such new technology and benefiting technically and commercially.

Bifilms The oxide film on the surface of a liquid can become entrained in the bulk liquid. The entrainment action is an enfolding of the surface, leading automatically to the folding-in and submersion of its surface film (Fig. 1). Such surface turbulence leads to entrainment events that are usually rapid, requiring only milliseconds. The creation of a new surface and the time available for the formation of a new oxide film on this fresh surface is so limited that the entrained film can be very thin, of the order of nanometers (Ref 1). Most films on liquid metals are oxides (although nitrides and carbon films are also common), and most are solid at the temperature of the melt. When entrained, the dry surfaces of the films fold inward, necessarily facing each other and making a double film, hence the name bifilm. The opposing halves of the film cannot bond, being a ceramic-to-ceramic interface; thus, all entrained oxides effectively constitute a crack (feature “A-B” in Fig. 1) in the liquid. Most pouring actions involve such severe surface turbulence that they fill the liquid with bifilm cracks. The bifilms are usually invisible and often undetectable by most techniques because the surface oxides are usually only a few molecules thick, approximately 20 nm, even though they may be square millimeters or square centimeters in extent. Occasionally, bifilms that are fractions of a meter (3 ft) across are seen in large alloy steel and nickel-base alloy castings. If they extend across the wall of the casting, they can form efficient leak paths and are the main reason for the rejection of castings for leakage faults. If bifilms are so common and their effects are so serious, how is it that they have been universally overlooked until now? The fact is that they are difficult to detect with standard metallurgical techniques. Despite decades of experience with the scanning electron microscope (SEM) in examining fractures, no one during the early years was aware of the presence of bifilms, with the predictable consequence that none were observed. Now that the features of bifilms are known, it is possible to recognize these features

and find bifilms in profusion on nearly every metal fracture surface that is examined. Furthermore, the bifilms are sometimes so extensive that they cover 100% of the fracture surface. (An example in Fig. 7 is described later.) They can remain largely invisible even in the SEM because they are so thin that the structure of the metal is clearly seen through the film. The presence of a film is seen only at the highest magnifications, where the characteristic creases and folds point to the presence of a defect that can be identified as a film originally generated on the liquid surface. Its near-mirror image resides on the opposite fracture surface as the “butterfly wing” pattern, which can often be seen if the fracture surfaces are placed together. Its characteristic fluorescence in the electron beam and its chemical composition are further identifying characteristics. Bifilms are easily opened and observed when the metal is in the liquid state. Fox and Campbell (Ref 2) assumed that there would be a thin layer of air (or other residual gas) trapped between the halves of the bifilm. The gas would be retained in both the minute surface roughness and the larger irregular folds and creases. Therefore, they predicted that if the pressure above the melt were reduced by a factor of 10, the gas entrained between the halves of the bifilm would be expanded by a factor of 10. Fox reduced the pressure to 0.01 atm

Fig. 1

Schematic representation of surface turbulence causing the entrainment of bifilms and bubbles. Small entrained bubbles form pores that decorate the bifilm crack “A-B.” Large bubbles, “C” and “D,” are buoyant, creating bubble trails.

The Oxide Film Defect—The Bifilm / 363 (1 kPa) and thus expanded the bifilms by 100 times, sufficient to make them observable in a radiographic study (Fig. 2). In subsequent work, the factor-of-10 reduction in pressure (down to 0.1 atm, or 10 kPa, residual pressure) has been found to be adequate, allowing the reduced pressure test (RPT) to be a powerful technique for observing and quantifying bifilm size and number. Bifilms, when searched for, can usually be found in large numbers. In aluminum alloys, a concentration of 109 m 3 (1.0 per mm3, or 16,000 per in.3) appears average, and even 1012 m 3 (1000 per mm3, or 16 million per in.3) is relatively common. A bifilm concentration of 1015 m 3 (1.0 million per mm3 or 16 million per in.3) would be necessary if a separate bifilm were required to be in place to initiate the fracture of the microscopic eutectic silicon particles that are commonly seen on fracture surfaces of aluminum-silicon alloys. As an alternative quantifying measure, the total length of bifilms in a standard sample can be measured. A convenient standard sample is that produced by the RPT, weighing approximately 100 g (0.2 lb). When sectioned vertically through its center and polished, its area of approximately 35 by 35 mm (1.4 by 1.4 in.) can be scanned using an automated metallographic system. In this way, the number can be counted and the average maximum length of the defects can be determined. Such inspections may reveal the total length of bifilms on the sampled surface to be only 0 to 1 mm (0 to 0.04 in.) in a high-quality melt, whereas a much worse length of 50 mm (2 in.) is more common. However, the length sometimes approaches 500 mm

Fig. 2

(20 in.) (Ref 3). The latter figure corresponds approximately to a sobering content of 5 mm2 (0.008 in.2) of bifilm per cubic millimeter. This work illustrates not only the wide spectrum of numbers and sizes but also that populations of very different sizes of bifilms occur because of different melt conditions. In addition, different populations accumulated from different stages of manufacture often coexist in the same sample. It is important to keep in mind that bifilms are not always oxides. Sometimes, they are nitrides, sulfides, amorphous carbon, graphite, or other materials. They are not always solid but can be partly solid or even liquid and, in some cases, not present at all. A bifilm would not be present when casting pure gold or platinum, since these noble metals do not possess surface oxides. Alternatively, a metal may contain no oxide films if it has been filtered or handled extremely carefully so as to avoid entraining its surface. When bifilms are present, as is usually the case, the quantity can be controlled to some extent by careful handling of the liquid metal. Entire commercial casting operations have been revolutionized by attention to the metal handling and pouring practices. The production of scrap castings has been reduced while simultaneously improving properties, usually with a reduction in costs.

Entrainment The minimum criterion for entrainment is a critical Weber number (We > 1), in which the inertial forces in the liquid that are tending to

Comparison of x-ray radiographs taken from the same melt of Al-7Si-0.4Mg solidified at the same time under (a) 1 atm (100 kPa), showing diffuse images of furled bifilms, and (b) 0.01 atm (1 kPa), showing unfurled bifilms. Source: Ref 2

perturb the surface exceed the surface tension forces that tend to restrain the surface, holding it in a compact form. In practice, the condition works out to be equivalent to a critical velocity, above which the surface can be propelled upward, achieving a height sufficient that it may enfold its surface as it falls back under gravity. This critical velocity is in the region of 0.5 m/s (1.5 ft/s) for nearly all liquids and is a useful parameter that is now being built into many filling system designs for castings, with excellent results if carried out correctly (Ref 4). At the present time, practically every pouring event violates the critical velocity criterion and introduces oxides to the melt. This is simply the result of the acceleration under gravity causing the critical velocity of 0.5 m/s (1.5 ft/s) to be exceeded if the fall of the liquid exceeds approximately 10 mm (0.4 in.). This is the critical fall height. This small distance is the reason that castings must be bottom gated at all times. Gating at the joint, if part-way down the casting, or anywhere above the floor of the casting risks the entrainment of defects. A second mechanism for entrainment of bifilms is the contraction of the surface, which happens, for instance, as the surface expands and contracts because of random perturbations (Ref 1). Such disturbances occur continuously in many melting operations. Each time the surface expands, more film area is created as its area stretches. However, each time it contracts, the newly created oxide raft cannot contract and therefore must fold, crumpling concertina-fashion. With continuing disturbances of the surface, the folds lengthen irreversibly in a kind of natural ratcheting mechanism. Eventually, the folds become sufficiently long to be dragged as cracklike defects into the bulk liquid. Turbulent processes such as semilevitation melting, and possibly even conventional induction melting, are likely to introduce large densities of defects. Clearly, the casting industry needs techniques that avoid the introduction of such damage. At the least, pouring transfers must be reduced, and, if necessary, the damage introduced must be reduced or reversed by additional treatment. However, there are now casting systems that eliminate pouring and other turbulence in the whole melting and casting operation, with significant benefits to the properties and reproducibility of products (Ref 5). One of the central features of surface entrainment is that the events are cumulative. For instance, for aluminum alloys, the first turbulence suffered by the material often occurs at the aluminum primary smelter. This first addition of bifilms is supplemented by a second population introduced by the prealloying producer. Several more populations follow in the foundry by melting, pouring into a transfer ladle, and finally pouring into the mold. Naturally, some of the original defects are lost by sedimentation, flotation, or by attachment to the walls of containers. Some may be lost by

364 / Principles of Solidification filtration, although it is known that filters have only a limited ability to trap inclusions (Ref 1). The remelting of foundry returns, and especially machining chips, adds to the total bifilm concentration. However, in this way, aluminum alloys are special. They have a surface oxide film that is marginally more dense than the liquid, but the entrained air in the folds of the bifilm means that many bifilms are near-neutral buoyancy. This feature, combined with the nearly zero Stokes velocity because of the high drag factor provided by its film morphology, means that separation by floating or sinking is especially slow, taking hours or days in conditions where the melt is gently convecting or, in the case of induction melting, strongly stirring. In addition to the problem of the slow separation by buoyancy when suspended freely in a liquid, the situation becomes much worse when the density of defects is such that they form an interconnected latticework throughout the melt. Such a situation is seen in an aluminum alloy in Fig. 2. These frameworks in the liquid are expected to be rather stable with time, since the mutual contacts mean that flotation or sedimentation is now difficult. Most liquid metals and alloys (containing normal populations of defects of interest in this account) pour similar to water because the viscosity of most engineering metals is not significantly different from that of water. Unfortunately, the sight of a melt pouring like porridge, yet at a temperature well above its melting point, is not that uncommon. It is a sign of the presence of an extremely high density of entrained films. (The mechanical properties of such cast material have not been reported in the literature but, in the experience of the author, are disastrously poor.) Some stainless steels, particularly the superduplex stainless types, have especially strong and tough oxide films that create macroscopic problems when pouring; occasionally, bifilms are found to form a linked network completely through cast walls over 100 mm (4 in.) thick. These are first indicated by radiography as areas of apparent shrinkage or sponge porosity. However, when the material is excavated by grinding and tested by dye penetrant, it reveals the appearance of a spider’s web of cracks. If the casting is to be salvaged, such material must be completely dug out and refilled with weld metal. Most other metals are not so disadvantaged. For instance, in the case of carbon/manganese steels, entrained defects are very different in density, thus providing a significant driving force for separation by flotation. However, when generous quantities of aluminum are used for deoxidation, the entrained alumina films still have a high drag because of their aspect ratio, and so separate slowly. If the steel is deoxidized with calcium after the aluminum addition, the surface film of solid alumina is converted into a low-melting-point eutectic of mixed alumina and calcia. The folding-in of this surface film is then simply the wetting

of two liquids that immediately mutually assimilate, forming spherical droplets that float out with maximum speed. As a result, calciumtreated steels are much cleaner than steels simply deoxidized with aluminum (Ref 6). In addition, those residual oxides that do not manage to detrain constitute the favored spherical-shaped inclusions that yield improved workability (Ref 7). Other steels, particularly the common carbon and low-alloy steels, have surface oxides that are partly liquid because of their compositions and the high casting temperatures. An entrainment event now results in the surface film adhering to itself as it is furled up by the internal turbulence in the liquid. The entrained film becomes a compact, sticky ball. This bifilm is glued shut and cannot unfurl again. Having a compact form, it can float out quickly, forming the large drossy or slaglike defects sometimes known as reoxidation defects, which are often centimeters in diameter. These decorate the cope surfaces of large steel castings and must be laboriously dug out and back-filled with weld metal (Ref 8). The sharp difference in behavior between cast aluminum alloys and cast steels is probably highly significant. The rapid detrainment of bifilms in most types of steels explains their generally high levels of cleanliness and their high ductilities, usually between 25 and 50%. Cast aluminum alloys frequently have high levels of bifilms because they are so sluggish to separate, resulting in ductilities often measured in low single figures, or even 0%. Only in the last few years, by using extremely careful handling in a specially designed plant, have higher ductilities in cast aluminum alloys been reproducibly produced in the range of 10 to 20%. The prospects of aluminum alloys matching the ductilities of steels are beginning to seem achievable.

Furling The ravelling (or furling) of the entrained bifilm into a compact mass is almost certainly a standard feature of the history of each bifilm. When a bifilm is entrained, the internal turbulence beneath the surface of the liquid will be strong. Because the bifilms are thinner than tissue paper, they have little rigidity of their own, so they are easily scrambled up by the turbulent eddies in the heavy flowing liquid. As they tumble along channels into a mold, they are expected to continuously ravel and unravel in different ways (Ref 1). In this compacted, convoluted form, a large bifilm may be able to pass through a relatively fine filter, explaining to some extent the partial effectiveness of filters. Convoluted, tangled bifilms are commonly seen in micrographs of cast light alloys. Radiographs of such compacted bifilms are seen as faint, diffuse images in Fig. 2. Also, while the bifilm is furled in this way, it is still a crack but nothing like a familiar stress crack. Its convoluted form means that its

deleterious effect on mechanical properties of the solidified cast metal is minimized, its effective crack length is minimized, and the effect of the crack is reduced because of the mechanical interlocking of its complex form. The bifilm only develops its maximum damaging effect later, as described in the next section.

Unfurling Once the pouring phase of a transfer is finished, or a casting operation is complete so that the melt has arrived in the mold, internal turbulence subsides, and a period of quiescence follows. During this quiet period, the bifilm may experience a number of influences that encourage the bifilm to unravel, opening fold by fold until it forms a substantially planar cracklike defect close to its original size when first entrained. This unfurling of the compacted defect is a critically important stage. This process defines the properties of many cast materials. In the simplest case, if freezing occurs quickly after the filling of the mold, the bifilms are frozenin as convoluted cracks of overall small, compact shape before they have a chance to unfurl, rendering them relatively harmless. Mechanical properties are therefore usually good. Everyone in a light-alloy foundry knows that the properties of chilled castings are better than those of slowly cooled castings. However, if freezing occurs slowly, the bifilm has time to unfurl, expanding to take up its full size as a planar crack and thus regaining its potential to seriously reduce properties. There are a number of reasons why bifilms unfurl, tending to return to their starting size. An understanding of the unfurling mechanisms is key to the understanding of casting properties. The mechanisms are discussed subsequently.

Gas Precipitation from Solution (Formation of Gas Microporosity) It is worth emphasizing that porosity in liquid metals cannot occur as a result of any classical nucleation processes, such as those involving the rearrangement of atoms (alternatively stated, the condensation of vacancies) to create an embryonic bubble. The impossibility of homogeneous nucleation is, of course, widely accepted. What may be less evident is that classical heterogeneous nucleation, although clearly significantly easier, is still not sufficiently easy to make pore nucleation possible in cast products (Ref 1). Thus, nucleation of porosity by any conceivable process is almost certainly impossible. In other words, interfaces in atomic contact are probably impossible to separate in any practical circumstance. This is because, despite any degree of wetting (i.e., assessed macroscopically by such concepts as contact angle), the bonding between atoms on an atomic scale is sufficiently strong to be inseparable in any conceivable practical situation (Ref 9). It is worth reflecting that all solid phases

The Oxide Film Defect—The Bifilm / 365 that have precipitated and grown in the liquid have grown atom by atom and thus are necessarily in perfect atomic contact with the matrix. None of these can be favorable substrates for pore formation or decohesion. Conversely, if an interface is not in atomic contact with the liquid, the separation of this interface would be expected to be an easy initiation site for a pore or a decohesion event. However, such an interface cannot have been grown in the liquid; therefore, it must have been an entrained interface. As such, it will be separated from the liquid by an unbonded oxide film, together with a separating layer of residual entrained gas. Figure 3 illustrates how entrainment always introduces nonbonded interfaces, making such inclusions fundamentally different from those grown in the melt or in the solid. All such unbonded, entrained interfaces will constitute excellent initiation sites for pores. Therefore, bifilms are excellent initiation sites for porosity formation. It can be quickly appreciated that the separation of these unbonded films does not resemble nucleation in the classical sense. No nucleation barrier exists, as in the case of true nucleation phenomena; no critical embryo exists; and no surface tension or surface energy considerations apply, because no liquid surfaces are involved. The presence of the bifilm means that the pore simply and gradually inflates (i.e., the collapsed, folded balloon slowly becomes more like a sphere), as illustrated in Fig. 4. If the continued expansion of the bifilm occurs at a late stage of solidification, its expansion is constrained by the presence of surrounding dendrites, thus forming interdendritic porosity (Fig. 4d). It can be immediately appreciated that the conventional division of round pores being attributed to gas, and interdendritic pores to shrinkage, is not correct. The different morphologies merely result from the round pore opening early and the interdendritic pore opening later during the progress of freezing. One clear example of the variation in bifilm unfurling rates is seen in the fascinating morphology of subsurface porosity in castings. This

is a common condition in which an array of pores lies just 1 or 2 mm (0.04 or 0.08 in.) under the surface of the casting. They are grown by the flux of hydrogen diffusing into the casting from reaction between the metal and its mold environment as a result of the thermal degradation of its organic binder (Fig. 5a). The pores grow in a rather uniform environment of hydrogen flux and are therefore gas pores. The 1 to 2 mm (0.04 to 0.08 in.) depth is almost certainly the result of the increasing concentration of gas ahead of the advancing solidification front, similar to snow gradually piling ahead of the snowplow, until a critical concentration is reached at which the pore can form (Ref 1). Curiously, the pores do not grow to a uniform shape and size. Some display a round morphology like a bubble (Fig. 5b), while others have a dendritic morphology that may be assumed to be a shrinkage pore (Fig. 5c). Clearly, the round pores have nucleated early while still surrounded by liquid, but the dendritic pores have formed late, after the dendrites have formed. These differences in the time of pore initiation cannot be the result of differences in growth conditions, since the pores often lie adjacent to each other, and the hydrogen flux is uniform as a result of the high rate of hydrogen diffusion in aluminum. Therefore, it seems that the different morphologies result from variations in the ease of initiation, with some bifilms opening easily and some only with difficulty. This is in agreement with expectations based on the randomness of turbulent bifilm compaction leading to a random difficulty of unfurling. In general, it seems that gas pores form only from the opening of bifilms or other entrained

Fig. 4 Fig. 3

Inclusions, entrained via the surface, carrying their coating of surface film and remnant of oxide bifilm trail

Unfurling and inflating of bifilms. (a) Simply folded bifilm. (b) Tangled, convoluted bifilm. (c) Overinflation of a bifilm, resulting in a spherical pore. (d) Overinflation of a bifilm later in solidification, resulting in an interdendritic pore

unbonded interfaces. In the absence of suitable substrates such as bifilms, the gas cannot precipitate and thus remains in supersaturated solution. For many years, it has been a well-known fact in casting operations that filtration can

Fig. 5

SEM images of subsurface porosity in an Al-7Si0.4Mg alloy around a sand core bonded with a phenolic urethane binder. (a) general view. Closeups of (b) spherical pore and (c) dendritic pore. Source: Ref 1

366 / Principles of Solidification improve the soundness of castings (Ref 10). (Occasionally this has been interpreted as the ability of a filter to filter out hydrogen!)

Shrinkage due to Solidification Contraction Clearly, shrinkage on solidification, if occurring in an enclosed volume, will reduce the local pressure. The combined atmospheric and metallostatic pressures may be reduced to zero and possibly even further, becoming negative as in a tensile stress. Such a reduction in the hydrostatic pressure will open the bifilm (exactly as in the RPT example described previously). In a trapped volume of liquid, the accompanying shrinkage during solidification will simply expand the bifilms to conserve volume. If the poor feeding condition is not severe, the opening may be so slight as to be overlooked, and the metal will appear perfectly sound but will exhibit reduced mechanical properties. This property loss for no apparent reason is widely experienced. Further opening will lead to the typical scattering of microshrinkage pores and a further reduction in properties. If the shrinkage is severe, macroshrinkage porosity can result, and the strength may be zero. The circumstances are the mirror image of those in which gas drives the unfurling; an increase of pressure on the inside is physically precisely equivalent to a reduction of pressure on the outside of the bifilm. Thus, it is logical that both gas and shrinkage can act in combination; one pushes from the inside, and the other pulls from the outside, leading to the unfurling of the bifilm to make a planar crack and possibly further inflation to make a pore.

straightened between the advancing dendrites, similar to a carpet unrolled across a floor. Such flat cracks are common in many fields of casting and appear to be the cause of hot tears. An example of dendrite straightening is seen on the fracture surface of an Al-4.5Cu alloy in Fig. 6(a). In this image, the large bifilms straightened by dendrites were observed by video radiography to be entrained during the filling of the mold. No metal-to-metal contact could be observed on this fracture surface even under close examination by the SEM, explaining the lamentable elongation to failure of only 0.3%. The whole surface was found to be covered with a transparently thin oxide film (Fig. 6b). The main dendrite rafts in Fig. 6(a) are seen to progress from each side of the casting, pushing the bifilms ahead and finally meeting in the middle. The light, slightly fluorescing regions making the north/south ridge across the diameter of the image were found to be heaps of spare oxide film, pushed ahead of the dendrites. In contrast, in a different part of the same casting that had been observed in the same radiographic study to enjoy a nicely laminar fill profile, the excised test bar produced 3.0% elongation to failure. The fracture surface (Fig. 6c) was poor, exhibiting many shrinkage cavities and implying the presence of a high

population of smaller bifilms. However, there were some metal-to-metal contact areas on this fracture surface, as evidenced by the patches of ductile dimples that were observed (Fig. 6d). Examples of dendrite straightening in other alloy systems have been reported elsewhere (Ref 1). These include spheroidal graphite cast irons, creating the so-called plate fracture effect (Ref 12). The latter is particularly disturbing because so-called ductile iron, marketed especially for its ductility, can unexpectedly fracture in a brittle manner, with the crack appearing to propagate along large planar grain boundaries. In industrial practice, but unfortunately not in the scientific literature, there is now much evidence for this mechanism. Hot tears in castings appear to be universally caused by the presence of bifilms, which invariably can be eliminated by using a well-designed filling system (Ref 4). (See also the article “Filling and Feeding System Concepts” in this Volume.) The use of poorly designed filling systems also explains the wide scatter in results that has long troubled the scientific study of hot tearing phenomena. Little attention has been devoted to the filling system design for cast testpieces for such research. Thus, the random entrainment effects from the turbulent fill have introduced serious quantities of bifilms that have clouded results with “noise.”

Linear Contraction in the Solid State (Hot Tearing) If the cast alloy is stretched, as often happens to parts of a casting during freezing, it is clear that the bifilms can easily be opened. The phenomenon of hot tearing during the solidification of castings has been shown to be eliminated by the removal of bifilms in the liquid. In the absence of bifilms, the casting simply stretches as a result of its low strength and high ductility at elevated temperature. This is achieved by improving the quality of the melt and providing a filling system design for the casting that does not reintroduce bifilms (Ref 1, 4).

Dendrite Pushing The advancing solidifying dendrites cannot grow through a bifilm because of its layer of air. Consequently, the bifilm is pushed by the advancing dendrites. This is why the majority of bifilms appear concentrated in grain boundaries or in interdendritic regions. If a bifilm is overtaken, becoming trapped inside an advancing array of dendrites like a growing dendrite forest, and one end remains attached to the wall of the casting (the forest floor), the bifilm is

Fig. 6

SEM images of the fracture surfaces from a single Al-4.5Cu alloy casting showing (a) the part that suffered a turbulent back-wave and consequent entrained bifilm flattened by dendrite growth and yielding 0.3% elongation. The nearly invisible oxide film covering the whole surface is seen in (b) in a folded area (arrow). (c) View of a more distant point of the same casting, which enjoyed a more tranquil fill, thus exhibiting a fracture surface influenced only by the small but more numerous original bifilms in the rather poor melt. (d) Area of ductile dimples confirming the true metal-to-metal contact in the tensile fracture from this second part of the casting, which yielded 3% elongation. Source: Ref 11

The Oxide Film Defect—The Bifilm / 367 Nucleation and Growth of Intermetallics It seems that bifilms, at least the oxide film variety that has been mainly observed so far, are favored sites for the nucleation and growth of a wide variety of intermetallics in a wide variety of matrices. In fact, the author has found that of the intermetallics precipitating from the liquid thus far studied, each one has formed on an oxide bifilm substrate. Such precipitation occurs on one or both of the wetted outside surfaces of the bifilm that originally grew on the melt surface. The growth of the original film occurred atom by atom, resulting in the external interface of the bifilm (the interface between the melt and the oxide) being in perfect atomic contact (effectively bonded to the melt), in contrast to its dry inner, unbonded surfaces. Cao and Campbell (Ref 11) were the first to report the association between iron-rich intermetallics in aluminum alloys and bifilms as substrates. An excellent example of an intermetallic grown on an oxide substrate is the beta-iron precipitate Al5FeSi seen in Fig. 7. Particles of silicon eutectic in this image also appear to have formed on bifilms, explaining their central cracks. Strontium-rich intermetallics have been observed growing on the underside of surface oxide films (Ref 14), as have carbides in vacuum-cast nickel-base superalloys (Ref 15). The concept of the presence of bifilms in aluminum melts has allowed a new explanation to be put forward for the mechanism of silicon modification. It is based on the finding that the elements sodium and strontium appear to deactivate bifilms as favored substrates for the nucleation and growth of silicon. As a result, the silicon phase no longer forms as large (apparently) brittle plates by easy nucleation and growth on bifilms in suspension in the liquid but is now forced to grow at the lower temperature of the eutectic, with its spacing dictated by diffusion and therefore on a significantly finer scale. The detailed account is outlined elsewhere (Ref 16).

Fig. 7

SEM image of intermetallics (eutectic silicon and b-iron platelets) nucleated and grown on bifilms, causing the bifilms to straighten and create long central cracks. Source: Ref 13

The precipitation of a rigid intermetallic crystal on a rather flimsy film of substrate has interesting consequences. The progressive growth of the crystal can force straightening of the film. However, if the film is thicker or unable to unfurl because its folds are glued by, for instance, low-melting-point patches or perhaps effective spot welded by high-temperature treatment of the liquid (Ref 17), the film sometimes dictates that the crystal grows in a compact and/or irregular form (Ref 16). Thus, the influence of the substrate on the precipitate or vice versa depends on the relative rigidities of the two, and this can be highly variable and mixed, even within one sample. Interestingly, there is evidence that the growing intermetallics can be broken up, fragmenting into smaller sections in a form of the particle refining process (Ref 18). This appears to be the result of the simultaneous precipitation of hydrogen into the bifilm, causing it to inflate and unfold, while the intermetallic is attempting to grow on this mechanically unstable substrate. This refinement mechanism can be easily visualized. The hydrogen is in solution in the liquid metal. Thus, it will attempt to equilibrate with any void, filling the void until the internal pressure in the void leads to re-solution in the melt at the same rate as hydrogen evolves into the void. In this case, the void is the minute air gap between the folded-over films. Hydrogen will diffuse through the films and into the air gap, gradually increasing the pressure of hydrogen in the gap. The gap will widen, inflating the bifilm, first changing from a practically closed crack to an open crack; eventually, if there is enough hydrogen, the bifilm takes on the appearance of a balloon. Now, it has effectively become a hydrogen pore. During the early part of this inflation, the bifilm starts as a convoluted, compact defect. As hydrogen is taken up, it unfurls, opening outward fold by fold (a large, limp balloon does exactly this). Any crystal attempting to grow on the outer surfaces of the bifilm will find the ground moving beneath its feet. The crystal is likely to fragment, appearing to grow as many separate small crystals instead of a single large crystal. Southin and Romeyn (Ref 19) suggest that an analogous process is happening in the case of dendrite solidification on a smoke-deposited, and therefore fragile, carbon surface. (This effect is frequently used by aluminum casters to increase the apparent fluidity of the melt, because the normal hindrance to flow provided by the dendrite array growing from the surface of the casting is prevented; the dendrites formed on the weak substrate continuously break away from the walls of the mold, leaving the channel clear for continued flow of the liquid.) Sometimes, the central bifilm is not seen in the intermetallic particle, which has caused some investigators to assume it is not present. However, conditions during freezing are not always conducive to opening the bifilm to render it visible. These conditions are easily

arranged, but most casting practices naturally go to some lengths to avoid such conditions by using well-degassed metal, low iron contents (for aluminum alloys), and a good feeding system, keeping the melt pressurized during freezing to maintain tightly closed bifilms. Summarizing the unfurling mechanisms, it is interesting to note that they are already generally well known to casting metallurgists as deleterious to ductility. These deleterious factors are otherwise known as gas porosity, shrinkage porosity, tensile prestrain, large grain size, and (for aluminum alloys) high iron content.

Reliability and Reproducibility of Mechanical Properties The simple fact that castings can vary erratically in their properties has never been satisfactorily explained, particularly when all steps to eliminate variation have apparently been taken. The variation from casting to casting, day to day, and week to week in a closely monitored production plant has been nothing short of baffling. The presence or absence of bifilms in the starting material and the random variations in the pouring of the castings are features that have remained unsuspected. One of the first clear demonstrations of property variation due to pouring was the experiment carried out by Green and Campbell (Ref 20) in which the manner of pouring was varied. The turbulence of the different systems clearly had a major impact on the reliability of the casting. In this work, the use of Weibull statistics was introduced to quantify the reliability. Different reliabilities were found for essentially identical alloys poured at identical temperatures, the only difference being the degree of turbulence during the final transfer of metal and the final content of bifilms. (It is noted that other damaging features that would have influenced the results, such as porosity, the precipitation of beta-iron phase, shrinkage, and so on, are consequences of the presence of bifilms and thus are secondary influences.) Interestingly, Green and Campbell’s approximation of the Weibull distribution in two-parameter form has recently been challenged (Ref 21). It seems that the full three-parameter form of the Weibull analysis is more valid and yields additional key information about the potential safety implications of the scatter. Green and colleagues (Ref 22) went on to show that the fatigue properties of Al-7Si0.4Mg alloy were controlled by its bifilm content. In fact, in 98% of all fatigue fractures closely observed by SEM in this work, bifilms were the initiators of cracks. The remaining 2% exhibited up to 100 times greater fatigue lives. These were originally thought to fail from fractures initiated by slip planes. However, on a recent re-examination of these results, it seems that the slip planes are more likely to have been planar fractures resulting from the straightening of oxide bifilms by dendrite pushing. If this new

368 / Principles of Solidification interpretation is correct, it is not clear at this time how far fatigue life can be extended if bifilms are eliminated. The 100-times benefit seems an underestimate. There has been much discussion to claim that pores are more important than bifilms in initiating failure, particularly fatigue failure (Ref 23, 24). In a sense, the differentiation is, in practice, almost irrelevant since both are entrainment defects, the pore being simply a rather more open type of bifilm. In addition, the pore definitely has zero cross-sectional tensile strength. In contrast, the bifilm has mixed strength, on average perhaps 50% of the strength of the matrix (Ref 1). This is the result of mechanical interlocking of folds and the interference of “jogs” in the film when subjected to shear, as well as variations in alignment. A planar bifilm aligned with the tensile direction will cause no observable reduction in strength, whereas when perpendicular to the tensile axis, it could behave as having zero strength. There are genuine reasons why pores may be more important defects than bifilms. However, the films are usually vastly greater in area (for instance, in Fig. 1, the pores are seen merely to “decorate” the bifilm), have a sharp aspect ratio, and are so thin that they are routinely overlooked by observers. When an observer claims to have identified a pore as the origin of a fatigue crack, he has probably failed to notice that the pore is almost certainly a part of a much larger-area bifilm, whose presence is difficult to detect because of its thinness and consequent transparency. Planar fracture regions are often observed in association with pores. These are often interpreted as cleavage surfaces, grain-boundary surfaces, or slip planes. Close examination often reveals that these features are dendrite-straightened oxide films. Careful experiments on a nickel-base superalloy, vacuum cast and subjected to hot isostatic pressing (HIP), demonstrated that these features only occurred in those castings that were poured turbulently (Ref 25). Open or partially open bifilms have practically zero strength. The application of pressure to the casting can therefore produce significant benefit by closing bifilms, even if the stability of the oxide interfaces resists bonding. Thus, pressure applied immediately after casting, prior to solidification, has been repeatedly demonstrated over the years to be beneficial to mechanical properties (Ref 26). Similarly, in the solid state, the mechanism of HIP can be easily understood to increase mechanical properties by closing bifilms, even if the defects are not closed by bonding (i.e., welded) (Ref 27).

Central Role of Bifilm Defects in Fracture Brittle failure appears to be accurately described using some variant of the Griffiths

crack concept, in which the elastic energy built up in the material exceeds the energy to propagate a pre-existing crack. Where there is no preexisting defect, however, fracture is probably not possible. The arguments for this are not presented here; see the review article in Ref 26. Essentially, the arguments emphasize the fact that metals have theoretical strengths in the region of 10 to 20 GPa (1450 to 2900 ksi), which is so high that such bonds cannot normally be broken. As correctly conceptualized by the Griffiths crack model, it follows that only a solid or liquid phase containing a defect can fracture. For instance, an inclusion or second-phase particle precipitated in a solid matrix that has nucleated and grown by diffusion of solutes will effectively be in perfect atomic contact with the matrix and therefore incapable of creating a crack. Either the inclusion, the matrix, or both will yield plastically, but a cavity or crack cannot open. Interestingly, silicon particles in aluminum-silicon alloys would not be expected to fracture during tensile testing. However, it is widely known that solids often appear to fail by decoherence from inclusions, second-phase particles, or grain boundaries. There are many examples, including the opening of grain boundaries in creep in such materials as nickel-base superalloys and superplastic forming in aluminum-zinc alloys. From these arguments, it follows that the metallurgical features, such as the grain boundaries alone, cannot be the origin of the failures. Such failures most likely originate from bifilms lying in the grain boundaries, where they will often occur as the result of dendrite pushing or grain growth that is expected to be arrested at a bifilm. As usual, the bifilms will be invisible to casual inspection, so that the incorrect identification of the cause of failure is common. In summary, failure is not expected to occur from any inclusion or other feature that has grown within the metal, whether in the liquid or solid phase, because its boundaries are expected to be in more-or-less perfect atomic contact with the matrix, giving such interfaces great tensile strength. Conversely, failure is expected from features entrained in those liquid metals that have a solid surface film. These features are necessarily associated within the “paper bag” of oxide film dragged into the matrix from the liquid surface (Fig. 3), creating an unbonded interface around the entrained particle. The practical outcome of the quest for highquality metal, low in bifilm content, has been the development of exciting new filling system designs for castings, with great benefits to integrity, soundness, enhanced properties, and the simultaneous reduction of costs on the industrial scale (Ref 5). The improvement in reliability and repeatability has been most significant. These benefits are highly prized targets in the marketplace and represent the new vision for future generations of casting production.

REFERENCES 1. J. Campbell, Castings, 2nd ed., Butterworth Heinemann, Oxford, 2003 2. S. Fox and J. Campbell, Scr. Mater., Vol 43 (No. 10), 2000, p 881–886; Int. J. Cast Met. Res., Vol 14 (No. 6), 2002, p 335–340 3. D. Dispinar and J. Campbell, Int. J. Cast Met. Res., Vol 17 (No. 5), 2004, p 280–294 4. J. Campbell, Casting Practice, Butterworth Heinemann, Oxford, 2004 5. J. Campbell, Eng., Vol 221 (No. 3), 1981, p 185–188 6. J. Finardi, Trans. AFS, Vol 108, 2000, p 409–417 7. R. Kiessling, Non-Metallic Inclusions in Steel, The Metals Society, London, 1978, p 42–86 8. J.M. Svoboda, R.W. Monroe, C.E. Bates, and J. Griffin, Trans. AFS, Vol 95, 1987, p 187–202 9. J. Campbell, Mater. Sci. Technol., Vol 22, 2006, p 127–145, 999–1008 10. K.J. Brondyke and P.D. Hess, Trans. Metall. Soc. AIME, Vol 230 (No. 7), 1964, p 1542–1546 11. X. Cao and J. Campbell, Metall. Mater. Trans. A, Vol 34, 2003, p 1409–1420 12. S.I. Karsay, Ductile Iron Production: The State of the Art 1992, QIT Fer et Titane Inc., Canada, 1993 and 2000, p 196–198 13. J. Mi, R.A. Harding, and J. Campbell, Int. J. Cast Met. Res., Vol 14 (No. 6), 2002, p 325–334 14. S.M. Miresmaeili, J. Campbell, S.G. Shabestari, and S.M.A. Boutorabi, Metall. Mater. Trans. A, Vol 36 (No. 9), 2005, p 2341–2349 15. A.K.M.B. Rashid and J. Campbell, Metall. Mater. Trans. A, Vol 35, 2004, p 2063– 2071 16. J. Campbell, Shape Casting: The John Campbell Symposium (San Francisco, CA), M. Tiryakioglu and P.N. Creapeau, Ed., TMS, 2005, p 3–12 17. R. Ahmad and R.I. Marshal, Int. J. Cast Met. Res., Vol 15 (No. 5), 2003, p 497–504 18. H. Sina, M. Emamy, M. Saremi, A. Keyvani, M. Mahta, and J. Campbell, The Influence of Ti and Zr on Electrochemical Properties of Al Sacrificial Anodes, Mater. Sci. Eng. A, Vol 431, 2006, p 263–276 19. R.T. Southin and A. Romeyn, 1980 Warwick Conference, Solidification Technology in the Foundry and Cast House, Metals Society, 1983, p 355–358 20. N. Green and J. Campbell, Trans. AFS, Vol 102, 1994, p 341–347 21. H. Zahedi, M. Emamy, A. Razaghian, M. Mahta, J. Campbell, and M. Tiryakiog˘lu, Metall. Mater. Trans. A, Vol 38, March 2007, p 659–670 22. C. Nyahumwa, N.R. Green, and J. Campbell, Trans. AFS, Vol 106, 1998, p 215– 223; J. Mech. Behavior Mater., Vol 9 (No. 4), 1998, p 227–235

The Oxide Film Defect—The Bifilm / 369 23. Q.G. Wang, P.N. Crepeau, D. Gloria, and S. Valtierra, Advances in Aluminum Casting Technology, M. Tiryakioglu and J. Campbell, Ed., ASM Symposium, 2002, p 209–218 24. M.J. Couper, A.E. Neeson, and J.R. Griffiths, Int. J. Fatigue Fract. Eng. Mater. Struct., Vol 13, 1990, p 213–227

25. M. Cox, personal communication based on M. Cox, M. Wickins, J.P. Kuang, R.A. Harding, and J. Campbell, Mater. Sci. Technol., Vol 16, 2000, p 1445–1452 26. J.T. Berry, R. Luck, and R.P. Taylor, Shape Casting: The John Campbell Symposium (San Francisco, CA), M. Tiryakioglu and P.N. Creapeau, Ed., TMS, 2005, p 113–122

27. J.T. Staley, Jr., M. Tiryakiog˘lu, and J. Campbell, Shape Casting: Second International Symposium, P.N. Crepeau, M. Tiryakiog˘lu, and J. Campbell, Ed., TMS, 2007

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 370-374 DOI: 10.1361/asmhba0005222

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Shrinkage Porosity and Gas Porosity Qingyou Han, Purdue University

IN METAL CASTINGS, two types of porosity are identified: shrinkage porosity and gas porosity. Shrinkage porosity occurs when the shrinkage on solidification cannot be compensated (Ref 1). Gas porosity occurs when the gas concentration in the liquid is higher than its solubility (Ref 2). The causes of porosity formation for these two types of porosity are inadequate feeding and gas evolution. Under some conditions, one pore formation mechanism is dominant so that one type of porosity occurs, but in most situations, both of the causes play important roles in porosity formation. Inadequate feeding causes a local pressure drop so that the gas becomes more supersaturated, resulting in earlier formation of the gas porosity. During the solidification of local,

isolated liquid pools, shrinkage generates pores filled with gases. It is the interplay of these two mechanisms that gives rise to porosity in a casting. Figure 1 illustrates gas porosity and shrinkage porosity in an aluminum AA 5182 remelt secondary ingot. Gas porosity is usually spherical, and the shrinkage porosity is irregularly shaped. This distinction is based on the assumption that the irregular shape is due to the interdendritic fluid being drawn away from a partially formed microstructure to feed the volume contraction elsewhere.

Shrinkage Porosity Shrinkage. The formation of shrinkage porosity is strongly affected by the shrinkage behavior of the alloy and the feeding conditions for the casting. In general, liquid contracts on solidification because of the random arrangement of atoms in the liquid and the orderly arrangement of atoms (lattice structure) in the

solid state. Table 1 (Ref 3) lists the solidification shrinkage of some pure metals. The most dense solids are those that have cubic closepacked (face-centered cubic and hexagonal close-packed) symmetry. Contraction on solidification for these metals is in the range of 3.2 to 7.2%. The solidification shrinkage for the less closely packed body-centered cubic lattice is in the range of 2 to 3.2%. The exceptions to this general pattern are those materials that expand on freezing. These include cerium, silicon, and bismuth. No shrinkage porosity has been observed in these metals, provided the liquid contraction can be compensated by using risers. The shrinkage behaviors of pure metals and alloys are quite different. For pure metals, solidification occurs at the melting temperature (the liquidus temperature equals the solidus temperature). The freezing front is usually planar. During solidification, the freezing front progresses from the casting surface toward the center, forming centerline porosity. By using proper thermal control and risers, most of the solidification shrinkage of a pure metal can be collected in the risers as a shrinkage cavity.

Table 1 Solidification shrinkage for some pure metals Melting point Metal

Al Au Co Cu Ni Pb Fe Li Na K Rb Cs Tl Cd Mg Zn Ce In Sn Bi Sb Si

Fig. 1

(a) Gas porosity and (b) shrinkage porosity in an AA 5182 remelt secondary ingot

Crystal structure(a)

fcc fcc fcc fcc fcc fcc bcc bcc bcc bcc bcc bcc bcc hcp hcp hcp hcp fct Tetragonal Rhombohedral Rhombohedral Diamond




C

660 1,063 1,495 1,083 1,453 327 1,536 181 97 64 39 29 303 321 651 420 787 156 232 271 631 1,410

F

Liquid density, kg/m3

Solid density, kg/m3

1,220 1,945 2,723 1,981 2,647 621 2,797 358 207 147 102 84 577 610 1,204 788 1,449 313 450 520 1,168 2,570

2,368 17,380 7,750 7,938 7,790 10,665 7,035 528 927 827 1,437 1,854 11,200 7,998 1,590 6,577 6,668 7,017 6,986 10,034 6,493 2,525

2,550 18,280 8,180 8,382 8,210 10,020 7,265 ... ... ... ... ... ... ... 1,655 ... 6,646 ... 7,166 9,701 6,535 ...




Volume change, %

(a) fcc, face-centered cubic; bcc, body-centered cubic; hcp, hexagonal close-packed, fct, face-centered tetragonal. Source: Ref 3

7.14 5.47 5.26 5.30 5.11 3.22 3.16 2.74 2.6 2.54 2.3 2.6 2.2 4.0 4.1 4.08 0.33 1.98 2.51 3.32 0.64 2.9

Shrinkage Porosity and Gas Porosity / 371 For an alloy, solidification shrinkage occurs throughout its freezing temperature range. The shrinkage amount varies with the temperature, depending on the fraction of solids precipitated at given temperatures and the nature of the phases formed. Figure 2 illustrates the temperature versus fraction solid curves of three ironcarbon alloys: Fe-0.8wt%C, Fe-3wt%C, and Fe-4wt%C. The Fe-0.8wt%C alloy represents an alloy that solidifies into a single phase. The temperature versus solid fraction curve of this alloy is fairly straight, indicating the shrinkage is fairly constant throughout solidification. Most wrought alloys have temperature versus solid fraction curves similar in shape to that of the Fe-0.8wt%C alloy, except for a small amount of eutectic phase formation at the late stage of solidification. For the Fe-3wt%C alloy, 58% of the solid forms as primary austenite dendrites in the temperature range between 1150 and 1300
C

(2100 and 2370
F), and 42% of the solids form at 1150
C (2100
F) as the eutectic phases. The resultant solidification shrinkage of this alloy is highly nonuniform. During precipitation of the austenite dendrites, the shrinkage behavior of the alloy is identical to that of the Fe-0.8wt%C alloy. However, a large amount of shrinkage occurs at the eutectic temperature because of the precipitation of the eutectic phases. Most of the casting alloys, such as aluminum-copper, aluminum-silicon, aluminummagnesium, copper-base alloys, zinc alloys, and magnesium alloys, have behavior identical to Fe-3wt%C, except that the amount of eutectic phases and the eutectic temperature ranges may vary. In fact, the shrinkage behavior of the alloys containing eutectic phases is strongly affected by the types of eutectic phase formed. In the iron-carbon alloys, the formation of austenite and carbide (Fe3C) phases leads to a volume contraction, but the formation of

Fig. 2

Temperature versus fraction solid curves of three iron-carbon alloys: Fe-0.8wt%C, Fe-3wt%C, and Fe-4wt%C. These curves are obtained using a lever rule.

Fig. 3

Schematic representation of the feeding mechanisms in a solidifying casting. Source: Ref 3

graphite phase leads to a volume expansion. As a result, gray cast irons and ductile irons with composition close to or higher than the eutectic composition usually expand during solidification. This is because the volume fraction of the graphite phase is large enough that volume expansion due to graphite formation is larger than the volume contraction due to the formation of the other phases. A temperature versus fraction solid curve for Fe-4wt%C is also shown in Fig. 2. The eutectic fraction can be as high as 0.84, and the alloy expands during eutectic solidification. Most of the alloys shrink during solidification, so porosity usually forms in the castings of these alloys if the solidification shrinkage is not fed. When alloys expand during solidification, shrinkage porosity is unlikely to occur except for ductile irons. For ductile irons, a graphite nodule is usually surrounded by a shell of eutectic austenite during eutectic solidification. As a result, graphite expansion works against the austenite shell and pushes the austenite dendrite apart, leading to the formation of porosity among eutectic cells and dendrites (Ref 4). Feeding. Risers, or appendages, are added to a casting in order to feed the shrinkage in a casting. Solidification takes place simultaneously in both the casting and risers, and liquid flows from the risers to feed the casting. The flow is driven by solidification shrinkage, contraction of the liquid and the solid as they cool, and gravity. The possible feeding mechanisms are illustrated in Fig. 3 (Ref 3). Liquid feeding precedes other forms of feeding and is normally the only method of feeding for skinfreezing materials such as pure metals or alloys with a narrow freezing range. The liquid has low viscosity, and for most of the freezing process, the feeding pass is wide, so the pressure difference required for liquid feeding is negligible. Mass feeding involves the movement of a slurry of solid grains and residual liquid. It usually occurs in an alloy with a long freezing range when the fraction solid is smaller than the dendrite coherence point. Similar to that of liquid feeding, the pressure difference required for mass feeding is also small since the feeding pass is wide. Liquid feeding and mass feeding are the main feeding mechanisms for an alloy with a large fraction of eutectic phases at the end of solidification, such as the two high-carbon alloys shown in Fig. 2. During interdendritic feeding, liquid flows through tiny tortuous channels between dendrites to feed the shrinkage at the far end of the mushy zone. Inadequate feeding normally occurs in the interdendritic feeding region where the dendritic network exhibits a large resistance to fluid flow. Especially when the fraction liquid is so small that the remaining liquid in the mushy zone may be isolated by the solids, the solidification shrinkage of this liquid will inevitably produce pores, since feeding becomes impossible. Experimental

372 / Principles of Solidification measurements indicated that feeding stops before the end of solidification at a solid fraction of 0.9 to 0.99, equivalent to a liquid fraction of 0.01 to 0.1 (Ref 5). Taking a volumetric shrinkage of 6.7% for Al-8wt%Cu, for instance, the solidification of a fraction liquid of 0.1 will give 0.67% of porosity if all the pores are formed after feeding stops. Interdendritic feeding is important for shrinkage porosity formation in an alloy such as Fe0.8wt%C, shown in Fig. 2, that has a long freezing range. For such an alloy, the mushy zone is large and the fraction liquid at the point of final solidification can be extremely small. As a result, most long-freezing-range alloys exhibit poor pressure tightness due to porosity formation. Interdendritic feeding is also affected by the shrinkage behavior of the alloy. A large amount of shrinkage associated with eutectic formation at the end of solidification makes interdendritic feeding more difficult, especially when the fraction of the eutectic liquid is small. However, as the fraction of eutectic liquid increases, the feeding path becomes wider and feeding becomes easier.

Fig. 4

Hydrogen solubility in pure aluminum as a function of temperature at 101 kpa (1 atm) pressure. Source: Ref 6

Gas Porosity Gas porosity is generally caused by the evolution of gases during the casting and solidification processes. Gas porosity occurs when the gas concentration in the molten metal is higher than its solubility. The gases in molten metal may be the result of a reaction between the metal and the gases in atmosphere or in the solid materials that contain the molten metal. The type of gases that are responsible for porosity formation during casting varies from metal to metal.

Gases in Metal Aluminum. The only gas that has a measurable solubility in liquid aluminum alloys is hydrogen. Moisture from the refractory lining or atmosphere can react with molten aluminum: 3H2 O þ 2½Al ¼ Al2 O3 þ 3H2

(Eq 1)

The hydrogen that is released in this reaction can dissolve in liquid aluminum as atomic species according to: H2 ¼ 2½H

(Ref 6). The lines of best fit to the solubility in liquid, SL, and solid, SS, aluminum are: 2760 þ 2:796 and T 2080 log SS ¼  þ 0:788 T

log SL ¼ 

(Eq 4)

The unit for solubility in the previous equations is mL/100 g, and temperature is given in degrees Kelvin. Alloying elements affect the solubility of hydrogen (Ref 7). Generally, the common alloying elements decrease the solubility of hydrogen. Copper and Tin. The solubility of hydrogen in molten copper, tin, and copper-tin alloys is shown in Fig. 5 (Ref 8). In pure copper, hydrogen solubility is greater in the liquid than in the solid state (approximately 5.5 mL/100 g versus 2 mL/100 g), which is why gas porosity is a problem in copper castings. In addition to hydrogen, oxygen is also soluble. Reaction between hydrogen and oxygen produces water vapor according to: 2½H þ ½O ¼ H2 O

(Eq 5)

(Eq 2)

The gas pressure of water vapor is given by: The solubility of hydrogen in liquid aluminum is related to the partial pressure of hydrogen, PH2 , using Sievert’s law: ½H2 ¼ kPH2

(Eq 3)

where the constant k is the activity of hydrogen in the metal at that temperature and at 101 kpa (1 atm) pressure. Figure 4 shows the hydrogen solubility in pure aluminum as a function of temperature at 101 kpa (1 atm) pressure

½H2 ½O ¼ kH2 O PH2 O

(Eq 6)

High levels of dissolved oxygen lead to low levels of dissolved hydrogen, and vice versa. It is beneficial to melt copper in an oxidizing atmosphere and then deoxidize to minimize porosity formation in the casting. Other complex gases in copper-base alloys include CO in nickel-bearing copper alloys or SO2 if high-sulfur fuel is used for melting.

Fig. 5

Solubility of hydrogen at atmospheric pressure in pure copper, pure tin, and copper-tin alloys. Source: Ref 8

Iron and Steel. In irons and steels, gases that are responsible for porosity formation in casting include CO, H2, and N2. Hydrogen and nitrogen are soluble in molten steels. Oxygen is also soluble and reacts with carbon to form carbon monoxide, CO, by the reaction in steel as: ½C þ ½O ¼ CO

(Eq 7)

In steelmaking practice, the CO reaction is used to reduce the high carbon levels in pig iron produced by blast furnaces. The CO reaction is so vigorous and violent that it is called a carbon boil. After the carbon content is reduced into the specified range, the excess oxygen in molten metal is reduced by using a deoxidizing addition. The vigorous carbon boil will reduce both hydrogen and nitrogen in solution to negligible levels. In many steel foundries where steel is simply remelted, dissolved hydrogen and nitrogen will contribute to porosity formation in castings.

Gas Evolution during Solidification As illustrated in Fig. 4, hydrogen solubility in pure aluminum is greater in the liquid than in the solid state (approximately 0.65 mL/100 g versus 0.034 mL/100 g). The solubility falls by an order of magnitude upon solidification. Generally, there is a large solubility drop of the dissolved gases during phase transformation from liquid to solid. On solidification, the dissolved gases are rejected by the growing solid and are enriched in the remaining liquid in the mushy zone. Most of the solute elements in an alloy are also enriched in the remaining liquid, thus decreasing the solubility of the dissolved

Shrinkage Porosity and Gas Porosity / 373 gases in the mushy zone. When the gas contents are higher than their solubility, gas porosity forms. Based on the fact that the diffusion coefficient of the dissolved gas, such as hydrogen, is a few orders of magnitude higher than that of the other solute elements in the alloys, the lever rule can be used for calculating the local gas concentration in the liquid in the early stage of solidification, according to: CS fS þ CL fL ¼ C0

(Eq 8)

where CS and CL are the concentrations of a dissolved gas in the solid and in the liquid, respectively, C0 is the initial gas concentration in the molten alloy before solidification starts, and fS and fL are the local solid fraction and liquid fraction in the mushy zone, respectively. In Eq 8, CS is less than or equal to the solubility of dissolved gas in the solid. Thus, the concentration of the dissolved gas in the remaining liquid can be many times higher than the initial gas concentration in the molten metal at the end of solidification. In fact, the use of the lever rule for the prediction of dissolved gas concentration in the remaining liquid usually underestimates the dissolved gas concentration at the point of final solidification, such as the eutectic front. Because of their limited diffusion in the liquid, the dissolved gases are highly enriched at the eutectic front, and their concentrations can be a few orders of magnitude higher than that predicted using the lever rule (Ref 9). This accounts for the observation that gas porosity occurs in alloys with initial gas concentrations lower than the solubility of the gas in the liquid.

Porosity Formation Homogeneous Nucleation. Similar to the nucleation of any phase, the free energy, DG, for the formation of a spherical vapor bubble of radius r is (Ref 10): 4 G ¼ 4pr2  þ pr3 ðP  PV Þ 3

(Eq 9)

where P is the hydrostatic pressure in the liquid, and PV is the vapor pressure, which is negligible in comparison with P since a very large negative pressure is required. The surface tension at the gas-liquid interface is represented by s. The maximum free energy, DG*, is calculated to be: G ¼

16ps3 3ðP  PV Þ2

(Eq 10)

for pores with a critical radius, r*: r ¼ 

2s P  PV

(Eq 11)

Hence, there is a critical radius for the bubble, below which it is not capable of surviving and above which it will tend to grow.

For the nucleation of a soluble gas in a liquid, the pressure required to nucleate a bubble is generally given by: Pg ¼ Pm þ

2s r

(Eq 12)

where Pg is the partial pressure of dissolved gas, and Pm is the local pressure. Assuming local equilibrium, the partial pressure of the dissolved gas, Pg, is related to the mass of the gas, n, in the pore by: Pg ¼ nRT

(Eq 13)

where R is the gas constant, and T is the temperature in the pore. To form a pore of radius r, n moles of gas are required. The pressure, Pg, needed to create a pore in metal is calculated to be on the order of 30,000 atm (3  109 Pa) (Ref 11). This means that homogeneous nucleation of pores is extremely difficult and is unlikely to occur in practice. Heterogeneous Nucleation. In castings, nucleation of pores can be expected to occur primarily at heterogeneous nucleation sites. The classical theory of homogeneous nucleation is extended to heterogeneous nucleation. The maximum free energy is: 16 3 2 ps P SðyÞ 3

(Eq 14)

SðyÞ ¼ ð2 þ cos yÞð1  cos yÞ2 =4

(Eq 15)

G ¼

where:

and y is the equilibrium contact angle of the bubble on the solid. The pressure required to nucleate a bubble is given by: Pg ¼ Pm þ

2s pffiffiffiffiffiffiffiffiffiffi SðyÞ r

(Eq 16)

When the contact angle is zero, the free energy for the formation of a pore will be zero so the pore can grow directly on the heterogeneous site. When the contact angle is 180
, Eq 14 and 16 reduce to Eq 10 and 12, which means the nucleation sites are ineffective. Heterogeneous sites are inherent in almost all normal castings, such as the solid-liquid interface and inclusions. Calculations indicate (Ref 12, 13) that the nucleation of a bubble on the planar interface is difficult. The cellular or dendritic interface provides a much more favorable nucleation site, partly because the interface is concave and partly because the soluble gases are concentrated in the interdendritic region. Inclusions are probably the most important sites for heterogeneous nucleation. It has been well demonstrated that the presence of inclusions greatly enhances the porosity formation in aluminum alloys. The easiest sites for bubbles to nucleate are the pre-existing gas bubbles entrapped in the cavities of the inclusions (Ref 3). No nucleation event is required, only growth. Such a site can act as a multiple

nucleation site if the bubbles detach, leaving a small bubble behind in the cavity. Equations 12 and 16 indicate that both Pm and Pg provide the driving forces for the nucleation of a pore. Porosity forms when: Pg > Pm þ 2s=r

(Eq 17)

It is the interplay of Pm and Pg that determines the formation of porosity in a casting. In the previous equation, Pg is a measure of the amount of dissolved gases in the liquid as described by Eq 3. If the concentration of a dissolved gas is higher than its solubility at Pm, which is the local pressure, a pore is likely to form. On the other hand, Pm is a measure of the ease of feeding the shrinkage locally under given external and hydraulic pressure. Difficulty in feeding occurs at locations where the local Pm is extremely low or negative. Under such a condition, the dissolved gases can escape from the liquid and form porosity to feed the solidification shrinkage. Neglecting the hydraulic pressure, the local pressure is given by the external pressure minus a pressure drop defined using Darcy’s law: V ¼

K @P mfL @x

(Eq 18)

where V is the velocity of the shrinkage-driven flow, @P/@x is the pressure gradient, m is the viscosity of the liquid, and K is the permeability of the mushy zone. For a mushy zone containing equiaxed grains, K follows the BlackKozenny equation (Ref 14): K¼

fL3 kC SV2

(Eq 19)

where kC is a constant, and SV is the solid-liquid interfacial area per unit volume. The local pressure, Pm, decreases with decreasing external pressure and increasing pressure drop or pressure gradient. The pressure drop increases with increasing velocity of the shrinkage-driven flow, solid-liquid interfacial area, and decreasing liquid fraction. Thus, it is extremely difficult to feed regions with a small liquid fraction that have a large solidification shrinkage.

Factors Affecting Porosity Formation Factors that affect porosity formation include the following: Dissolved Gases in the Melt. The higher the dissolved gas concentration in a molten alloy, the higher the Pg value. As a result, porosity will be more likely to occur. When porosity forms, the mass of the gas, f, in pores per 1 g of alloy is calculated using (Ref 15): 100f ¼ ðCL  SL ÞfL þ ðCS  SS ÞfS

(Eq 20)

374 / Principles of Solidification Thus, the amount of porosity increases with increasing concentrations of the dissolved gases, CL, in a molten alloy. Degassing is effective in minimizing porosity formation. Alloy Composition. Apart from affecting the initial concentrations of the dissolved gases in the melt, alloy composition also determines the solidification interval and the shrinkage characteristics of the alloy. Usually, porosity level increases with increasing solidification interval, reaches a peak value at the composition when the nonequilibrium eutectic phases start to occur, and decreases as the eutectic fraction increases. This phenomenon is associated with a decreased permeability due to an increased SV and a decreased fL as the solidification interval increases. When a small fraction of eutectic phase forms, a large demand for feeding occurs, resulting in a large shrinkage-driven flow and a large pressure drop. As the eutectic fraction is increased, the feeding channel becomes open, and feeding becomes easier. External pressure can be used to increase the local pressure, Pm, in the mushy zone and to minimize porosity formation. Squeeze casting is a casting process using external pressure during the solidification stage of a casting to achieve porosity-free components. Cooling Conditions. Fast cooling is usually beneficial in reducing porosity level if risers and chills are properly used for feeding the liquid shrinkage. Increasing the cooling rate can

increase the temperature gradient at the freezing front, making the feeding channel open and wide. Furthermore, the radius of a pore is usually proportional to the secondary dendrite arm spacing, which decreases with increasing cooling rates. A pore with smaller radius will have higher internal pressure (Eq 12) and contain more gas molecules for a given volume (Eq 13). As a result, the volume fraction of pores decreases with increasing cooling rate.

REFERENCES 1. T.S. Powonka and M.C. Flemings, Pore Formation in Solidification, Trans. Metall. Soc. AIME, Vol 236, 1966, p 1157–1165 2. E.J. Whittenberger and F.N. Rhines, Origin of Porosity in Castings of MagnesiumAluminum and Other Alloys, J. Met., 1952, p 409–420 3. J. Campbell, Castings, Butterworth-Heinemann, 1991 4. M. Burditt, Ductile Iron Handbook, American Foundrymen’s Society, 1992 5. V.L. Davies, Feeding Range Determined by Numerically Computed Heat Distribution, AFS Cast Met. Res. J., Vol 11, June 1975, p 33–44 6. C.E. Ransley and H. Neufeld, The Solubility of Hydrogen in Liquid and Solid Aluminum, J. Inst. Met., Vol 74, 1948, p 599–620

7. G.K. Sigworth and T.A. Engh, Chemical and Kinetic Factors Related to Hydrogen Removal from Aluminum, Metall. Trans. B, Vol 13, 1982, p 447–460 8. M.T. Roeley, Casting Copper-Based Alloys, American Foundrymen’s Society, 1984 9. Q. Han and S. Viswanathan, Hydrogen Evolution during Directional Solidification and Its Effect on Porosity Formation in Aluminum Alloys, Metall. Mater. Trans. A, Vol 33, 2002, p 2067–2072 10. J.C. Fisher, The Fracture of Liquids, Appl. Phys., Vol 19, 1948, p 1062–1066 11. R. Cole, Boiling Nucleation, Adv. Heat Trans., Vol 10, 1974, p 85–166 12. W.R. Wilcox and V.H.S. Kuo, Gas Bubble Nucleation during Crystallization, J. Cryst. Growth, Vol 19, 1973, p 221–228 13. P.M. Wilt, Nucleation Rates and Bubble Stability in Water-Carbondioxide Solutions, J. Colloid Interface Sci., Vol 112 (No. 2), 1986, p 530–538 14. A.J. Duncan, Q. Han, and S. Viswanathan, Measurement of Liquid Permeability in the Mushy Zones of Aluminum-Copper Alloys, Metall. Mater. Trans. B, Vol 30, 1999, p 745–750 15. K. Kubo and D.R. Pehlke, Mathematical Modeling of Porosity Formation in Solidification, Metall. Trans. B, Vol 16, 1986, p 359–366

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 375-378 DOI: 10.1361/asmhba0005223

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Castability—Fluidity and Hot Tearing Lars Arnberg, Norwegian University of Science and Technology Asbjrn Mo, SINTEF Materials and Chemistry

CASTABILITY of alloys is a measure of their ability to be cast to a given shape with a given process without the formation of casting defects. These defects can be cracks, pores, segregations, inclusions, or other internal or external faults that impair the properties and/or appearance of the casting. This article examines two phenomena closely related to castability: fluidity and hot tearing.

Fluidity Fluidity is usually defined as the maximum fluidity length, which is the distance a melt will flow in a standard mold with a constant cross section before the flow is terminated by solidification. It should be noted that this is different from a physicist’s definition of fluidity, which is the inverse of viscosity. Fluidity limits the geometry of a casting that can be successfully filled, and insufficient fluidity can result in misruns and other defects. Fluidity has been studied for more than a century, and theories about factors that limit fluidity as well as experimental methods for measuring fluidity have been developed. The mechanisms for flow termination have been studied by Flemings (Ref 1) and others and were found to be different for alloys with short freezing ranges than for long-freezing-range alloys with a mushy solidification mode. Shortfreezing-range alloys form columnar crystals on the mold wall, pinching off the flow as they grow, as shown in Fig. 1(a). For long-freezing-range alloys, equiaxed crystals or dendrite fragments are carried with the melt, as shown in Fig. 1(b), and choke the flow at the leading tip when the crystals become coherent and form a network. The solid fraction at which this phenomenon occurs depends on the dendrite morphology but will typically be 0.3 to 0.5 (Ref 3). The choking mechanism is much more efficient in terminating melt flow, and fluidity is therefore low in alloys with long freezing ranges.

 Metal variables

a. Chemical composition b. Freezing range c. Viscosity d. Heat of fusion  Mold and mold/metal variables a. Heat transfer b. Mold and metal thermal conductivity c. Mold and metal mass density d. Specific heat e. Surface tension  Casting variables a. Metal head b. Channel diameter c. Casting temperature d. Inclusion content Many of these variables are interdependent: Freezing range, heat of fusion, and specific heat depend on alloy composition; mold variables depend on the casting process; and so on. A short freezing range will, as mentioned, improve fluidity. This is shown in Fig. 2, which shows how fluidity varies with alloy composition for lead-tin alloys. It can be seen that fluidity has maxima for the pure metals and at the eutectic composition where the freezing point is at a minimum. Similar composition versus fluidity curves have been constructed for many

alloys (Ref 2). The viscosity, and thus the fluidity, of liquid metals is strongly influenced by impurities; it has been shown that high-meltinclusion content significantly reduces fluidity (Ref 5). A higher heat of fusion will improve fluidity, and alloying can be used to improve fluidity. An example is the addition of silicon to aluminum alloys. Silicon has a heat of fusion that is approximately five times that of aluminum, and therefore, aluminum-silicon alloys have excellent fluidity. Molds with low thermal conductivity and low mass density can be used when fluidity is a problem. Similarly, insulating mold coatings can reduce heat transfer and improve fluidity (Ref 6). The geometry of the sprue and gating will influence fluidity. A larger metal head will give an increased pressure and a higher velocity of the melt stream, which will improve fluidity. A higher melt temperature will strongly improve the fluidity. For example, it has been found for aluminum alloys that an increase of

Factors Influencing Fluidity Fluidity is a complex technological property and depends on many factors (Ref 4). These can be categorized as follows:

Fig. 1

Stoppage of metal flow. (a) Pinching of metal flow in a skin-freezing alloy with a narrow freezing range. (b) Choking of flow at the leading tip in an alloy with a long freezing range. Source: Ref 2

Fig. 2

Influence of alloy composition on fluidity in lead-tin alloys. Source: Ref 2

376 / Principles of Solidification melt superheat of 1 K (2  F) typically increases the fluid length 1% (Ref 6). The effects of other melt variables, such as gas content, grain refinement, and various trace elements, have been investigated (Ref 4); however, their effects are unclear or are small compared to other effects such as melt superheat.

Fluidity Measurements Fluidity data are important in order to optimize alloys and casting processes, and there are many tests available to obtain such data (Ref 4). Unfortunately, data from the various tests cannot always be directly compared, although they normally give similar trends. Another difficulty with fluidity tests has been poor precision and reproducibility. Fluidity tests are based on pouring metal into one or several channels and then measuring the fluidity length. Three such tests are shown in Fig. 3. It has been demonstrated that the precision of the spiral test in Fig. 3(b) can be improved to a relative error of approximately 5% by using an automatic pouring system that gives a constant pouring temperature and metal head (Ref 6).

and inadequate interdendritic melt feeding of the solidification shrinkage. More specifically, at the later stages of solidification, the dendritic network grows to an extent whereby it has sufficient strength to transmit thermal contractions, and tensile stresses will build up in a restrained casting that is not free to contract. Because the grain boundaries are still covered by a thin liquid layer, the material is still quite weak, and melt feeding to compensate for the solidification shrinkage is simultaneously taking place. Various investigators have explained hot tearing as a too-severe thermally induced deformation and/or an inadequate melt feeding of the solidification shrinkage. In addition to being controlled by these two macromechanisms, a series of microscopic effects and phenomena play a significant role in hot tearing susceptibility, such as grain morphology, dendrite bridging (coalescence) (Ref 10), the local interplay between the coalescing dendrites and the localized liquid feeding (Ref 11), and the alloy composition, including the effect of various trace elements. Some conditions must be met for a hot tear to develop:

Modeling of Fluidity

 The

Fluidity can be predicted with specialized software that simultaneously addresses the heat transfer in the casting and mold, and the solidification and fluid flow in the casting. Provided that accurate data are available for thermophysical properties, coherency, and surface and interface heat-transfer coefficients, the fluid length can be calculated, assuming that dendrite coherency terminates the flow at a given solid fraction (Ref 7).



Hot Tearing Hot tearing occurs in some alloys during the final stages of solidification. It is a serious defect and occurs in castings that are restrained from shrinking in the mold. The characteristics of a hot tear are summarized by Campbell (Ref 2):  Ragged branching crack  Main tear with offshoots along intergranular

paths

 Oxidized fracture surface with dendritic

morphology  Location at a hot spot  Apparent randomness in appearance  Highly alloy-specific The mechanism of hot tearing has been extensively investigated; see the overviews by Sigworth (Ref 8) and Eskin et al. (Ref 9). It is commonly believed that hot tears start to develop in the mushy zone at solid fractions close to 1, and the phenomenon is associated with excessive thermally induced deformations caused by the nonuniform cooling contraction of the casting

  

casting must be restrained from contracting. A coherent network must be established. It should be noted that dendrite coherency in this context; that is, when a network is withstanding tensile forces, is different from, and occurs later than, dendrite coherency in the context of fluidity when the shear strength of the network is important. A liquid film must still be present on the dendrite or grain boundaries. A significant temperature decrease must occur after coherency. The solid network must be sufficiently dense to prevent feeding of liquid into the tear.

Bridges between dendrites must be established in order to withstand the tensile stress buildup. In aluminum alloys, it has been found that bridging between dendrites is established at liquid fractions of approximately 0.1, where typically 90% of the grain boundaries are still covered by liquid films (Ref 12). At this stage, the permeability of the solid network is so low that feeding of liquid into the dilating network is prevented. On the other hand, if the solid dendrite network is bridged to a large degree and a low fraction of the grain boundary is covered by liquid, the network will deform plastically rather than tear. A hot tear will start as a pore in the bulk material or at the surface. Nucleation of a hot tear will occur under conditions similar to a pore, most likely an inclusion. Removal of inclusions or other defects will reduce the hot cracking tendency. Solidifying material in a hot spot is particularly vulnerable to hot tearing, because the surrounding material is colder, stronger, and contracting.

Fig. 3

Fluidity tests. (a) Vacuum fluidity test. (b) Fluidity spiral in sand mold. (c) Multichannel fluidity test in metal mold

Alloy Dependence Figure 4 shows how hot cracking susceptibility varies with alloy concentration for aluminumcopper alloys. It can be seen that the curve peaks at approximately 1wt% Cu. Similar l-shaped curves have been constructed for other alloys (Ref 2). The curves can be understood from the solidification paths of aluminum-copper alloys

Castability—Fluidity and Hot Tearing / 377 develop during the vulnerable stage when the grain boundaries are still largely covered by liquid. This alloy is therefore more susceptible to hot tearing. Finally, at even lower alloy concentrations, the fraction liquid at grain boundaries will be too low, and the material will again become more ductile and thus less prone to hot tearing.

Hot Tearing Tests

Fig. 4

Influence of alloy composition on hot tearing susceptibility in aluminum-copper alloys. Source: Ref 2

Various tests have been designed for measuring the hot tearing tendency in alloys (Ref 9). Most of these are based on casting in a mold that restrains the metal from contracting, typically, ring-shaped castings contracting onto a rigid core or straight “dog-bone” castings restrained at the ends. The tests are evaluated by inspection for hot tears. A variant of the straight-type casting has several straight sections of different lengths, where the longest sections are the most likely to tear. This test can be used to assign a hot tearing sensitivity index to alloys. High-temperature tensile testing has been used to assess the strength of alloys close to their solidus, and the results have been used to evaluate the hot tearing tendency. A problem with the tests, however, is that most of them are designed to be carried out during melting. Their relevance for evaluating the behavior of solidifying alloys is thus questionable. More recently, tests have been reported for measuring the development of tensile stresses in castings (Ref 13), and it is expected that these will be important for the further understanding and quantification of hot tearing phenomena.

Modeling of Hot Tearing and Hot Tearing Criteria Fig. 5

Temperature dependence of solidification of Al-1wt%Cu (dashed curve) and Al-20wt%Cu (solid curve)

shown in Fig. 5, which shows the temperature versus fraction solid curves for Al-20wt%Cu and Al-1wt%Cu. The alloy with the higher copper content has a liquidus temperature of 592  C (1095  F) and a freezing range of 44  C (80  F). The temperature decrease during solidification occurs early, and the eutectic temperature is reached at a fraction solid of 0.45. The solid phase at this point is not coherent and therefore will not transmit any tensile stresses. Furthermore, the structure is open enough to feed any opening between dendrites that may occur. After reaching the eutectic temperature, the alloy will solidify isothermally without any thermal contraction. Thus, this alloy will not hot tear. However, the Al-1wt%Cu alloy has a liquidus temperature of 657  C (1215  F) and will solidify almost isothermally until a fraction solid of approximately 0.9. At this stage, the solid network is coherent and will transmit stress. The temperature will then decrease more than 90 until the eutectic temperature is reached, and a significant stress will

Most models and criteria attempting to relate hot tearing susceptibility to controllable material and process parameters address one of the previously mentioned main mechanisms: excessive thermally induced deformations or inadequate feeding of the solidification shrinkage. Some investigators argue that hot tearing will occur if the liquid is no longer able to fill the intergranular openings caused by the solidification shrinkage. Others define a vulnerable region for high solid fractions in which the permeability is low and only thin continuous films of interdendritic liquid exist that are not able to sustain possible thermal stresses. Based on continuum mechanics models that quantify thermally induced deformations and associated stresses, various criteria are proposed that consider overcritical values of stress, strain, or strain rate as the triggering factor for hot tearing. Again, the overview by Eskin et al. (Ref 9) is recommended for details; only a few main points are summarized here. Rappaz et al. (Ref 14) were the first investigators to simultaneously model feeding difficulties and excessive deformations. They considered microporosity formation as the initiating

mechanism for hot tearing and included the additional need for interdendritic melt feeding caused by the tensile deformation of the solidified material in the direction transverse to the columnar dendritic growth. Further elaborations on this idea can be found in Ref 15 to 17. Volume-averaged two-phase conservation equations for mass and momentum in the coexisting liquid and solid phases in a coherent and isotropic mushy zone are formulated in Ref 18 to 20. Both the interdendritic feeding and the thermally induced deformations were simultaneously taken into account, and the possibility of dilatation (opening up) of the dendritic network was included. A hot tearing criterion was then introduced (Ref 20) by calculating a weighted sum of shear and dilatation deformation in a vulnerable solidification interval defined by a limit pressure for solidification shrinkage feeding and a limit solid fraction above which the material obtains sufficient strength. The main challenge with all modeling approaches so far is to formulate, by means of variables included in the model, hot tearing criteria that accurately differentiate the hot tearing susceptibility for different choices of material and process parameters in real casting operations. The formalistic approach in Ref 18 to 20 furthermore reveals the need for accurate constitutive relations for the rheology of the deforming dendrite network and for the interdendritic melt flow permeability. Although many models and criteria correctly reflect the trends in hot tearing susceptibility caused by changes in these parameters, reality shows that significant research and development work is still needed before modeling tools of real predictive power are available and widely implemented into industrial daily practice. (See also the article “Modeling of Stress, Distortion, and Hot Tearing” in this Volume.)

REFERENCES 1. M.C. Flemings, Solidification Processing, McGraw-Hill, 1974, p 219 2. J. Campbell, Castings, Butterworth-Heinemann, 1991 3. L. Arnberg, L. Backerud, and G. Chai, Solidification Characteristics of Aluminium Alloys, Vol 3, Dendrite Coherency, AFS, 1996 4. M. Di Sabatino, “Fluidity of Aluminium Foundry Alloys,” NTNU Doctoral thesis, 2005, p 185 5. M. Di Sabatino, L. Arnberg, S. Rrvik, and A. Prestmo, Mater. Sci. Eng. A, Vol 413–414, 2005, p 272 6. M. Di Sabatino, L. Arnberg, S. Brusethaug, and D. Apelian, Int. J. Cast Met. Res., Vol 19 (No. 2), 2006, p 94 7. M. Di Sabatino, L. Arnberg, and F. Bonollo, Mater. Sci. Technol., Vol 23 (No. 1), 2005, p 3 8. G.K. Sigworth, AFS Trans., Vol 106, 1999, p 1053

378 / Principles of Solidification 9. D.G. Eskin, Suyitno, and L. Katgerman Prog. Mater. Sci., Vol 49, 2004, p 629 10. M. Rappaz, A. Jacot, and W.J. Boettinger, Metall. Mater. Trans. A, Vol 34, 2003, p 467 11. S. Vernede, P. Jarry, and M. Rappaz, Acta Mater., Vol 54, 2006, p 4023 12. Y. Ju and L. Arnberg, Int. J. Cast Met. Res. Vol 16 (No. 6), 2003, p 522 13. C. Davidson, D. Viano, L. Lu, and D. St. John, Int. J. Cast Met. Res., Vol 19 (No. 1), 2006, 59

14. M. Rappaz, J.-M. Drezet, and M. Gremaud, Metall. Mater. Trans. A, Vol 30, 1999, p 449 15. M. Braccini, C.L. Martin, M. Sue´ry, and Y. Bre´chet, Modelling of Casting, Welding and Advanced Solidification Processes— IIX, Shaker Verlag, 2000, p 18 16. J.F. Grandfield, C.J. Davidson, and J. A. Taylor, in Continuous Casting, W. Schneider and W. Ecke, Ed., DGM, Frankfurt, Nov 13, 2000, p 205

17. Suyitno, W.H. Kool, and L. Katgerman, Mater. Sci. Forum, Vol 396–402, 2002, p 179 18. I. Farup and A. Mo, Metall. Mater. Trans. A, Vol 31, 2000, p 1461 19. M. Mhamdi, A. Mo, and C.L. Martin, Metall. Mater. Trans. A, Vol 33, 2002, p 2081 20. M. M’Hamdi, A. Mo, and H.G. Fjær, Metall. Mater. Trans. A, Vol 37, 2006, p 3069–3083

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 379-381 DOI: 10.1361/asmhba0005224

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Semisolid Metal Processing Qingyue Pan and Diran Apelian, Worcester Polytechnic Institute

SEMISOLID METAL (SSM) PROCESSING, also known as SSM casting or semisolid forming, is a special die casting process where a partially solidified metal slurry (typically, 50% solid and 50% liquid instead of a fully liquid metal) is injected into a die cavity to form die cast components. The key to SSM processing is to generate an SSM slurry that contains a globular primary phase and exhibits thixotropic behavior. Semisolid metal processing not only produces complex dimensional details (e.g., thin-walled sections), as in high-pressure die casting, but also produces high-integrity castings currently attainable only with squeeze or lowpressure permanent mold casting. Semisolid metal processing offers several distinct advantages over conventional near-net shape forming technologies. These advantages include low manufacturing cycle time, increased die life,

reduced porosity and solidification shrinkage, improved mechanical properties, and so on.

Flow Behavior of SSM Viscosity Evolution During Continuous Cooling The origin of SSM can be traced to the experiments conducted by David Spencer at Massachusetts Institute of Technology in 1971 (Ref 1, 2). Spencer’s work originally was directed toward understanding hot tearing in alloy castings; however, in the course of his viscosity measurements of semisolid Sn-15wt %Pb slurries, he discovered that if a semisolid slurry is sheared above the liquidus temperature and slowly cooled into the two-phase region,

it can have an apparent viscosity that is orders of magnitude less than a slurry that is cooled to the same temperature (solid fraction) before shear, as illustrated in Fig. 1. Detailed microstructural analysis revealed that the difference in apparent viscosity values comes from the significant difference in morphology of the solid phase in the slurries (spherical versus dendritic). Moreover, at a given solid fraction, the apparent viscosity decreases with increasing shear rate or decreasing cooling rate, as illustrated in Fig. 2. This discovery has led to the development of semisolid processing.

Rheology of SSM An SSM slurry is a mixture of liquid metal and solid particles that have a predominantly

Fig. 1

Experimentally determined viscosity and shear stress versus fraction solid of Sn-15%Pb alloy cooled at 0.006 K/s with a shear rate of 200 s 1. Source: Ref 2

Fig. 2

Effect of shear rate on apparent viscosity of semisolid alloys. Source: Ref 3

380 / Principles of Solidification globular shape. The rheological behavior of SSM slurries is similar to that of viscoplastic materials, such as concentrated suspensions, pastes, foodstuffs, emulsions, and foams. This class of materials is characterized by the existence of a yield stress, which implies that the material behaves as a solid if the applied stress is below the yield stress; however, if the applied stress exceeds the yield stress, the material will flow and exhibit liquidlike behavior. The current understanding of the complex rheology of SSM slurries is based on the fact that under steadystate conditions, the semisolid slurry behaves as a shear-thinning (pseudoplastic) fluid, wherein the effective viscosity decreases with increasing shear rate. However, under rapid transient conditions, the semisolid slurry behaves as a shear-thickening fluid; that is, effective viscosity increases with increasing shear rate.

Yield Stress of Semisolid Slurries The yield stress of an SSM is the result of a continuous solid skeleton matrix formed by welded particles, physical crowding of particles, or other interparticle interactions during processing. The presence of a distinct yield stress in a semisolid material significantly modifies its rheological behavior. Both experiments and numerical simulations show that, depending on the flow velocity and yield stress of a semisolid slurry, there are five distinct die-filling patterns during semisolid processing: shell, disk, mound, bubble, and transition. The distinctive filling patterns are illustrated in Fig. 3. The measured yield stress of A356 and 357 aluminum alloys as a function of temperature in the two-phase region is shown in Fig. 4 and 5. It is found that:

 The yield stress of commercial semisolid

aluminum alloys is a strong function of temperature (solid fraction). In the commercial forming temperature region, as temperature increases, yield stress decreases dramatically, varying between 103 and 10 1 kPa (150 and 0.015 psi).  At a given solid fraction, the yield stress of semisolid slurries depends on microstructural indices (i.e., entrapped liquid content, shape factor of the alpha phase, and the alpha particle size). Low entrapped liquid content and small, round alpha particles tend to decrease the stress needed to initiate the flow, that is, the yield stress of the slurry.

Semisolid Microstructural Formation Solidification Microstructure Under Forced Convection Nearly all commercial alloys solidify dendritically with either a columnar or an equiaxed dendritic structure. During solidification, there are a number of processes taking place simultaneously in the two-phase region. These include ripening, interdendritic fluid flow, and solid movement. The dendritic structure is greatly affected by convection during the early stages of solidification, whereas under vigorous convection and slow cooling, the solidification structure tends toward a rosette or a spherical morphology. Figure 6 illustrates current understanding of microstructual formation under forced convection, specifically:  Forced convection promotes crystal growth

Fig. 3

Flow patterns as a function of Bingham number, Bi, and Reynolds number, Re. The figure illustrates five distinct filling patterns (bubble, mound, shell, disk, and transition) in semisolid casting. Source: Ref 4

Fig. 4

Yield stress of GR A356 and MHD A356 alloys as a function of temperature. Source: Ref 4

Fig. 5

Yield stress of GR 357 and MHD 357 alloys as a function of temperature. Source: Ref 4

due to an enhanced mass transport in the two-phase region.  Under forced convection, a laminar flow changes dendritic growth morphology to a rosette, whereas a turbulent flow changes growth morphology from a rosette to a sphere.  Several mechanisms have been proposed to explain the observed nondendritic

Fig. 6

Schematic illustrating morphological transition under forced convection. Source: Ref 5

Semisolid Metal Processing / 381 REFERENCES 1. D.B. Spencer, Ph. D. thesis, MIT, Cambridge, MA, 1971 2. D.B. Spencer, R. Mehrabian, and M.C. Flemings, Metall. Trans., Vol 3, 1972, p 1929–1932 3. M.C. Flemings, Metall. Trans. B, Vol 22 (No. 3), 1991, p 269–293 4. Q.Y. Pan, D. Apelian, and A.N. Alexandrou, Yield Behavior of Commercial Aluminium Alloys in the Semi Solid State, Metall. Mater. Trans. B, Vol 35, Dec 2004, p 1187–1202 5. Z. Fan, Int. Mater. Rev., Vol 47 (No. 2), 2002, p 49–85

SELECTED REFERENCES  A.N. Alexandrou and F. Bardinet, J. Mater.

Process. Technol., Vol 96, 1999, p 59–72

 A.N. Alexandrou, E. Duc, and V. Entov,

Fig. 7

Schematic of semisolid metal (SSM) processing routes: thixocasting versus rheocasting. MHD, magnetohydrodynamic; GR, grain refinement

morphology. These include dendrite arm fragmentation, dendrite arm root remelting, and growth-controlled mechanisms. It seems that the observed rosette and spherical morphologies are more likely related to a growth phenomenon rather than mechanical fragmentation.

Globular Structure via Copious Nucleation Originally, to achieve an SSM slurry, it was thought necessary to cool the liquid into the two-phase region and to shear off and break the dendrites (i.e., melt agitation via mechanical or magnetohydrodynamic stirring). However, recent research and development show that it is not necessary to use shear forces to produce the globular semisolid structure. Instead, a slurry with an ideal semisolid structure can be produced directly from the melt if the temperature of the melt is such that many nuclei (copious nucleation) are generated, and the nuclei do not grow past a certain point (i.e., suppression of dendritic growth), nor melt back into the bulk liquid. This controlled nucleation and growth concept is the

genesis of various processes and methodologies to generate semisolid slurries from the liquid state.

Semisolid Processes: Thixocasting versus Rheocasting There are two primary semisolid processing routes: thixocasting and rheocasting. Thixocasting begins with a nondendritic solid precursor material that is specially prepared by the manufacturer using continuous casting methods. Upon reheating the material into the mushy (i. e., two-phase) zone, thixotropic slurry forms and becomes the feed for the casting operation, as illustrated in Fig. 7. Rheocasting (also known as slurry-on-demand) starts with material in the liquid state, and the thixotropic slurry is formed directly from the melt via special thermal treatment/management of the system; the slurry is subsequently fed into the die cavity. Of these two routes, rheocasting is favored in that there is no premium added to the billet cost, and the scrap recycling issues are alleviated.



      



J. Non-Newton. Fluid Mech., Vol 96, 2001, p 383–403 D. Apelian, J. Jorstad, and A.M. de Figueredo, Science and Technology of Semi-Solid Metal Processing, NADCA, Rosemont, IL, 2001, p 7–1 to 7–20 G.R. Burgos, A.N. Alexandrou, and V. Entov, J. Rheol., Vol 43, 1999, p 463–483 P.A. Joly and R. Mehrabian, J. Mater. Sci., Vol 11 (No. 8), 1976, p. 1393–1418 D.H. Kirkwood, Int. Mater. Rev., Vol 39 (No. 5), 1994, p 173–189 P. Kumar, C.L. Martin, and S. Brown, Metall. Trans. A, Vol 24 (No. 5), 1993, p 1107–1116 P. Kumar, C.L. Martin, and S. Brown, Acta Metall. Mater., Vol 42 (No. 11), 1994, p 3595–3614 Q.Y. Pan, D. Apelian, G.R. Burgos, and A.N. Alexandrou, Alum. Trans., Vol 2 (No. 1), 2000, p 37–56 C.J. Paradies and M. Rappaz, Proceedings of the Eighth International Conference on Modeling of Casting and Welding Processes, June 7–12, 1998 (San Diego, CA), p 933–940 S. Sannes, H. Gjestland, L. Arnberg, and J.K. Solberg, Proceedings of Third International Conf. on Semi-Solid Processing of Alloys and Composites (Tokyo, Japan), 1994, p 271–280

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 382-385 DOI: 10.1361/asmhba0005225

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Spray Casting P.S. Grant, Oxford University

SPRAY CASTING—also known as spray forming—is a niche casting process for the manufacture of preforms. The primary advantage of the spray casting process is the ability to manufacture alloy compositions problematic by conventional processes, such as ingot casting, direct chill casting, powder metallurgy, and so on. The process is shown schematically in Fig. 1 for the most widely employed spray casting configuration used in the production of cylindrical preforms. Other configurations allow for the manufacture of large-diameter rings and tubes or wide strip (Ref 1). Spray casting operates at flow rates from 3 to 5 kg/ min (7 to 11 lb/min) for aluminum alloys up to 20 kg/min (44 lb/min) for steels. Commercial examples of alloys manufactured by spray casting include:
Hypereutectic aluminum-silicon-base alloys

for cylinder liners (Peak, Germany). Spray

cast preforms, such as that shown in Fig. 2, are extruded and then swaged to parallel side tubes that are used directly as cylinder liners in engine blocks for a number of mass-production automotive customers. The refined, high volume fraction of primary silicon yields a low coefficient of thermal expansion, low wear rate, and allows elevatedtemperature operation.
Silicon-aluminum alloys for thermal management in electronic packaging applications (Sandvik-Osprey, United Kingdom). Spray casting allows for large preforms with silicon concentrations up to 70 wt% to be manufactured successfully. Packages for high-value electronics and other electrical components are then machined from consolidated preforms. Changing the proportion of silicon to aluminum offers a balance of low density, high thermal conductivity, and low coefficient of thermal expansion that

may be matched to various circuit board and passive electronic materials.
Aluminum-neodymium alloy preforms for sputtering targets that are used to deposit coatings onto various components in display screen and information storage applications (Kobelco, Japan). Spray casting provides levels of microstructural homogeneity unachievable by conventional casting, which is essential for a high-quality sputtering source.
Speciality steel preforms of up to 600 mm (24 in.) diameter with low macrosegregation and a refined and homogeneous carbide distribution (Danspray, Denmark)
Spray cast thick steel coatings on dissimilar steel feedstock preforms for reinforcing bar applications (CMC, United States). Preforms are spray cast with up to a 6 mm (0.2 in.) thick outer layer (cladding) of stainless steel at up to 16 tonnes/h (18 tons/h), as shown in Fig. 3. The spray-coated preform is rolled to a stainless-steel-clad rebar that offers extended lifetimes when subsequently incorporated into an aggressive concretebased environment.

Principles The spray casting process is comprised of the following sequential steps: 1. Continuous gas atomization of a melt stream using high-pressure, high-flow-rate nitrogen or argon to produce a spray of 10 to 500 mm diameter alloy droplets 2. Subsequent droplet cooling at typically 102 to 104 K/s (180 to 18,000  F/s) and acceleration to 50 to 100 m/s (160 to 300 ft/s) under the action of the atomizing gas 3. Deposition of droplets at an incrementally forming preform surface 4. Relatively slow cooling at 0.1 to 10 K/s (0.2 to 20  F/s) and solidification of any residual liquid in the spray cast preform

Fig. 2

Fig. 1

Schematic of the spray casting process for the manufacture of billets

Spray cast hypereutectic aluminum-silicon-base alloy produced by Peak, Germany, for cylinder liner applications in the automotive sector. Courtesy of Peak

The metallurgical advantages of spray casting arise through the unusual solidification conditions and are manifest in the microstructure through:

Spray Casting / 383

Fig. 3

Spray casting plant at CMC Steel, United States, for the manufacture of spray-clad reinforcing bar. Courtesy of CMC Steel

established optimal and well-controlled processing approaches (Ref 1). Since the original conception of spray casting by Richard Singer at Swansea University, United Kingdom, in the 1970s and its subsequent commercialization by Osprey Metals (now SandvikOsprey) through the 1980s and 1990s, a number of further developments, beyond the basic configuration in Fig. 1, have been made. First, the use of twin melt atomization from a single, very large melt has increased spray casting flow rates for steels up to 20 kg/min (44 lb/min) and preform diameters up to 600 mm (24 in.). Secondly, the reinjection of overspray from previous casts of the same composition is now used in all commercial spray casting plants; screened solid overspray particles are continuously injected into the atomization region during spray casting and co-deposited with the droplet spray at the preform. Because of the large amount of droplet particle remelting in spray casting (discussed later), no remnants of the overspray powder microstructure remain in the spray cast preform, the properties are unaffected, and overall process yields can exceed 90%.


Equiaxed/polygonal

grains of diameter typically in the range of 20 to 50 mm and the absence of columnar/dendritic grain morphologies
High levels of microstructural homogeneity and low levels of macrosegregation, regardless of position in the preform
Ability to produce the characteristic spray cast microstructure in all important engineering alloys, such as aluminum-, iron-, nickel-, titanium-, and copper-base alloys

Fig. 4

(a) Typical 25 kg (55 lb) development-scale spray cast Al-5.35Mg-1.2Li-0.28Zr preform. (b) Corresponding as-cast microstructural orientation map obtained by electron backscattered diffraction showing the noncolumnar/dendritic equiaxed polygonal grains characteristic of the spray casting process

Figure 4(a) shows a 25 kg (55 lb) low-density spray cast Al-5.3Mg-1.2Li-0.28Zr aerospace alloy preform, and the corresponding characteristic polygonal as-cast microstructure is shown in the electron backscattered diffraction orientation map in Fig. 4(b). The as-spraycast grain size was  15 mm, and the porosity was <1 area%, which was subsequently closed by hot isostatic pressing and then forged to shape. The disadvantages of spray casting include as-cast porosity, which, for most industrial applications, requires closing by extrusion, rolling, hot isostatic pressing, or other downstream processes, and process losses, since not all droplets created by atomization end up in the spray cast preform; this material is termed overspray. Spray cast aluminum, steel, and nickel alloys have shown material properties that are equivalent to or better than the same alloys processed conventionally. Early concerns regarding structural properties—the possibility of oxidation during processing undermining fatigue performance, or entrapped atomizing gas undermining creep performance due to an enhanced thermally induced porosity response—have largely been overcome through experience in commercial-scale spray casting plants that have

Atomization and the Droplet Spray Gas atomization is a chaotic, stochastic process that always produces a wide range of droplet diameters. The resulting droplet mass or volume distribution can usually be closely approximated to a log-normal distribution, such as is shown in Fig. 5(a). A consequence of this unavoidable spread of droplet diameters is that, at the point where the droplet trajectories intersect the growing preform surface, typically having traveled a distance of up to 0.7 m (2.3 ft), the spray comprises droplets with a broad range of velocities, temperatures, and solid fractions. This behavior has been studied extensively by modeling and specialized experimental measurements. As shown in Fig. 5(a), droplet diameters less than 50 mm may be fully solid, droplets with a diameter greater than 200 mm may be fully liquid, and droplets with intermediate diameters may be mushy. Therefore, while the average or equilibrated solid fraction of the spray at deposition is in the range of 0.3 to 0.6 for the production of low-porosity preforms with the characteristic spray cast microstructure (Ref 2), the millions of individual droplets have a wider range of solid fractions, from 0 to 1. Figure 5(b) shows the same droplet diameter distribution as in Fig. 5(a), but on a number rather than mass or volume basis. The distribution of droplet solid fractions is now dominated by the very large number (but small volume) of solid droplets. As discussed later, the presence of a large number of solid and mushy particles, but a large volume of substantially or fully liquid droplets, is critical in promoting the characteristic spray cast grain structure.

384 / Principles of Solidification

Fig. 6

Fig. 5

(a) Typical log-normal droplet diameter probability density distribution on a mass or volume basis obtained by gas atomization, with superimposed assumed boundaries between solid, mushy, and liquid droplets at the point of deposition during spray casting. (b) The same log-normal droplet diameter probability density distribution on a number basis, with the same superimposed boundaries

Deposition and Grain Multiplication When droplets/particles arrive at the mushy preform top surface, some solid particles bounce from the preform surface and are lost as overspray, while substantially liquid droplets deform and may splash at the preform surface, generating a host of smaller secondary droplets that either redeposit or are lost as overspray. The timescale for droplet spreading is in the micro- to millisecond range. Once spreading is complete, droplets then equilibrate to the preform average surface temperature, Teq (which remains approximately constant throughout spray casting as the preform grows and once an initial thermal transient is passed). The time for equilibration, tTeq , can be estimated from tTeq  x2 =a, where x is the equilibration distance, and a is the thermal diffusivity. For x of up to several hundreds of micrometers for the largest droplets and assuming a typical

Schematic representation of the changes in solid fraction during the spray casting process and the accompanying change in microstructure prior to and after droplet deposition

value of a = 105/m2/s (104 ft2/s) for metals, then tTeq <0:1 s. Consequently, large droplets deposited with a temperature greater than the average surface temperature will be quenched on deposition; solid particles with temperatures lower than the average surface temperature will be reheated rapidly, and droplets with intermediate temperatures may partially remelt. For the mushy droplets, the abrupt change in droplet cooling rate from the spray of typically 102 to 104 K/s (180 to 18,000  F/s) to that in the preform of typically 0.1 to 10 K/s (0.2 to 20  F/s), combined with the high shear forces during spreading, destabilizes the droplet dendrite/ cell network and promotes dendrite/cell network breakup. Therefore, a second or so after deposition, the conditions in the mushy material close to the preform surface (which continues to grow) will comprise residual liquid from the larger mushy and liquid droplets in the spray and a very high number density of partially remelted solid particles and broken dendrite fragments from mushy droplets. With such a high number density of efficient nuclei, substantial undercoolings are impossible and further nucleation events unlikely. These conditions facilitate the rapid spheroidization of remnant solid fragments in an attempt to minimize the solid/liquid interfacial area, since dendrite fragments from solid or mushy droplets are not observed in the final microstructure. The average grain diameter, d0, in the mushy material just below the top surface is dependent on the number of solid particles per unit volume, NV, in the mushy material, with average solid fraction, fs, immediately after deposition and equilibration, and can be estimated from:

 d0 

fs NV

1=3 (Eq 1)

Attempts have been made to calculate NV (Ref 3) under spray casting conditions but rely on various assumptions or calculations regarding the distribution of solid, mushy, and liquid droplets in the spray, the grain multiplication effects, and other factors. Even though many of these factors cannot be quantified reliably in practice, Eq 1 and the very high nucleation number density in the preform suggest that:
Regardless of specific spray casting condi-

tions, there will always be a very large number of small solid particles in the spray that, while contributing little to the preform mass, dominate grain nucleation behavior in the preform.
Even when changes in fs and NV in Eq 1 may be realized—for example, by changing the spray casting conditions (such as melt flow rate, atomizing pressure, spray distance, etc.)— the resulting average grain diameter has only a relatively weak cube root dependency. Figure 6 shows schematically the changes in solid fraction during the spray casting process as a function of axial distance from the point of droplet atomization. The droplets equilibrate in the preform top surface to a common solid fraction, fse, and the preform solid fraction (ignoring any radial variations across the preform) then gradually increases as the preform is slowly withdrawn to maintain a constant spray distance. The presence of a significant liquid fraction in the preform near the top surface region and the

Spray Casting / 385 large amount of remelting distinguish spray casting as a semicontinuous casting process distinct from incremental spray deposition processes such as arc or plasma spraying, where there is little remelting, and discrete droplet deposition events are preserved in the final microstructure.

Microsegregation The deposition of mushy droplets with different solid fractions will contain residual liquid of varying composition. Similarly, partially remelted solid particles will release relatively concentrated liquid. The time for compositional equilibration of the liquid, tCeq , can be estimated from tCeq  x2 =DL , where DL is the coefficient of diffusion for solute atoms in the liquid. Again, assuming x is up to several hundred micrometers and DL = 108 m2/s (107 ft2/s), then tCeq is typically of the order of seconds. Equilibration of liquid compositions will necessarily involve remelting and homogenization as concentrated, relatively hot liquid is diluted by mixing with cooler, remelted solid of lower solute concentration, helping to reduce microsegregation.

Macrosegregation and Coarsening As shown in Fig. 6 for the production of large preforms, liquid persists in the spray cast preform for many minutes, and macrosegregation profiles may develop in spray cast preforms under some conditions, for example, limited copper inverse macrosegregation in various high-strength aluminum alloys driven by shrinkage at the preform periphery (Ref 4). Diffusion-controlled coarsening of the solid particles/grains in the relatively flat temperature gradient of the slowly cooling preform may be expected to occur, following:

d3  d30 ¼ KðT Þt

(Eq 2)

where d is the grain diameter at time t, d0 is the starting grain size at t = 0, and K(T) is the temperature- (and solid fraction) dependent coarsening constant. Second-phase particles formed early in solidification and perhaps microporosity from small amounts of insoluble atomizing gas have been suggested to play an important role in inhibiting grain growth at higher solid fractions, where the thickness of liquid films delineating grains approaches the length scale of the second phases (Ref 5). Because coarsening involves the transport of solute by diffusion, similar to the remelting and equilibration behavior described previously, it can also be expected to play a role in reducing microsegregation. The developing spheroidal/polygonal spray cast grain structure allows the flow of liquid to feed solidification shrinkage in the spray cast preform better than in columnar/dendritic morphologies expected in conventionally cast preforms of similar size. Since approximately half of the alloy latent heat has been removed prior to deposition, spray casting is able to produce cast preforms in compositions considered problematic by conventional casting because of their large freezing ranges and tendency for shrinkage-induced macrosegregation and defects.

Summary Spray casting has developed from a rapid solidification process, apparently related to powder metallurgy, into a niche casting process for producing exceptionally homogeneous preforms in thousands of tonnes per year of otherwise difficult-to-cast alloys, such as hypereutectic aluminum-silicon alloys and specialty steels and claddings, used primarily in structural applications. More recently, spray casting has found successful commercial application

as the primary production route for silicon-aluminum- and aluminum-neodymium-base alloys for higher added value in the electronics and telecommunications sectors. The key phenomena that distinguish spray casting from other solidification processes are the enormous amounts of solid remelting, compositional mixing, and equilibration that occur at and close to the preform top surface. Spray casting is an intrinsic or self-grain-refining casting process in which the purpose of the distinctive atomization step is to generate a large number density of solid fragments and particles that will undergo subsequent partial remelting and grain multiplication and to remove a significant fraction of the alloy latent heat prior to reconstituting the particles/droplets into a standard metallurgical preform so that important but problematic alloys with wide freezing ranges can be cast successfully.

REFERENCES 1. P.S. Grant, Spray Forming, Prog. Mater., Sci., Vol 39, 1995, p 497–545 2. K.P. Mingard, P.W. Alexander, S.J. Langridge, G.A. Tomlinson, and B. Cantor, Direct Measurement of Sprayform Temperatures and the Effect of Liquid Fraction on Microstructure, Acta Mater., Vol 46 (No. 10), 1998, p 3511–3521 3. P.S. Grant, Solidification in Spray Forming, Mater., Trans. A, Vol 38, 2007, p 1520–1529 4. K.P. Mingard, B. Cantor, I.G. Palmer, I.R. Hughes, P.W. Alexander, T.C. Willis, and J. White, Macro-Segregation in Aluminium Alloy Sprayformed Preforms, Acta Mater., Vol 48, 2000, p 2435–2449 5. S. Annavarapu and R.D. Doherty, Inhibited Coarsening of Solid-Liquid Microstructures in Spray Casting at High Volume Fractions of Solid, Acta Metall. Mater., Vol 43, 1995, p 3207–3230

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 386-389 DOI: 10.1361/asmhba0005226

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Rapid Solidification W.J. Boettinger, National Institute of Standards and Technology

RAPID SOLIDIFICATION is a tool for modifying the microstructure of alloys from that obtained by ordinary casting. Three general classes of processing methods that may produce such changes are shown in Fig. 1: 1) melt spinning to yield thin (approximately 100 mm) ribbon or fiber, 2) atomization to produce powder (100 to 200 mm), and 3) surface melting/deposition and resolidification to produce thin surface layers (0.1 to 200 mm). Each of these classes includes methods with a variety of names reflecting the significant activity of inventors in this field: splat cooling, melt quenching, planar flow casting, laser annealing, and so on. Thin plasma-sprayed deposits may have the characteristics of rapidly solidified samples. All of these techniques may be thought of as casting where at least one physical dimension is small so that heat is extracted more rapidly than in conventional casting. For each processing class, the two insets in Fig. 1 show a range of how microscopic solidification can occur. The left inset shows the case when a relatively smooth solidification front traverses the samples. The front may also be cellular, dendritic, or eutectic. The right inset shows cases where solid is forming ahead of, or instead of, a smooth front. The microscopic response to the heat extraction rate can vary considerably depending on alloy system, composition, and the heterogeneities present in the melt. Due to rapid release of heat of fusion, solidification causes a large change in the thermal field, unlike solid-state transformation, where the enthalpy change on transformation is generally smaller. Thus, cooling rate prior to solidification is not a good predictor of metallurgical solidification processes. The exception is the case of glass formation. Factors such as temperature prior to nucleation or solidification-front velocity are natural to solidification theory. Detailed information on solidification theory can be found in this Handbook in the Section “Principles of Solidification.” The next Section of this Handbook, “Modeling and Analysis of Casting Processes,” includes articles on the application of mathematical modeling and computer simulation to solidification processes. Because of the limited size of rapidly solidified products, many samples have been prepared only for research to probe the fundamental nature

of crystal growth at rapid rates and/or the amorphous state. Other direct uses for the materials are inexpensive wire manufacture and amorphous tapes and ribbons for brazing and magnetic applications. The latter two constitute the most successful commercial application of this technology. Some rapidly solidified materials have been subjected to subsequent consolidation into bulk samples by a myriad of methods that may require shredding, encapsulation (canning), and consolidation by hot extrusion or hot isostatic pressing. Microstructural changes in the solid state from this subsequent processing are often significant and must be considered seriously.

Microstructural Modification The change from a traditional casting method to a rapid solidification method may considerably alter the microstructure of an alloy. The level of modification depends most sensitively on the alloy system and composition. Microsegregation. The least dramatic of these changes, but not the least important, is elimination of macrosegregation and refinement of the scale of microsegregation. The phases produced may be the same as in a conventional casting, but the size of the sometimes brittle secondphase particles will be reduced. The reduced microsegregation scale can shorten homogenization times. These effects can be exploited for improvements in mechanical and corrosion properties. Change in Primary Phase. The next level of modification is a change in the identity of the primary (first) phase to solidify upon cooling, in addition to the refinement mentioned previously. The change in identity is important because the primary phase produced by solidification in general is the coarsest. Many examples are found among hypereutectic aluminum-transition metal alloys, where a coarse intermetallic would normally be the primary phase followed by solidification of the aluminum solid solution in a finer eutectic product. Rapid solidification of such hypereutectic alloys often produces a dendritic or cellular structure of aluminum as the primary phase, followed by refined intermetallic.

Because the primary aluminum phase frequently contains excess transition metal, these alloys can be manipulated by heat treatment during consolidation to produce dispersion-hardened alloys with high fractions of hardening phase that are resistant to coarsening. Such alloys cannot be heat treated again because the hardening particles cannot be put back into solid solution in the solid state. New Primary or Secondary Phase. The identity of the primary phase and/or secondary phase may be different from the one(s) encountered in a normal casting process and may not be present in the equilibrium phase diagram. For example, aluminum-rich aluminum-iron alloys contain Al3Fe at slow cooling rates but Al6Fe at high rates (Ref 1, 2). The discussion that follows on metastable phase diagrams is important here. Formation of Noncrystalline Phases. Quasi-crystalline and/or amorphous (metallic glass) states may be achieved for special compositions. The entire microstructure may be noncrystalline, or the noncrystalline phase may be a secondary phase in combination with a crystalline phase. In recent years, some alloy compositions have been found to form quasi-crystalline and/or amorphous structures without rapid cooling. Some of these materials have been transformed by subsequent heat treatment/consolidation to produce extremely fine-grained materials. To understand the fundamentals of these four microstructural changes, normal crystal growth and solidification theories and models have been extended and modified to treat more rapid cooling conditions. This requires a more careful consideration of three factors: heat flow, thermodynamic constraints/conditions at the liquid-solid interfaces, and diffusional kinetics/ microsegregation.

Heat Flow Rapid solidification of crystalline alloys clearly involves the rapid removal of superheat and latent heat of fusion by an appropriate heat sink. For application of the solidification theory, it is often useful to estimate the liquidus

Rapid Solidification / 387

Fig. 1

The three types of processes used to create rapidly solidified materials (not to scale). Shown below each are two examples of how microscopic solidification may take place.

isotherm velocity for a given heat-transfer situation. This quantity plays a much more fundamental role than cooling rate in understanding rapid solidification microstructures. Heat sinks can be classified as external or internal.

External heat sinks are massive (relative to the casting size) metal substrates such as the wheel in melt spinning; large volumes of highvelocity cooling gas, such as found in atomization processes; and the underlying cold, unmelted solid in surface melting. The effectiveness of these heat sinks is characterized by a heat-transfer coefficient, h, between the metal and the external heat sink. For surface melting, infinite values of h are appropriate because the melt is in contact with its own solid. For melt spinning, heat-transfer coefficients depend greatly on the wheel material, surface finish, and atmosphere. Their determination poses the same problem as for any mold/casting interface. For atomization, the heat-transfer coefficient is often dominated by convective/conductive cooling by the gas, which dramatically increases with decreasing powder size. Internal heat sinks are intrinsic to the material in the sense that the solidifying casting is its own heat sink. This occurs when the alloy is undercooled below its liquidus temperature prior to the initiation of solidification. The release of latent heat can then be absorbed by the sample itself, a process that is quite efficient but may cause a reheating (recalescence) of the sample. When the undercooling of the liquid prior to nucleation is spatially uniform, the effectiveness of this heat sink is expressed by the dimensionless undercooling, DY = DT/(L/C), where C is the heat capacity per unit volume of liquid, and L is the latent heat per unit volume (L/C for aluminum is 364 K). For powder solidification, if DY = 0.5, one-half of the particle can freeze rapidly, and the remainder must rely on additional external heat transfer through the powder exterior, which is usually much slower. Nucleation usually starts on the powder surface. If the solidification front is relatively smooth, the particle may exhibit large spatial variations in microstructure as the solidification front changes speed across the powder due to the recalescence (Ref 2). Alternatively, if solidification occurs through the propagation of a fine dendritic network across the powder, significant coarsening of the entire network will occur during recalescence. While it is clear that most nucleation processes are heterogeneous, rarely have the heterogeneities been identified with any certainty. Thus, a prior knowledge of the nucleation temperature of an alloy is as difficult to determine as the heat-transfer coefficient, h. One trend is apparent, however. Initial undercooling generally increases with decreasing particle size. By dispersing a liquid sample into a large number of small, independent drops, only a small fraction of these drops will contain the potent nucleants, so that the majority of the drops can display a large undercooling. Based on experience with droplet emulsion samples (Ref 3), supercooling effects become measurable for size refinement below approximately 100 mm diameter and can become appreciable for powder sizes less than approximately 10 mm. This general trend is also reflected (Ref 2) in atomized aluminum-iron powder.

Thermodynamic Constraints/LiquidSolid Interface Conditions The process of solidification cannot occur without a slight deviation from equilibrium. However, it is clear that different degrees of departure from equilibrium constitute a hierarchy that corresponds to increasing solidification rate (Ref 4): full diffusional (global) equilibrium, local interfacial equilibrium, metastable local interfacial equilibrium, and interfacial nonequilibrium. Full Diffusional (Global) Equilibrium. For global equilibrium, chemical potentials and temperature (and pressure) are uniform throughout the system. Global equilibrium is invoked at each time instant during cooling for descriptions of solidification that apply the lever rule. At each instant, the system has a uniform temperature and each phase has a uniform composition given by the phase diagram. As time progresses, the temperature drops, the compositions of the phases change, and the fractions of liquid and the various solid phases are given by the lever rule, a simple consequence of mass balance. Local Interfacial Equilibrium. During most solidification processes, gradients of temperature and composition exist within the phases. Thermal and solute diffusion models are then used to describe the spatial and temporal changes in temperature and composition within each phase. The equilibrium phase diagram is employed to give the possible temperatures and compositions for boundaries between the phases, for example, at the solidification interface. The widely used liquidus slope, mL, and equilibrium partition coefficient, kE, are defined in the context of the local equilibrium assumption. The Gibbs-Thomson effect (Ref 5) is included in the local equilibrium assumption. It describes shifts in equilibrium conditions due to the curvature of the liquid-solid interfaces and typically becomes more important when fine microstructures develop from the melt, as in rapid solidification. Interface equilibrium is widely used to model the majority of solidification processes that occur in castings. For example, under the assumptions of fast diffusion in the liquid phase, negligible diffusion in the solid phases, and local equilibrium at the liquid-solid interface (ignoring curvature), the Scheil equation (see the article “Thermodynamics and Phase Diagrams” in this Handbook) gives a reasonable first approximation to the nonequilibrium dendritic coring or microsegregation in conventional castings. Standard dendritic and eutectic theories use this condition with the inclusion of the Gibbs-Thomson effect. Metastable Local Interfacial Equilibrium. Metastable equilibrium can also be relevant globally or locally and is important in ordinary metallurgical practice as well as in rapid solidification (Ref 6). For example, one can understand the microstructural change of cast iron from the stable gray form (austenite and graphite) to the metastable white form (austenite and cementite) with increasing solidification rate

388 / Principles of Solidification

T Ma (A)

L

a

a (B) TM

L a

Temperature

simple eutectic system where the eutectic reaction is L ! b + g. For example, the composition marked “a” may freeze with a primary b phase, followed by a eutectic of b and g. Another possibility would be if the b phase failed to form instead. The peritectic reaction, L + a ! b, would be absent, and primary solidification of a would continue until the eutectic forms between a and g. For alloy “b” in the case where b fails to form, the microstructure may be a eutectic structure of a + g forming at a lower temperature than the stable eutectic temperature. Interfacial Nonequilibrium. For extremely rapid liquid-solid interface freezing rates (1 m/ s velocity, or 3 ft/s), the local interfacial equilibrium assumption breaks down. Solute can be trapped into the freezing solid at levels exceeding the equilibrium value of solid for the corresponding liquid composition present at the interface, as given by the phase diagram. The chemical potential of the solute increases upon being incorporated into the freezing solid in a process called solute trapping. This increase in chemical potential of the solute across the interface is offset by a decrease in

Temperature

using information from the stable and the metastable phase diagrams combined with a eutectic growth analysis (Ref 7). The eutectic temperature and composition for white cast iron are well-defined thermodynamic quantities, just as they are for gray cast iron. Metastable equilibrium at any temperature is given by a common tangent construction to the molar free energy versus composition curves for the liquid, austenite, and cementite phases and thus minimizes the free energy as long as graphite is absent. When solidification is complete, a two-phase mixture of austenite and cementite can exist in a global metastable equilibrium. The concept of local metastable equilibrium is especially important during rapid solidification because some equilibrium phases, especially those with complex crystal structures, have sluggish nucleation and/or growth kinetics and are absent in rapidly solidified microstructures, exposing other phases in the free-energy landscape. Figure 2(a) shows a hypothetical phase diagram involving four phases: liquid, L, and three solid solution phases, a, b, and g. Figure 2(b) shows the extrapolation of the liquidus and solidus to the pure components A and B. Six melting points are indicated for pure A and for pure B in the a, b, and g structures. If the a phase has difficulty nucleating and/or growing, the relevant phase diagram would exclude the liquidus and solidus curves for this phase. Solidification then could be modeled and understood with the (metastable) phase diagram based on the remaining liquidus and solidus lines. This is a

g

Tg

g Composition

b

(a)

L B

Composition

(a) a

Temperature

T Ma (A) TMb (A)

L

T Ml (B)

b

a

g

Diffusional Kinetics/Microsegregation Tg

TMb (B)

b

Composition (b) T Ma (B) Composition

B

(b)

Fig. 2

a

g

TMg (A) A

Temperature

A

(a) Stable phase diagram of a hypothetical system involving four phases: L (liquid), a, b, and g. (b) A superposition of three metastable phase diagrams: L-a, L-b, and L-g. Alloy compositions “a” and “b” are discussed in the text.

Schematic representation of T0 curves (dotdashed line) for the liquid-to-solid (crystal) transformations without solute segregation, L ! a and L ! b. For (a) a transition to partititionless solidification during cooling is thermodynamically possible before the liquid reaches the glass transition temperature (Tg). Supersaturation of the a or b phases is more likely than glass formation. For (b), a transition to partititionless solidification during cooling is impossible for a range of compositions before the liquid cools to the glass transition temperature. Glass formation is more likely in (b).

Fig. 3

chemical potential of the solvent in order for crystallization to occur, that is, to yield a net decrease in free energy (Ref 8). Additional kinetic assumptions may be used to define a liquidus slope and a partition coefficient that depend on interface velocity for model development. In particular, the partition coefficient may go to unity so that a solid may form with the same composition as the liquid without any need for solute diffusion in the liquid (partitionless solidification). This extreme case is only possible if the interface temperature is below the T0 curve, that is, the temperature and composition where the molar free energies of liquid and solid are equal. It places a bound on this solute trapping and can prevent the partition coefficient from going to unity, no matter what the velocity is. The T0 curve can be superimposed on the phase diagram and lies between the liquidus and solidus, as shown in Fig. 3. Crystalline solid phases cannot be formed from the melt with concentrations beyond the T0 curve (to the liquid side), no matter how much nonequilibrium is involved. If the T0 curves of the two solid phases of a binary eutectic plunge steeply, leaving a region of composition where no single crystalline phase can form (Fig. 3(b)) a liquid alloy in the center of such a phase diagram can only crystallize into a mixture of solid phases with different compositions, regardless of the departure from equilibrium. The diffusion kinetics of this separation from the liquid phase (eutectic growth) frequently is slow compared to partitionless solidification and depresses the solidification temperature to near the glass transition temperature, Tg. Here, an increased liquid viscosity effectively halts crystallization, and a metallic glass will form. In contrast, alloys with T0 curves that are only slightly depressed below the stable liquidus curves make good candidates for solubility extension and unlikely ones for glass formation. Partitionless solidification is thermodynamically possible in this case. Silver-copper is an example of an alloy system where alloys of eutectic composition can form a single facecentered cubic phase of eutectic composition.

Morphological Stability. The avoidance of constitutional supercooling for dilute alloys by solidification at slow velocity and high thermal gradients in various crystal growth and/or directional solidification apparatuses is well known. Solidification takes place with a planar interface and no microsegregation. Less well known is the fact that the full morphological stability theory of the liquid-solid interface predicts a similar result for freezing at very high solidification velocity independent of the temperature gradient. This is called absolute stability. Potentially unstable wavelengths of the perturbed interface shape become very small, inducing large curvatures where the Gibbs-Thomson effect stabilizes a

Rapid Solidification / 389 planar interface. For Ag-1 and 5 at.% Cu alloys, a transition from cellular structures to plane-front growth was obtained by increasing the growth velocity beyond 0.3 and 0.6 m/s (1 and 2 ft/s), respectively (Ref 9). With increasing freezing range on the phase diagram, corresponding to different alloy compositions, the velocity needed for plane-front solidification increases. These silver-copper alloys have relatively small freezing ranges, and the velocities are low enough that it was not necessary to invoke nonequilibrium interface conditions in the theory. In order to obtain plane front for alloys with larger freezing ranges, a modification of the theory to include solute trapping was necessary (Ref 10). Indeed, due to the thermodynamic constraints mentioned previously (position of T0 curves), plane-front solidification into crystalline material with no microsegregation of some alloys is not possible. Cellular and Dendritic Growth. At intermediate growth velocities, cellular and/or dendritic growth remain common occurrences in rapidly solidified material. An example of an analysis assuming local equilibrium is given by Ref 7, which computes how the dendrite tip radius, r, changes with dendrite tip velocity, V (or bulk supercooling), as well as how the tip temperature and tip liquid and solid concentration vary with dendrite tip velocity. The general trend for the tip radius follows Vr2 = constant, the constant depending on the alloy composition, C0. However, at both low and high velocities, that is, near the constitutional supercooling and absolute stability velocities, the tip radius approaches infinity (planar interface) and deviates from the Vr2 = constant law. At intermediate velocities, typical of casting situations, the tip temperature is only slightly below the liquidus temperature for C0 and approaches the solidus temperature for C0 at low and high velocities. Also intermediate velocities, the tip solid composition (which is also the concentration at the center of a solid dendrite) is very close to the solidus composition, kEC0. This latter situation is the basis of the nonequilibrium Scheil approach to modeling microsegregation. However, the tip composition increases toward C0 as the velocity approaches the constitutional supercooling or absolute stability velocities. Composition at the core of the dendrite. This effect reduces the overall microsegregation and the amount of second solid phase that needs to form in the interdendritic region to complete solidification. In the limit of extreme rates, dendrites can fail to grow, and a fine cellular or even plane-front solidification can occur. Eutectic Growth. In traditional eutectic theory, the solution to the diffusion equation in the liquid ahead of the growing two-phase solid is valid only when the Peclet number, lV/2D, is much smaller than unity (eutectic spacing, l; velocity, V; and liquid diffusion coefficient, D).

A result of the traditional approach is that l2V = constant. Trivedi et al. (Ref 11) recomputed the solute field in the liquid in front of a growing eutectic when the Peclet number is large while maintaining the local equilibrium assumption. The analysis destroys the constancy of l2V in a way that depends strongly on the Peclet number, the volume fractions of the solid phases, the shape of the metastable extensions of the liquidus and solidus curves below the eutectic temperature, and the partition coefficients, kE. Two cases are distinguished, depending on whether the two kE values are (a) close to unity or (b) close to zero. As the velocity is increased in case (a), the eutectic interface temperature is found to approach the metastable extension of the solidus curve for one of the constituent solid phases. As the velocity is increased, the eutectic spacing actually increases with increasing velocity. Indeed, eutectic solidification is replaced by single-phase planar growth at high velocities. In case (b), the interface temperature cannot reach a metastable solidus curve of either phase. The supercooling becomes so large that the temperature dependence of the diffusion coefficient has a major influence, and the spacing decreases with velocity faster than predicted by a constant l2V value. In both cases, there exists a maximum velocity for coupled eutectic growth. The eutectic microstructure is replaced in case (a) by single-phase growth of one of the phases, whereas in case (b), glass formation is possible if the interface temperature reaches the glass transition temperature, where the melt viscosity (diffusion coefficient) plummets (Ref 12). In fact, the cases where the kE’s are close to zero are the same as those that would exhibit plunging T0 curves and lead to glass formation, as described previously. Case (b) may also lead to the formation of a metastable crystalline phase if the eutectic interface temperature drops below the liquidus for such a phase. Another important factor in alloys containing a eutectic is the shifting of the coupled zone at elevated solidification speed. The alloy composition where fully eutectic microstructures can be found may shift to compositions not including the eutectic composition. Such analysis is required to describe the hypereutectic aluminum alloys mentioned previously (Ref 7). REFERENCES 1. I.R. Hughes and H. Jones, J. Mater. Sci., Vol 11, 1976, p 1781 2. W.J. Boettinger, L.A. Bendersky, and J.G. Early, An Analysis of the Microstructure of Al-8 wt%Fe Rapidly Solidified Powders, Metall. Trans. A, Vol 17, 1986, p 781–790

3. J.H. Perepezko and I.E. Anderson, in Synthesis and Properties of Metastable Phases, E.S. Machlin and T.S. Rowland, Ed., TMSAIME, 1980, p 31 4. W.J. Boettinger and J.H. Perepezko, Fundamentals of Solidification at High Rates, Rapidly Solidified Alloys—Processes, Structure, Properties and Applications, H. Lieberman, Ed., Marcel-Decker, 1993, p 17–78 5. G. Lesoult, Fundamentals of Growth, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 109–113 6. J.H. Perepezko and W.J. Boettinger, Use of Metastable Phase Diagrams in Rapid Solidification, Alloy Phase Diagrams, L.H. Bennett, T.B. Massalski, and B.C. Giessen, Ed., Mater. Res. Soc. Symp. Proc., Vol 19, Elsevier North Holland, NY, 1983, p 223–240 7. H. Jones and W. Kurz, Metall. Trans. llA, 1980, p 1265 8. J.C. Baker and J.W. Cahn, The Thermodynamics of Solidification, Solidification, American Society for Metals, 1971, p 23 9. W.J. Boettinger, D. Shechtman, R.J. Schaefer, and F.S. Biancaniello, The Effect of Solidification Velocity on the Microstructure of Ag-Cu Alloys, Metall. Trans. A, Vol 15, 1984, p 55–66 10. W.J. Boettinger, S.R. Coriell, and R.F. Sekerka, Mechanisms of Segregation-Free Solidification, Mater. Sci. Eng., Vol 65, 1984, p 27–36 11. R. Trivedi, P. Magnin, and W. Kurz, Acta Metall., Vol 35, 1987, p 971 12. W.J. Boettinger, The Effect of Alloy Constitution and Crystallization Kinetics on the Formation of Metallic Glass, Rapidly Quenched Metals IV, T. Masumoto and K. Suzuki, Ed., The Japan Inst. of Metals, 1982, p 99

SELECTED REFERENCES  H. Biloni and W.J. Boettinger, Solidification,

Physical Metallurgy, 4th ed., P. Haasen and R.W. Cahn, Ed., North Holland, Amsterdam, 1996, p 669–842  W.J. Boettinger, L.A. Bendersky, R.J. Schaefer, and F.S. Biancaniello, On the Formation of Dispersoids during Rapid Solidification of an Al-Fe-Ni Alloy, Metall. Trans. A, 19, 1988, p 1101–1107  W. Kurz, B. Giovanola, and R. Trivedi, Acta Metall. Vol 34, 1986, p 823  J.H. Perepezko, in Casting, Vol 15, Metals Handbook, 9th ed., ASM International, 1988, p 101

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 390-397 DOI: 10.1361/asmhba0005227

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Solidification During Casting of Metal-Matrix Composites Pradeep Rohatgi and Benjamin Schultz, University of Wisconsin—Milwaukee Nikhil Gupta, Polytechnic Institute of New York University Atef Daoud, Central Metallurgical Research and Development Institute

METAL-MATRIX COMPOSITES (MMCs) are engineered combinations of two or more materials in which reinforcing phases are dispersed in a metal or an alloy. Structurally, cast MMCs consist of continuous or discontinuous fibers (designated by the subscript f), whiskers (w), or particles (p) in a metal or an alloy that solidifies in the restricted spaces between the reinforcing phase (or phases) to form the bulk of the matrix. There are several cast materials, such as aluminum-silicon alloys and cast irons, that exhibit two-phase microstructure in which the volume and shape of the phases are governed by phase equilibria and that have a long history of foundry production. Modern cast MMCs differ from these traditional materials in that any selected volume, shape, and size of reinforcement can be introduced into the matrix, even beyond the permissible limits presented by the phase diagrams. By carefully controlling the size, shape, surface properties, volume fraction, and distribution of the reinforcement and by controlling the solidification conditions, MMCs can be synthesized having a tailored set of useful engineering properties. For example, as shown in Table 1, combinations of very high specific strength and specific modulus, beyond those of conventional monolithic alloys, are achievable. Table 2 summarizes the applications of several commonly used MMCs and the properties that make them suitable for

their applications. Microstructural design and synthesis procedures have been developed to achieve unique combinations of properties, including improved elevated-temperature and fatigue strengths, increased damping ability, tailored electrical and thermal conductivities,

reduced wear rates, and targeted coefficients of thermal expansion. These tailored properties provide opportunities for a variety of new applications for MMCs that were not possible using conventional materials. For more information about MMC types and their applications, see the

Table 1 Specific strength and specific modulus of some metal matrices, reinforcements, and metal-matrix composites Material

Amount of fiber reinforcement, vol%

Specific strength, N
m/kg

60 60

20,000 4,986–6,000

7.59  10 7 4.406  10

35 35 38 30 ... ... 100 100 100 100 100 100 100

45,337 10,622 28,333 28,163 11,481 17,143 78,431 5 6.67  10 50,000 5 1.538  10 5 1.618  10 59,459 14,974

7.77  10 ... ... 7 6.53  10 7 2.53  10 7 2.59  10 8 1.567  10 8 2.19 to 3.29  10 8 1.175  10 8 1.62  10 8 1.35  10 8 1.68  10 7 1.79  10

50 50

56,604 5,283

7.92  10 7 5.66  10

50 50

8,803 3,697

1.092  10 ...

Al-Li/Al2O3 0 90 Ti-6Al-4V/SiC 0 90 Mg/carbon (Thornel) Al/carbon 6061 Al 2014 Al SiC(f) SiC(w) Al2O3(f) B(f) C(f) Be(f) W(f) Al/boron 0 90 Al/SiC 0 90

Specific modulus, N
m/kg 7

7

7

8

Table 2 Selected potential applications of cast metal-matrix composites Composite

Applications

Special features

Aluminum/graphite Aluminum/graphite, aluminum/a-Al2O3, aluminum/SiC-Al2O3 Copper/graphite Aluminum/SiC Aluminum/glass or carbon microballoons Magnesium/carbon fiber

Bearings Automobile pistons, cylinder liners, piston rings, connecting rods Sliding electrical contacts Turbocharger impellers ... Tubular composites for space structures

Aluminum/zircon, aluminum/SiC, aluminum/silica Aluminum/char, aluminum/clay

Cutting tools, machine shrouds, impellers

Cheaper, lighter, self-lubricating, conserve Cu, Pb, Sn, Zn, etc. Reduced wear, antiseizing, cold start, lighter, conserves fuel, i mproved efficiency Excellent conductivity and antiseizing properties High-temperature use Ultralight materials Zero thermal expansion, high-temperature strength, good specific strength and specific stiffness Hard, abrasion-resistant materials

Low-cost, low-energy materials

...

Solidification During Casting of Metal-Matrix Composites / 391 article “Synthesis and Processing of Cast MetalMatrix Composites and Their Applications” in this Volume. The matrix alloy in cast MMCs solidifies in the restricted space between the reinforcing phase to form the bulk of the matrix. The solidification microstructure of the matrix is refined and modified due to the presence of fibers and/or particles, providing a possible means of controlling solute microsegregation within the microstructure, macrosegregation of reinforcements in the casting, and the grain size in the matrix. The microstructure of the matrix surrounding the reinforcements and the nature of the interface between reinforcements and the matrix are important contributors to the overall MMC properties. The reactions at the solid-liquid interface and the nucleation and growth of primary and secondary phases around reinforcements influence the microsegregation and nanosegregation of reinforcements within the matrix microstructure.

Incorporation of Reinforcements Most ceramics are not wetted or are poorly wetted by molten metals, so the intimate contact between reinforcement and alloy can only be promoted by artificially inducing wettability or by using external forces to overcome the thermodynamic surface energy barrier and viscous drag effects. Mixing techniques generally used for introducing and homogeneously dispersing a discontinuous phase in a melt are:

and long contact times promote wettability due to interfacial reactions, resulting in reduced contact angle between the ceramic phase and the melt. The work of adhesion between a ceramic and a melt decreases with increasing heat of formation of carbides. A high energy of formation for a stable carbide implies strong interatomic bonds and correspondingly weak interaction with melts. This leads to a high interfacial energy and a small work of immersion, resulting in poor wetting. High valence electron concentration generally implies lower stability of carbides and improved wettability of ceramics by metals. Hence, optimization of processing parameters based on thermodynamics and kinetics can result in higher-quality MMCs. Magnesium/graphite, aluminum/graphite, and several other fiber-reinforced composites are valuable structural materials because they combine high specific strength and stiffness with a near-zero coefficient of thermal expansion and high electrical and thermal conductivities. Wetting and bonding between the fiber and the metal in these systems is induced by the deposition of a thin layer of titanium and boron or SiO2 onto the fibers (Ref 9).

Solidification Processing of MMC In solidification processing of composites, liquid metal is combined with the reinforcement phase and solidified in a mold. The solidification processes of composite synthesis can be divided into two main classes: stir mixing and melt infiltration. (For more information, see Composites, Volume 21, ASM Handbook.)

Stir Mixing. In stir mixing, the reinforcement phase is added to the melt, and the mixture is stirred with the help of either a mechanical stirrer (Ref 10) or a high-intensity ultrasonic device (Ref 11, 12). This action disperses the reinforcing phase throughout the melt. The melt-particle slurry can be cast either by conventional foundry techniques, such as gravity, pressure die, or centrifugal casting, or by novel techniques such as squeeze casting (liquid forging), spray codeposition, melt spinning, or laser-engineered net shaping processing. Many of these casting techniques are covered in detail in other articles in this volume some typical squeeze casting parameters are shown in Table 3. Rohatgi et al. (Ref 7) have studied the thermodynamics and kinetics of transfer of particulate and fibrous reinforcements into the melt and the solidification of melt into a composite solid phase. This information has been used to design processes that are used to make different cast MMCs. Figure 1 shows typical microstructures of particle-dispersed composites produced by the stir-casting method (Ref 8). Stir mixing and casting is now used for largescale production of particle-reinforced MMCs (Ref 14, 15). Various metals, such as aluminum, magnesium, nickel, and copper, have been employed as the matrix, and a wide variety of reinforcements, such as SiC, graphite, SiO2, Al2O3, Si3N4, and ZrSiO4, have been added as reinforcements. The study of cast composites has added considerable understanding to the solidification of conventional monolithic castings, especially issues related to the effects of inclusions on the fluidity, viscosity, nucleation, growth, particle settling, and particle pushing.

 Addition of particles to a vigorously agitated,

fully or partially molten melt (Ref 1–4)

 Injection of discontinuous phase into the     

melt with an injection gun (Ref 5) Dispersion of pellets or briquettes into a mildly agitated melt (Ref 6) Addition of powders to an ultrasonically agitated melt Addition of powders to an electromagnetically stirred melt Centrifugal dispersion of particles in a melt Infiltration of loose or packed beds/preforms of reinforcements by molten alloys

Table 3 Typical casting conditions for SiC fiber-reinforced aluminum and aluminum alloy (Al-4.5Cu, Al-11.8Si, and Al-4.8Mg) composites Type of composite

Plate type Tubular type

Metal mold temperature, K

Time between pouring and pressurizing, s

573 573

10 13

Applied pressure MPa

ksi

Pressure time, s

Casting temperature, K

49 40

7.1 5.8

90 90

1073 1073

Source: Ref 13

Reinforcement-Metal Wettability Wetting between molten alloys and dispersoids is desirable from the standpoint of ease of fabrication, dispersion of particles, and property-performance relationships (Ref 7, 8). Improvement in wetting between the matrix and the reinforcing phase is also important to minimize or eliminate porosity and to help the reinforcement function as a nucleation catalyst for the matrix microstructure. The wetting properties of ceramics by liquid metals are governed by a number of variables, including heat of formation, stoichiometry, and valence electron concentration. High temperature

Fig. 1

Microstructures of typical cast metal-matrix composites made by stir casting. (a) A356/20wt%Al2O3. (b) A356/ 15wt%SiCp

392 / Principles of Solidification threshold pressure needed to initiate infiltration (Young’s equation) and the kinetics of infiltration (Darcy’s law) are in reasonable agreement with the experiments. Table 4 lists the wetting and threshold pressures for Al/SiC and Al/B4C composite systems. Recently, techniques of pressureless infiltration of ceramic preforms have been developed (Ref 25, 26) that allow casting of net-shaped composites.

In stir-mixed slurries of melts and reinforcements, it is important to retain adequate enough fluidity in the melts to make sound castings. Generally, the fluidity of alloys decreases with increasing additions of reinforcements, and this can require changes in mold design. Reciprocal relationships between fluidity and the apparent viscosity of melt-particle suspensions have been noted in aluminum-silicon alloys, ironcarbon-sulfur alloys, and Al-4.5Cu-1.5Mg alloy/2.5 wt% flaky mica composites. The fluidity values of particle-filled composite melts are generally adequate for making gravity-cast composites at low volume fraction of particles, up to approximately 30 vol%. Additions of silicon carbide, alumina, graphite, mica, and other ceramic particles to aluminum alloys cause a reduction in spiral fluidity. The spiral fluidity of these alloys decreases linearly with increasing particle surface area per unit weight. The fluidity of composite slurries decreases with decreases in the temperature. The fluidity of composite melts also depends on the shape, size, flocculation, and segregation of particles in the melt. Infiltration Process. In this process, liquid metal is infiltrated through the narrow crevices between the fibers or particulate reinforcements, which are either in a packed bed or arranged in a preform and fixed in space (Ref 16–18). High pressure or vacuum can be used to assist the infiltration process. As the liquid metal enters between the fibers or particles during infiltration, it cools and then solidifies, producing a composite. In general, the infiltration technique is divided into three distinct operations. The first step is the preform preparation, which involves assembling the reinforcement elements together into a porous body. The second step is the infiltration process during which the liquid metal permeates the preform. The last step is the solidification of liquid metal throughout the preform. Melt infiltration can be achieved with the help of mechanical pressure (Ref 19, 20), inert gas pressure, or vacuum (Ref 21, 22). The calculations of

Solidification Fundamentals

the chemical free energy change due to solute segregation near the particles. This finding suggests that the nucleation of primary phases on the particles is more favorable for the hypereutectic compositions than for hypoeutectic compositions. This is in agreement with the presence of primary silicon phases on the particles in hypereutectic aluminum-silicon alloy melts containing SiC or graphite particles (Fig. 2) (Ref 29) and the absence of primary a-aluminum on particles in hypoeutectic aluminum-silicon alloy melts.

Nucleation and Growth Effects

Dendrite Formation and Microsegregation

Aluminum-silicon and aluminum-copper alloys have been extensively used as matrix materials in a wide variety of cast MMCs containing graphite and ceramic particles, carbon and glass microballoons, and discontinuous or continuous ceramic fibers (Ref 27, 28). The microstructures of these MMCs show that the primary aluminum phase tends to nucleate in the interstices between the ceramic phase, unless surface modification is carried out to promote heterogeneous nucleation on the particle or the fiber surface. The thermal conductivity and heat diffusivity of particles are generally less than those of the melt, and during the cooling process, the temperature of the particles will be higher than that of the melt. In such cases, it is difficult for the primary phase to nucleate at the particle surfaces. Thermal analysis showed that the unreinforced alloys exhibited undercooling for primary-phase nucleation, whereas the composites generally did not show any significant undercooling. The grain size of the composites is often smaller than that of the unreinforced alloy under identical casting conditions. Solute diffusion is impeded during growth due to the barrier effects of particles. Therefore, the delayed growth from the melt gives additional time for the formation of nuclei, which can yield a refined structure. The potential of nucleation on SiC and graphite particles in aluminum-silicon alloy melts can be calculated by

The primary phase morphology and the distribution of second phases depend on the relative magnitudes of dendrite arm spacing and interfiber spacing. Figure 3(a) demonstrates the effect of fiber and platelet-type reinforcements on the solidification of aluminum-copper alloys (Ref 30). In the fiber-reinforced region, it can be seen that primary aluminum does not nucleate on the surface of fibers as expected. Primary aluminum dendrites grow outward from the region between the fibers, depositing the last freezing eutectic liquid on the fiber surfaces. A similar phenomenon occurs in SiC platelet-reinforced aluminum-copper alloys (Fig. 3b, c), where the Al-CuAl2 eutectic is enriched at the SiC platelet/matrix interface, and the primary phase nucleates away from the platelets (Ref 31). Microstructures in fiber-reinforced MMCs can be modulated in a predetermined manner by controlling interfiber spacing and cooling rate. Figure 4 (Ref 7) demonstrates microsegregation in the interfiber regions of an aluminum/ carbon fiber composite, in which the carbon fibers were chilled outside of the mold. In this case, very fine-sized a-phase grains were in contact with the surfaces of graphite fibers. If the cooling rate is sufficiently high or if fiber volume fraction is sufficiently low, the matrix alloy solidifies without influence from the fibers. At sufficiently slow cooling rates, when

Table 4 Wetting characteristics and threshold infiltration pressures for aluminum/SiC and aluminum/B4C systems Al/SiC Temperature Alloy

Pure aluminum

Al-2Cu

Al-4.5Cu Al-2Mg

Al-4.5Mg Al-2Si

Al-4.5Si Source: Ref 23, 24



C

700 800 900 700 800 900 800 700 800 900 800 700 800 900 800



F

1290 1470 1650 1290 1470 1650 1470 1290 1470 1650 1470 1290 1470 1650 1470

Al/B4C

Threshold infiltration pressure (Pth) 2

3

Surface energy (g), J/m  10

851 840 830 843 832 822 831 767 757 747 652 847 836 826 831

Threshold infiltration pressure (Pth)

Contact angle (u), degrees

kPa

psi

Contact angle (u), degrees

kPa

psi

106.3 102.3 101.2 106.7 103.7 100.2 103.0 104.6 101.2 99.3 102.1 105.7 103.6 99.1 103.2

917 686.1 620.6 934.3 758.5 558.5 717.1 744.7 565.4 462.0 524.0 882.6 737.8 503.3 730.0

133 99.5 90 135.5 110 81 104 108 82 67 76 128 107 73 106

120.0 115.6 109.4 116.7 114.3 110.1 112.9 115.1 98.9 95.1 92.9 ... ... ... ...

800 751.7 572.4 786.2 710.3 586.2 668.9 675.9 241.4 137.9 68.9 ... ... ... ...

116 109 83 114 103 85 97 98 35 20 10 ... ... ... ...

Solidification During Casting of Metal-Matrix Composites / 393  Quality of the melt-particle slurry prior         

Fig. 2

Optical micrographs showing nucleation of silicon phase (a) on graphite particle and (b) on nickel-coated graphite particles in hypereutectic aluminum-silicon alloys

to casting Cooling rate Viscosity of the solidifying melt Shape, size, and volume fraction of particles Specific gravities of particles and the melt Thermal properties of particles and the matrix alloy Chemistry and morphology of crystallizing phases and their interactions with particles Nucleation of primary phases on reinforcements Entrapment or pushing of particles by solidifying interfaces Flocculation of particles and the presence of any external forces during solidification

Figure 5 shows typical microstructures of particle-dispersed composites where the particles are segregated in the last freezing interdendritic regions as a result of pushing by the primary solid. The segregation of particles into the interdendritic regions causes severe agglomeration and interparticle contact, impairing the mechanical properties. Segregation is more severe at smaller particle sizes and at larger dendrite arm spacing (Ref 23). Research on particle pushing by solidifying interfaces in cast MMCs has been focused on modeling the effects of thermal properties (Ref 24), solute diffusion (Ref 13), gravity, and microgravity (Ref 36). Other areas of focus have included the effects of convection (Ref 37), external fields such as centrifugal force, morphology of the solidification front (Ref 38), and surface energies (Ref 39, 40) (Table 5). Kim and Rohatgi (Ref 41) have proposed models on particle pushing, taking into account the effect of interface shape, which is initially planar, and the presence of solute. The shape of the interface becomes curved after the interaction. The critical interface velocity in a pure melt, above which the particles are engulfed by the moving interfaces, is given by the following equation (Ref 13): Vc ¼

Fig. 3

(a) Transverse section of an SCS-2SiC fiber in an Al-4.5%Cu matrix. (b) and (c) Solidification microstructure of discontinuously reinforced SiC-Al alloy composites showing the influence of the spacing between SiC platelets on microsegregation pattern in aluminum-copper alloys

the secondary dendrite arm spacing in the unreinforced alloy is comparable to interfiber spacing, the grain size becomes large compared to interfiber spacing. In this case, fibers do not enhance the nucleation of the solid phase. With a further decrease in the cooling rates, the extent of microsegregation is reduced, and at sufficiently slow cooling rates, the matrix can be rendered free of microsegregation. The underlying mechanisms range from restricted

diffusion in the solid state to dendrite coalescence during solidification (Ref 32–35).

Distribution of Reinforcements The spatial arrangement of the discontinuous ceramic phase in the cast structure principally determines the properties of the cast composite. The distribution of phases depends on:

  sa0 kR1 þ 1 18m R1

(Eq 1)

where Ds is the surface energy difference, a0 is the atomic diameter, m is the viscosity, k is the curvature of the interface, and R1 is the particle diameter. During the rejection of particles by growing crystals and pushing of reinforcements ahead of the advancing interface, a viscous force is generated that prevents the pushing of the particle. Therefore, it is the balance of these counteracting forces that determines the rejection or engulfment of the particle. The shape of the solidification front and the magnitude of these forces are affected by several parameters, such as relative density difference, particle size, relative difference in thermal conductivity and heat diffusivity between the particle and the matrix melt, and alloy composition. Figure 6 shows the variation in critical velocity for engulfment as a function of particle size,

394 / Principles of Solidification

Fig. 4

Solidifying microstructure of thermally managed Al-9%Cu alloy composite (a) with external cooling of graphite rod extending out of the melt and (b) without external cooling of the graphite rod

suggesting that smaller particles will require higher velocities for engulfment. The interaction between particles and the solidification fronts was observed to lead frequently to segregation of particles in the last freezing interdendritic regions. This happens when the velocity of solidification fronts is lower than the critical interface velocity, an event that depends on various factors, including particle size and its density, melt viscosity, and the shape of the interface and its velocity. An interface velocity greater than the critical interface velocity leads to the engulfment of particles by the interfaces, resulting in a uniform particle distribution due to the presence of the particles within the matrix, an outcome that is likely to greatly enhance the matrix mechanical properties. The critical interface velocities have been experimentally measured in Al-Si/ SiC composites; however, it has not been possible to engulf the majority of the reinforcement particles within the dendrites in most conventional casting processes used today (2008).

and quantified its dependence on the thermal conductivities of the particle and the melt. More recently, Garvin et al. (Ref 49, 50) developed a multiscale method in which they coupled the dynamics of heat and fluid transport at the microscale (i.e., particle scale) with the nanoscale intermolecular interactions acting across the thin melt gap, without imposing any assumptions on the geometry of the front/particle or on the force laws. The solution requires solving a lubrication equation (with disjoining pressure effects for the intermolecular forces included as a body force) for the melt layer and Navier-Stokes equations for the overall particle-front system. The results from these studies allow the critical velocity for particle engulfment to be obtained from the coupled dynamics. The sensitive interaction between the particle and the melt front leads to the determination of the critical velocity for pushing/ engulfment in particle-front interactions.

Modeling of Particle-Pushing Phenomenon

In the work of Yang et al. (Ref 51–53), interactions of fronts in undercooled pure melts and directionally solidified (constitutionally supercooled) alloy melts have been studied. These results indicate that the dynamics of particlefront systems in such morphologically unstable situations may be very different from those in the often-studied stable planar solidification front cases. The thermal conductivity ratio of the particle to the melt (kp/kl) seems to play an opposite role in the case of an undercooled melt (pure material) versus a directionally solidified pure material (Ref 53). It appears that in dendritic systems with kp/kl < 1, particles are likely to be engulfed or become trapped in the interdendritic spaces as opposed to being pushed ahead of the front. As a dendrite approaches a particle in an undercooled pure melt, it becomes more convex as kp/kl increases. In the directional solidification case of a pure material, the solidification front becomes less convex as kp/kl increases.

Particle/solidification front interactions in metal-ceramic systems have been modeled using computational methods. The studies showed a distinct difference between the results obtained from a full dynamic analysis and those from previous steady-state analyses (Ref 42–45), particularly when premelting effects were included in the calculations. A significant finding was that when premelting effects are included in the study, the particle pushing/engulfment transition could be determined directly from the dynamics of the problem (Ref 46). The premelted film induced changes in the curvature of the solidification front underneath the particle (i.e., becomes more concave), which promoted engulfment of the particles. Garvin and Udaykumar (Ref 47, 48) computationally examined the drag force acting on the particle as it is approached by the front

Dendrite and Particle Interaction Fig. 5

Microstructures of composites showing pushing of ceramic particulates in the last freezing region liquid. (a) 7074 Al-20vol%Al2O3 particle matrix. (b) A356 alloy-based composite with Al2O3 (c) A356 alloy-based composite with ZrO2. Source: Ref 60

Nanoparticles and Solidification Front Interactions Solidifying interfaces interacting with nanoparticles will likely require a different numerical approach than the ones used for microparticles. The lubrication approach to finding the drag on the particle may not be valid for the case of a nanoparticle. In order for the lubrication theory to hold, the ratio of the gap thickness to the particle radius must be small (d/Rp  0.1). In addition, the curvature of the nanoparticle will have a first-order effect on the solidification-front morphology. At high curvatures, the GibbsThomson condition will become very important and tends to flatten the solidifying interface. This will aid in the engulfment of particles.

Solidification During Casting of Metal-Matrix Composites / 395 Table 5

Selected theoretical models of particle pushing

Study

Reported parameters and outcomes

Uhlmann et al. Bolling and Cisse Hoekstra and Miller Chernov et al. Stefanescu et al. Neumann et al. Gilpin Catalina and Stefanescu Sasikumar et al. Kaptay Kim and Rohatgi Rempel and Worster Hadji

Repulsive (surface energy) and attractive (viscous drag) forces are introduced; front shape assumed a priori More rigorous determination of front shape than Uhlmann et al.; surface roughness considered; introduced gravity and curvature-dependent attractive (drag) force; some ill-defined length scales; obscure nature of rejection force Pushing by ice front modeled; rejection force is due to temperature-dependent transition layer on size; temperature gradient is included Disjoining pressure as repulsive force and fluid drag as attractive force; shape-preserving paraboloidal growth front; considers smooth particles and thermal conductivities of particle and melt; neglects kinetic undercooling; solute screening is considered separately Repulsive surface energy and attractive fluid drag; successive approximation-type approach to incorporate thermal conductivities of particle and melt, buoyancy, and volume fraction; a later refinement addresses growth of perturbations on front. Good agreement with some experimental observations Thermodynamic model of pushing; postulates an equation-of-state to determine surface energies. Eminently useful for low-energy (e.g., organic) systems Disjoining pressure as the repulsive interaction and drag as attractive force; temperature gradient effects included Proposed a dynamic model of pushing. The model considers acceleration of an initially stationary particle by the front to a steady state at which a thin film is retained. The critical velocity is the limiting velocity at steady state. Approach is similar to Chernov et al. but a more rigorous numerical solution of the problem; thermal conductivities of particle and melt are considered; front shape under both steady and nonsteady conditions is determined; assumes repulsive van der Waals and attractive fluid drag forces Determines conditions for force and spontaneous engulfment by accounting for the interfacial force. Neglects gravity, buoyancy, lift forces, and interfacial energy gradients. Derives an equation for the forces between two solids separated by a thin liquid film Steady-state model of interaction between a spherical particle and a curved interface. Considers thermal conductivity difference between particle and liquid, and the temperature gradient. Also incorporates solute effects in deducing the engulfment velocity Considers a power law dependence between film thickness and undercooling based on intermolecular interactions. Predicts the interface shape and critical velocity Analyzes the interface instability of a planar front confronted by a particle during unidirectional solidification of dilute suspensions. Defines a segregation coefficient that is modified by particles due to solute and thermal shielding

In addition, small particles are more easily pushed; hence, the solidification rates needed to engulf small particles could be quite high (Fig. 6). Brownian motion may also play a role in how a nanoparticle behaves as it interacts with the solidification front. Traditional continuum mechanics approaches may not be applicable for modeling nanoparticles interacting with solidification fronts, due to the small length and time scales. However, molecular dynamics approaches can be applied to such systems. Furthermore, while molecular dynamic calculations can reveal the physics at the nanoscale, a multiscale approach will be necessary to predict the particle distributions at the scales of practical interest, that is, at the macroscale.

Fig. 6

Fig. 7

Variation of the critical interface velocity with particle size

Transmission electron micrograph of A206/2vol %Al2O3 (47 nm) nanocomposite produced by the authors at the University of Wisconsin—Milwaukee

Nanocomposites A variety of metal-matrix nanocomposites (MMNCs) are being developed, with properties that far exceed the limits for metals or composites that contain microscale reinforcements. For example, carbon nanotubes have been shown to exhibit ultrahigh strength and modulus. When included in a matrix, they could impart significant property improvements to the resulting nanocomposite (Ref 54). Solidification processes have been applied for synthesizing nanocomposites (Ref 55). These processes can be divided into three categories: rapid solidification, stir mixing followed by solidification, and infiltration of melt into a preform of reinforcement followed by solidification. Rapid solidification, such as melt spinning, or spray atomization can lead to nanosized grains as well as amorphous metals, from which nanosized reinforcements can be precipitated in the amorphous matrix during heating to form nanocomposites (Ref 56, 57). The stir-mixing methods that have been applied to synthesize MMNCs include use of a high-temperature impeller to stir the meltreinforcement slurry and use of ultrasonic

mixing, where an ultrasonic horn is used to create cavitation in the melt that disperses the particulate reinforcements. Figure 7 shows a microstructure exhibited by a cast nanocomposite synthesized at the University of Wisconsin— Milwaukee (Ref 58, 59). This MMNC was made using a unique method combining the use of stir mixing and ultrasonic mixing, with a wetting agent added to the molten alloy to incorporate nanoparticles in a metallic matrix. This process resulted in the incorporation of nanoparticles within microscale grains of aluminum and formed a bimodal microstructure. Infiltration methods that have been used to cast MMNCs include use of ultrahigh-pressure and pressureless infiltration techniques. In the case of MMNCs, incorporation of as little as 1 vol% of nanosized ceramic particles has led to a much greater increase in the strength and wear resistance of aluminum- and magnesium-matrix composites than was achieved at much higher loading levels of microsized particles. Such improvements have great implications for the automotive, aerospace, and, in particular, defense industries due to the significant weight savings and exceptional properties that can be achieved.

Conclusions  Considerable progress has been made in the

field of cast MMCs since cast aluminumgraphite particle composites were first synthesized in 1965. Stir mixing and casting as well as pressure infiltration have emerged as the two major processes to make composites.  Presence of reinforcements in the melt influences solidification of castings in several ways, including the changes in the patterns of microsegregation that can reduce the heat treatment time required for castings. The knowledge base in the solidification of MMCs has enhanced the understanding of solidification of monolithic castings.

396 / Principles of Solidification  MMCs have unique properties, including high











specific strength, high specific modulus, wear resistance, low coefficient of thermal expansion, and high specific conductivity, which make them very attractive for automotive, electronic packaging, and space applications. Several MMC components are currently being used as automotive components, such as brake rotors, pistons, cylinder liners, drive shafts, and engine pulleys. They are also used in aerospace applications such as space shuttle orbiter struts, antenna wave-guide mast, microwave thermal packaging, power semiconductor base, ventral fins, eurocoptor blade sleeves, fuel access doors, and fan exit guide vanes. Emerging cast MMCs with considerable potential include aluminum-graphite, aluminum-fly ash, aluminum-silicon carbide-graphite, leadfree copper-graphite, and open-cell and syntactic composite foams. There are several research challenges in the field of cast MMCs, including development of special matrices and reinforcements, decreasing the cost and facilitating machining of composites, and generating handbook-grade data for design. Several fundamental issues include nucleation and growth amid reinforcements and interactions between solidifying interfaces. Micro- and nanosized reinforcements need further investigation to achieve the desired distribution of reinforcements in cast composites. In the future, the technology of casting MMCs can enable foundries to produce higher-performance castings and other advanced materials, including functionally gradient materials, nanocomposites, smart composites, biomedical implants, superconducting composites, and porous and cellular solids. There are exciting opportunities for producing exceptionally strong, wear-resistant cast MMNCs with acceptable ductility by solidification processing. However, low-cost bulk processing methods must be developed to synthesize these materials with little to no voids or defects and with improved ductility.

ACKNOWLEDGMENT Nikhil Gupta would like to acknowledge the National Science Foundation grant CMMI0726723. The authors thank Dr. Justin Garvin at the Air Force Research Laboratory for useful discussions. REFERENCES 1. A. Banerjee, M.K. Surappa, and P.K. Rohatgi, Preparation and Properties of Cast Aluminum Alloy-Zircon Particle Composites, Metall. Trans. B, Vol 14, 1983, p 273 2. P.K. Rohatgi, B.C. Pai, and S.C. Panda, Preparation of Cast Aluminum-Silica Particulate Composites, J. Mater. Sci., Vol 14, 1979, p 2277

3. M.K. Surappa and P.K. Rohatgi, Preparation and Properties of Cast AluminumCeramic Particle Composites, J. Mater. Sci., Vol 16, 1981, p 983 4. D. Nath, R.T. Bhat, and P.K. Rohatgi, Preparation of Cast Aluminum Alloy-Mica Particle Composites, J. Mater. Sci., Vol 15, 1980, p 1241 5. F.A. Badia and P.K. Rohatgi, Dispersion of Graphite Particles in Aluminum Castings Through Injection of the Melt, Trans. AFS, Vol 77, 1969, p 402 6. B.C. Pai and P.K. Rohatgi, Production of Cast Aluminum-Graphite Particle Composites Using a Pellet Method, J. Mater. Sci., Vol 13, 1978, p 329 7. P.K. Rohatgi, R. Asthana, and S. Das, Solidification, Structures and Properties of Cast Metal-Ceramic Particle Composites, Int. Met. Rev., Vol 31, 1986, p 115–139 8. M. Emamy, A. Razaghian, H.R. Lashgari, and R. Abbasi, The Effect of Al-5Ti-1B on the Microstructure, Hardness and Tensile Properties of Al2O3 and SiC-Containing Metal-Matrix Composites, Mater. Sci. Eng. A, Vol 485 (No. 1–2), 2008, p 210–217 9. H.A. Katzman, Carbon-Reinforced MetalMatrix Composites, U.S. Patent 4,376,808, 1983 10. D.M. Schuster, M.D. Skibo, and W.R. Hoover, Light Met. Age, Feb. 15, 1989 11. M.A. Liquin, C. Feng, and S. Guangji, J. Mater. Sci. Lett., Vol 14, 1995, p 649 12. V.G. Gorbunov, V.D. Parshin, and V.V. Panin, Russ. Cast Product., Vol 8 1971, p348 13. J.K. Kim and P.K. Rohatgi, An Analytical Solution of the Critical Interface Velocity for the Encapture of Insoluble Particles by a Moving Solid-Liquid Interface, Metall. Mater. Trans. A, Vol 29, 1998, p 351–358 14. Alcan Aluminum Corporation (Duralcan), U.S. Patent 4,786,467, 1988 15. T. Sridharan, Extended Abstract of Conference on Metal-Matrix Composites: Property Optimization and Application, U.K. IOM, 1989 16. A. Mortensen, J.A. Cornie, and M.C. Flemings, JOM, Vol 40 (No. 2), 1988, p 12 17. H. Fukunaga and K. Goda, Bull. JSME (Japan), Vol 28, 1985, p 235 18. A.A. Das and S. Chatterjee, Metall. Mater. Technol, March 1981, p137 19. P.S. Cook and P.S. Werner, Mater. Sci. Eng. A, Vol 144, 1991, p 189 20. S.K. Verma and J.L. Dorcic, Cast Reinforced Metal Composites, S.G. Fishman and A.K. Dhingra, Ed., ASM International, 1988, p115 21. Y. Kun, V. Dollhopf, R. Kochendorfer, Compos. Sci. Technol., Vol 46, 1993, p1 22. J. Yang and D.D.L. Chung, J. Mater. Sci., Vol 24, 1989, p3605 23. L. Kurian and R. Sasikumar, Acta Metall. Mater., Vol 40, 1992, p 2375 24. D.M. Stefanesfu and B.K. Dhindaw, Behavior of Insoluble Particles at the

25. 26. 27.

28. 29.

30.

31. 32. 33.

34.

35.

36.

37.

38.

Solid-Liquid Interface, Casting, Vol 15, Metals Handbook, 9th ed., ASM International, 1988, p 142–147 E. Breval, M.K. Aghajanian, J.P. Biel, and S. Anolin, J. Am. Ceram. Soc., Vol 76 (No. 7), 1993, p1865 M.K. Aghajanian and J.T. Burke, in Tomorrow’s Materials Today, SAMPE series, Vol 34, 1989, p 817 Y. Liu, P.K. Rohatgi, and S. Ray, Tribological Characteristics of Aluminum-50 vol pct Graphite Composite, Metall. Mater. Trans. A, Vol 24 (No. 1), 1993, pp 151–159 O. Couteau and D.C. Dunand, Creep of Aluminum Syntactic Foams, Mater. Sci. Eng. A, Vol 488 (No. 1–2), 2008, p 573–579 P.K. Rohatgi, C.S. Narendranath, and D. Wang, in Microstructure Formation during Solidification of Metal Matrix Composites, P.K. Rohatgi, Ed., TMS, 1992, p 149–160 A. Mortensen, M.N. Gungor, J.A. Cornie, and M.C. Flemings, Alloy Microstructure in Cast Metal-Matrix Composites, J. Met., March 1986, p 30 R. Asthana and P.K. Rohatgi, Z. Metallkd., Vol 84, 1993, p 44 H. Nakae and S.Wu, Key Eng. Mater., Vol 127–131, 1997, p503 J.A. Cornie, A. Mortensen, M. Gungor, and M.C. Flemings, The Solidification Process During Pressure Casting SiC and Al2O3 Reinforced Al-4.5% Cu Metal-Matrix Composites, Proceedings of the Fifth International Conference on Composite Materials (ICCM-V), W.C. Harrigan, J. Strife, and A.K. Dhingra, Ed., American Institute of Mining, Metallurgical, and Petroleum Engineers, 1985, p 809 L.J. Masur, A. Mortensen, J.A. Cornie, and M.C. Flemings, Pressure Casting of Fiber Reinforced Metals, Proceedings of the Sixth International Conference on Composite Materials (ICCM-VI), American Institute of Mining, Metallurgical, and Petroleum Engineers, 1987 A. Mortensen, M.C. Flemings, and J.A. Cornie, “Columnar Dendritic Solidification in a Metal-Matrix Composite,” Annual Report of the Centre for the Processing and Evaluation of Metal and Ceramic Matrix Composites, Massachusetts Institute of Technology, Feb 1987 D.M. Stefanescu, F.R. Juretzko, B.K. Dhindaw, A. Cartalina, S. Sen, and P.A. Curreri, Particle Engulfment and Pushing by Solidifying Interfaces, Part I: Microgravity Experiments and Theoretical Analysis, Metall. Mater. Trans. A, Vol 29, 1998, p 1697–1706 S. Sen, B.K. Dhindaw, A. Catalina, D.M. Stefanescu, and P.A. Curreri, Melt Convection Effects on the Critical Velocity of Particle Engulfment, J. Cryst. Growth, Vol 173, 1997, p 574–584 J.A. Sekhar and R. Trivedi, Solidification Microstructure Evolution in the Presence of Inert Particles, Mater. Sci. Eng. A, Vol 147, 1991, p 9

Solidification During Casting of Metal-Matrix Composites / 397 39. A.W. Neumann, R.J. Good, C.J. Hope, and M. Sejpal, An Equation of State Approach to Determine Surface Tensions of Low Energy Solids from Contact Angle, J. Colloid Interface Sci., Vol 69 (No. 2), 1974, p 291–302 40. R. Asthana and S.N. Tewari, Compos. Manuf., Vol 4, 1993 p 3 41. J.K. Kim and P.K. Rohatgi, The Effect of the Diffusion of Solute between the Particle and the Interface on the Particle Pushing Phenomenon, Acta Metall. Mater., Vol 46 (No. 4), 1998, p 1115–1123 42. G.F. Bolling and J. Cisse, A Theory for the Interaction of Particles with a Solidifying Front, J. Cryst. Growth, Vol 10, 1971, p 56–66 43. A.A. Chernov, D.E. Temkin, and A.M. Mel’nikova, Theory of the Capture of Solid Inclusions during the Growth of Crystals from the Melt, Sov. Phys. Crystallogr., Vol 21 (No. 4), 1976, p 369–373 44. D.K. Shangguan, S. Ahuja, and D.M. Stefanescu, An Analytical Model for the Interaction between an Insoluble Particle and an Advancing Solid/Liquid Interface, Metall. Mater. Trans. A, Vol 23, 1992, p 669–680 45. A.W. Rempel and M.G. Worster, The Interaction between a Particle and an Advancing Solidification Front, J. Cryst. Growth, Vol 205, 1999, p 427–440 46. J.W. Garvin and H.S. Udaykumar, Effect of a Premelted Film on the Dynamics of

47.

48.

49.

50.

51.

52.

Particle-Solidification Front Interactions, J. Cryst. Growth, Vol 290 (No. 2), May 2006, p 602–614 J.W. Garvin and H.S. Udaykumar, Drag on a Particle Being Pushed by a Solidification Front and Its Dependence on Thermal Conductivities, J. Cryst. Growth, Vol 267 (No. 3–4), July 2004, p 724–737 J.W. Garvin and H.S. Udaykumar, Drag on a Ceramic Particle Being Pushed by a Metallic Solidification Front, J. Cryst. Growth, Vol 276 (No. 1–2), March 2005, p 275–280 J.W. Garvin, Y. Yang, and H.S. Udaykumar, Multiscale Modeling of Particle-Solidification Front Dynamics, Part I: Methodology, Int. J. Heat Mass Trans., Vol 50 (No. 15–16), July 2007, p 2952–2968 J.W. Garvin, Y. Yang, and H.S. Udaykumar, Multiscale Modeling of ParticleSolidification Front Dynamics, Part II: Pushing-Engulfment Transition, Int. J. Heat Mass Transf., Vol 50 (No. 15–16), July 2007, p 2969–2980 Y. Yang and H.S. Udaykumar, Sharp Interface Cartesian Grid Method, Part III: Solidification of Pure Materials and Binary Solutions, J. Comput. Phys., Vol 210 (No. 1), Nov 2005, p 55–74 Y. Yang, J.W. Garvin, and H.S. Udaykumar, Sharp Interface Numerical Simulation of Directional Solidification of Binary Alloy in the Presence of a Ceramic

53.

54. 55. 56. 57. 58.

59.

60.

Particle, Int. J. Heat Mass Transf., Vol 51 (No. 1–2), Jan 2008, p 155–168 Y. Yang, J.W. Garvin, and H.S. Udaykumar, Sharp Interface Simulation of Interaction of a Growing Dendrite with a Stationary Solid Particle, Int. J. Heat Mass Transf., Vol 48 (No. 25–26), Dec 2005, p 5270–5283 J.P. Salvetat, G. Andrew, D. Briggs, T.M. Bonard, R.R. Basca, and A.J. Kulik, Phys. Rev. Lett., Vol 82, 1999, p 944 P. Rohatgi, N. Gupta, and B. Schultz, submitted to Int. Mater. Rev., 2007 S.F. Hassan, and M. Gupta, J. Alloy. Compd., Vol 429, 2007, p 176 B. Cantor, J. Metastable Nanocryst. Mater., Vol 1, 1999, p 143 B.F. Schultz and P.K. Rohatgi, Structure and Properties of a Stir Cast Aluminum Alloy A206-Al2O3 Nanoparticles Composite, submitted to Mater. Sci. Eng. A, 2007 B.F. Schultz, P.K. Rohatgi, S. Alaraj, and J.B. Ferguson, Influence of Processing Parameters on the Hardness of A206 Alloy-Al2O3 Nanoparticle Composites, submitted to Compos. Sci. Technol., 2007 A. Daoud, M. Abo-Elkhar, Journal of Material Processing Technology, Volume 120, Issues 1–3, 15 Jan 2002, p 296–302

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 398-401 DOI: 10.1361/asmhba0005228

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Solidification Research in Microgravity Martin E. Glicksman, University of Florida

GRAVITY has profound influences on most solidification and crystal growth processes. Gravity—the ubiquitous body force permeating all matter—is responsible for diverse fundamental phenomena, including hydrostatic pressure, sedimentation, and natural, that is, buoyancydriven, melt convection. The results of solidification and single-crystal growth are often assessed through the as-cast structure or crystal quality. Cast alloys, for example, are modified by altered patterns of nucleation, solute microand macrosegregation, degree and severity of gas and shrinkage porosity, fragmentation of dendrites, the as-cast grain size, grain shapes, and crystallographic texture. Even the detailed distributions of second-phase particles, including equilibrium and nonequilibrium precipitates, as well as the dispersion of mold wall particles and dross are affected by the magnitude and direction of gravity. Single crystals, by contrast, are judged by their structural perfection at different length scales, their spatial uniformity of electronic dopant levels, or refractive indexes at appropriate optical wavelengths. The force of gravity, g, encountered on Earth normally equals approximately 9.8 m/s2 (32 ft/s2), with its direction defining the local vertical. Modification of gravity over practical time scales for the purposes of modifying or controlling solidification was traditionally accomplished only in shot or spray towers, where molten alloys were solidified rapidly during free-fall, or when cast in vertical and horizontal centrifugal machines, where Earth’s gravity can be arbitrarily increased for the purposes of casting pipe, pressure vessels, and axisymmetric liners. On the other hand, decreasing the force of gravity, especially for extended times, as discussed in this article, proves to be a far more daunting and expensive technological challenge.

commercial and academic solidification experiments. These ground-based facilities include instrumented vacuum research drop towers/ tubes providing reduced gravity for approximately 4.5 s at 10 4 g. In the United States, much deeper drop shafts were built at the National Aeronautics and Space Administration (NASA) Glenn Research Center in Cleveland, Ohio (150 m, or 500 ft), and in Japan at the Kamisunagawa mine shaft, Hokkaido (710 m, or 2330 ft). Ground facilities, including deep shafts, are limited, nevertheless, to approximately 10 s of reduced gravity by free-fall. Specially equipped piloted aircraft, flying sequential parabolic paths, are also available for experiments that require no more than approximately 20 s at 10 2 g. Suborbital sounding rockets, a rather expensive option, include NASA’s earlier Space Processing Applications Rocket and the European Space Agency’s (ESA) TEXUS sounding rockets, launched from northern Sweden, which extend access to reduced gravity with recoverable payloads to nearly 300 s at 10 4 g. All of these microgravity options, however, preclude all but the briefest solidification studies in reduced gravity. Solidification researchers needing extendedduration microgravity access require rocketdriven space flight into low-Earth orbits— defined as orbits with apogees between 150 and 300 nautical miles above the surface of the Earth. America’s first space station, Skylab,

which was operated by NASA in the 1970s, accommodated small-volume, low-power materials and fluids experiments and set the stage for the Space Shuttle era beginning in the 1980s. Specifically, three of NASA’s Space Shuttle orbiters, Challenger, Columbia, and Discovery, were outfitted with a variety of hardware configurations allowing autonomous, semiautonomous, and crew-mediated microgravity research for times extending to 10 to 16 days. Shuttle-based platforms provided quasi-static (steady) gravity levels as low as 10 6 g, which constitutes true microgravity. Experimental payloads having small upmass, operating on self-contained battery power, and needing minimal telemetry for command and control functions were accommodated on Shuttle missions as “locker” experiments and so-called “get-away specials.” Small microgravity experiments could also be operated by trained astronauts in the Middeck Glovebox. The most robust configurations for microgravity solidification experiments, however, were 1) the United States Microgravity Laboratory (USML) residing within the SpaceLab, a pressurized Shuttle module built by ESA and manned by trained astronauts and mission specialists (Fig. 1, left); or 2) the United States Microgravity Payload (USMP), which consisted of a group of three to five stand-alone experiments fastened to a trusslike frame, called the multipurpose experiment support structure

Availability of Reduced Gravity: Past, Present, and Future Short-duration free-fall research facilities, known as drop towers and drop tubes, are available worldwide in heights between a few meters and up to as high as 100 m (330 ft) for

Left: Astronauts at work on orbit in Spacelab. Right: United States Microgravity Payload-4 configuration of microgravity solidification experiments MEPHISTO, AADSF, and IDGE in the open Shuttle bay during STS-87 (see text for a description of experiments). Columbia is flying in a tail-down orientation in low-Earth orbit.

Fig. 1

Solidification Research in Microgravity / 399 (EMPESS), located in the unpressurized Shuttle bay (Fig. 1, right). The EMPESS provided kilowatts of power, triaxial accelerometers, and wideband telecommunications for all the on-board USMP experiments. The USML and USMP, as dedicated microgravity platforms, were both capable of accepting larger, much more complex solidification experiments, some employing programmable high-temperature furnaces, others requiring up-linked ground controls plus K-band video and S-band data down-links. The USML and USMP each flew four operational missions aboard Columbia between 1992 and 1997. Sadly, just five years after these programs were completed, orbiter Columbia, flying as STS-107, was destroyed, with loss of all crew, upon re-entry into the atmosphere on February 1, 2003. Lastly, the International Space Station (ISS), after a decade of phase 2 orbital assembly, comprises a net mass of more than 225,000 kg flying continual 90 min orbits around the Earth. The ISS allows access to more than 14,000 ft3 (400 m3) of pressurized, on-orbit working space and habitat. In principle, there is continuous access to microgravity environments aboard the ISS. Safety and cost concerns with NASA’s aging Shuttle fleet, however, now limit the ISS crew size to 120-day rotations or increments (Ref 1), each consisting of three astronauts, who remain busy mainly with construction of, and repairs to, the ISS. Construction costs of the ISS and its severely delayed completion schedule have resulted in elimination of NASA’s planned Materials Science Solidification Rack and the ISS furnace facility, without which most microgravity solidification experiments were cancelled after USMP and USML missions were completed some ten years ago. Excepting a few solidification studies to be discussed subsequently, all of which were carried out in the ISS Microgravity Science Glovebox in 2002, NASA-sponsored microgravity solidification research has all but ceased. With the planned phase 3 additions to ISS of the European Columbus module during 2008, and, eventually, attachment of Japan’s experimental

Fig. 2

The International Space Station depicted in its phase 3, fully operational configuration, with components assembled from the United States, Europe, Canada, Japan, and Russia

module, opportunities may arise again to pursue microgravity solidification research (Fig. 2).

Solidification Research in Microgravity Microgravity solidification research prior to 1990 is detailed in a comprehensive ASM Handbook review by P.A. Curreri, “Low-Gravity Effects During Solidification,” Casting, Vol 15, ASM Handbook, ASM International, 1988, p 147–158. Experiments in solidification science and technology performed subsequent to that review were carried out on Space Shuttle-based USML and USMP missions and, most recently, on the ISS. These studies consist of a variety of complex microgravity solidification experiments that involve pure metals, alloys, and semiconductors. Most, but not all, of the selected experimental studies were carried into orbit as primary space flight experiments that were granted precedence in the flight rules. That is, mission protocols and flight time-lines were arranged specifically to accommodate the microgravity levels, trajectories, antennae allocations, and gravitational orientations demanded by the scientific requirements of the individual experiments. A selective review of these solidification studies follows. In lieu of individual citations, the reader is instead referred to the voluminous on-line archives and the microgravity search engine maintained by NASA for every microgravity mission. The solidification experiments discussed here are therefore fully named, and the official NASA acronyms for them are included for the purposes of searching the world-wide web and, especially, NASA’s Microgravity Science and Applications Program Tasks and Bibliography data files (Ref 2, 3).

Key Solidification Experiments MEPHISTO. A series of solidification experiments named MEPHISTO (Material pour l’Etude des Phenomenes Interessant la Solidification sur Terre et en Orbite) were developed by the French Space Agency. MEPHISTO experiments were initially flown in 1992 on USMP-1, then in 1994 on USMP-2, in 1996 on USMP-3, and finally, in 1997 on USMP-4. The MEPHISTO facility was designed to study the influence of natural convection during directional solidification of metallic alloys. Microgravity reduces both sedimentation and convection, permitting observation of subtle interfacial phenomena normally masked, or complicated, by gravity. A comparison of space- and ground-based results obtained with MEPHISTO allowed specific evaluation of the effects of convection on solute redistribution during the directional solidification of alloys and the observation of transitions in interface morphology from planar to cellular. By using three identical samples that were alternately

melted and solidified together, simultaneous measurements could be made of the solid-melt interfacial growth speed and temperature, allowing probing of the interfacial attachment kinetics—a nonequilibrium measurement rarely accomplished prior to these experiments. Advanced Gradient Heating Facility. Several researchers conducted solidification experiments aboard Columbia using the Advanced Gradient Heating Facility (AGHF), a multizone furnace developed by ESA. Three aluminumindium alloys were directionally solidified in the AGHF to study the physics of solidification processes in immiscible monotectic alloys. Three samples, which differed in indium content, were processed at the same growth rate to permit comparison of their microstructures and to determine how indium was distributed during freezing in the aluminum matrix. Two of these alloys had compositions that could not be processed under steady-state conditions on Earth, because of gravitationally-driven convective instabilities and subsequent sedimentation of agglomerated molten indium. Another AGHF experiment used commercial aluminum-base samples to obtain insight into the mechanism of precipitate redistribution during solidification. Additional studies were geared toward enhancement of the fundamental understanding of the dynamics of insoluble particles at solid/liquid interfaces. The physics of the problem is of direct relevance to such areas as solidification of metal-matrix composites, management of casting defects such as inclusions and porosity, development of high-temperature superconductor crystals with superior current-carrying capacity, and the directional solidification of monotectic alloys. TEMPUS. The second flight of the International Microgravity Laboratory supported experiments from the United States and 12 other countries. The German-built levitation facility, Tiegelfreies Elektromagnetisches Prozessieren Unter Schwerelosigkeit (TEMPUS), permitted measurements of the thermophysical properties of high-temperature glass-forming and crystallizing metallic melts. Specifically, in space, TEMPUS allowed multigram specimens of alloys to be electromagnetically levitated, heated, and melted without any physical contact with container walls. Measurement of the static and dynamic shapes of molten samples levitated in microgravity provided precise noncontact high-temperature calorimetry, along with determination of the molten alloy mass density, viscosity, and surface tension. The TEMPUS levitation facility also allowed observation and quantitative assessment of such important solidification phenomena as crystal nucleation rates, glass formation temperatures, dendritic growth speeds, and grain size variations arising from melts solidified at extreme values of thermal supercooling (hundreds of degrees Kelvin). Particle Engulfment and Pushing by Solidifying Interfaces. The behavior of solid-melt interfaces upon encountering inert particulates is an important issue in casting technology.

400 / Principles of Solidification A microgravity solidification experiment named Particle Engulfment and Pushing by Solidifying Interfaces (PEP) was initially designated for flight aboard the ISS. PEP was flown in an early flight opportunity that developed on the Life and Microgravity Science Mission on-board Columbia in 1996. PEP experiments allowed observation of 0.5 mm (0.02 in.) diameter zirconia particles being pushed, or engulfed, by a moving solid-melt interface in directionally solidifying high-purity aluminum. The critical transition speed separating particle rejection from particle entrapment occurred in microgravity at approximately half the value (1 mm/s) compared with observations based on terrestrial solidification. The PEP experiments suggest that melt convection occurring adjacent to the moving solid-melt interface significantly influences particle-trapping dynamics. The Isothermal Dendritic Growth Experiment (IDGE) flew three times on Columbia missions USMP-2, 3, and 4. Its purpose was a fundamental evaluation of the physics of dendritic growth. The IDGE data included acquisition of high-resolution photographs and near-realtime images and videos of growing dendrites that were transmitted to the ground. Command and control signals for changing experimental parameters were sent via telemetry from a dedicated ground laboratory, greatly expanding the flexibility and capability of the orbiting unmanned hardware. The speed, size, and morphologies of dendritic crystallites growing from supercooled melts were measured postflight. The supercooling levels that were studied in the IDGE series covered the range typically encountered in industrial practices, such as casting, welding, and ingot solidification. Two test substances, body-centered cubic succinonitrile and face-centered cubic pivalic acid, were studied in detail. Data from several hundred experiments allowed assessment of solidification, including the steady-state dendritic tip radius, growth speed, and supercooling. Specifically, the basic conduction/diffusion theory of heat and mass transfer from a growing dendrite was confirmed. Surprisingly, scaling laws relating dendritic speed, size, and supercooling were shown to be independent of the magnitude of gravity, with terrestrial and microgravity experiments yielding indistinguishable results. Dendritic growth speeds, which are sensitive to gravity, proved to be insensitive to vibratory accelerations, termed “g-jitter,” arising from crew activities and machinery operating on the Shuttle. This finding contrasts with the extreme sensitivity to quasi-static gravity levels found in some slow crystal-growth experiments on metals and semiconductors. Advanced Gradient Heating Facility (AGHF). The AGHF, an ESA-provided Shuttle-based facility, was used to study the solidification of several monotectic aluminum-indium alloys containing different concentrations of indium. Microgravity influences the effective diffusion rate of a solute by slowing natural melt convection in both monotectic and eutectic

alloys. Convection normally enhances solute diffusion, thereby influencing the conditions under which coupled phase growth may occur, and also affects the as-solidified microstructural perfection in polyphase alloys. Solidification and Pore Formation. As mentioned earlier, budgetary exigencies eliminated both the ISS furnace facility and the materials science research rack, curtailing execution of many planned microgravity solidification experiments. Now, some ten years later, only three solidification experiments have been performed by ISS astronauts using the Microgravity Science Glovebox, (MSG), which provides a sealed environment aboard the ISS that allows manipulation of relatively small, self-contained experiments. The first two experiments carried out on ISS, increment 5, were Solidification Using a Baffle in Sealed Ampoules (SUBSA), and Pore Formation and Mobility Investigation (PFMI). SUBSA sought to determine how microgravity and a flow-suppressing mechanical baffle combine to reduce melt convection and affect dopant diffusion of tellurium and zinc during the steady-state solidification of indium-antimony crystals. SUBSA, which used transparent quartz ampoules to contain the semiconductor samples, also permitted both astronauts and ground-based scientists to observe semiconductor interface shapes during convection-free steady-state crystal growth and allowed visualization of ampoule wetting and melt flow. In addition, fluid mechanical computer simulations of the flows expected within the ampoules under both terrestrial and orbital conditions were compared to the flight observations. PFMI observed the formation, motion, and interactions of bubbles nucleated and grown in a solidifying ampoule of a well-characterized transparent alloy (succinonitrile and water). In microgravity, any gas bubbles produced during solidification are not swept to the melt surface by buoyancy forces and released but are free to remain within the melt, cling to the solidmelt interface, and, perhaps, be incorporated into the cast structure. The presence of such pores would be detrimental to the properties of the casting. Knowledge of solidification-produced bubble dynamics in microgravity could prove useful for understanding how to design casting processes, perhaps to be carried out in the future in reduced-gravity environments. Coarsening in Solid-Liquid Mixtures. The second solidification-related experiment performed in the MSG aboard ISS, increment 7, was named Coarsening in Solid-Liquid Mixtures (CSLM-2). Precursor experiments (CSLM-1) were flown on Space Shuttle missions STS-83 and STS-94. The primary purpose of the CSLM series was quantitative evaluation of phase coarsening rates under long-term microgravity during partial solidification of tin-lead alloys processed isothermally at 185  C (365  F). Kinetic and morphologic coarsening data were obtained in microgravity over a wider range of volume fractions than would

be possible on Earth, because terrestrial gravity causes sedimentation and gross separation of solid-liquid phase mixtures undergoing coarsening. Solid-melt alloy mixtures were partially melted and then held under carefully controlled isothermal conditions for extended times in ISS microgravity. Since all the key thermophysical characteristics of this binary alloy were already well characterized, the CSLM experiment allowed direct tests of predictions based on kinetic theories of Ostwald ripening.

The Future of Microgravity Solidification Microgravity science has matured since the early days of space flight, when researchers first recognized that they may be able to take advantage of the reduced gravitational environment of orbiting spacecraft to study, among other subjects, solidification and crystal growth phenomena. Spacelab-, Russian Mir station-, and Shuttle-based experiments included investigations of fluids, biotechnology, combustion, fundamental physics, and materials science and provided researchers a better understanding of some basic phenomena in these fields, some of which are difficult or impossible to study under terrestrial conditions. A deeper probing of these phenomena demands that critical microgravity experiments be performed for longer periods of time. The ISS in its final phase 3 configuration was designed as the premier long-term microgravity research platform in low-Earth orbit. The budgetary justification and design of the ISS, including its rich complement of research facilities, included the requirement that scientifically complex, long-duration experiments would be performed on-board under high-quality microgravity conditions. The ISS, now approaching completion, should provide researchers from many nations with almost continuous microgravity conditions in month-long periods, between which spacecraft dockings, orbital reboosts, maintenance, and repairs would be performed. The planned increase in experimental capability (mass, power, communications, microgravity quality, and time-on-orbit) should be conducive to gaining improved understanding of both terrestrial and microgravity phenomena. Although the ISS earned high marks for its programmatic purpose, design, and strategic planning, it has been criticized for its lack of results and accountability (Ref 4). Whether reality will eventually match expectation for the ISS remains to be seen.

REFERENCES 1. International Space Station Payload Information Source, http://pdlprod3.hosc.msfc. nasa.gov/, NASA, accessed May 2008

Solidification Research in Microgravity / 401 2. NASA Technical Reports Server, http:// naca.larc.nasa.gov/index.jsp?method=aboutntrs, NASA, accessed May 2008

3. NASA Experiment List—Alphabetical, www. nasa.gov/mission_pages/station/science/experiments/, NASA, accessed May 2008

4. ExpectMore.gov: International Space Station, www.whitehouse.gov/omb/expectmore/detail/ 10000348.2004.html, accessed May 2008

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 402-403 DOI: 10.1361/asmhba0005229

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Homogenization Sujoy Krishna Chaudhury, Honeywell International, Inc.

HOMOGENIZATION, in a broad sense, refers to processes designed to achieve uniform distribution of solutes or phases in a given matrix. During solidification of a melt, microsegregation occurs due to the redistribution of solutes as solute is rejected into the liquid. Uniform distribution of the solute elements is very important since it plays a vital role in the subsequent thermomechanical processing of the alloy. It is a common industrial practice to homogenize the cast product prior to hot working. Homogenization is a diffusion-controlled process, and its kinetics depend on several factors, such as temperature, diffusing distance (i.e., secondary dendrite arm space), diffusivity of solutes, dissolution rate, and so on. In general, cast nonferrous alloys (both heat treatable and non-heat treatable) and ferrous alloys are homogenized at temperatures 20 to 100  C (36 to 180  F) below the solidus temperature. Care should be taken to avoid incipient melting during homogenization, since it could cause severe distortion of parts and degrade the properties of the casting. Homogenization temperature and time are selected based on the composition of the alloy and thermal history of the casting. Slow cooling after homogenization ensures better workability. Homogenization literally refers to equalization of solute concentration in the single phase; however, it is often accompanied by one or several of the following diffusion-related processes: grain coarsening or recrystallization, precipitation of supersaturated elements, dissolution of unstable phases or precipitates, coarsening/ spheroidization of stable intermetallic phases, surface oxidation, hydrogen degassing, and pore generation and agglomeration. Before discussing various issues pertinent to homogenization, it is imperative to address the root cause for inhomogeneities in cast components.

Sources of Inhomogeneity Castings are inhomogeneous due to microsegregation (also commonly referred to as coring) and macrosegregation. The cause of microsegregation in an alloy is the difference between the thermodynamic equilibrium solubility of alloy elements in different phases that coexist subsequent to solidification. The solubility of

solute in the solid phase follows the solidus line under equilibrium conditions and increases with the decreasing temperature. This is combined with the inability of diffusion of solute in solid state to fully return the composition to its equilibrium level. As a result, during the early stage of solidification, the solid matrix phase is depleted of solute, and the liquid phase is enriched with solute. An example of microsegregation in an aluminum alloy is shown in Fig. 1(a), and the subsequent homogeneous microstructure (i.e., uniform solute distribution) after homogenization is shown in Fig. 1(b). The index of microsegregation (@) is a useful parameter to qualitatively evaluate homogeneity in a given matrix phase, where @ = (CM Cm)/ 0 0 Cm ); CM and Cm are the maximum and (CM minimum solute concentrations after homogenization treatment, respectively, and the superscript 0 refers to the solute concentration before homogenization treatment. Prior to any homogenization treatment, @ = 1, and after complete homogenization, @ = 0. The index of microsegregation depends on extrinsic parameters, such as temperature and time of homogenization, and intrinsic parameters, such as diffusivity and characteristic diffusion distance for the solute. In addition, homogenization depends on several factors, such as the relative stability of intermetallic phase(s) and the thermal history of the casting (i.e., solidification rate, casting process, etc.). A common approach to limiting microsegregation is through nonequilibrium processing, such as rapid solidification of melt. Processes that have relatively high solidification rates and result in reduced microsegregation include spray forming, melt spinning, atomization, vapor phase deposition, and so on.

 Increase reproducibility of thermomechani-

cal processing Some of the intermetallic phases may spheroidize or transform into more stable phases. In addition, controlled precipitation from the supersaturated matrix during homogenization plays an important role in subsequent processing of alloys to control recrystallization and grain growth. Other benefits of homogenization treatments are increased resistance to pitting corrosion, increased resistance to stress-corrosion cracking, improved ductility, and uniform precipitate distribution during subsequent aging.

Characterization Methods There exist several methods to investigate homogenization, both qualitatively and quantitatively. One simple way to observe microsegregation is through proper chemical etching of

Benefits of Homogenization Treatment It is nearly standard industrial practice to homogenize alloys before thermomechanical processing. Some key objectives of homogenization are to:  Improve workability  Improve mechanical properties by dissolving

nonequilibrium intermetallic phases(s)

Fig. 1

Optical micrographs of an aluminum alloy showing (a) microsegregation and (b) homogenized structure. Source: Ref 1.

Homogenization / 403 distance (i.e., arms spacing, grain size, etc.), and cooling conditions  Establishing the correlation between the fluid flow and associated microsegregation. This is complicated because flow occurs within the interdendritic spaces, between grains, and in the bulk liquid.  Accounting for phase transformations such as peritectic transformation in steel, eutectic transformation in aluminum alloys, and precipitation

Fig. 2

Simulated (line) and experimental (scatter point) data showing microsegregation of (a) chromium and (b) molybdenum in Fe-13.2Co-11.1Ni-3.1Cr-1.2Mo-0.23C steel. Source: Ref 2.

the alloy. Because of compositional variations, the regions of microsegregation etch differently from the matrix, thus producing contrast when viewed under an optical microscope. A common method to quantitatively determine microsegregation is the use of a scanning electron microscope with an x-ray microanalyzer (energy-dispersive x-ray spectroscope, or EDS). Figure 2 shows typical EDS scattered data of solute distributions across a dendrite due to microsegregation of chromium and molybdenum. However, this method has a limitation and becomes unreliable when the solute concentration in the matrix phase is low. The most accurate tool for determining microsegregation or homogeneity in an alloy is the local electrode atom probe (LEAP) microscope. It gives a three-dimensional distribution of atoms in a matrix phase and accurate solute concentration at specified locations. However,

there are only a few LEAP microscopes available, and the cost of the analysis is relatively high.

Computational Modeling Accurate quantitative prediction of microsegregation is challenging because of the following difficulties:  Quantifying the equilibrium solubility of each

phase as a function of temperature. This is traditionally done using partition coefficients, which is reasonable for low-solute concentrations. However, multicomponent phase diagrams are required for complex systems.  Solving for diffusion transport within the solid phase, which requires knowledge of diffusion coefficients for each element, diffusing

Despite the barriers, some success has been achieved at predicting the homogenization behavior in a few alloys. Reasonably accurate simulation of microsegregation of chromium and molybdenum in AerMet (Carpenter Technology Corporation) steel has been predicted using commercial software (Ref 2). The advantage of computer simulation is that it can save money and time. However, the major limitation is that the accuracy of the computational models has been validated only on a few selected alloys. In addition, it is necessary to provide thermodynamic, diffusivity, and interface mobility data for each alloy. These data are not easily available for all metallic systems. REFERENCES 1. R.K. Gupta, N. Nayan, and B.R. Ghosh, Met. Sci. Heat Treat., Vol 47 (No. 11–12), 2005, p 522–525 2. H.E. Lippard, C.E. Campbell, T. Bjorklind, U. Borggren, P. Kellgren, V.P. Dravid, and G.B. Olson, Microsegregation Behavior during Solidification and Homogenization of AerMet100 steel, Met. Mater. Trans. B, Vol 29, 1998, p 205–210

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 404-407 DOI: 10.1361/asmhba0005230

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Heat Treatment Sujoy Krishna Chaudhury, Honeywell International, Inc.

HEAT TREATMENT in the broadest sense refers to the postsolidification thermal processing (both heating and cooling cycles) to achieve a desired set of material properties. Fortunately, the properties of most commercial alloys can be tailored by selecting a suitable heat treatment. In this article, the discussion is limited to heat treatment of nonferrous castings (for heat treatments of ferrous castings, readers are advised to refer to Heat Treating, Volume 4, ASM Handbook).

Designations of Heat Treatment Various heat treatment designations for common cast alloys are listed in Table 1. The asfabricated form of the casting is designated as the F temper. Castings are used in the F-tempered condition for noncritical applications, that is, where strength of the casting is not a deciding factor and low manufacturing cost is a prerequisite. Among the various T tempers, T5, T6, and T7 are most commonly practiced in casting industries. The T5 temper refers to the artificial aging of castings, while T6 and T7 tempers refer to solution heat treatment, quenching, and either peak aging (T6) or overaging (T7). All other T tempers, such as T2, T3, T8, T9, and T10, designate heat treatments for wrought alloys subjected to cold work. In this article, the discussion is limited to heat treatment of cast alloys. The choice of heat treatment depends on various factors, such as application, dimensional and microstructural

Table 1 Various heat treatment tempers commonly practiced for nonferrous castings Suffix

F T T1 T4 T5 T6 T7 W

Treatment and/or condition (sequence)

As-fabricated Heat treated to stable condition, excluding annealing Cooled from an elevated temperature and naturally aged Solutionized, quenched, and naturally aged Rapidly quenched from elevated temperature and artificially aged Solutionized, quenched, and artificially aged Solutionized, quenched, and artificially overaged Unstable temper for alloys that age spontaneously at room temperature after solution heat treatment

stability, mechanical properties, sensitivity to environmental degradation, and so on. A typical heat treatment regimen for commercial cast nonferrous alloys is comprised of solution heat treatment, quenching, and aging. Each of these treatments is discussed with reference to cast aluminum alloys.

Solution Heat Treatment Solution heat treatment, or solutionizing, involves raising the temperature of the casting to just below the eutectic temperature. The primary objective of solution heat treatment is to attain the maximum solute concentration in solid solution. To achieve this, the casting is held at very high temperature for a sufficiently long time to produce a nearly homogeneous solid solution. The temperature at which solution heat treatment should be carried out depends on the composition of the alloy and maximum solid solubility. Normally, castings are solutionized near but below their eutectic temperature. It is of paramount importance that the solution heat treatment temperature does not exceed the eutectic temperature of the alloy. If the eutectic temperature is exceeded, solution heat treatment may result in incipient melting (i.e., local melting), which has deleterious effects on mechanical properties and the structural integrity of the casting. Incipient melting can be avoided by slowly heating the casting to dissolve phases that are susceptible to melting at low temperature. Normally, temperature variations within +/
6  C (+/
10  F) are allowed during heat treatment. Recently, efforts have been made with limited success to carry out solution heat treatment in steps to avoid incipient melting. It is particularly useful for alloys with low eutectic temperatures to solution heat treat in two steps. In the first step, they are solutionized at a low temperature to dissolve low-melting-point phases, and in the second step, the temperature is raised above the eutectic temperature to rapidly solutionize the casting. Heating the casting at too low a temperature should be avoided because it results in incomplete dissolution and a relatively high degree of microsegregation, thereby reducing the strengthening potential of the alloy. The time

needed to solutionize an alloy depends on the initial microstructure (i.e., processing method) of the alloy, section thickness, furnace loading, dissolution rate, diffusivity of solute elements, and so on and can vary from a few minutes to 20 h. In general, the required soak times are longer for castings with coarse microstructures (as in the case of sand casting) compared to those with refined microstructures (as in the case of thin-walled permanent mold casting). In addition to dissolution and elemental homogenization, solution heat treatment results in thermal modification of phases that will not dissolve completely in the matrix. Thermal modification often results in fragmentation and subsequent spheroidization of phases that are present in excess of their solid solubility limit. For example, in the case of common aluminum-silicon-base cast alloys, the large, unmodified eutectic silicon flakes native to the aluminum-silicon alloys are broken up, coarsen, and eventually spheroidize during solution heat treatment. Spheroidization of eutectic silicon in cast aluminum-silicon-base alloys significantly improves the ductility of the alloy. It usually takes a long time (24 h or more) to spheroidize unmodified eutectic silicon. It is important to reduce the net solutionizing time and lower manufacturing costs without compromising properties. One way to reduce solutionizing time of cast alloys is through chemical modification. Aluminum-silicon-base cast alloys are often chemically modified by adding strontium or sodium in trace amounts (i.e., in levels of tens to hundreds of parts per million). The addition of strontium or sodium to aluminum-silicon-base cast alloys results in the modification of the silicon phase structure of the eutectic composition to a refined, fibrous morphology, as opposed to a coarse, flakelike structure in unmodified alloys. The silicon phase of modified eutectic compositions undergoes rapid fragmentation and spheroidization in a relatively short time compared to silicon particles in unmodified compositions, where complete spheroidization is never achieved, even after 24 h of solution heat treatment. Some products may undergo recrystallization and grain growth during solution heat treatment, which may restrict the maximum temperature and time allowed for solution heat

Heat Treatment / 405 treatment. Solution heat treatment is normally performed in air, but molten salt baths or fluidized beds can be used to facilitate rapid heating of castings. However, it should be noted that some alloys undergo severe distortion when rapidly heated and should be solutionized by heating slowly. Distortion may occur due to creep during solution heat treatment and must be considered while designing the loading of castings in the furnace. Blistering is another problem often observed with castings during solution heat treatment. During casting, gases are adsorbed and often trapped beneath the surface. During solution heat treatment, these trapped gases expand and result in blistering.

Quenching The next step after solutionizing is quenching. Normally, water temperatures between 65  C (145  F) and boiling (100  C, or 212  F) are used so that the thin layer of steam that forms at the surface of the casting on quenching in room-temperature water is disrupted. The casting should be taken from the solutionizing furnace and plunged into the quench as rapidly as possible to avoid a delay in quench. A delay of less than 10 s is ideal. The objective of the quench is to rapidly cool the alloy before any of the dissolved solutes can precipitate out. In addition, rapid quenching results in high vacancy concentration in the matrix. Vacancy concentration plays a vital role in the subsequent aging step. In general, high vacancy concentration results in faster aging of precipitates and thus can reduce the aging time significantly. The result of quenching is a supersaturated solid solution. Castings are stabilized when left at room temperature after quenching. In this condition, the casting is referred to as being in the T4 temper. Quench media other than water are sometimes used. Air blast quenching or water mist sprays may be used if the objective is to reduce quench distortion and cost, provided that the lower mechanical properties that result are acceptable for the application in question. Slow cooling rates during quenching could result in precipitation of coarse particles that are

Fig. 1

detrimental to mechanical and corrosion properties. It also reduces the level of supersaturation that adversely affects the subsequent aging response. In general, high cooling rates result in the highest strength and best corrosion resistance. The selection of quench medium is often determined by several factors, such as application, quench sensitivity of the alloy, residual stress, structural integrity, and so on. Among them, quench sensitivity of the alloy is the most important factor. Quench sensitivity is generally high in alloys containing higher solute levels and containing dispersoids that act as nucleation sites for coarse precipitates. Alloys should be quenched so as to avoid the nose of the continuous cooling transformation curve. Water is the most commonly used quenching medium. Conventionally, castings are quenched in water at temperatures between 60 and 80  C (140 and 175  F). The quench rate in water is relatively high, and a significant amount of residual stress is generated, causing part distortion. Blistering may also occur. Problems due to quenching of components in water can be overcome by using quenching media with a reduced cooling rate. Unfortunately, doing so may reduce the mechanical properties of the alloy. Components with complex shapes and nonuniform thickness are quenched at slow rates to avoid distortion and excessive residual stress. This can be achieved by using water at temperatures between 70 and 85  C (160 and 185  F), boiling water, polyalkylene glycol, fluidized beds, forced air, and so on. The most common approach is to add polyalkylene glycol to the quench medium in varying proportions in order to control the quench rate and therefore quench distortion. A promising quenching medium is the fluidized bed, a particle bed that is fluidized using helium, nitrogen gas, or air. The cooling rate can be varied by adjusting processing parameters such as bed particle diameter, fluidizing agent, bed material, and so on. The relative cooling rates in various furnaces at the surface and center of a specimen are shown in Fig. 1 (a and b), respectively.

Aging Aging refers to the controlled breakdown of supersaturated solid solution at room temperature (natural aging) or at elevated temperature (artificial aging) to form nanosized precipitates in the matrix. These precipitates strengthen materials by a mechanism known as precipitation strengthening by pinning the dislocations and inhibiting their free motion. After the casting has been quenched, it may undergo artificial aging immediately, or there may be an intermediate room-temperature natural aging step. The former (i.e., artificial aging) is known as T6 or T7 temper (depending on aging time), while the latter (natural aging followed by artificial aging) is known as T61 temper. In practice, foundries frequently refer to either as T6. The advantage of the T61 temper is that there is time to straighten the casting if needed. While proper racking of the casting during heat treatment can minimize it, there is generally always some quench distortion or warpage of the casting during the steps leading up to the T4 temper. While in the T4 temper, the casting exhibits its maximum softness and ductility and thus is most easily straightened mechanically. Both manual and automated systems may be employed to accomplish this. Another advantage to the T61 temper is flexibility. Delays in going from the quench tank to the aging ovens are much less important and can be absorbed into the natural aging time. Castings can be aged after solution heat treatment to T6 (peak age) or T7 (overage) temper or can be aged directly without solution heat treatment (referred to as the T5 temper). It is a common practice to age the casting to strengthen the alloy and preserve its dimensional stability. Aging is done by soaking castings for a required time at one or sometimes two temperatures (known as step aging). The decomposition of the supersaturated matrix is usually a complicated process and takes place in several precipitation stages. The finely dispersed precipitates have a dominant effect on properties such as yield strength, tensile strength, ductility, corrosion resistance, and structural integrity. These properties can be tailored for

Relative cooling rates ( C/s) between 900 and 750  C (1650 and 1380  F) in various quenching media for a bar with maximum cross section of 200 mm (8 in.) at the (a) surface and (b) center of the specimen. Source: Ref 1

406 / Principles of Solidification specific applications by selecting the proper processing temperature and time. Both temperature and time of aging determine the type, size, shape, and size distribution of precipitate in the matrix. On a typical precipitation strengthening curve, strength increases with increasing aging time, and the curve has one or more peaks (maximum values) followed by a decrease in the strength. Each peak is due to the precipitation of a specific type of precipitate; for example, in the case of aluminum-copper-base cast alloy, the first peak is due to the formation of Guinier-Preston (GP) zones, while the second peak is due to the precipitation of y0 or y00 precipitates (depending on the aging temperature). The maximum strength condition is referred to as peak aging, and the subsequent decrease in strength with longer times is known as overaging. During overaging, coarse precipitates are formed that are less effective in pinning down the dislocation motion. Precipitation of phase(s) (stable or metastable) may occur at room temperature (natural aging) or at high temperature (artificial aging).

Natural Aging Natural aging refers to the decomposition of the supersaturated solid solution in the matrix at room temperature. Most alloys age harden at room temperature to yield stable and reasonable properties for many applications. During the natural aging process, the casting undergoes incubation, a period where prenuclei, or clusters, form throughout the casting. These are the places where precipitates will subsequently form and grow during the artificial aging or precipitation hardening operation. Because the natural aging takes place at room temperature, diffusion rates are very low and the prenuclei are finely distributed, the ideal condition from which to artificially age. Some alloys can be naturally strengthened in a few days, while others take years to achieve the same strength level. During natural aging, the high level of supersaturation and high vacancy retention cause rapid formation of clusters or coherent stable GP zones. For most applications, the natural aging should be no longer than 24 h and no shorter than 8 h. The incubation process proceeds fairly rapidly during the 8 h following quenching. Starting the artificial aging during this period can cause some variability in the end result unless exact timing is possible. The casting will still be perfectly serviceable, but it will be harder to maintain tight controls on the process if the natural aging time is variable. This is especially so if no set incubation period is adhered to in the 0 to 8 h range. The period between 8 and 24 h is more forgiving.

Artificial Aging Artificial aging involves raising the temperature of the casting back to the point where the elements in supersaturated solid solution begin

to come out of solution in the form of precipitates at a relatively faster rate. The actual formation of these precipitates is a multistage process, and the end result is a fine dispersion of stable or metastable precipitates in the matrix. Precipitates form, grow, and then, if the casting is kept at the aging temperature, begin to ripen, with the smaller precipitates disappearing to feed the growth of the larger precipitates (also known as Ostwald coarsening). The strength of the alloy increases as artificial aging proceeds, while the ductility decreases. The process can be stopped at any point, and thus, a wide range of combinations of strength and ductility can be generated from the same casting. It should be pointed out that the aging temperature chosen determines the speed with which the precipitation process occurs. At relatively high aging temperatures, the reaction would proceed more quickly than at low temperatures. It obviously becomes more difficult to exactly hit a given point on the aging curve as the speed of reaction is increased. Conversely, “time is money,” and many manufacturers have optimized their heat treatment cycles to both minimize cost and optimize properties. The temperature-time combination chosen by a given foundry is based on their experience with their heat treatment equipment and will vary from foundry to foundry. There are, of course, limits to how far the temperature-time-property relationship can be pushed. High aging temperatures will cause the reaction to proceed rapidly, and precipitates will begin to coarsen due to ripening during the entire aging time. This heat treatment is referred to as overaged, or the T7 temper, and may be used when high ductility is required at low strength in a part that must be dimensionally stable at elevated temperatures. This temper, called a stabilization heat treatment, is not uncommon where engine components are concerned but is rarely used for A356/357 suspension components. Artificial aging from the as-cast condition may also be employed to stabilize castings without solutionizing and quenching; this is the T5 temper. One of the side effects of the heat treating operation is that there is a tendency for the castings to expand as they are aged. This can cause dimensional changes, which must be accounted for by the foundry via pattern shrink allowance and proper racking or straightening operations.

Emerging Heat Treating Technologies Over the past few decades, several heat treating furnaces have been developed. Some commonly used heat treating furnaces are listed in Table 2. The most widely used heat treating furnace for heat treatment of cast aluminum components is the forced-air circulation electric resistance furnace. An emerging technology for rapid heat treatment of aluminum components

Table 2 Relative heat-transfer coefficient (h) ranges for various heat treatment furnaces Heating medium or furnace type

h, Wm
2K
1

Lead Salt bath Fluidized bed Radiant fluidized bed (above 800  C, or 1470  F) Forced circulation furnace Radiant atmosphere furnace (above 750  C, or 1380  F) Vacuum furnace

1200–1800 500–1200 500–700 200–300 150–200 120–220 120–200

Source: Ref 1

is the fluidized bed furnace (Ref 1). The fluidized bed consists of a bed of particles (usually sand) that behaves like a fluid when air is passed through orifices at the bottom of the bed. The gas flow rate is sufficiently high that the terminal velocity of the solid particles is exceeded, and the bed goes into motion. Fluidized bed furnaces can be used to solutionize, quench, or age the workpiece; they can be either batch-type or in-line continuous-type furnaces. Other emerging heat treating methods are infrared heating, induction heating, salt baths, and so on, each of which has its own advantages and limitations. The selection of heat treating equipment is based on factors such as cost, efficiency, environmental effect, maintenance cost of the equipment, geometry of parts, and so on. Relative heat-transfer coefficient ranges for various furnaces are given in Table 2.

Microstructural Changes due to Heat Treatment of Cast Alloys During solution heat treatment, several microstructural changes occur. These include fragmentation and spheroidization of eutectic silicon, dissolution of intermetallics, morphological changes of iron-rich intermetallics, recrystallization, grain growth, and so on. Among them, silicon fragmentation and spheroidization and dissolution of intermetallics are of paramount importance in cast aluminum-silicon alloys.

Dissolution During solution heat treatment, soluble intermetallics are dissolved, and the matrix is supersaturated with solute on subsequent quenching. Only those intermetallic phases dissolve that have high solubility at elevated temperature. The primary factor governing dissolution of intermetallic phases is the relative diffusivity of solutes in the matrix. In addition, other factors that control the dissolution rate of intermetallic phases are interface mobility and curvature effect. Commercial cast aluminum alloys often contain iron as an impurity. Solution heat treatment results in the partial dissolution of iron-rich

Heat Treatment / 407

Fig. 2

Comparison of the iron-rich phase morphology in modified A356. Dendrite arm spacing, 30 to 45 mm; solution heat treated at 538  C (1000  F) for 30 min and 10 h. Source: Ref 2

Fig. 4

Transmission electron micrographs of Al-Si-CuMg (319) alloy heat treated to T6 temper showing distribution of precipitates (a) near the periphery of the aluminum dendrite and (b) at the center of the aluminum dendrite. Source: Ref 3

Fig. 3

Deep-etched morphologies of eutectic silicon particles in an aluminum alloy. As-cast and after 10 h solution heat treatment at 538  C (1000  F), Source: Ref 2

intermetallics, as shown in Fig. 2(a and b). The morphology of iron-rich intermetallics plays an important role in the performance of the casting. In general, the sharp edge of iron-rich intermetallic phases may act as a stress concentrator, which should be avoided. However, despite prolonged solution heat treatment, complete dissolution of iron-rich intermetallics is not possible because of the limited solubility of iron in the aluminum matrix.

Spheroidization An example of spheroidization of silicon in an unmodified and strontium-modified Al-Si-Mg alloy resulting from solution treatment is shown

in Fig. 3(a to d). Spheroidization of eutectic silicon increases the ductility of the alloy.

Precipitation Aging results in the breakdown of the supersaturated matrix to form nanosized particles. The breakdown of the supersaturated matrix takes place as follows: Supersaturated matrix ! GP zones ! Metastable phase(s) ! stable precipitate Typical precipitate distribution in the aluminum matrix (Al-Si-Cu-Mg alloy) is shown in Fig. 4(a and b). The time and temperature of aging treatment determines the number density,

size, and size distribution of precipitates. In general, precipitate size increases with the increase in aging time. The strength of the alloy increases to a peak value (known as peak aging), at which the precipitate size attains a certain critical value. When the precipitate size exceeds the critical value, the strength of the alloy decreases; this is called overaging. REFERENCES 1. R.W. Reynoldson, Heat Treatment in Fluidized Bed Furnaces, ASM International, 1993 2. F. Major and D. Apelian, A Microstructural Atlas of Common Commercial Al-Si-X Structural Castings, Proceedings of AFS Conference on Structural Aluminum Castings (Orlando, FL), 2003 3. S.K. Chaudhury, D. Apelian, J. Keist, P. Meyer, D. Massinon, and J. Morichon, Microstructure and Mechanical Properties of Heat Treated 319 Alloy Diesel Cylinder Heads, Metall. Trans. A, 2008, in press

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 408-416 DOI: 10.1361/asmhba0005293

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Hot Isostatic Pressing of Castings Stephen J. Mashl, Bodycote HIP—North America

HOT ISOSTATIC PRESSING, a process that is often referred to by its acronym, HIP, is used to eliminate porosity in castings. This is accomplished by surrounding the castings with a pressurized fluid, usually argon gas, while simultaneously heating the castings to a temperature that is below the solidus but high enough to maximize plastic flow and enhance atom/vacancy diffusion within the material. Under the simultaneous influence of hydrostatic pressure and elevated temperature, a pore will collapse and shrink, initially by plastic flow, then by diffusional mechanisms. The pressures employed in commercial HIP applications typically range from 105 to 310 MPa (15,000 to 45,000 psi). The lower pressure, 105 MPa (15,000 psi), is appropriate for most metal castings, while higher pressures are used in special applications, such as processing creep-resistant alloys. Figure 1 illustrates the effect of HIP on an A356-T6 aluminum alloy casting. Shrinkage porosity and possible gas porosity are observed in the casting that went through a T6 heat treat in the as-cast condition (Fig. 1a). The HIP of the casting prior to the T6 treatment yields the microstructure shown in Fig. 1(b). Other than the complete elimination of porosity, no significant change in microstructure is observed. Figure 2 shows two valuable aerospace castings being examined prior to HIP. Castings used in aircraft applications represent a large portion of the market for HIP of castings.

Fig. 1

Microstructures of an A356 aluminum casting. (a) In the as-cast and T6 condition. (b) In a cast, HIPed, and T6-treated condition. The as-cast sample exhibits shrinkage porosity and a possible gas pore. Following HIP, the porosity is eliminated, but otherwise, the microstructure remains unaffected.

History of HIP Hot isostatic pressing was developed in the mid 1950s at Battelle Memorial Laboratories in Columbus, Ohio, in response to a problem encountered by the United States’ nascent atomic energy program. Uranium oxide fuel elements were corroding at an unacceptable rate in the reactors. To find a solution, the U.S. Atomic Energy Commission contacted Battelle Memorial Institute and asked the research organization to develop a method of bonding Zircaloy to UO2 fuel elements while maintaining strict dimensional control. This challenge led Dr. Russell Dayton, Henry Saller, Stan Paproki, and Edwin Hodge to invent the process originally named

Fig. 2

Two titanium aerospace castings are examined prior to HIP.

Hot Isostatic Pressing of Castings / 409 gas pressure bonding and documented under U.S. patent 687,842 (Ref 1–3). The technique combined the use of elevated temperature and the application of a hydrostatic pressure via pressurized gas. This combination created an environment that was very well suited to solid-state bonding. Soon after discovering this cladding technique, the potential of using this process to consolidate metal powders was realized, and the process became known as hot isostatic pressing, or HIP (Ref 4). On the earliest HIP vessel designs, the pressure vessel was located within a resistance furnace, a design choice probably dictated by the use of commercially available hardware. This type of HIP unit became known as a hot wall vessel. An example of one of the earliest HIP systems is shown in Fig. 3. The limitations of this design were quickly realized, and in 1956, Charles Boyer, with the assistance of technician James Stone, built the first cold wall HIP vessel, that is, a unit in which the furnace resides within the pressure vessel. By keeping the pressure vessel relatively cool, the strength of the pressure vessel remained high throughout the cycle, and much higher pressures could be achieved. The cold wall HIP vessel design remains in use today (2008) (Ref 5–8). The first investigation into using HIP to remove internal porosity in castings took place in the mid-1960s when it was tested on aluminum alloys (Ref 9, 10), but the first commercial success came about as HIP began to be used to improve the service life of fatigue-critical parts,

such as jet engine compressor disks and turbine blades. The levels of improvement were such that, between 1970 and 1975, the HIP of castings quickly grew from a laboratory project to standard commercial practice for aerospace applications (Ref 11–17). The integration of HIP into the processing of high-value, fatigue-critical parts increased, finding application in powergeneration components, orthopedic implants, structural castings, and aerospace components. The densification of castings quickly grew to become the largest segment of the commercial HIP industry, accounting for approximately 60% of sales in the United States (Ref 9–12).

Reasons for Using HIP Even very small amounts of porosity can have a large detrimental effect on the mechanical properties of metal (Ref 18–26), and a desire to improve mechanical properties is, by far, the primary reason for HIPing castings. The effect of HIP on mechanical properties is expanded on later in this article, but there are other reasons to employ HIP. In all cases, performance or appearance of the finished part is improved by the elimination of porosity via HIP. Examples include:  Scrap recovery: A well-controlled foundry











process may experience a process deviation that results in parts that contain an unacceptable amount of porosity. Parts that have 

failed x-ray inspection or other nondestructive testing (NDT) may be restored to a passing condition by subjecting the failed parts to HIP. Reduction of inspection requirements: With the consistent high quality achievable through the use of HIP, it is possible that the frequency of NDT inspection can be decreased from a very high level to one of intermittent spot checks. Improved machinability: During machining of a casting, the interruption of the cutting process that occurs when the cutting tool contacts a region of subsurface porosity will result in tool deflection and a poor surface finish. Improvements in surface finish: With parts that are polished to a smooth surface finish, subsurface porosity that becomes exposed during finishing operations can become a visible defect on the surface, diminishing the aesthetic value of the finished part. Achievement of vacuum-tight, metal-to-metal seals: When machining the surface of a casting that will become part of a metal-tometal seal in a high-vacuum environment, the exposure of subsurface porosity during machining often results in a seal that cannot retain the desired vacuum level. Medical safety: When castings are used in biomedical applications, such as surgical instruments, a subsurface pore that becomes exposed during finishing operations may harbor bacteria. Improved plating quality and yields: When a casting goes through an electroplating process, any porosity exposed during the postcast finishing of that part will produce a defect in the as-plated surface that is larger than the size of the pore where it intersects the surface. By eliminating subsurface porosity through the use of HIP prior to any machining or finishing operations, electroplating quality will be higher and yields improved.

Overview of a HIP System

Fig. 3

Duplicate of an early hot wall autoclave at Battelle Memorial Institute. Maximum pressure: 34 MPa (5000 psi); maximum temperature: 815  C (1500  F). Source: Ref 32

As illustrated in Fig. 4, a HIP system is conceptually simple; the castings being processed rest on trays or in containers that, in turn, reside within a resistance furnace that is surrounded by a pressure vessel. Specially designed ports and feed-throughs, built into either the top or bottom closure of the pressure vessel, allow the passage of gas and electricity while maintaining the pressure integrity of the vessel. A mantle or heat shield is located between the furnace and the pressure vessel. In addition to providing the layer of insulation that limits heating of the pressure vessel, channels within the mantle are designed to promote convective flow of the hot gas, helping to ensure temperature uniformity within the working zone of the unit. Typical HIP Cycle. During a typical HIP cycle, the unit is loaded with parts and sealed.

410 / Principles of Solidification

Fig. 5

Schematic of a conventional HIP cycle. The HIP cycles are conventionally identified by the dwell conditions; thus, this schematic shows the timetemperature-pressure profile for a 1120  C (2050  F), 100 MPa (1000 bar), 2 h cycle. Much of the total elapsed time for a HIP cycle is taken up in heating/pressurization, cooling, and venting. The cooling rate on this graph is on the order of 3  C/min (5  F/min), a rate typical of older HIP equipment. Some modern HIP units are capable of rapid cooling, at rates 20 or greater than that shown here.

to plant and personnel associated with improper maintenance and careless operation of HIP equipment is very high. Selecting HIP Conditions. Temperature, time, and pressure are the three critical process variables to be determined when choosing the processing parameters for a new material. In some cases, heating and cooling rates also must be considered. When selecting HIP process parameters, it is desirable to:  Choose a temperature that will lower the

Fig. 4

Schematic of a top-loading, monolithic-wall HIP system. Components to be HIPed are loaded onto racks or placed in an open container, which is then placed within the interior of a resistance furnace. The furnace resides within a pressure vessel. A mantle or heat shield serves as an insulating layer between the furnace and the pressure vessel. The mantle and an external cooling water circuit act to keep the temperature of the pressure vessel low during a HIP cycle, thus maximizing the vessel strength. A vacuum system (not shown) and an inert gas supply and gas compressors are used to purge and fill the vessel with inert gas, for example, argon. After evacuating and filling with the inert gas, power is supplied to the furnace, and temperature and pressure rise to the targeted values used for that material.

Once the vessel is closed, a vacuum system removes air from the vessel, and then the vessel is filled with inert gas, typically argon. Several evacuate-argon backfill cycles may be carried out to eliminate all residual air from the vessel interior. The unit is then pressurized with inert gas while the furnace is heating. An increase in pressure occurs from both the mechanical action of the compressor and the temperatureinduced pressure rise that occurs when heating gas in an enclosed volume. Once the desired temperature and pressure are achieved, the dwell portion of the HIP cycle begins. Since plastic deformation via yielding occurs very rapidly, the pore shrinkage that takes place due to plastic flow is complete once the dwell temperature and pressure are reached. The remaining portion of the dwell time allows additional pore shrinkage via diffusion as well as solid-state bonding across the collapsed pore interface. Upon completion of the dwell portion of the cycle, the power to the furnace is turned

off, and, after the gas has cooled sufficiently, the gas within the vessel is vented. These steps and the relative timing of a standard process cycle are illustrated in Fig. 5. Cooling rates achieved during the final portion of the cycle vary from system to system. While older HIP units cool slowly (2 to 5  C/min, or 4 to 9  F/min), some modern HIP vessels can cool at rates in excess of 100  C/min (180  F/min). With some alloy systems, this rapid-cool capability allows a degree of microstructural control. Compared to other metal-processing operations such as forging and hot rolling, a HIP system is very quiet and seems quite tame during operation. It can be a bit disconcerting, however, to calculate the amount of energy that is stored within a 1.5 m (60 in.) diameter HIP vessel at peak pressure and learn that this stored energy is roughly equivalent to the energy released by the detonation of 225 kg (500 lb) of 2,4,6-trinitrotoluene, or TNT (Ref 27). Like many pieces of commercial equipment, the risk

 





flow stress of the metal as much as possible, thus maximizing the amount of densification that occurs by plastic flow Select a temperature at which self-diffusion will occur at an acceptable rate, thereby enhancing the final stages of densification Identify a process temperature that is safely below the solidus temperature, taking into account the possibility of elemental segregation and incipient melting within the casting Select heating and cooling rates that strike a balance between short cycle times and other metallurgical factors that are affected by heating or cooling rate. For example, on slow cooling, duplex stainless steels form a brittle intermetallic, sigma phase. Rapid cooling from the soak temperature of the HIP cycle is not sufficient to provide solutionized and quenched properties, but it does make the part less susceptible to cracking during post-HIP handling and heat treat. The material(s) that are in contact with a casting during HIP must also be considered. Interdiffusion between a casting and a tray on which the casting sits can alter the chemical composition of the surface material, resulting in contamination of the casting or even localized melting.

Table 1 lists a number of common HIP conditions for a wide range of cast metals.

Hot Isostatic Pressing of Castings / 411 Mechanisms of Pore Closure During HIP. There are four main mechanisms by which pores are eliminated during HIP:    

Plastic flow Power law creep Coble (grain-boundary) creep Nabarro-Herring (lattice) creep

Given appropriate temperature and pressure conditions, all densification mechanisms will be active simultaneously, but the rate at which densification occurs by these different mechanisms varies significantly. Densification via Plastic Flow. The bulk flow stress (yield stress if at room temperature) of a porous material varies as a function of pore fraction, with the flow stress increasing as pore fraction decreases. When a hydrostatic pressure exceeding the bulk flow stress of a material is applied to a material having isolated internal porosity, pores shrink until the bulk flow stress of Table 1 Common hot isostatic pressing conditions for various cast metal alloys Temperature Material



C

Aluminum alloys 510 (e.g., A355, A357, A201) Titanium and 900 titanium alloys Plain carbon and 1065 low-alloy steels Nickel-base 1185 superalloys—I ´ (IN-718, Rene 77) Cobalt-base 1220 alloys (F75) 1185 Nickel-base superalloys—II (Mar-M 247, Rene´ 125)

Fig. 6



Pressure

F

MPa

ksi

Time, h

950

100

15

2

1650

100

15

2

1950

100

15

4

2165

100

15

4

2200

100

15

4

2165

175

25

4

the material matches that of the applied pressure. At that point, pore shrinkage ceases. Since dislocations in crystalline metals move at the speed of sound, densification via plastic flow can be assumed to occur instantly, but, because the flow stress of a metal is strongly influenced by temperature, pore shrinkage via plastic flow can be considered the dominant densification mechanism during the heating segment of a HIP cycle. Densification via Power Law Creep. During densification via plastic flow, the movement of some dislocations through the crystalline lattice may be hindered by precipitates or other small microstructural obstacles. Diffusion of atoms and vacancies to and from a pinned dislocation will allow it to climb around the obstacle and continue on its course through the lattice. The amount of diffusion needed for this to take place is relatively small, so, while this introduces a time-dependency into the densification process, pore shrinkage can still be assumed to occur at a relatively rapid rate. Diffusional Densification via Coble and Nabarro-Herring Creep. Of the various densification mechanisms, densification via diffusion takes place at the slowest rate; thus, these mechanisms only become dominant in the final stages of densification, essentially after densification via plastic flow and power law creep has ceased. The surface energy associated with porosity provides a driving force for pore shrinkage via diffusion. The movement of atoms to the pore surface and the movement of vacancies from the pore surface into the bulk metal and the exterior of the sample serve to complete the final stages of densification. It is worth noting that some aerospace materials are designed to be creep resistant, giving them improved performance in a gas turbine environment. The same features that serve to improve operating performance in these alloys make them more difficult to HIP. Some of these

alloys must be processed at higher-than-normal pressures to achieve full densification within a reasonable period of time.

Effect of HIP on Mechanical Properties As stated earlier, the primary reason for employing HIP is to improve mechanical properties. While most mechanical properties are improved with the elimination of internal porosity, the degree of improvement varies, depending on the property being considered. Figure 6 displays the relationship between ultimate tensile strength, tensile ductility measured as percent elongation, and impact resistance as a function of pore fraction in sintered iron powder compacts (Ref 18). It should be noted that while these results are from a powder metallurgy study, the relationship between internal porosity and mechanical properties also holds true for castings when considering the effect of microporosity on properties. Consider the relative steepness of the curves in the region immediately adjacent to the broken line that represents the fully dense state, that is, 0 to 10% porosity on the x-axis. The slope of the line in this region provides a good indication of the relative sensitivity of that specific property to small amounts of porosity. Of the three properties examined in this figure, it is apparent that tensile strength is the least sensitive to the presence of small amounts of porosity. Tensile ductility shows a stronger response to low pore fractions, while impact resistance is extremely sensitive to microporosity. Fatigue properties are also extremely sensitive to small amounts of porosity, and, as shown in Fig. 7, HIP of castings will generally produce significant improvements in high-cycle fatigue life at any given stress level. The fatigue

Three graphs showing the relationship between tensile strength, ductility measured as percent elongation, and impact resistance and relative density in pressed and sintered iron powder metallurgy (PM) samples, consolidated to varying density levels. The data indicate that for materials with very low fractions of porosity, for example, most commercial castings, variations in the pore fraction at levels near full density will result in minor variations in tensile strength, moderate variations in ductility, and significant variations in fracture toughness. While these data represent the behavior of a PM material, the same relationships generally hold true for castings. Graphs produced using data from Ref 18.

412 / Principles of Solidification limit, that is, the maximum cyclic stress that can be applied without inducing crack initiation in a high-cycle fatigue environment, is also increased with HIP. The relationships between porosity and mechanical properties described here generally hold true for all cast metals, provided that porosity, and not some other type of defect, is limiting property improvement. Figure 8 shows the effect of HIP on the mechanical properties of a group of AB2, nickel-aluminum bronze alloy castings. No significant improvement in yield stress is observed, but the ultimate tensile strength is increased by 170 MPa (25,000 psi), or 34%, and tensile ductility, measured as percent elongation, more than doubles. This degree of improvement in ultimate tensile strength is unusual; however, these castings contained a high fraction of porosity, 10 to 20%. If one examines the trends shown

in Fig. 6 and takes into account the large amount of porosity in these parts, it is seen that the observed dramatic improvement in UTS and percent elongation follows the predicted trends. An example of the effect of microporosity on impact toughness is displayed in Fig. 9. In this case, the concentration of microporosity in the castings varies from 1 to 12 vol%, and the impact toughness improves from approximately 110 to 150 J/cm2 as the pore fraction decreases. The HIP of these castings should yield parts having a fracture toughness similar to that of the wrought material. Table 2 is valuable in that it demonstrates how the response of any given group of castings to HIP can vary. The data provided show the results of stress-rupture testing of five nickel-base superalloy castings. The Rene´ 80 castings show dramatic improvement in time

to failure and ductility; the Rene´ 77 samples tested at 980  C (1800  F) show only moderate improvement in life and ductility, and that same alloy tested at 815  C (1500  F) shows an increase in time to failure and no significant change in ductility. It is not unusual, when comparing the as-HIPed properties of one group of castings to another, to encounter this apparently inconsistent response to HIP processing. The varied response can be caused by multiple factors. Perhaps one alloy solidifies over a wider temperature range and is thus more susceptible to interdendritic shrinkage. Having more porosity in the as-cast condition, this alloy would likely see greater property improvement with HIP. Perhaps an alloy has a microstructure in which one microconstituent influences mechanical properties as much or more than the amount of microporosity present in the as-cast material. It is possible that this alloy would show minimal improvement with HIP.

Effect of Inclusions on As-HIPed Properties

Fig. 7

Stress-number of cycles fatigue plot for a gravity die-cast automotive diesel engine block. The alloy is AlSi7MgCu0.5. Fatigue life was determined using rotating beam equipment. At a given stress level, a threeto ten-fold increase in fatigue life is seen when HIP is used. The fatigue limit of the HIPed samples is seen to increase from 70 to 100 MPa (10 to 15 ksi), an increase of 43%. The gray lines through the data represent a second order regression; The broken lines represent 95% Confidence limits.

Fig. 8

Bar chart showing the change in mechanical properties following HIP. The material was a cast nickelaluminum bronze alloy (AB2) that contained 10 to 20% porosity in the as-cast condition. The dramatic change in the ultimate tensile strength (UTS) and tensile ductility measured as percent elongation is due to the very high pore fraction in the as-cast material. Source: Ref 23

When considering HIP as a means of achieving improved mechanical properties in castings, it must be realized that HIP only eliminates subsurface porosity from a casting. Other defects, such as oxide inclusions or brittle phases, can also have a strong detrimental effect on mechanical properties, and these defects will likely be unaffected by the HIP process (Ref 28–30). For example, if a casting has a sufficiently high concentration of nonmetallic inclusions, the inclusions may play a greater role in limiting fatigue life than does the porosity present in the part. In this case, HIPing the cast part may not improve mechanical properties. This concept is supported by the experimental data of Wang et al. (Ref 22). In that work, the researchers poured castings in a laboratory environment, controlling the concentration of both oxides and porosity. Then, HIP was used to produce pore-free samples. After fatigue testing, the fracture surfaces of the fatigue test samples were analyzed to determine the type of defect that caused the initial crack formation. The results of those tests, shown in

Fig. 9

Influence of microporosity on the impact toughness of 1wt% Cr-0.25wt%. Mo cast steel. Source: Ref 24

Hot Isostatic Pressing of Castings / 413 Table 2

Stress-rupture properties of superalloy castings Temperature

Material(a)

IN-738, as-cast IN-738, HIP Rene´ 77, as-cast Rene´ 77, HIP Rene´ 77, as-cast Rene´ 77, HIP IN-792, as-cast IN-792, HIP Rene´ 80, as-cast Rene´ 80, HIP



C

980 980 815 815 980 980 870 870 870 870



Stress

F

MPa

ksi

Average life, h

Elongation, %

Reduction in area, %

1800 1800 1500 1500 1800 1800 1600 1600 1600 1600

152 152 138 138 152 152 310 310 310 310

22 22 20 20 22 22 45 45 45 45

19.0 52.5 77.0 165.0 51.0 68.0 175.0 283.0 41.5 141.0

11.8 20.5 10.0 9.0 19.4 22.0 9.2 12.1 2.5 11.5

20.0 20.6 20.0 18.0 37.0 55.0 6.5 22.0 2.5 17.0

(a) HIP, hot isostatic pressing. Source: Ref 17

Fig. 10, indicate that, for a given probability of failure, porosity had the greatest detrimental effect on fatigue life. If, however, the data are separated into vertical zones by drawing lines through the first and last data point for each type of failure, there is a range (region II on the graph) where either porosity or oxide inclusions could be the fatigue-life-limiting defect. Another conclusion that can be drawn from Fig. 10 is that optimal fatigue properties can only be obtained in an aluminum casting if both pores and oxide inclusions are eliminated from the material. When a foundry produces a new casting for the first time, if one or more mechanical properties fail to meet target specifications, the question may be raised; “Will HIP bring the mechanical properties up to the desired level?” Unfortunately, the degree to which mechanical properties can be improved is impossible to predict, because too much depends on other factors. If, however, the foundryman or engineer has an understanding of the level of porosity in their castings and the amount and type of nonmetallic inclusions, some reasonable predictions can be made. For example, consider the following two hypothetical case studies. Case 1. A production line produces small investment castings; process control is excellent, and the shape of the casting allows a reliable and efficient gating and riser system. In such a case, it can be assumed that gross porosity will not be a problem, and only a small fraction of shrinkage porosity will exist within the parts. The HIP will totally eliminate all of the shrinkage porosity, but, because the total pore fraction was low in the as-cast state, improvement in yield and ultimate tensile strength will be small and perhaps unnoticeable in test-totest variations. Tensile ductility measured as percent elongation to failure may increase a small amount, that is, a few percent, and improvement in fracture toughness and fatigue life should be significant. (Note: Sometimes an improvement in ductility by a “few percent” means the difference between meeting or not meeting an end-user’s specification. Case 2. A foundry produces parts in a manner similar to that described in case 1, but the part is of a shape that makes it difficult to design a foolproof molten metal feed system. In addition, the alloy has a narrow pouring temperature range and a process specification that

is challenging for the foundry to meet. The foundry finds that most of their production meets the customer-designated strength and ductility requirements, but the occasional casting falls below the minimum limit. A study shows that the castings that exhibit low tensile strength contain regions of centerline porosity. The problem is intermittent; thus, the facility must perform 100% inspection of this part. In this case, the addition of postcast HIP will eliminate the centerline porosity and increase the tensile strength of the intermittent bad castings while having little effect on the strength of the good castings. If as-cast properties are monitored on a large-lot basis, the average tensile strength may only increase a small amount, but the castings that would normally test poorly because of centerline porosity will now display good strength, in line with that of the best castings coming off of the production line. Statistical analysis will show that the standard deviation of the mean will decrease. While, in this case, the number of parts that significantly improve in strength may be small, when the liability associated with one bad part getting into the field is high, the ability to eliminate the rare lowstrength part is very valuable.

Effect of HIP on the Shape and Structure of Castings Volumetric Shrinkage and Distortion. Distortion is always a concern during the processing of components downstream of the casting operation. If a casting contains internal porosity, volumetric shrinkage during HIP is inevitable, the percentage of shrinkage being equal to the volume percent porosity present in the casting prior to HIP. In castings where porosity is primarily from interdendritic shrinkage, the overall fraction of porosity within the material will be small and evenly disbursed; thus, the volumetric shrinkage during HIP will be small, and no gross distortion of the part will occur. An example is shown in Fig. 11 where the intricate vanes of a turbocharger impeller cast of an aluminum alloy are seen in Fig. 11 (a). Figure 11(b) shows micrographs of the cross section of an impeller vane before and after HIP. While the shrinkage porosity is evident in the as-cast part, the pores are small

Fig. 10

Two-parameter Weibull plot showing probability of failure versus the natural log of the number of cycles to failure. In this study, a large number of aluminum castings were produced in a laboratory setting in which porosity, oxide inclusions, and other microstructural features were precisely controlled (Source: Ref 22). Fatigue specimens were cycled to failure, and the fracture initiation site was characterized according to the type of defect (pore or oxide inclusion) or microstructural feature (slip plane) present at the initiation site. The graph indicates that, for these castings, pores can be the most detrimental defect, that is, causing the greatest reduction in fatigue life. The data can be divided into four zones by drawing a vertical line through the fist and last data point for each category of failure. In zone I, fatigue life is controlled solely by porosity. In zone II, either pores or oxides can be the controlling defect. In zone III, oxides alone control fatigue life. It is not until both porosity and oxide inclusions are eliminated that these aluminum castings achieve the highest fatigue performance, shown in zone IV. Source: Ref 30

and evenly dispersed throughout the cross section. As a result, there is no obvious distortion in the as-HIPed section when the porosity is eliminated. In contrast, when pores are larger and located near the surface, it is common to see small dimples in the surface where metal from the near-surface region has flowed into the pore. If gross porosity is present within a casting and the pores are isolated from the surface, HIP can still produce a pore-free part; however, HIPing parts having large volume fractions of porosity will result in large volumetric shrinkage within the part. If the gross porosity is localized to a specific region within the casting, distortion can occur. Other Microstructural/Physical Changes. Typically, HIP exposes a part to elevated temperatures for two or more hours. As a result, all the microstructural changes that take place at elevated temperature can occur during HIP. Positive side benefits of HIP can be the homogenization and stress relief that take place in a conventional HIP cycle. If a process stream originally included a homogenization or stressrelief anneal, those process steps can be omitted if HIP is used. On the negative side, grain growth and precipitate coarsening can occur during HIP, leading to an undesirable effect on some mechanical properties. Figure 12 shows micrographs of a steel casting before and after HIP. While the shrinkage porosity is

414 / Principles of Solidification obviously eliminated, coarsening of the microstructure and loss of the dendritic microstructure are also apparent. Employing higher pressures, lower temperatures, or shorter cycle times may provide a way to decrease the effects of high-temperature exposure, but there is no avoiding the fact that HIP is a high-temperature process that relies on diffusion to achieve the desired end result.

Problems Encountered in HIP

Fig. 11

(a) Cast aluminum alloy turbocharger impeller showing the intricately shaped vanes that are critical to part performance. (b) Microstructure of the vane cross section before and after HIP. Because the porosity tends to be small and evenly dispersed, the volumetric shrinkage is minimal, and no significant change in shape occurs during HIP.

Fig. 12

(a) Porosity caused by interdendritic shrinkage during the solidification of a cast 4340 steel. The post-HIP microstructure of this component, shown in (b), provides evidence that HIP eliminated the microporosity. These two micrographs also show the potential for HIP to alter the microstructure of a material beyond the elimination of porosity. Comparing (a) and (b), it is apparent that both microstructural coarsening and significant homogenization took place during the HIP cycle. Nital etch

Fig. 13

Schematic showing the effect of HIP on a pore that is connected to the surface of a casting. The inert gas used as a pressurization medium easily penetrates the internal pore network. Because the pressure of the gas within the pore interior is equal to the applied external pressure, no pore shrinkage occurs.

Surface-Connected Porosity. The presence of surface-connected porosity is the most common cause for dissatisfaction when a foundry or end user decides to employ HIP to solve a porosity problem. The simple fact is that there must be a gastight layer of material surrounding the pores for HIP to work. The vast majority of a pore network can be subsurface, but if there is one pinhole penetration of the surface that opens the pore network to the external highpressure gas, the gas will penetrate the pore network. Once the gas pressure within the pore network has equalized with the external pressure, no densification will occur. Figure 13 illustrates the concept of surface-connected porosity, while Fig. 14 shows a HIPed bronze casting that, after machining, exhibited a spongy pore network that was obviously surface connected during the HIP cycle. The best solution to a surface-connected porosity problem is to eliminate its occurrence in the casting operation through modification of the mold design, by incorporating chills into the mold, or by using any other technique that will help to ensure that a solid layer of metal is formed on all surfaces. There are also some postcasting solutions to surface-connected porosity. If porosity is exposed during the removal of the gates and risers, leaving a sufficient amount of gating material or HIPing the part with risers attached often ensures that any porosity that crosses from these attachments into the part itself will be eliminated. In some cases, a surface-sealing process, such as weld overlay, can be used to seal surface-connected porosity. The sealed pore will contain air, so HIP could create a network of oxides in the region where the pore had been, but, if the end result is acceptable, sealing off a surfaceconnected pore network is one way to salvage an expensive casting that would otherwise become scrap. Thermally induced porosity occurs when a subsurface pore contains a gas that has both a low solubility and a low diffusivity within the metal. Illustrated schematically in Fig. 15, the trapped gas is unable to migrate away from the pore during HIP; thus, during pressurization, the pore shrinks, and the pressure of the gas within the pore rises until the combination of the pressure of the gas within the pore and the flow stress of the metal is in balance with the external hydrostatic stress acting on the part, limiting any further densification. After HIP,

Hot Isostatic Pressing of Castings / 415

Fig. 14

Bronze casting with sponge porosity that was not removed by HIP. The immediate proximity of the pores to the as-cast surface indicates that this is an interconnected network of porosity that broke through to the surface of the casting. As a result, high-pressure argon was able to penetrate the pore network during the pressurization portion of the HIP cycle. With gas pressure within the pores equal to the external pressure, no densification took place.

the gas-filled pores will have shrunk to the point where they may be difficult to resolve with a light microscope, but they still contain gas under a very high pressure. Any subsequent heating of that material in a lower-pressure environment, such as in a heat treating operation, will soften the metal to the point where the pores containing the pressurized gas expand. The post-HIP heating seems to create porosity where none had existed before, thus giving the phenomenon its name. In fact, the porosity was always there; the HIP only decreased the size of the pores to the point where they were “invisible” for a period of time. Thermally induced porosity is much more common in HIPed powder metallurgy materials, but it has been observed in high-pressure die castings. The common belief is that the vaporization and cracking of die lubricants, combined with turbulence during die fill and rapid solidification, results in porosity that contains low-diffusivity, high-molecular-weight gases. As a result of this behavior, high-pressure die castings are the one type of casting where HIP cannot routinely eliminate porosity.

Economics of HIP Like heat treating, HIP is one of many postcasting processes that can provide enhanced mechanical properties in a cast component. It has been shown that HIP combined with a good-quality traditional casting technique, such as sand, investment, or permanent mold casting, can produce parts that meet or exceed those properties obtained using the newer casting processes that employ vacuum or pressurized inert gas as a means of minimizing pore formation during solidification. For foundries with an installed sand, investment, or permanent mold operation, the incorporation of HIP into the postcasting process may yield a high-end

Fig. 15

Schematic showing the phenomenon of thermally induced porosity. If an isolated pore contains a gas that has both a low solubility and a low diffusivity in the metal, then during HIP the pore shrinks due to the applied hydrostatic pressure, but the gas remains in the pore. The pore will shrink until the pressure of the gas within the pore is approximately equal to the externally applied hydrostatic pressure. When the HIP vessel retains pressure while the process atmosphere cools, then the flow stress of the metal increases and the pores remain small, often indistinguishable from small inclusions, albeit still containing pressurized gas. Subsequent heating at nearatmospheric pressures, for example, during heat treating, lowers the flow stress of the metal, and the pore expands.

product without having to purchase capitalintensive and complex foundry equipment (Ref 31). The cost of installing a commercial HIP system with a working diameter greater than 1 m (3.3 ft) can easily run into the tens of millions of dollars. While some companies have sufficient internal demand to justify in-house HIP, most companies only HIP a fraction of their total production. This fact, possibly combined with the operating costs and risks associated with running a high-pressure process, makes outsourced HIP attractive for many foundries and end users. It is estimated that approximately 70% of all castings HIPed in North America are HIPed using outsourced services.

time at elevated temperature to the overall thermal history of a casting, microstructural changes such as grain growth and precipitate coarsening can occur. With respect to the production of cast components for an application, HIP is an intermediate process that, like many heat treating techniques, is capable of providing mechanical properties significantly better that those achieved in the as-cast state. As such, HIP should be treated as one of several tools available to the foundryman or end user—a tool that can be used when the performance requirements dictated by the application demand superior mechanical properties from a cast component. REFERENCES

Summary Hot isostatic processing can eliminate internal porosity in castings, thus providing significantly improved ductility, fatigue strength, and impact properties in most castings. If the pore fraction in a casting is high, HIP may also yield an increase in strength. At times, other factors, such as decreased inspection costs and improved surface finish, make HIP attractive to an end user. Because of the complex relationship between porosity, other microstructural features (especially nonmetallic inclusions), and mechanical properties, HIP does not guarantee an improvement in mechanical properties. What is generally assured is that porosity within a casting will be eliminated, provided that the pores do not penetrate the surface of the casting and they do not contain a low-solubility, low-diffusivity gas that would inhibit densification. Distortion during HIP is usually unnoticeable, with the exception being castings that have gross porosity in localized regions. The thermal profile of a HIP cycle is such that the process can eliminate the need for a stressrelief or homogenization anneal. By adding

1. H.A. Saller, S.J. Paprocki, R.W. Dayton, and E.S. Hodge, A Method of Bonding, U.S. Patent 687,842 (classified) and Canadian Patent 680,160, Feb 18, 1964 2. R.W. Dayton and H.A. Saller, “The Fabrication of Zirconium Alloy SIR Fuel Elements,” Report BMI-941, Battelle Memorial Institute, Columbus, OH, Aug, 13, 1954 3. A.H. Clauer, K.E. Meiners, and C.B. Boyer, “Hot Isostatic Pressing,” Report MCIC-82–46, Metals and Ceramics Information Center, Battelle Columbus Laboratories, Columbus, OH, Sept 1982 4. E.S. Hodge, Elevated-Temperature Compaction of Metals and Ceramics by Gas Pressure, Powder Metall., Vol 7 (No. 14), 1964, p 168–201 5. C.B. Boyer, The Evolution and Future of HIP Systems, Proceedings of Second International Conference on Isostatic Pressing (Stratford-Upon-Avon), Met. Powder Rep., Sept 21–23, 1982, p 7–1 to 7–40 6. C.B. Boyer, F.D. Orcutt, and R.M. Conaway, High Temperature High Pressure Autoclaves, Chem. Eng. Prog., Vol 62 (No. 5), May 1966 p 99–106

416 / Principles of Solidification 7. C.B. Boyer, private conversations, June 2005 8. C.B. Boyer, The Development of a High Pressure Idea, High Pressure—Codes, Analysis and Applications, ASME-PVP Vol 263, A. Khare, Ed., 1993 9. D.L. Kerr, R.C. Lemon, and E.E. Stonebrook, Castings, U.S. Patent 3,496,624, Feb 20, 1970 10. “Casting Densification Process,” TMD Report 5, Alcoa Technology Marketing Division, Alcoa 11. J.C. Hebeisen, HIP Casting Densification, Proceedings from ASME-PVP Conference. Aug 1999 (Boston, MA) 12. H.D. Hanes and J.M. McFadden, HIP’ing of Castings: An Update, Met. Prog., April 1983 13. R. Widmer, Densification of Castings by Hot Isostatic Pressing, Proceedings of the 26th Annual Meeting of the Investment Casting Institute, Oct 1978 (Phoenix, AZ) 14. “Process for High Integrity Castings,” Report AEG-896, Materials and Process Technology Laboratories—Aircraft Engine Group, General Electric Company, Cincinnati, OH, Sept 1974 15. R.W. Widmer, HIP’ing Makes Inroads in Casting Densification, Met. Prog. Foundry Technol., Aug 1982 16. H.D. Hanes and J.M. McFadden, HIP’ing of Castings: An Update, Met. Prog., April 1983 17. R. Irving, Hipping Is One Way to Check Porosity in Cast Components, Iron Age, Nov 19, 1982

18. A. Squire, Density Relationships of Iron Powder Compacts, Trans. AIME, Vol 171, 1947, p 485–505 19. D.A. Gerard and D.A. Koss, The Dependence of Crack Initiation on Porosity During Low Cycle Fatigue, Mater. Sci. Eng. A, Vol 129, 1990, p 77–85 20. J.F. Major, Porosity Control and Fatigue Behavior in A356T61 Aluminum Alloy, AFS Trans., Vol 105, 1997, p 901–906 21. J. Campbell, The New Metallurgy of Cast Metals, Proceedings from the Second International Aluminum Casting Technology Symposium, Oct 7–9, 2002 (Columbus, OH), ASM International, 2002 22. Q.G. Wang, D. Apelian, and D.A. Lados, Fatigue Behavior of A356-T6 Aluminum Cast Alloys, Part I: Effect of Casting Defects, J. Light Met., Vol 1, 2001, p 73–84 23. H.V. Atkinson and S. Davies, Fundamental Aspects of Hot Isostatic Pressing: An Overview, Metall. Mater. Trans. A, Vol 31, Dec 2000, p 2981–3000 24. T. Crousatier, Fonderie, Vol 347, 1975, p 265–280 25. H.V. Atkinson and B.A. Rickinson, Hot Isostatic Processing, Adam Hilger, Bristol, 1991 26. S.J. Mashl, J.C. Hebeisen, D. Apelian, and Q.G. Wang, “Hot Isostatic Pressing of A356 and 380/383 Aluminum Alloys: An Evaluation of Porosity, Fatigue Properties and Processing Costs,” SAE Tech. Pub. 2000–01–0062, SAE Inc., 2000

27. Stored Energy of a Pressurized Gas Vessel, Appendix E, Pressure Safety and Cryogenics, Chap. 7, Lawrence Berkeley Health and Safety Manual (Publication 3000), Lawrence Berkeley National Laboratory, http://www.lbl.gov/ehs/pub3000/ (accessed Apr 30, 2008) 28. M.M. Diem, S.J. Mashl, and R.D. Sisson, Evaluating a Combined HIP and Solution Heat Treat Process for Aluminum Castings, Heat Treat. Prog., Vol 3 (No. 4), 2003, p 52–55 29. S.J. Mashl and M.M. Diem, “The Response of Two Al-Si-Mg Castings to an Integrated HIP and Heat Treat Process,” SAE Tech. Pub. 2004–01–1019, SAE Inc., 2004 30. S.J. Mashl, The Influence of Oxide Inclusions on the Post-Hip Fatigue Life of Two Al-Si-Mg Castings, Proceedings: Shape Casting: The Second International Symposium, P.N. Crepeau, M. Tiryakioglu, and J. Campbell, Ed., The Minerals, Metals and Materials Society, 2007 31. J.C. Hebeisen and B.M. Cox, “The Effect of HIP Processing on the Properties of A356 T6 Cast Aluminum Steering Knuckles,” SAE Tech. Pub. 2004–01– 1027, SAE Inc., 2004 32. S.J. Paprocki, E.S. Hodge, and P.J. Gripshouer, Gas-Pressure Bonding, DMIC Report 159, Defense Materials Information Center, Battelle Memorial Inst, Columbus, Sept. 1961

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 419-424 DOI: 10.1361/asmhba0005233

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Numerical Methods for Casting Applications* SIMULATION OF CASTING is motivated by the need to minimize casting defects by understanding and controlling the fluid flow and thermal processes throughout the casting process. However, modeling of casting processes is difficult because computational models must account for a variety of complex phenomena simultaneously, including complex fluid flows, heat transfer with solidification, electromagnetic fields, nonlinear solid mechanics, complex threedimensional mold geometries, and microstructural and defect evolution. Additionally, the simulation must yield results within a reasonable period of time using a reasonable allocation of computing resources. Computational fluid dynamics (CFD) is one of the tools available for understanding and predicting the performance of thermal-fluids systems. It is used to provide insight into thermal-fluids processes, to interpret experimental measurements, to identify controlling parameters, and to optimize product and process designs. Computational fluid dynamics is one discipline falling under the broad heading of computer-aided engineering, which together with computer-aided design and computer-aided manufacturing comprise a mathematics-based approach to engineering design, analysis, and fabrication. Since casting is a fluid-based manufacturing process, CFD is a natural tool to use. In this article, the basic principles of CFD are qualitatively described. The numerical methods used to solve the CFD equations are described, including geometry description and discretization. Finally, the application of CFD to a few casting problems is demonstrated.

Fundamentals of CFD Computational fluid dynamics has as its objective the numerical solution of fluid-flow equations. The calculus problem of solving a coupled system of nonlinear partial differential equations (PDEs) for the variables of interest (e.g., velocity, pressure, and temperature) is transformed into an algebra problem of solving

a large system of simultaneous linear equations for discrete unknowns that represent the state of a thermal-fluids system; the latter is amenable to numerical solution on a digital computer. This is a somewhat abstract description of CFD, but it is necessary to speak in general terms when introducing a subject that encompasses such a wide variety of solution techniques. There are several numerical methods for solving CFD equations, and this overview focuses on the most common: finite difference, finite volume, and finite element. Other methods include spectral and some computational particle methods. In this article, the terminology computational fluid dynamics is reserved for computationally intensive three-dimensional simulations of thermal-fluids systems where nonlinear momentum transport plays an important role. It does not encompass all branches of numerical analysis as applied to fluid-dynamics problems. In particular, consideration of zero- or quasi-dimensional analysis of fluids systems (Ref 1, 2) and linear heat conduction or potential flow problems (Ref 3, 4) is excluded. For a more rigorous description and treatment of CFD and solution techniques, see “Computational Fluid Dynamics” in Materials Selection and Design, Volume 20 of ASM Handbook (Ref 5). The advent of three-dimensional calculations has increased the engineering relevance of CFD, but many obstacles must be overcome before CFD realizes its full potential as an engineering design tool. The most limiting factor is spatial resolution. Most fluids of practical interest have features whose relevant spatial and temporal scales span many orders of magnitude. For example, modeling of structure formation in castings involves several length scales ranging from the atomic level (10
10 to 10
9 m), to the microstructural level (10
7 to 10
4 m) for describing dendritic or eutectic structure, to the macroscopic dimensions of the casting (10
2 to 1 m). Computer performance is not sufficient to fully resolve these phenomena at all length scales. Thus, the effects of small-scale, unresolvable features on

the large-scale, average flow features of interest are “modeled” through modifications to the governing PDEs. Other examples include turbulence models, combustion models, and multiphase flow models. All models necessarily introduce imprecision, and an ongoing goal of research is to improve the accuracy of these models. Other issues for three-dimensional engineering CFD include geometry acquisition and grid generation, numerical accuracy, and diagnostics to extract the physical information of interest from the computations.

Governing Principles The equations of fluid dynamics can be derived from kinetic theory or continuum points of view (Ref 6–8), each of which complements the other. Kinetic theory regards the fluid as made up of molecules subject to collisions and intermolecular forces. Kinetic theory derivations are valid only for dilute gases but give detailed information about how such transport phenomena as stresses and heat fluxes arise from molecular fluctuations, which in turn are related to the average molecular properties that the fluid equations solve. Continuum derivations regard the fluid as a continuous medium and show the applicability of the fluid equations to a much broader class of media than dilute gases but do not capture how fluctuations at the atomic scale affect transport phenomena. Continuous, Compressible Media. Three basic physical principles, applicable to any continuous medium, are used in continuum derivations:  Conservation of mass  Newton’s second law that force equals mass

times acceleration

 The first law of thermodynamics, which

states that total energy, in all its forms, must be conserved These three principles lead to the basic three equations of motion: the mass, or continuity, equation; the momentum equation; and the total

*Adapted from P.J. O’Rourke, D.C. Haworth, and R. Ranganathan, “Computational Fluid Dynamics,” Materials Selection and Design, Vol 20, ASM Handbook, ASM International, 1997

420 / Modeling and Analysis of Casting Processes energy equation. Each of the articles that follow in this Section develops these fundamental balance equations in the manner appropriate to the specific modeling task. Constitutive Relations of Fluid Flow. To complete the balance equations, the stress and heat flux must be expressed in terms of known fluid variables and their derivatives. These expressions are known as constitutive relations. A fluid is a medium for which the nonhydrostatic part of its stress depends only on its rate of deformation. It has no memory of its previous configurations. For most casting simulations, the liquid metal is modeled as a Newtonian fluid (a viscous fluid whose shear stresses are a linear function of the fluid strain rate) and represented by the Navier-Stokes equations (Ref 9). However, casting processes may involve non-Newtonian flow (Ref 9). For certain flow situations, considerable computer time can be saved by solving simplified forms of the compressible flow equations, for example, by solving for the steady-state solution or assuming incompressibility (Ref 5). Turbulence. There are many flow situations in which fluids undergo changes in their properties, such as their velocities, in which the property varies dependently at two different size or time scales that differ by many orders of magnitude. Examples are the seemingly chaotic motions in a turbulent flow or in a multiphase flow such as a liquid spraying into a gas. For instance, classical theories of turbulence predict that the ratios of the largest to the smallest fluctuation length scales of turbulent flows are dependent on the velocity and size scales of the largest turbulent eddies (Ref 10). Mathematical strategies for calculating average flow fields can be found in Ref 5.

Numerical Solution of the Fluid-Flow Equations Numerical analysis methods are used to solve CFD equations for a continuously varying flow field, which has an infinite number of degrees of freedom, by representing the flow field with a finite set of data. Time-dependent terms in the governing equations are solved incrementally by a determined time step. On the other hand, spatially-dependent terms are solved by dividing the computational domain into nonoverlapping cells, a process known as discretization, and generating an array of cells, referred to as a grid or mesh. This article introduces the discretization techniques used by finite-difference, finite-volume, and finite-element methods of numerical analysis. The discrete equations of the finitedifference, finite-volume, and finite-element techniques look similar and are referred to generically as difference approximations. The application of arbitrary Lagrangian-Euler discretization techniques is addressed briefly. There are other models and methods for discretization that are outside the scope of this article. This article is

only a general introduction, and the reader should consult Ref 3 and 11 to 15.

Finite-Difference Methods (FDMs) In FDMs, the entire fluid region of interest is divided into nonoverlapping cells, and approximate values of the fluid variables are stored in each cell. This subdivision is called a grid or a mesh. Derivatives are approximated by taking differences between the variable values in neighboring cells, using the idea of a Taylorseries expansion. Order of accuracy is one measure of the accuracy of an FDM. To test the accuracy of a finitedifference solution, one can refine the grid by reducing the cell size. When cell size is reduced by a factor of 2, numerical errors will be reduced approximately by a factor of 4 when using a second-order method, but only by a factor of 2 with a first-order method. Thus, it may seem desirable to use only methods with a very high order of accuracy. In practice, however, it is difficult to define high-order methods near boundaries, and often, numerical solutions using high-order methods have oscillations in regions of steep gradients. Because of these difficulties, most modern FDMs have second- to fourth-order accuracy and sometimes drop to first-order accuracy in regions of steep gradients.

Finite-Volume Methods (FVMs) As with FDMs, FVMs subdivide the computational region into a mesh of cells, but finitevolume cells can be arbitrary quadrilaterals in two dimensions, hexahedra in three dimensions, or indeed any shape enclosed by a set of corner points. In contrast, FDMs are defined on grids that are obtained using orthogonal curvilinear coordinate systems. Thus, FVMs have much more geometric flexibility than FDMs. Finitevolume methods approximate forms of the fluid equations that are integrated over these cells, which are also called control volumes. Using FVMs, one can easily construct discrete approximations that have the conservative property; that is, the discrete approximations can mimic the physical laws from which the fluid equations were derived by conserving properties such as computed mass, momentum, and energy. A conservative approximation has the property that if two cells share face, the contributions due to fluxes through the common face cancel each other. Conservative difference approximations have a desirable accuracy. For example, it can be shown that difference approximations that conserve mass, momentum, and energy will calculate the correct jump conditions across shocks without having to resolve shock structure (Ref 3). A problem with FVMs is that it is difficult to formulate higher-order FVMs. When a FVM is specialized to a finite-difference grid, the difference approximations look very much like finitedifference approximations, and one can perform

Taylor-series expansions and determine the order of accuracy of the method. When more general meshes are used, however, it is unclear whether the same accuracy can be expected.

Finite-Element Methods (FEMs) Finite-element methods (Ref 11) use a consistent spatial interpolation when evaluating all the spatial derivative terms in the fluid dynamics equations. These methods have long been popular in stress-analysis problems and have recently been gaining popularity in CFD problems because of advances in the methodology. With the FEM, functions of continuous quantities (temperature, displacement, etc.) are represented by approximations; that is, heterogeneous structures are divided into small elements, each of which is approximated as homogeneous. For example, using the FEM, a circle can be represented by a circumscribed polygon, where each side of the polygon is a finite element. Similar to FVMs, the computational domain is subdivided into either arbitrary hexahedra or tetrahedra (Fig. 1, 2). Each discrete, connected cell is a finite element, and the points at which the finite elements are connected are the nodes. When the finite-element grid is refined in such a way that the dimensions of elements are reduced by a factor of 2, then the difference between the computed and exact solutions, as measured by a global integral of this difference, is reduced by a factor of 4. Higher-order FEMs can be constructed by adding midside nodes to the elements and using nonlinear basis functions.

Arbitrary Lagrangian-Euler Equations Another approach is to discretize the flow domain using arbitrary Lagrangian-Euler (ALE) techniques. The ALE approach allows the free surface and mesh to deform together. The advantage is that the free surface is represented directly. However, large surface distortions require frequent remeshing, and the method cannot support situations where the free surface is not defined by a single value, such as when waves or splashing occur (Ref 9).

Grid Generation for Complex Geometries Most castings of industrial importance have complex geometries similar to the casting simulation shown in Fig. 3. Before applying most of the CFD methods described, a computational grid must be generated that fills the flow domain and conforms to its boundaries. For complex domains with curved or moving boundaries, or with embedded subregions that require higher resolution than the remainder of the flow field, grid generation can be a formidable task requiring more time than the flow solution itself. Two general approaches are

Numerical Methods for Casting Applications / 421

Fig. 2

Principal cell or element types for computational fluid dynamics. (a) Tetrahedron: four vertices or nodes, four faces, and six edges for each element. (b) Hexahedron: eight vertices or nodes, six faces, and twelve edges for each element. (c) Sampling of possible edge and/or face degeneracies for hexahedral elements. Source: Ref 5

Fig. 1

Examples of grids used in computational fluid dynamics calculations. Two-dimensional examples are shown for clarity. (a) Structured grid. (b) Block-structured grid. (c) Unstructured hexahedral (quadrilateral) grid. (d) Unstructured tetrahedral (triangular) grid. (e) Local mesh refinement via a transition region on an unstructured hexahedral grid. (f) Chimera grid. Source: Ref 5

available to deal with complex geometries: use of unstructured grids and use of special differencing methods on structured grids. Meshing Complexity. Figure 1 shows examples (in two dimensions) of several possible CFD grid arrangements. In a structured threedimensional grid (Fig. 1a), one can associate with each computational cell an ordered triple of indices (i, j, k), where each index varies over a fixed range, independently of the values of the other indices, and where neighboring cells have associated indices that differ by
þ1. Thus, if Ni, Nj, and Nk are the number of cells in the i-, j-, and k- index directions, respectively, then the number of cells in the entire mesh is NiNjNk.

Additionally, each interior vertex in a structured grid is a vertex of exactly eight neighboring cells. In an unstructured grid (Fig. 1c, d), on the other hand, a vertex is shared by an arbitrary number of cells. Unstructured grids are further classified according to the allowed cell or element shapes (Fig. 2). In the case of FVMs in particular, an unstructured CFD code may require a mesh of strictly hexahedral cells (Fig. 2b), hexahedral cells with degeneracies (Fig. 2c), strictly tetrahedral cells (Fig. 2a), or may allow for multiple cell types. In any case, the cells cannot be associated with an ordered triple of indices as in a structured mesh.

Intermediate between structured and unstructured meshes are block-structured meshes (Fig. 1b), in which “blocks” of structured grid are pieced together to fill the computational domain. By using a block-structured approach, complex geometries can be represented while still using locally-structured mesh solvers (Ref 9). There are three advantages of unstructured meshes over structured and block-structured meshes. First, unstructured meshes do not require that the computational domain or subdomains be topologically cubic. This flexibility allows construction of unstructured grids in which the cells are less distorted and therefore give rise to less numerical inaccuracy, compared to a structured grid. Second, local adaptive mesh refinement (AMR) is naturally accommodated in unstructured meshes by subdividing cells in flow regions where more numerical resolution is required (Fig. 1e, f). Such subdivisions cannot be performed in structured meshes without destroying the logical (i, j, k) indexing. Third, in some cases, particularly when the cells are tetrahedra, unstructured grid generation can be automated with little or no user intervention (Ref 17). Thus, generating unstructured grids can be much faster than generating block-structured grids. However, unstructured-mesh CFD codes generally demand higher computational resources. Additional memory is needed to store cell-to-cell and vertex-to-cell pointers on unstructured meshes while this information is implicit for structured meshes. Also, the implied

422 / Modeling and Analysis of Casting Processes connectivity of structured meshes reduces the number of numerical operations and memory accesses needed to implement a given solution algorithm compared to the indirect addressing required with unstructured meshes. The relative advantages of hexahedral versus tetrahedral element shapes remain subjects of debate in the CFD community. Tetrahedra have an advantage in grid generation, since any arbitrary three-dimensional domain can be filled with tetrahedra using well-established methodologies (Ref 17). By contrast, it is not mathematically possible to tessellate an arbitrary three- dimensional domain with nondegenerate six-faced convex volume elements. Thus, each of the various automatic hexahedral grid- generation approaches that have been proposed (e.g., Ref 18, 19) either yields occasional degeneracies or shifts the location of boundary nodes, thus compromising the geometry. Mathematical Complexity. A second general approach to computing flows in complex geometric configurations shifts the emphasis from complexity in grid generation to complexity in the differencing (mathematical) scheme (Ref 20–22). Structured and block-structured grids are used, but one of three numerical strategies is used to extend the applicability of these grids. The first strategy is to use so-called chimera grids (Ref 20) that can overlap in a fairly arbitrary manner (Fig. 1f). Solutions on the multiple grids are coupled by interpolating the solution from each grid to provide the boundary conditions for the grid that overlaps it. This is a very powerful strategy that naturally compensates for problems in which two flow regions meet at a boundary with a complicated shape or where one object moves relative to another. The second numerical strategy is to use so- called embedded boundaries (Ref 21). Again, structured meshes are used, but the complicated boundary of the computational domain is allowed to cut through computational cells. Special numerical methods are then used in the partial cells that are intersected by the boundary. In the third strategy, local AMR is allowed by using a nested hierarchy of grids (Ref 22). The different grids in the hierarchy are structured and have different cell sizes, but the cells in the more finely resolved grids must subdivide those of the coarser grids. Much development remains before these specialized differencing techniques have the robustness, generality, and efficiency to deal with the variety of problems presented in engineering applications. For the near future, then, the use of various unstructured-mesh approaches is expected to dominate in engineering applications of CFD.

Casting Applications

Fig. 3

Metal casting simulation. (a) Schematic sand casting configuration. (b) Mesh (five million elements) for casting and cooling channels. (c) Computed local solidification times ranging from 1 to 3000 s. Source: Ref 16

In terms of software, it would be impossible to fully catalog all available packages that are suitable for modeling of casting processes. However, Table 1 summarizes a few of the

Numerical Methods for Casting Applications / 423 Table 1 Software packages for numerical solution of computational fluid dynamics problems Software package

Discretization method(a)

Meshing

Web resource

MAGMASOFT Procast FLOW-3D PHYSICA THERCAST

FV FE FV FV FE

www.magmasoft.com www.esi-group.com www.flow3d.com www.physica.co.uk www.transvalor.com

CFX FLUENT

FV FV

Block-structured Fully unstructured Block-structured Unstructured Arbitrary Lagrangian-Euler Block-structured Unstructured

www.ansys.com www.fluent.com

(a) FV, finite volume; FE, finite element

Fig. 4

Simulated flow of fluid through a fan-shaped gate (a) with a top connection to the runner and (b) with a bottom connection to the runner. Source: Ref 27

codes that have been mentioned in the literature (Ref 9). Casting is a process in which parts are produced by pouring molten metal into a cavity having the shape of the desired product. Figure 3(a) is a schematic of a typical sand casting configuration. The fluid properties may be highly temperature dependent and non-Newtonian, including phase changes. Fluid flow plays two important roles in the casting process. First, and most obviously, the flow of molten metal is necessary to fill the mold. Second, and less obvious, is the effect of convective fluid flow during solidification of the casting. It is the task of the foundry engineer to design gating and riser systems (Fig. 3a) that ensure proper filling and solidification. The CFD is an ideal tool for tracking the location of the three- dimensional solid-liquid interface as it moves with time and the heat transfer accompanying the solidification process. Proper designs result in lower scrap and less casting repair at the foundry. An example of a computational mesh and computed solidification times is given in Fig. 3(b) and (c) (Ref 5). Recent references from the literature give ample evidence of the vast amount of CFD activity that is taking place in this area (Ref 9, 16, 23–26). Gate Design. An example of the usefulness of CFD modeling applied to gate design is shown in Fig. 4 (Ref 5). In this example, the goal was to reduce the speed of the flow of melt into the mold cavity. Expanding the horizontal runner proved to be ineffective, so adding a vertical gate to the end of the runner was investigated. Figure 4(a) shows the pattern of fluid flow into the mold when the fan-shaped gate is placed on top of the runner. The simulation shows that the fluid is initially constrained by the vertical sides of the runner, causing the liquid to jet vertically into the mold before falling back to fill the gate from above. When the fanshaped gate is placed at the bottom of the runner (Fig. 4b), the fluid expands smoothly into the expanding volume, filling the gate from below and thus reducing the fluid speed before entering the mold cavity and resulting in a more uniform liquid front. Defect Simulation. Foundry engineers put considerable energy and resources into minimizing casting defects that arise from filling the mold. Many articles in this Volume discuss specific defects, such as gas porosity, shrinkage porosity, oxide bifilms, and so on, in detail. Figure 5 shows an example simulation of internal macroporosity (left) and surface porosity (right). The differing distribution of the porosity is a function of the different freezing ranges of the alloys used. Figure 6 shows the development, transport, and breakup of a large bubble in a liquid column, simulating the surface entrainment of gas during mold filling (Ref 9). Fluid Flow in Tundishes. In the continuous casting of steels, the tundish is the reservoir between the ladle and the mold, and its purpose is to control the pouring head of the steel and to provide a constant casting speed. Fluid flow

424 / Modeling and Analysis of Casting Processes

Fig.

5

Simulated development of internal macroporosity (left) and surface porosity (right). Source: Ref 9

within the tundish is complex and turbulent; modeling requires the solution of three-dimensional Navier-Stokes equations and equations that describe turbulent flow. In addition to the obvious application of using modeling to design tundish geometry, modeling can be used to represent more common situations, such as the change in fluid flow through the tundish resulting from a new chemistry during a grade change. An example simulation of fluid flow in a tundish with a conventional baffle is shown in Fig. 7(a). A longitudinal view of the computed fluid flow when a double weir and dam configuration is used is shown in Fig. 7(b) (Ref 9).

Fig. 6

Fig. 7

10. 11. 12.

REFERENCES 1. Flowmaster USA, Inc., Fluid Dynamics International, Evanston, IL 2. WAVE Basic Manual: Documentation/ User’s Manual, Version 3.4, Ricardo Software, Burr Ridge, IL, Oct 1996 3. P.J. Roache, Computational Fluid Dynamics, Hermosa Publishers, 1982 4. N. Gru¨n, “Simulating External Vehicle Aerodynamics with Carflow,” Paper 960679, SAE International, 1996 5. P.J. O’Rourke, D.C. Haworth, and R. Ranganathan, Computational Fluid Dynamics, Materials Selection and Design, Vol 20, ASM Handbook, ASM International, 1999 6. F.H. Harlow and A.A. Amsden, “Fluid Dynamics,” Report LA-4700, Los Alamos Scientific Laboratory, June 1971 7. W.G. Vincenti and C.H. Kruger, Introduction to Physical Gas Dynamics, Robert E. Krieger Publishing, 1975 8. P.A. Thompson, Compressible-Fluid Dynamics, McGraw-Hill, 1972 9. M. Cross, K. Pericleous, T.N. Croft, D. McBride, J.A. Lawrence, and A.J. Williams, Computational Modeling of Mold Filling and Related Free-Surface Flows in

Simulated transport, development, and breakup of a large bubble. Source: Ref 9

13. 14.

15.

16. 17.

18. 19.

20.

Shape Casting: An Overview of the Challenges Involved, Metall. Trans. B, Vol 37, (No. 6), Dec 2006, p 879–885 L.D. Landau and E.M. Lifshitz, Fluid Mechanics, Pergamon Press, 1959 R. Peyret and T.D. Taylor, Computational Methods for Fluid Flow, Springer-Verlag, 1983 G.D. Smith, Numerical Solution of Partial Differential Equations, 2nd ed., Oxford University Press, 1978 R.D. Richtmyer and K.W. Morton, Difference Methods for Initial-Value Problems, 2nd ed., Interscience Publishers, 1967 C.A.J. Fletcher, Computational Techniques for Fluid Dynamics, Vol I, Fundamental and General Techniques, 2nd ed., Springer-Verlag, 1991 C.A.J. Fletcher, Computational Techniques for Fluid Dynamics, Vol II, Specific Techniques for Different Flow Categories, 2nd ed., Springer-Verlag, 1991 R.H. Box and L.H. Kallien, SimulationAided Die and Process Design, Die Cast. Eng., Sept/Oct 1994 M.C. Cline, J.K. Dukowicz, and F.L. Addessio, “CAVEAT-GT: A General Topology Version of the CAVEAT Code,” Report LA-11812-MS, Los Alamos National Laboratory, June 1990 HEXAR, Cray Research Inc., 1994 G.D. Sjaardeam et al., CUBIT Mesh Generation Environment, Vol 1 and 2, SAND941100/-1101, Sandia National Laboratories, 1994 W.D. Henshaw, A Fourth-Order Accurate Method for the Incompressible Navier-

Simulated fluid flow in a steel continuous casting tundish with (a) a conventional baffle configuration and (b) a double weir and dam configuration. Source: Ref 26

21.

22.

23. 24.

25.

26.

27.

Stokes Equations on Overlapping Grids, J. Comput. Phys., Vol 133, 1994, p 13–25 R.B. Pember et al., “An Embedded Boundary Method for the Modeling of Unsteady Combustion in an Industrial Gas-Fired Furnace,” Report UCRL-JC-122177, Lawrence Livermore National Laboratory, Oct 1995 J.P. Jessee et al., “An Adaptive Mesh Refinement Algorithm for the Discrete Ordinates Method,” Report LBNL-38800, Lawrence Berkeley National Laboratory, March 1996 L. Karlsson, Computer Simulation Aids V-Process Steel Casting, Mod. Cast., Feb 1996 C. Li and B.G. Thomas, Thermomechanical Finite-Element Model of Shell Behavior in Continuous Casting of Steel, Metall. Trans. B, Vol 35 (No. 6), Dec 2004, p 1151–1172 H.J. Thevik, A. Mo, and T. Rusten, A Mathematical Model for Surface Segregation in Aluminum Direct Chill Casting, Metall. Trans. B, Vol 30 (No. 1), Feb 1999, p 135–142 R.D. Pehlke, Computer Simulation of Solidification Processes—The Evolution of a Technology,” Metall. Trans. A, Vol 33, Aug 2002, p 2251–2273 J. Campbell, Casting Practices: The 10 Rules of Castings, Elsevier Butterworth Heinemann, 2004

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 425-434 DOI: 10.1361/asmhba0005234

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Modeling of Transport Phenomena and Electromagnetics Matthew John M. Krane, Purdue University Vaughan R. Voller, University of Minnesota Ben Q. Li, University of Michigan

A COMPLETE MODEL of the solidification of a metal alloy involves coupling of energy, species, mass, and momentum conservation equations (Ref 1–7). This article examines critical features of four key areas of modeling transport phenomena associated with casting processes:





Heat and species transport in a metal alloy Flow of the liquid metal Tracking of the free metal-gas surface Inducement of metal flow via electromagnetic fields

The basic interactions between the phenomena are illustrated in Fig. 1. The modeling of the solidification of an alloy requires the coupling of equations that describe the conservation of energy and solutal species with auxiliary relationships controlled by the alloy phase equilibria. These conservation equations are significantly influenced by the flow of the liquid metal. Modeling the liquid metal flow requires solution of the Navier-Stokes equations that couple momentum and mass conservation. The solution of these equations is coupled to the heat and species transport because it requires an accounting of the momentum transfer between the solid and liquid phases due to the dependence of buoyancy and thermophysical properties on local temperature and composition. Determination of the location and movement of the liquid metal-gas free surfaces requires the coupling of computational front tracking algorithms with the solution of the Navier-Stokes equations. Finally, liquid metal flows can be induced by electromagnetic fields. Accounting for this feature requires the solution of the electromagnetic field equations coupled to appropriate source terms in the Navier-Stokes equations.

Conservation of Thermal Energy

conservation equation for energy transport can be written in terms of either the enthalpy of the solid-liquid mixture or the temperature of the system, or a combination of both. In general, the domain in a solidification process will consist of multiple phases distributed among solid, liquid, and gas fractions. A simplified—but reasonably realistic—system to consider is an alloy in which (a) the transformation from liquid to solid occurs across a temperature range, DTls = Tl (liquidus temperature)  Ts (solidus temperature); (b) the solid phases are lumped into a single solid phase; and (c) no significant gas voids form, that is, the sum of the volume fractions of the solid (s) and liquid (l) can be taken as unity, gs þ g1 ¼ 1. A conservation equation describing the thermal energy transport in this system, dating back to the middle of the last century (Ref 8), is obtained by writing the heat balance in terms of enthalpy. The solid (Hs) and liquid (Hl) phase enthalpies can be defined as:

r ¼ gl rl þ gs rs

(Eq 5)

With these definitions, an appropriate mixture enthalpy balance for the system under consideration can be written:

 

@rH ¼  r ðgs rs Hs Vs þ gl rl Hl V l Þ @t þ r ðkrT Þ þ ST

(Eq 6)

where k ¼ gs ks þ gl kl is a mixture thermal conductivity; Vl and Vs are the velocities of the liquid and solid phases, respectively; and ST is a generic heat source term that may include Joule heating or viscous dissipation. The first term of Eq 6 represents the storage of enthalpy, which is determined by the balance of the other terms, which respectively model the advection and conduction of thermal energy.

Conservation of Solute

ðT Hs ¼

cs da

(Eq 1)

cl da þ Lf

(Eq 2)

Tref

and ðT Hl ¼

A solution of Eq 6 requires a relationship between the enthalpy, temperature, and liquid fraction. This function can be derived from a consideration of the thermodynamics of the system in the form of the phase diagram. To reach this point, however, the local composition

Tref

where c is the specific heat at constant pressure, T is temperature, Tref is a reference temperature, and Lf is the local latent heat at the reference temperature, that is, the difference between the phase enthalpies at T = Tref. A mixture enthalpy can be defined as a mass fraction ( fi) weighted average of the liquid and solid enthalpies: H ¼ fl Hl þ fs Hs

The primary driving force for the solid-liquid phase change in casting processes is the extraction of heat from the liquid melt. To describe the thermal response to this heat removal, the

where the mixture density is defined as:

(Eq 3)

or, in terms of volume fractions (gi): rH ¼ gl rl Hl þ gs rs Hs

(Eq 4)

Fig. 1

Schematic of relationships among various transport phenomena and associated physics

426 / Modeling and Analysis of Casting Processes must be known, so an equation describing the conservation of solute in the system must be developed and solved. For a given solute species, a mixture concentration can be defined by: ð h i 1 ð1  Þrs Cs þ rl Cl dV rC ¼ Vrev

(Eq 7)

Vrev

where Ci is the composition in phase i, Vrev is a representative elemental volume (rev) in the solidification domain, and i is a microscopic phase marker (i = 1 in the liquid, i = 0 in the solid). If the nature of the solid and liquid phases in the rev can be characterized by the volume fraction of liquid: gl ¼ V1rev

Ð

dV ¼ 1  gs

Vrev

and the local solute transport in the liquid phase of the rev is rapid, then Eq 7 can be more conveniently written as: r C ¼ gs þgl rl Cl

where: gs ¼

(Eq 8) Ð gs 0

rs Cs da

In such a system, the conservation of solute can be written as:

 

@rC ¼ r ðgs V s þ gl rl Cl V l Þ @t þr ðgl rl Dl rCl Þ

(Eq 9)

The derivation of Eq 9 neglects interdiffusion among elements and diffusion in the solid phase. These assumptions are justified by noting that a scaling analysis shows that advection dominates species transport throughout the process (Ref 9) and that, for most metal alloys, especially if the alloy is a substitutional solution, the diffusion coefficient in the liquid is much larger than that in the solid (Dl  Ds) (Ref 10).

Explicit Numerical Solution for Energy and Composition A numerical solution of Eq 6 and 9 is required. If a field discretization method is used, for example, finite differences, control volumes, or finite elements, the essential features of the solution are the same. The problem domain is covered by a grid of node points, and, through the conservation statements, the discretization scheme of choice is used to generate sets of algebraic equations that relate each nodal value of a given independent variable to the values at its neighboring nodes. The key nature of the coupled solution of Eq 6 and 9 is well illustrated by considering a time-explicit solution. Assuming that, at a given time, the velocity field is provided and the nodal fields of temperature, liquid fraction, and phase enthalpies are known, the most straightforward computational solution of Eq 6 is to use explicit time integration. In this approach, the Pth nodal mixture volume enthalpy, at the next time step (new), is calculated from:

  ½rHnew ¼ rH P þ
t½r ðgs rs Hs V s þ gl rl Hl V l Þ P  (Eq 10) þr ðkrT Þ þ ST P





where Dt is the discrete time step, and the nomenclature d[
]P implies that the quantity in brackets is represented by a suitable space discretization at the node point “P.” In a similar manner, the updated mixture concentration can be obtained from:  new   r C P ¼ r C P þ
t½r ðgs V s





þgl rl Cl V l Þ þ r ðgl rl Dl rCl ÞP

(Eq 11)

At the end of each time step, in order to proceed to the next time step, the temperature, liquid fraction, phase enthalpy, and phase concentration nodal fields must be extracted from the updated mixture volume enthalpy and mixture concentration nodal fields. In the general case, where macrosegregation is important (see the article in this volume “Modeling of Microsegregation and Macrosegregation”), this calculation could involve intricate calculations to couple thermal and species concentrations through the phase diagram (Ref 7). Two example calculations are outlined here. Example 1. The first example calculation involves a binary alloy, with a known and defined liquidus line. If the Gulliver-Scheil assumption (complete local solute mixing in the liquid, no solute diffusion in the solid) holds for the solidification of this alloy, a relatively straightforward iteration scheme, applied at each node in turn, can be developed to extract the required values from the predicted mixture fields. The key steps in this iteration, for a node located in the phasechange range, are given in Table 1. The obvious place to modify this calculation is in the description of the microsegregation, the transport of the solute in the solid phase. Different microsegregation models will give different results for the integral of composition over the solid volume in step 2 in Table 1. A simple alternative is to assume a lever rule (complete solute mixing in both the liquid and solid), such that: ð1gold l Þ

Ð

old rs Cs da ¼ gold s rs Cs ¼ gs rs ko Cl

(Ref 11)

0

More detailed treatments of the microsegregation that account for finite and nonzero solid

solute diffusion can be incorporated via exact (Ref 12), approximate (Ref 13–16), or numerical (Ref 16, 17) models. Example 2. The second example calculation of mixture composition and enthalpy fields considers the case of a general alloy when macrosegregation is not important, that is, when the mixture concentrations, at a given point in the casting, remain close to the initial concentrations in the molten alloy (rC ffi constant). In this case, the appropriate phase diagram information can be calculated and stored a-priori through the use of a thermodynamics calculation package, for example, JMatPro (Ref 18). This prior information could include variations with temperature of the solid mass fraction, fs(T ) = 1  fl(T ) (calculated by assuming a lever or Gulliver-Scheil rule) (Ref 18); the phase enthalpies, Hs(T ) and He(T ); and the phase densities, rs(T ) and rl(T). The mass fractions are related to the volume fractions in phase k through fk ¼ rs gk =r, so Eq 6 can be rewritten as: rcm eff

 

@T ¼ r ðgs rs Hs V s þ gl rl Hl Vl Þ @t þ r ðkrT Þ þ ST

(Eq 12)

where, for a given temperature T, the modified effective specific heat: cm eff ¼

dH H dr þ r dT dT

(Eq 13)

can be determined from the stored thermodynamic data. The standard definition of the effective specific heat is ceff ¼ dH=dT , and the additional term on the right of the modified form in Eq 13 allows for Eq 12 to be applied to problems where the densities vary with temperature. An explicit time integration of Eq 12: ½T new P ¼ ½T P þ






t m  ½r ðgs rs Hs V s þ gl rl Hl V l Þ rceff

þ r ðkrT Þ þ ST P

(Eq 14)

immediately provides the new time-step value for the nodal temperature field. Subsequent use of these temperature values in the stored thermodynamic data directly provides the nodal field information required for the next time-step calculations.

Table 1 Closure iterations for thermal-solutal coupling in mushy region 0
new

rs Hs gl ¼ ½rH r Hl r Hs l

s

old ð1g   Ðl Þ rC ¼ rs Cs d þ gold l  gl ko Cl þ gl rl Cl (a)

0

3. Estimate a current temperature from the liquidus line. 4. With the estimated liquid fraction, liquid concentration, and temperature, update the phase enthalpies and densities. 5. Repeat steps 1 through 4 to convergence.

T = T(Cl) Eq 1 and 2

(a) The integral over the old solid fraction is known, and it is assumed that all the solid formed in the time step is at the interface equilibrium value. Assumptions: Binary alloy with liquidus line, T = T(Cl). At local scale, complete solute mixing in the liquid and no solute diffusion in the solid (Gulliver-Scheil). Equilibrium at solid/liquid interface (*), Cs ¼ k0 C1 (ko = partition coefficient)

Modeling of Transport Phenomena and Electromagnetics / 427

Equations Cast in AdvectionDiffusion Form The advantage of the explicit calculations in Eq 10, 11, and 14 is that they can easily incorporate complexities in the thermodynamics of the phase change. The explicit nature of the calculation, however, can impose several restrictions on the size of the time step, leading to inefficiencies in calculations. Further, the form of the governing equations, Eq 6 and 9, are not suitable for immediate implementation in most commercial or general application computer codes. Often, a requirement of using these codes is to rewrite the transport equations as general advection-diffusion equations:





@ðrfÞ þ r ðrfVÞ ¼ r ð rfÞ þ Sf @t

(Eq 15)

rH V ¼ r ck rH rf ðH  H ÞV  V  k þr rðH  HÞ þ S c s

s

l

s

s

(Eq 16)

T

s

s

Here, V ¼ fl Vl þ fs Vs . The algebraic shuffling of terms from Eq 6 to 16 has led to an equation that includes two source terms that arise not from the physics but from the manipulation into the form of Eq 15. This form still depends on temperature in the source terms, but the H  T  fl relationships can be found as described previously. Under the assumption of constant specific heats in each phase and a scaling of temperature to set Tref = 0, the energy of Eq 6 can be rewritten, in the form of Eq 6, with f = T:





@ðr cT Þ þ r ðr cT VÞ ¼ r ðkrT Þ @t @ðrfl Lf Þ   r ðrVfl Lf Þ  r  @t   ½rfs ðcl  cs ÞT þ Lf V  V s þ ST ðEq 17Þ



rC V ¼ rrf D rC rf ðC  ÞV  V  þ r rf D rðC  CÞ

@rC þr @t r

l

s l

where f is the transported variable, G is the diffusion coefficient, and Sf is a source/sink term. The advantage of the form in Eq 15 is that it allows for a time-implicit solution using standard algorithms (e.g., Ref 19–22). The manipulation of the energy and species governing equations into the form in Eq 15 can lead to new terms that are typically lumped into the source Sf. An example of this formulation, using f ¼ H in Eq 6, is from Ref 11: @ðrHÞ þr @t r

with heat extraction. On the other hand, the temperature formulation is much easier to relate to the equilibrium phase diagram to obtain the fraction solid, an advantage that can be important in multicomponent alloys with complicated relations for phase equilibria. In a general casting model, Eq 16 and 17 need to be accompanied by solute species equations, for example, Eq 9. Such equations can also be written in the advection-diffusion form. For example, assuming that the solid density is a function of temperature, Eq 9 can be written in the form of Eq 15 for mixture composition by setting f ¼ C ¼ fs þfl Cl :



The choice of an enthalpy (Eq 16) or temperature (Eq 17) formulation is influenced by the behavior of the system and the ease with which H and T can be related to the fraction liquid. During isothermal phase changes, convergence of the solution to the phase change temperature can be difficult, because the latent heat is absorbed or released only when the system crosses that point. In this case, tracking mixture enthalpy as the primary variable eliminates the convergence problem, because H , unlike T, varies continuously

l

l

s

since these effects directly control the microstructure and final quality of a cast part, the prediction of velocity and pressure in the liquid metal is also required because they can significantly alter the temperature and composition fields. The basic equations for the flow and pressure are the Navier-Stokes equations that express conservation of mass and linear momentum. Derivation of the general forms of these equations can be found in many standard texts (e.g., Ref 24, 25). For a Newtonian fluid, these equations can be written as:



@r þ r rV ¼ 0 @t

and

l





@ðrVÞ þ r ðrVVÞ ¼ r ðmrVÞrP þ SM (Eq 20) @t

s

l

(Eq 19)

(Eq 18)

As noted previously, Eq 16 to 18 are in a form that is suitable for a time-implicit solution using common numerical algorithms. This alternative frequently has a computational advantage over the previously discussed explicit scheme because accuracy, not stability restrictions, is the limit on the size of the simulation time step. The coupled implicit solution of Eq 16 or 17 and 18 also can follow the template provided in Table 1. In this case, however, an additional outer iteration in a time step will be required to recalculate discrete forms of the governing equations in light of the updated values from the inner iterations.

A Comment on Two-Phase Modeling The continuum mixture models just described treat the entire domain as one phase and predict mixture quantities for the field variables (e.g., velocity, enthalpy), combined with different closure relations to calculate the fraction solid and the values of those field variables in the separate phases. Another approach is a two-phase model (Ref 4, 23), in which a separate transport equation is solved for each phase present. For example, there are energy conservation equations for each phase, which are coupled by terms that describe the flow/transfer of energy between the two phases. This approach makes possible a more complete description of the phenomena at the microscopic level and links those phenomena explicitly to the macroscopic scale. In practice, a direct use of a two-phase model requires significant effort to determine and model the interactions at the phase boundaries. The concepts in this modeling approach, however, are invaluable in constructing appropriate, valid, and tractable mixture models.

The left side of Eq 20 represents the local and convective acceleration of the fluid, while the right side contains the forces affecting fluid momentum. The first two terms on the right are the viscous shear and a pressure gradient. The last term accounts for body forces, for example, gravity and electromagnetic forces, and additional viscous terms not expressed by r  (mrV ) (Ref 19). The central task in using Eq 19 and 20 to predict the fluid flow and pressure in a casting requires the adaptation of the Navier-Stokes equations to account for physical phenomena specific to the two-phase (solid + liquid) solidifying systems. A formal way to achieve this is to use a two-phase volume-averaging approach (Ref 23). In this approach, separate solid and liquid equations, expressing mass and momentum conservation at a “microscopic point” in the liquid and solid phases, are developed. These equations are then averaged over a representative elementary volume, containing both the solid and liquid phases, to arrive at macroscopic statements of mass and momentum conservation in each phase. The critical modeling component in this approach is the construction of appropriate interphase transfer terms to describe the momentum and mass exchanges between the phases. The conservation of momentum in the liquid phase can be written in the form of Eq 19:





@ðrl gl V l Þ þ r ðrl gl V l V l Þ ¼ r ðgl mrV l Þ  gl rP @t þ gl rl g þ SM (Eq 21)

Basic Equations for Mass and Momentum Conservation

where, in addition to the terms previously noted, the general source (SM) will also contain extra terms arising from the volume-averaging process (for example, see Table 2 in Ref 4). This statement of the fluid momentum conservation also includes an explicit accounting of the buoyancy force due to gravity, g. A suitable mixture mass conservation that can be used with Eq 21 is:

While attention typically is turned first to the prediction of the transport of heat and species,

@r þ r ðrl V l Þ ¼ r ðrs gs V s Þ @t





(Eq 22)

428 / Modeling and Analysis of Casting Processes With this system, it is important to note that in the fully liquid domain, the equations revert to the basic form of the Navier-Stokes equations given in Eq 19 and 20. Arriving at a computationally tractable model for the flow and pressure in a given system requires efforts focused on judicial use of auxiliary relationships and of assumptions to construct the source terms in the previous equations. The key tasks that must be undertaken include (a) the specification of source terms that can model the momentum transfer between the solid and liquid phases, (b) the accounting of velocity fluctuations arising from turbulence, (c) the appropriate accounting of other forces that drive or restrain the fluid flow, for example, density variations, (d) the construction of approaches that allow for the tracking of gasliquid free surfaces, and (e) the development of equations that allow for the specification of forces resulting from electromagnetic fields. Elements in these tasks are outlined in more detail in the following sections.

Modeling the Momentum in the Solid + Liquid Mushy Region A key part of the source term SM in Eq 21 is a drag force that accounts for the exchange of momentum between the solid and liquid phases in the mushy region. In regions where the solid dendritic matrix is rigid and moves at a prescribed velocity (either zero or a specified casting speed), there is a significant amount of shear transmitted at the solid-liquid interface, and it is worthwhile to consider how this drag will depend on the morphology of this region. At high liquid fractions, where dendrites are unconnected, the increased drag on the liquid is due to the no-slip condition on the very high solid surface area per unit volume. While the flow is significantly retarded by this friction, even at very small amounts of solid, there is still enough potential for the flow to cause the exchange of fluid between the mushy zone and the melt pool. Closer to the solidus, where the dendrites are much more interconnected, the surface area/volume ratio is a less important contributor to the frictional drag than the tortuous and narrow passages through the solid structure. Here, the drag is so large that typical levels of buoyancy and electromagnetic forces cannot induce significant velocities, and the very weak flow is controlled primarily by shrinkage. In order to predict this frictional drag effect, which acts throughout a volume rather than on a surface, it is modeled as a volumetric sink of momentum in the Navier-Stokes equations. Given the slow speed of flows in typical mushy zones, the relation for the drag is usually based on a modified Darcy’s law (Ref 26), which gives the drag contribution to the source term in Eq 21 as (Ref 4): SD ¼ g2l mK1 ðV l  V s Þ

(Eq 23)

where K is a permeability tensor. While this formulation for the interfacial drag looks simple, there is a great complexity lurking in the treatment of the permeability. A common approach is to use the Blake-Kozeny model for permeability in terms of solid volume fraction ðg1 Þ and dendritic arm spacing (d). Assuming the mushy region morphology forms an isotropic medium: K¼

d2 g3l I 180 ð1  gl Þ2

(Eq 24)

This expression was derived for flow perpendicular to a bundle of cylindrical tubes of uniform size and spacing and is only valid at low liquid fractions, which makes it less attractive than it first appears. As noted, with the exception of shrinkage-induced flows, all of the interesting flow effects in the mushy zone occur near the liquidus temperature at high liquid fractions. Also, the geometry for which it is used is much different than a bundle of parallel tubes, because real dendritic microstructures are not evenly spaced and have side branches that vary in size. Several studies extended permeability models to high liquid fraction, where Eq 24 gives values of permeability that are much too high. Due to experimental difficulties, numerical studies have been used to predict flow behavior and relate it to permeability (Ref 27, 28). General correlations are not provided, but the plotted results show that data from a range of Reynolds numbers and simulated dendrite geometry collapse to one curve of a nondimensional, isotropic permeability (K*) as a function of g1 . To accomplish this simplification, the proper length scale for the permeability was found to be not the dendritic arm spacing, as for Eq 24, but the inverse of the surface area/ volume ratio (Sn), so: K  ¼ KSn2

(Eq 25)

The effect of microstructural anisotropy on the permeability model should also be considered. Flow parallel to the axis of an array of columnar dendrites is obstructed mainly by the secondary dendrite arms, while the liquid moving perpendicular to that axis must move around the primary arms, giving a very different flow pattern. One example of different permeability functions for flow parallel and perpendicular to the primary dendrite axis for liquid fraction above 0.66 can be found in Ref 29. When g1 is near 1, the ratio of the permeabilities given by these correlations is approximately ½, and both are much smaller than the prediction of the Blake-Kozeny model (Eq 24). The permeability in equiaxed microstructures is much more isotropic. To solve the mass and momentum equations in the solid + liquid region, the solid velocity must be solved or prescribed. An obvious assumption is to assume a fixed solid structure

such that Vs = 0. In many solidification processes, such as directionally solidified turbine blade casting or electroslag remelting of steels and superalloys, most or all of the solid forms in a rigid structure with little free-floating solid. In such cases, this assumption is sufficiently realistic and results in a much simpler set of equations than if the solid is not rigid. In solving Eq 21 and 22 for the case of a fixed solid structure, a simplification is often made to assume that drag friction SD is the dominant term in the source SM, that is, the additional averaging terms present in this source are neglected. In contrast to the fixed solid, in processes that rely on grain refinement to produce a more uniform, equiaxed microstructure or in cases that have a columnar-to-equiaxed transition, there will be a considerable number of solid particles formed in the melt or detached from the mold walls and the rigid dendritic matrix. These solid particles, moving with the liquid flow or settling toward the bottom of the liquid pool, can substantially alter the flow as well as the macrosegregation and heat-extraction patterns. Models to predict the motion of these free-floating solid particles during solidification have been developed (Ref 30–32). When these solid particles are present, the difficulties and uncertainties in the model multiply rapidly. The predictions of solid transport will depend on the size and shape distributions of the particles, as well as the mechanisms by which the dendrites detach from and attach to the rigid structure. Examples of the effect of solid motion are found in Ref 33 and 34.

Modeling Turbulence While the liquid velocities and melt pools are generally small in casting processes, the flows are not always laminar. In large-scale liquid metal processes, such as vacuum arc or plasma arc remelting, or direct chill or continuous casting, the flows can transition from laminar to turbulent and relaminarize in different regions of the melt pool. Turbulence is also more likely to develop in cases in which the flows are driven by electromagnetic forces, which can produce much higher velocities than buoyancy alone. The hallmark of a turbulent flow is the appearance of a broad spectrum of local velocity fluctuations, resulting in the formation of flow eddies over a large range of length scales. There is a plethora of turbulence models that can be used to simulate the effects of the velocity fluctuations and eddies on the flow, temperature, and composition fields (Ref 35). The basic approach is to time-average the NavierStokes equations, removing the details of the fluctuating velocities and replacing them with a time-averaged source term in the momentum equation. This term has the form of the shear stress term in a Newtonian fluid and so is referred to as the Reynolds stress tensor. A common way to represent this term is to use

Modeling of Transport Phenomena and Electromagnetics / 429 the eddy viscosity concept that models the Reynolds stress via an additional shear viscosity. Typically, the parameters to define this turbulent viscosity are obtained using variations of the k-e model (Ref 36, Ref 37, Ref 38), which includes two additional transport equations for turbulent kinetic energy (k) and dissipation (e). These two values are also used to modify the heat and species diffusivities (a and D) to account for the effect of turbulence on transport of these quantities. In contrast to the time-averaged approach, there is also an emerging interest in developing and using turbulence models that can resolve the fluctuating velocity components. Two examples are direct numerical simulation (DNS), which solves the Navier-Stokes equations directly, and large eddy simulation (LES), which uses a filtering equation to explicitly account for the momentum transport due to larger eddies, while the momentum in the smaller eddies is included implicitly via subgrid models. The performance of these more computationally intensive methods has been compared to the k-e approach in the modeling of the continuous casting of steel (Ref 39). (This process has relatively high momentum compared to other solidification processes and is more likely to undergo transition to turbulence.) In this work, Thomas et al. show that the time-averaged predictions of flow patterns from the k-e, DNS, and LES methods are very close to each other, although, by its nature, the k-e model cannot account for transient fluctuations. Proper use of the k-e model also relies on empirically derived constants that heavily depend on the specific flow conditions. In spite of the fact that use of the incorrect constants in the k-e model can lead to very wrong results, the DNS and LES represent a huge increase in computational effort (due largely to the necessary increase in grid resolution) and so are less likely to be used in the near future except in research codes.

Phenomena Restraining or Driving Flow The source term (SM) in Eq 20 can be made up of several phenomena that either restrain or drive the liquid flow in a solidifying alloy. A key restraining force, discussed previously, is the drag in the mushy zone. Other driving/ restraining forces include electromagnetic effects (discussed in the last section), buoyancy, and the shrinkage of the alloy during phase changes. Shrinkage. The shrinkage-induced flows arise in Eq 21 and 22 not from the source terms, SM, but from the temporal variation of mixture density in Eq 22. Most metal alloys shrink by 2 to 10% when they solidify (Ref 10). As the liquid turns to solid, a pressure gradient will develop in the liquid at whatever level is necessary to fill the volume deficit formed. While the liquid velocities generated can be quite small and have little effect on macrosegregation

(see Ref 40 for a scaling analysis of this issue and Ref 41 and 42 for some exceptions), the attendant volume change can cause the casting to pull away from the mold wall, and the pressures induced can cause defects such as hot tears and porosity. The modeling of the evolution of gas porosity requires an additional advection-diffusion equation to track the dissolved gas in the metal, as well as a pore nucleation model. The choice of whether or not to include this effect depends on the purpose of the model. If macrosegregation levels are low, and porosity and solid deformation are not an issue, then the advantage of including this effect must be weighed against the increase in computational complexity. Because the volume of the metal is not constant throughout the process, some sort of open boundary, where fluid is allowed to flow in and out based on the local pressure (which will be affected by the shrinkage-induced flow), will be necessary. This approach is the easiest, but it lacks information on the alloy beyond the boundary and so incorporates uncertainties in the behavior near that region (Ref 40). Another option is to develop a more realistic model that simulates the rise and fall of a free surface in a riser. As seen in the next section, this approach may give better predictions for the fluid flow, but it also makes the calculations more difficult. Buoyancy is a driving force for flow that occurs in most castings. Both nonuniform cooling and liquid composition differences due to elemental partitioning during solidification cause density gradients in the liquid. Unless these density variations are stable (cooling from below and interdendritic liquid heavier than the bulk), they will cause natural convection flows in the liquid. In the absence of electromagnetic forces, and once the effects of pouring die out, buoyancy will be the main driver of flow in most static casting processes. Equation 21 includes a buoyancy source term that requires a model for the variation of liquid density with temperature and composition. The most common approach is to use the Boussinesq approximations, which assume that the temperature and composition dependence of density appears only in this buoyancy source term and that those density changes are much less than the value of the nominal liquid density (Drl/rl 1) (Ref 43). Making these assumptions allows the buoyancy source term to be written as a linear function of temperature and composition changes in the liquid:  SB ¼ gl rl g ¼ gl rl;o T ðT  Tref Þ þ

N 1 X

   S;i Cl;i  Cl;i;ref g

(Eq 26)

i¼1

where rl,o is the liquid density at the reference temperature and composition, and N is the number of species in the alloy. The reference values are usually taken as the initial value of temperature and the nominal alloy composition. The thermal and solutal expansion coefficients are:

bT ¼ 

1 @rl rl @T

and bS;i ¼ 

1 @rl rl @Cl;i

(Eq 27)

In cases of large composition or temperature differences in the liquid metal or other cases in which the Boussinesq approximations break down, the liquid density can be better represented by (Ref 44): r1 l ¼

N X i¼1

Cl;i   rl;i þ i T  Tl;i

! (Eq 28)

where the Li is related to the thermal expansion coefficient and can be found in Ref 44. After the local liquid composition and temperature are found, Eq 28 can be used to obtain the local liquid density and the buoyancy source term: SB ¼ gl rl g

(Eq 29)

Dealing with Fronts and Free Surfaces Although basic computational tools for heat and mass transfer calculations based on the equations introduced previously are well established, their application in the modeling of casting processes is challenging. In particular, many casting models require the transient tracking of a front or free surface, for example, the solid-liquid front in microstructural evolution models or the metal-air free surface during simulation of mold filling (pouring). Because computational methods only provide discrete information/data at specified points (i.e., the node points on the computational mesh), special procedures must be developed to handle problems that require the tracking of a front or free surface.

Overview of Computational Front-Tracking Schemes Very broadly speaking, in the computational mechanics field there are two methods that can be used to track fronts. The first approach is to use a deforming mesh. In this approach, a line of nodes in the computational mesh is assigned to always lie on the front of interest. As the calculation continues in time, physical conditions on the front (e.g., a heat or mass balance) are used to update the position of the front nodes. In turn, this updating leads to a deformation of the computational mesh. Although high accuracy in front tracking can be achieved, deforming mesh techniques are not suitable for general applications. Excessive distortion of the mesh will require a computationally expensive remeshing to retain accuracy. Further, the method cannot be readily applied to problems with diffusive interfaces (e.g., a moving mushy region), where a specific front cannot be identified. The alternative to deforming grids is to use an index method. In these methods, the nodal

430 / Modeling and Analysis of Casting Processes mesh remains fixed in space during the calculation, and the progress of the front over this mesh is tracked via an index factor stored at the nodes, for example, 1 f 1. In this example, if the front separates a fully solid region from a fully liquid region, then a nodal value of f = 1 indicates a node in the solid, f = 1 represents a node in the liquid, and a node with a value strictly in the range 1 < f < 1 is in the vicinity of the interface. In this way, changes in the nodal values of f can be used to track the movement of the front. The calculated nodal values of f can also be used to modify the discretizations of the governing equations and estimate property changes across the front. Index methods are much more flexible than deforming grid methods; in particular, they do not suffer from grid distortion and can readily deal with diffusive interfaces. For these reasons, index methods have found a wide application in casting simulations. Again, very broadly speaking, there are two classes of index methods: order parameter methods and volume-of-fluid (VOF) methods. In order parameter methods, a separate equation for the evolution of the index factor is established in addition to the governing equations. In the case of modeling the evolution of microstructures, an appropriate evolution equation can be derived from thermodynamic considerations; this is the basis of phase-field models (Ref 45). Alternatively, where an appropriate flow velocity can be constructed, the order parameter can be a distance function that measures the normal distance from a given node point to the front of interest. This distance function is referred to as a level set function, and the zero level set f = 0 coincides with the interface. The values of the level set are evolved in time by a pure nonconservative advection equation. Level set methods (Ref 46, 47) have been successfully used to resolve many fronttracking problems. These methods can readily track the interface by identifying the f = 0 isoline. In addition, due to the smooth field of f(x, t) in the vicinity of the interface, they can, through a finite difference approximation, provide estimates of the interface normal direction: n^¼ rf=jrfj

(e.g., a liquid fraction, 0 F 1, for tracking solid-liquid interfaces), and its evolution equation is derived explicitly from the governing heat and mass transfer equations. The VOF methods are relatively easy to apply and can deal equally well with sharp and diffusive interfaces. In operation, however, the spatial region of nodes in which F is strictly in the range 0 < F < 1 is usually limited to one-grid spacing. The essential difference between a level set and VOF method is illustrated in Fig. 2. The panel in Fig. 2(a) shows the tracking of the front in terms of a VOF variable of liquid fraction associated with the midpoint of each cell. The panel in Fig. 2(b) shows the level set distance function values at the cell centers. A more dispersed and smooth field is provided by the level set distance function. The limitation of the VOF parameter variation to one-grid spacing can lead to less local accuracy in tracking fronts (e.g., jumps and plateaus in tracking the melt front in a pure material) (Ref 48, 49) and in estimating front direction and curvature. Despite these drawbacks, however, the VOF method is widely used and is incorporated into several commercial general computational flow dynamics (CFD) and casting-specific codes. In closing this general discussion of modeling techniques for casting simulations, it is important to note that, in addition to the deforming grid (Lagrangian) and fixed grid index methods (Eulerian), there are also a number of hybrid schemes that combine elements of both Lagrangian and Eulerian concepts. A wellused idea is that of the immersed boundary (Ref 50, 51). These methods employ a fixed-background Eulerian mesh of nodes superimposed with a line of moving nodes attached to the interface of interest. Field calculations are carried out on the background mesh, employing a modified discretization in the vicinity of the interface to account for the moving line of nodes. These field calculations are then used in appropriate interface conditions to advance the interface nodes.

Application of VOF Models in Casting Simulations Because of their wide use in modeling of casting processes, a more detailed account of how VOF methods are used is provided here. As an example, modeling a basic mold-filling operation involves the tracking of the liquid metal-gas interface. In the original VOF method presented by Hirt and Nichols (Ref 52), this problem is dealt with by only solving for the momentum and mass continuity in the liquid domain. Following this solution for the velocity field, the VOF variable (the metal fraction, F) is calculated from an advection equation:



@F þ r ðV F Þ ¼ 0 @t

(Eq 32)

Assuming a constant liquid density, this equation is essentially a statement of metal conservation at the liquid-gas interface. Solution of Eq 32 involves locating the interface in any partially filled cells (i.e., nodal values in the range 0 < F < 1) with a simple line interface calculation that aligns the interface with the most appropriate coordinate direction. This stair-step reconstruction of the liquid-gas interface is then used in a donor-acceptor solution of Eq 32, an approach that ensures that no cell can donate more liquid than it has or can accept more liquid than it can accommodate. In the subsequent solution of the momentum and mass continuity for the flow field, either at the next time step (in explicit calculations) or iteration (in implicit calculations), the pressure can be forced in partially filled cells 0 < F < 1 to the value P = 0 (assuming adequate venting of the trapped gas). This approach works well for mold-filling processes that are free of highly convoluted free surfaces and involve liquids with small surface tensions. If the flow is turbulent, that is, entraps gas bubbles, or involves a liquid with large surface tension, alternative approaches must be found.

(Eq 30)

and curvature:



 ¼ r n^

(Eq 31)

In using this approach, one should keep in mind that (a) the accuracy depends on a sound construction of the field velocity based on the physical interface velocity; (b) as the solution progresses, the ability of the nodal field, f(x,t), to measure distance is degraded and requires reinitialization each time the front moves to reset the distance function, a step somewhat akin to remeshing in a deforming grid; and (c) basic methods may not conserve mass. In contrast to order parameter methods, in VOF methods, the index is a physical quantity

Fig. 2

Comparison of order parameter fields. (a) Liquid fraction field in a volume-of-fluid calculation. (b) Level set distance function field

Modeling of Transport Phenomena and Electromagnetics / 431 An alternative VOF that can address the surface tension and trapped gas uses a two-fluid model (liquid metal and gas). Assuming no slip between the fluids and incompressibility (r  V = 0), this model solves the mass and momentum equations throughout the whole domain for the single flow field, V. In this model, properties are written as a linear combination of liquid (l) and gas (g) properties; for example, a mixture density is: r ¼ gl rl þ gg rg

(Eq 33)

In this way, using the incompressibility condition in the continuity Eq 19, the following advection equation for the update of the liquid fraction can be obtained (Ref 53):



@F þ V rF ¼ 0 @t

(Eq 34)

This equation is identical in form to the nonconservative advection equation used to update a level set. Note, however, the metal fraction F in Eq 34 has the narrow band property illustrated in the panel in Fig. 2(a). Values with information, that is, strictly in the range 0 < F < 1, are restricted to a region approximately one computational cell wide. Another, more recent, two-fluid model can be found in Ref 54, and an example of work comparing VOF methods to real-time x-ray images during mold filling is found in Ref 55. From a given calculated metal fraction field, the liquid-gas interface in the two-fluid model can be reconstructed from a piecewise linear interface calculation (Ref 53). This calculation constructs a piecewise linear fit along the interface that splits the computational cell, associated with a nodal liquid fraction, into volume fractions of F and (1  F), respectively. This reconstructed surface is used in two ways. First, the local orientation of the surface is used in the solution of Eq 34 to partition the advection of F and preserve a sharp interface (Ref 54). Secondly, it can be used to estimate interface curvatures and calculate interface tractions due to surface tension; these tractions can subsequently be incorporated as source terms into the momentum balance equation (Ref 22). In addition to tracking the location of free surfaces of liquid metal during pouring or the interface between two fluids (e.g., a slag-metal surface), it is important to model surface energy forces on these interfaces. The imposition of the VOF field over a fixed computational grid, with the interface volumes containing different amounts of fluid, can lead to large differences in surface tension from one cell to the next. This numerical artifact gives rise to parasitic velocities destabilizing a surface, which is the opposite effect surface tension should have. Brackbill et al. (Ref 56) proposed a continuum surface force model to smooth out these local variations. This force is integrated over a small region of the free surface near each control volume on that surface: ð FST ðxÞ ¼ s

ðrÞnðrÞdðx  rÞdr Surface

(Eq 35)

where r is the vector between the point of interest (x) and the surrounding volumes, and d is a Dirac delta function. Discussion of methods of discretization of this function are in Ref 57. The method of Brackbill et al. (Ref 56) provides a smooth surface tension distribution over an interface on a fixed grid. To use their model (or any other surface energy model), the local normal and curvature must be found. The normals can be found from the surface reconstruction algorithms mentioned previously. There are many techniques of calculating interface curvature from the VOF variable, several of which are evaluated in Cummins et al. (Ref 58). These methods of calculating local surface normals and curvature are useful for crystal morphology modeling (Ref 59, 60), which also requires care to be taken to recover accurate estimates of surface energy.

Simulation of Electromagnetic Effects in Castings Many metal-casting processes employ electromagnetic fields to enhance the performance of casting operations and to improve the quality of casting products (Ref 61–66). The use of an electromagnetic field can either increase the melt flow motion or melt homogenization by providing a strong stirring force or suppress the undesired turbulence at the entrance of the melt into a caster to reduce impurity entrainments. To generate a stirring force, an alternating current (ac) magnetic field is often applied (Ref 62), while the static magnetic field is often used to produce the damping effects to reduce turbulence or flow disturbances at the entrance (Ref 63). An ac field may also be used for electromagnetic casting, in which the field takes the place of a mechanical mold to support the molten metal during casting (Ref 65). Modeling strategies of electromagnetic fields in casting applications thus depend on which effects (stirring or damping) are to be achieved. The electromagnetic field distributions are governed by Maxwell’s equations, which are four vector equations summarizing the basic laws governing the electromagnetic field behavior in a medium (Ref 67, 68): @B @t

(Eq 36)

@D þJ @t

(Eq 37)

rE¼

rH ¼

(Eq 38)



(Eq 39)

r D ¼ re

D ¼ eE

where E is the electric field, D is the electric displacement field, B is the magnetic induction (magnetic flux density) field, and H is the magnetic intensity. Also, re is “free” electric

(Eq 40)

and B ¼ mH

(Eq 41)

where e is the permittivity, and m is the permeability of the medium. These properties can be strong functions of the phase, composition, and temperature of the materials in the casting, and so those dependencies must be evaluated to determine if they should be included in the model. For stirring applications, the electric and magnetic fields induced in the melt by an applied ac or time-varying direct current magnetic field are calculated by solving the Maxwell equations or their time-harmonic forms with displacement currents neglected (Ref 67– 69). The finite element method is widely used in general-purpose commercial software to obtain the needed field distributions in the metal melt from the solution of the Maxwell equations. For special arrangements such as two-dimensional and axisymmetric configurations, simplified algorithms that use a hybrid boundary element and finite element method are computationally more efficient (Ref 70). Volume integral or inductance formulations may also be used (Ref 66, 71, 72). Software based on the hybrid methods, although efficient, is often unavailable in general-purpose form but can be obtained from active researchers in this area (Ref 70). Electromagnetic stirring may also occur in vacuum arc or plasma arc remelting processes as a result of nonuniform electric field distributions in the casting. For these operations, the electric field and the current density distribution can be derived from an electric potential, and the Maxwell equations reduce to a Laplace equation for which solutions can be obtained using any commercially available CFD software. The magnetic field distribution can then be obtained using the Biot-Savart law (Ref 73). With the electric and magnetic fields known, as described, the Lorentz force, F, responsible for stirring action can be obtained via the expression: F¼JB



r B¼0

charges, including both induced and impressed charges, but not bounded charges such as induced dipoles in dielectrics. The current density consists of the sum of the impressed Ji and induced Je contributions: J = Ji + Je. The relations between E and D and between B and H are specified by the constitutive equations:

(Eq 42)

where J is the current density, and B is the magnetic field in the solidifying melt. The stirring force, which is the time-averaged value of F in Eq 42, is then substituted into the NavierStokes equations as a momentum source (or force) term to obtain the velocity field distribution in the melt. For these applications, the Joule heating, Q = E
J, may also be added as a

432 / Modeling and Analysis of Casting Processes source term to the energy (or heat transfer) equation. For example, if the Joule heating is the only heat source, then ST = Q in Eq 6. For these systems, the electromagnetics and thermal/fluid flow fields may be modeled sequentially. First, the electric and magnetic fields can be calculated independent of the fluid flow and thermal fields, assuming the electrical properties are not temperature-dependent. The Lorentz force and the Joule heating source subsequently are obtained from the calculated electric and magnetic field variables. These source terms are then included in the Navier-Stokes equations and thermal energy balance equation to obtain the electromagnetically induced flow and thermal fields (Ref 66, 71, 72). For electromagnetic casting applications, the melt shape is unknown a priori and is determined as a balance between surface tension, gravity, and the electromagnetic pressure (or the potential component of the Lorentz force) along the surface (Ref 65, 70). Also, the induced current density and the magnetic field are dependent on the shape of the melt. Thus, an iterative procedure is required that involves the electromagnetic calculations and the melt shape calculation, using techniques such as VOF described in the previous section. With the shape and the electromagnetic field so determined, the flow and heat transfer calculations can proceed in the same fashion as described (Ref 71, 72). For magnetic damping or magnetic braking applications, a static magnetic field is applied to produce a Lorentz force always opposing the melt flow (Ref 63, 74–77). This opposing Lorentz force distribution is from Eq 42, where: J ¼ ðrf þ V  BÞ

(Eq 43)

and s is the electrical conductivity of the melt, f is the electric potential, V is the velocity field of the melt, and B is the applied static magnetic field. The electrical potential f may be obtained by solving: r2 f ¼ r
ðV  BÞ

(Eq 44)

For some special cases, rf is equal to zero, and the force expression reduces to: F ¼ J  B ¼ ðV  B  BÞ

(Eq 45)

This force can be directly implemented in the Navier-Stokes equations as a momentum sink to predict a reduced flow field (Ref 74). For a more general case, the solution of Eq 44 is required along with appropriate boundary conditions (Ref 75, 76). For these applications, there is no need to have specially designed computational electromagnetics software, and most computational packages for fluid flow and heat transfer can be used to solve the scalar equation for f. It is noted that the solution for f needs to be coupled with the fluid flow and heat transfer associated with the problem, and hence, an iterative procedure is needed. The final result will be the solution of f, velocity,

and temperature field, along with other field variables of interest to a casting system.

Summary The conservation equations that represent these important physical phenomena during casting processes are presented and an outline discussion of how they can be solved is provided. There is a well-established array of general and specific computational tools that can be readily applied to modeling casting processes. This article summarizes the key features of the conservation equations in these tools. The intention is to provide the relevant information on the operation, advantages, and drawbacks of the various modeling features to allow a user to make the wisest use of the available tools. REFERENCES 1. W.D. Bennon and F.P. Incropera, The Evolution of Macrosegregation in Statically Cast Binary Ingots, Metall. Trans. B, Vol 18, 1987, p 611–616 2. C. Beckermann and R. Viskanta, DoubleDiffusive Convection during Dendritic Solidification of a Binary Mixture, Physico. Chem. Hydrodyn., Vol 10, 1988, p 195–213 3. V.R. Voller, A.D. Brent, and C. Prakash, The Modelling of Heat, Mass and Solute Transport in Solidification Systems, Int. J. Heat Mass Transf., Vol 32, 1989, p 1719– 1731 4. C. Beckermann and C.Y. Wang, MultiPhase/-Scale Modeling of Transport Phenomena in Alloy Solidification, Annu. Rev. Heat Transf., Vol 6, 1995, p 115–198 5. C.R. Swaminathan and V.R. Voller, Towards a General Numerical Scheme for Solidification Systems, Int. J. Heat Mass Transf., Vol 40, 1997, p 2859–2868 6. M.J.M. Krane, F.P. Incropera, and D.R. Gaskell, Solution of Ternary Metal Alloys, Part I: Model Development, Int. J. Heat Mass Transf., Vol 40, 1997, p 3827–3835 7. V.R. Voller, Numerical Methods for PhaseChange Problems, Chap. 19, Handbook of Numerical Heat Transfer, 2nd ed., W.J. Minkowycz, E.M. Sparrow, and J.Y. Murthy, Ed., John Wiley, 2006 8. N.R. Eyres, D.R. Hartree, J. Ingham, R. Jackson, R.J. Sarjant, and S.M. Wagstaff, The Calculation of Variable Heat Flow in Solids, Philos. Trans. R. Soc. (London) A, Vol 240, 1946, p 1–57 9. M.J.M. Krane and F.P. Incropera, A Scaling Analysis of the Unidirectional Solidification of a Binary Alloy, Int. J. Heat Mass Transf., Vol 39, 1996, p 3567–3579 10. E.A. Brandes and G.B. Brook, Ed., Smithell’s Metals Reference Handbook, 7th ed., Butterworth-Heinemann, Ltd., 1992

11. W.D. Bennon and F.P. Incropera, A Continuum Model for Momentum, Heat and Species Transport in Binary Solid LiquidPhase Change Systems, Part 1: Model Formulation, Int J. Heat Mass Transf., Vol 30, 1987, p 2161–2170 12. M.C. Schneider and C. Beckermann, Numerical Study of the Combined Effects of Microsegregation, Mushy Zone Permeability and Flow, Caused by Volume Contraction and Thermosolutal Convection, on Macrosegregation and Eutectic Formation in Binary Alloy Solidification, Int. J. Heat Mass Transf., Vol 38, 1995, p 3455–3473 13. C.R. Voller and C. Beckermann, Approximate Models of Microsegregation with Coarsening, Metall. Mater. Trans. A, Vol 30, 1999, p 3016–3019 14. V.R. Voller, Semi-Analytical Model of Microsegregation in a Binary Alloy, J. Cryst. Growth, Vol 197, 1999, p 325–332 15. V.R. Voller, Semi-Analytical Model of Microsegregation and Coarsening in a Binary Alloy, J. Cryst. Growth, Vol 197, 1999, p 333–340 16. V.R. Voller, Model of Microsegregation during Binary Alloy Solidification, Int. J. Heat Mass Transf., Vol 43, 2000, p 2047– 2052 17. S.A. Cefalu and M.J.M. Krane, Microsegregation in Ternary Alloys Solidifying in an Open System, Mater. Sci. Eng. A, Vol 359, 2003, p 91–99 18. Z. Guo, N. Saunders, A.P. Miodownik, and J.-Ph. Schille´, Modelling of Materials Properties and Behaviour Critical to Casting Simulation, Mater. Sci. Eng. A, Vol 413– 414, 2005, p 465–469 19. S. Patankar, Numerical Heat Transfer and Fluid Flow, Hemisphere, 1980 20. C. Hirsch, Numerical Computation of Internal and External Flows, Vol I, Fundamentals of Numerical Discretization, Wiley, 1988 21. H.K. Versteeg and W. Malalasekera, An Introduction to Computational Fluid Dynamics: The Finite Volume Method, Longman, 1995 22. J.H. Ferziger and M. Peric, Computational Methods for Fluid Dynamics, 3rd ed., Springer, 2002 23. J. Ni and C. Beckermann, A Volume-Averaged Two-Phase Model for Transport Phenomena during Solidification, Metall. Trans. B, Vol 22, 1991, p 349–360 24. H. Schlichting, Boundary-Layer Theory, 7th ed., McGraw-Hill, 1979 25. R.L. Panton, Incompressible Flow, 3rd ed., Wiley, 2005 26. D.A. Nield and A. Bejan, Convection in Porous Media, 2nd ed., Springer, 1999 27. D. Nagelhout, M.S. Bhat, J.C. Heinrich, and D.R. Poirier, Permeability for Flow Normal to a Sparse Array of Fibres, Mater. Sci. Eng. A, Vol l91, 1995, p 203–208 28. M.S. Bhat, D.R. Poirier, and J.C. Heinrich, A Permeability Length Scale for Cross

Modeling of Transport Phenomena and Electromagnetics / 433

29.

30.

31.

32.

33.

34.

35. 36.

37.

38.

39.

40.

41.

42.

Flow Through Model Structures, Metall. Mater. Trans. B, Vol 26, 1995, p 1091– 1092 S.D. Felicelli, J.C. Heinrich, and D.R. Poirier, Simulation of Freckles during Vertical Solidification of Binary Alloys, Metall. Trans. B, Vol 22, 1991, p 847–859 C.Y. Wang and C. Beckermann, Equiaxed Dendritic Solidification, Part I: Multiscale Modeling, Metall. Mater. Trans. A, Vol 27, 1996, p 2754–2764 J. Ni and F.P. Incropera, Extension of the Continuum Model for Transport Phenomena Occurring During Metal Alloy Solidification, Part I: The Conservation Equations, Int. J. Heat Mass Transf., Vol 38, 1995, p 1271–1284 J. Ni and F.P. Incropera, Extension of the Continuum Model for Transport Phenomena Occurring During Metal Alloy Solidification, Part II: Microscopic Considerations, Int. J. Heat Mass Transf., Vol 38, 1995, p 1285–1296 C.Y. Wang and C. Beckermann, Equiaxed Dendritic Solidification, Part II: Numerical Simulations for an Al-4 wt pct Cu Alloy, Metall. Mater. Trans. A, Vol 27, 1996, p 2765–2783 C.J. Vreeman, J.D. Schloz, and M.J.M. Krane, Direct Chill Casting of Aluminum Alloys: Modeling and Experiments on Industrial Scale Ingots, ASME J. Heat Transf., Vol 124, 2002, p 947–953 D.C. Wilcox, Turbulence Modeling for CFD, DCW Industries, 2006 W.P. Jones and B.E. Launder, The Prediction of Laminarization with a Two-Equation Model of Turbulence, Int. J. Heat Mass Transf., Vol 15, 1972, p 301–314 W.P. Jones and B.E. Launder, The Calculation of Low-Reynolds-Number Phenomena with a Two-Equation Model of Turbulence, Int. J. Heat Mass Transf., Vol 16, 1973, p 1119–1130 B.E. Launder and B.I. Sharma, Application of Energy-Dissipation Model of Turbulence to the Calculation of Flow Near a Spinning Disk, Lett. Heat Mass Transf., Vol 1, 1974, p 131–138 B.G. Thomas, Q. Yuan, S. Sivaramakrishnan, T. Shi, S.P. Vanka, and M.B. Assar, Comparison of Four Methods to Evaluate Fluid Velocities in a Continuous Slab Casting Mold, ISIJ Int., Vol 41, 2001, p 1262– 1271 M.J.M. Krane and F.P. Incropera, Analysis of the Effect of Shrinkage on Macrosegregation in Alloy Solidification, Metall. Mater. Trans. A, Vol 26, 1995, p 2329–2339 A.V. Reddy and C. Beckermann, Modeling of Macrosegregation due to Thermosolutal Convection and Contraction-Driven Flow in Direct Chill Continuous Casting of an Al-Cu Round Ingot, Metall. Mater. Trans. B, Vol 28, 1997, p 479–489 D.G. Eskin, Q. Du, and L. Katgerman, Relationship Between Shrinkage-Induced

43.

44. 45.

46. 47.

48.

49.

50.

51.

52.

53. 54.

55.

56.

57.

58.

59.

Macrosegregation and the Sump Profile upon Direct-Chill Casting, Scr. Mater., Vol 55, 2006, p 715–718 B. Gebhart, Y. Jaluria, R.L. Mahajan, and B. Sammakia, Buoyancy-Induced Flows and Transport, Hemisphere Publishing Corporation, 1988 T. Iida and R. Guthrie, The Physical Properties of Liquid Metals, Clarendon Press, Oxford, 1988 W.J. Boettinger, J.A. Warren, C. Beckermann, and A. Karma, Phase-Field Simulations of Solidification, Annu. Rev. Mater. Res., Vol 32, 2002, p 163–194 J.A. Sethian, Level Set Methods and Fast Marching Methods, Cambridge University Press, Cambridge, 1999 S. Osher and R.P. Fedkiw, Level Set Methods: An Overview and Some Recent Results, J. Comput. Phys., Vol 169, 2001, p 463–502 P.H. Price and M.R. Slack, The Effect of Latent Heat on Numerical Solutions of the Heat Flow Equation, Br. J. Appl. Phys., Vol 5, 1954, p 285–287 V. Voller and M. Cross, Accurate Solutions of Moving Boundary Problems Using the Enthalpy Method, Int. J. Heat Mass Transf., Vol 24, 1981, p 545–556 H.S. Udaykumar, R. Mittal, and W. Shyy, Computation of Solid-Liquid Phase Fronts in the Sharp Interface Limit on Fixed Grids, J. Comput. Phys., Vol 153, 1999, p 535–574 D. Juric and G. Tryggvason, A Front Tracking Method for Dendritic Solidification, J. Comput. Phys., Vol 123, 1996, p 127–148 C.W. Hirt and B.D. Nichols, Volume of Fluid (VOF) Method for the Dynamics of Free Boundaries, J. Comput. Phys., Vol 39, 1981, p 201–225 W.J. Rider and D.B. Kothe, Reconstructing Volume Tracking, J. Comput. Phys., Vol 141, 1988, p 112–152 S.H. Garioch and B.R. Baliga, A PLIC Volume Tracking Method for the Simulation of Two-Fluid Flows, Int. J. Num. Meth. Fluids, Vol 52, 2006, p 1093–1134 F.-Y. Hsu, M.R. Jolly, and J. Campbell, Vortex Gate Design for Gravity Castings, Int. J. Cast Met. Res., Vol 19, 2006, p 38–44 J.U. Brackbill, D.B. Kothe, and C. Zemach, A Continuum Method for Modeling Surface Tension, J. Comput. Phys., Vol 100, 1992, p 335–354 M. Bussman, J. Mostaghimi, and S. Chandra, On a Three-Dimensional Volume Tracking Model of Droplet Impact, Phys. Fluids, Vol 11, 1999, p 1406–1417 S.J. Cummins, M.M. Francois, and D.B. Kothe, Estimating Curvature from Volume Fractions, Comput. Struct., Vol 83, 2005, p 425–434 R. Shao, “Numerical and Experimental Study of Constrained Solidification,” Ph.

60.

61.

62.

63.

64.

65. 66.

67. 68. 69. 70.

71.

72.

73.

74.

75.

D. dissertation, School of Materials Engineering, Purdue University, 2007 V.R. Voller, An Enthalpy Method for Modeling Dendritic Growth in a Binary Alloy, Int. J. Heat Mass Transf., Vol. 51, 2008 p 823–834 A.A. Tzavaras and H.D. Brody, Electromagnetic Stirring and Continuous Casting — Achievements, Problems and Goals, J. Met., Vol 36 (No. 3), Feb 1984, p 31–37 E. Takeuchi, M. Zeze, T. Toh, and S. Mizoguchi, Applied MHD in the Process of Continuous Casting, Magnetohydrodynamics in Process Metallurgy, J. Szekely, J.W. Evans, K. Blazek, and N. El-Kaddah, Ed., TMS, 1992, p 189–202 C. Vives, Effects of Magnetically Damped Convection during the Controlled Solidification of Metals and Alloys. Int. J. Heat Mass Transf., Vol 33, 1987, p 479 S. Asai, Metallurgical Aspects of Electromagnetic Processing of Materials, Liquid Metal Magnetohydrodynamics, J. Lielpeteris and R. Moreau, Ed., Kluwer Academic Publisher, 1989 Z.N. Getslev, Casting in an Electromagnetic Field, J. Met., Vol 23 (No. 10), Oct 1971, p 38–39 B.Q. Li, J.W. Evans, and D.P. Cook, An Improved Mathematical Model for Electromagnetic Casters and Testing by a Physical Model, Metall. Trans. B., Vol 22, 1991, p 121–134 J.D. Jackson, Classical Electrodynamics, Wiley, 1975 W.F. Hughes and F.J. Young, The Electromagnetohydrodynamics of Fluids, Wiley, 1966 B.Q. Li, The Fluid Flow Aspects of Electromagnetic Levitation Processes, Int. J. Eng. Sci., Vol 32, 1994, p 45–67 S.P. Song and B.Q. Li, Coupled Boundary/ Finite Element Analysis of Magnetic Levitation Processes: Free Surface Deformation and Thermal Phenomena, J. Heat Transf., Vol 120, 1998, p 492–504 J. Sakane, B.Q. Li, and J.W. Evans, Mathematical Modeling of Meniscus Profile and Melt Flow in Electromagnetic Casters, Metall. Trans. B, Vol 19, 1988, p 397–408 B.Q. Li, A Finite Element Analysis of Magnetically-Driven Recirculating Flow in Electromagnetic Near Net Shape Casting, J. Mater. Process. Technol., Vol 3, 1995, p 351–359 B.Q. Li, Simulation of Magnetically-Driven Flow in Solidification Processes Using FIDAP, Proceedings of the Fourth FIDAP Users Conference (Evanston, IL), 1991 G.M. Opeper and J. Szekely, The Effect of an Externally Imposed Magnetic Field on Buoyancy Driven Flow in a Rectangular Cavity, J. Cryst. Growth, Vol 64, 1983, p 505–515 H. Tozawa and S. Takeuchi, Flow Control of Molten Steel in Continuous Casting

434 / Modeling and Analysis of Casting Processes Mold by Static Magnetic Field, Magnetohydrodynamics in Process Metallurgy, J. Szekely, J.W. Evans, K. Blazek, and N. El-Kaddah, Ed., TMS, 1992, p 215–222

76. K. Li and B.Q. Li, Numerical Analyses of g-Jitter Induced Convection and Solidification in Magnetic Fields, J. Thermophys. Heat Transf., Vol 17, 2003, p 199–209

77. K. Li and B.Q. Li, A 3-D Model for g-Jitter Driven Flows and Solidification in Magnetic Fields, J. Thermophys. Heat Transf., Vol 17, 2003, p 498–508

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 435-444 DOI: 10.1361/asmhba0005236

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Direct Modeling of Structure Formation Ch.-A. Gandin, Ecole Nationale Superieure des Mines de Paris I. Steinbach, Ruhr-University Bochum

Introduction MODELING OF STRUCTURE FORMATION in casting of alloys involves several length scales, ranging from the atomic level (10
10– 10
9 m) to the casting dimensions, i.e. the macroscopic scale, (10
2–1 m). Intermediate length scales are used to define the microstructure of the growing phases (i.e., 10
7–10
4 m, e.g., dendritic or eutectic patterns) and the grain structure (i.e., 10
4–10
2 m, e.g., equiaxed and columnar grains). This article concentrates on these intermediate length scales, where transport phenomena govern the spatial and temporal evolution of the structure. To calculate this evolution, conservation equations are written for each individual phase (liquid or solid), with the most challenging aspect being the treatment of the interfaces between phases and the nonequilibrium condition of the liquid metal during the solidification state. The microstructure of a casting can consist of different grains of the same thermodynamic phase, distinct by orientation, or different phases of one alloy, distinct by composition and atomic structure. Within one grain metallic alloys also show a microstructure caused by segregation patterns due to the dendritic growth instability of the growth front. Figure 1(a) presents a longitudinal cross section of a 0.17 m (7 in.) Al-7wt%Si cylindrical ingot. The grain structure is revealed by the various gray levels. Elongated columnar grains are present at the bottom of the casting, while a more isotropic grain size is observed at the top. Only magnification of the grains toward the top of the casting reveals the dendritic nature of the equiaxed microstructure made of the a-aluminum phase (Fig. 1b). Further magnification would reveal the lamellas of the eutectic microstructure, where silicon is solidified as a second thermodynamic phase together with the a-aluminum phase. It is located between dendrite arms and formed from the interdendritic liquid at the end of solidification. As details of the microstructure cannot be resolved at the macroscopic scale of the casting, models have been developed to determine an average grain size of the equiaxed grains and the extent of a columnar zone (Ref 1, 2).

These models are based on averaging procedures to write conservation equations for a two-phase mixture made of solid and liquid. These equations are solved at the scale of the casting using a control volume (CV) or finite element (FE) method. The results of such models are average grain sizes, and possibly average dendrite arm spacings, within the casting. It consequently does not directly provide a map of the structure. The typical representative elementary volume (REV) of such macroscopic scale method is shown in Fig. 1. It contains several grains or phases that are not resolved on the macroscopic scale. The cellular automaton (CA) method was first developed to model the macrostructure in casting by resolving each individual grain. The grains are defined by envelopes, drawn around a dendritic structure, and the entire grain structure of a casting can be computed. This is possible by considering a smaller REV, whose typical size is shown by the large highlighted square in Fig. 2(a) and its magnified view in Fig. 2(b). As a consequence, the dendritic structure cannot be resolved and is approximated by microscopic models. Special algorithms have to be designed to integrate the evolution of the volume fraction of the phases present in the REV over time, as well as to integrate the kinetics of the development of envelopes defining the grains. An even smaller REV must be used to compute the detailed dendritic or eutectic microstructure within the grains. The typical REV of such a microscopic scale method is shown by small squares in Fig. 2(b). The phase field (PF) method is now very popular (Ref 9–15) because of its profound thermodynamic basis. Although only small portions of the grain structure can be computed, the PF method offers the possibility of simulating directly the development of the solid/liquid interface, including its kinetics, pattern formation, and microsegregation with a limited number of approximations. While this article focuses on a presentation of the CA and PF methods that represent the state of the art for modeling macrostructure and microstructure, several other methods are under development. One could cite the levelset method (Ref 16) and the pseudo front

tracking method (Ref 17) for modeling interface development, as well as other forms of the front tracking method to simulate the growth of the boundary defined between a mushy zone and the melt (Ref 18). Before examining the technical details, it is worth noting that all models for microstructure formation on different scales rely on an indicator function that gives the phase state at a specific point in space and time. On the macroscopic scale it is usually the fraction of solid (fs) that has the value 1 in solid, 0 in liquid and values 0 < fs < 1 in the two-phase region, the so-called mushy region. This indicator function can be mapped to the local temperature only under conditions close to equilibrium solidification. In most cases, however, it will be an independent variable, the time evolution of which depends on the nucleation and microscopic growth conditions (see, e.g., Ref 19). In the CA model, a discrete indicator function InS is used to distinguish between the mushy zone and the liquid (for details see below) that characterize the state of one REV, that is, one cell. A changing value of the indicator function between two adjacent REVs defines the boundary between the solidifying mush and liquid. Applying a kinetic model for the motion of this boundary then determines the growth of the grain structure. The liquid outside a grain envelope has to stay in an undercooled state until the local REV is captured by a growing envelope, or an appropriate nucleation condition is fulfilled. The PF method, finally, uses the PF variable F(x,t) as an analogue to fs, where the intermediate values 0 < F < 1 characterize the interface between solid and liquid. Again the phase state changes only at the position of the interface if no nucleation condition is defined in the bulk liquid or solid. Variation of the PF variable through the interface reflects the nature of a diffuse interface at the atomistic scale. Using these indicator functions, all models are able, with a different level of sophistication, to account for nonequilibrium phenomena that characterize a typical solidification process and are reflected in the complex rules of microstructure formation. This article starts from the microscopic scale with principles and applications of the PF

436 / Modeling and Analysis of Casting Processes

(b)

system can be described as a diffuse transition on a width of several atomic distances. In numerical calculations using the PF method, however, the width of the interface Z is mostly taken to be large compared to its true physical dimension, but small compared to the scale of the microstructure r that has to be resolved. The justification to do so lies in the invariance or the PF equation (for constant outer fields and to the first order in Z/r) against rescaling of the interface (Ref 11). In the case of coupling the PF to diffusion fields, the effect of nonconstant diffusion fields inside the interface on the kinetics of the transformation and the redistribution of solute has to be considered. A rigorous way to treat this problem was developed by Karma and coworkers (Ref 9, 10) for binary alloys in the dilute solution limit. For the general case of multicomponent multiphase alloys this correction scheme still deserves to be adapted. The general aspects of the PF method are only briefly outlined. For a more complete review see (Ref 21). In general, a PF model starts from an appropriate thermodynamic function. For solidification under constant pressure, the Gibbs free energy F is used. It is the integral over the free-energy density contributed by the interface f IF and by the bulk thermodynamics, f TD: ð ð   F ¼ fdx ¼ dx f IF þ f TD

(Eq 1)

In the solid state an elastic contribution will also be important (Ref 12) as well as magnetic or electric contributions for certain systems. From the principle of energy minimization, the evolution equation of the PF and the concentration are derived: _ a ¼


X b

c_i ¼

X

rM ij r

j

(a)

Fig. 1

Ingot structure (a) as observed in a longitudinal cross section of a 0.17 m (7 in.) height Al-7 wt%Si cylindrical casting. Distinction is possible between elongated columnar grains located in the bottom two-thirds of the casting and the isotropic equiaxed grains in the top region. Magnification in (b) reveals the dendritic microstructure of the alloy, with an interdendritic black region corresponding to a eutectic microstructure that is not resolved at this scale. The size of the highlighted square would be typical of the representative elementary volume used to solve conservation equations at the scale of the casting using a finite element method. Only one REV is represented while the number of REV is defined to fully cover the domain of the casting to be modeled.

Direct Microstructure Simulation Using the Phase Field Method A phase field Fa(x) is an indicator function of the real space coordinate x, indicating that the material at this coordinate exists in the state

of the phase a if Fa(x) = 1, or in some other (not a) phase if Fa(x) = 0.* The transition between one phase and another is assumed steep, but continuous. The transition region thus characterizes the interface between the phases that is “diffuse” instead of a sharp jump between the phases. During a phase transformation this interface moves by converting the thermodynamically unstable phase into a more stable phase; the description of this interface motion is the main goal of the method. The idea of a diffuse interface dates back to van der Waals (Ref 20), who intensively thought about the forces between atoms and molecules. It is well accepted today that the solid/liquid interface in a metallic

dF dF
da db



dF dcj

(Eq 2a)

(Eq 2b)

where mab is the mobility of the interface between phase a and phase b, ci is the concentration of component i and Mij the chemical mobility matrix. The different PF models differ in the particular formulation, but agree in general principle. The following formulation given in (Ref 13) is used to find the particular form: _ a ¼

method. Then, using the mesoscopic scale of individual grains, the CA model is introduced as a computationally efficient method to predict grain structures in castings. Finally, the coupling of the CA to macroscopic calculation of heat, flow, and mass transfers in castings and applications to realistic casting conditions is discussed.



mab

X b

  mab sab b r2 a
a r2 b

#  2   pffiffiffiffiffiffiffiffiffiffiffiffi a b Gab (Eq 3a)
2 a
b þ ab 2ab

 (Eq 3b) Gab ¼ fa
fb
mi cia
cib " !# X X c_i ¼ (Eq 3c) rDij rcj
cja ra j

a

*In the context of thermodynamics the expression “phase field” is also used to characterize a region of the composition space of an alloy, where one specific phase is stable. The concept therein is complementary; however, the approach to determine a “phase field” is different.

Direct Modeling of Structure Formation / 437 the seed particle (Ref 28). The growth of the nucleus is handled analytically, and the solid fraction is accumulated locally until its size can be treated by the PF (Ref 24).

Equiaxed Solidification of a Hypereutectic AlCuSiMg Alloy

(a)

(b)

Fig. 2

Schematic of a single grain (Ref 8) growing in a uniform temperature field is shown in (a). The square highlighted in (a) shows a typical length scale for representative elementary volumes (REV) used by the cellular automaton (CA) method. The two small squares in (b) show the typical scale for REV used by the phase field (PF) method. Only one CA REV is shown although the number of CA REV is defined to fully cover the domain of the casting to be modeled. Similarly, only two PF REV are shown while the number of PF REV is defined to cover a domain of the casting representative of the microstructure to be modeled.

where sab is the interfacial energy, Zab the interface width, DGab the free-energy difference, and mi the generalized chemical potential between phase a and b. The specific freeenergy density, fa, of phase a related to the concentration cia in phase a. The diffusivity matrix, Dij, can be calculated from the chemical mobilities of the individual components and the thermodynamic factor @ 2 f=@ cia @ cia . In the interface it is dependent on the properties of both phases that build the interface. For more details and a sound derivation see Ref 13. The term in square brackets in the PF (Eq 3a) penalizes high curvature of the interface while DGab drives the interface in direction of the thermodynamic stable phase. In the diffusion equation, the contribution proportional to the gradient of the PF variable accounts for solute redistribution at the interface. These equations are usually solved with standard finite element or control volume algorithms in the entire calculation domain with the advantage that no explicit tracking of the interface is necessary. As the thermodynamic driving force DGab in the PF equation depends on the local concentration and a moving interface redistributes solute via the DF term, both Eq 3a and 3b are closely coupled. The entire set of equations represents the interaction between growth and diffusion and interface curvature that naturally result in the formation of dendritic microstructure during metal solidification. Before giving examples, it is useful to examine two further issues related to solidification in casting processes: multicomponent alloy thermodynamics and nucleation.

Multicomponent Alloy Thermodynamics During solidification any material separates into two (or more) different phases: liquid and solid (or different solid phases). The clear reason

is the minimization of the total Gibbs energy as the sum of the Gibbs energies of the individual phases weighted by the phase fractions. Neglecting surface energy, the PF and concentration equations described previously will tend to the thermodynamic equilibrium as the minimum of Gibbs energies. It is therefore natural to combine a computational method for equilibrium calculation, the so-called CALPHAD method (Ref 22), and the PF method. In CALPHAD databases the Gibbs energies of the individual bulk phases fa are tabulated as functions of the alloy composition cia . The direct coupling between PF simulation and CALPHAD calculation was first demonstrated in Ref 23 and is described in more detail in Ref 13 and 24. For alternative approaches see Ref 25, 26. It must be stated clearly that only a reliable description of the alloy thermodynamics of the bulk phases enables a realistic PF simulation in technical alloys.

Nucleation Nucleation is not included in the general framework of PF theory. Some authors (Ref 12, 27) propose models with strong thermal noise to introduce random nucleation in a PF simulation. However, it is well known that an interface in a well-resolved PF simulation needs a minimum of four to five numerical cells; thus the size of a nucleus in an inoculated casting, the order of which is lower than a micrometer, cannot be resolved in a calculation of solidification structures of the order of several tens of micrometers. To overcome this problem, a deterministic seed density model is incorporated, where a distribution of heterogeneous seed particles is distributed over the calculation domain without resolving them explicitly. A seed is activated, if the local supercooling of a solid phase in the liquid matrix exceeds the supercooling needed for a solid particle to grow when its size corresponds to the size of

During commercial alloy solidification, a number of different phases nucleate. Nucleation and growth are closely correlated, and the whole process is governed by the competition between external heat extraction and release of latent heat. The latter may be considered in an average over the calculation domain, as the heat diffusion is typically 3 orders of magnitude higher than the solute diffusion in the liquid (Ref 24). Figure 3 shows four stages of solidification of the commercial piston alloy KS1295, where the main components Al, Cu, Si, and Mg have been considered. The alloy composition is hypereutectic with respect to the aluminum-silicon eutectic composition. Cooling was simulated by assuming a constant heat-extraction rate of 15 J/cm3, deduced from a macroscopic casting simulation of the piston, and considering the release of latent heat during solidification (Ref 21). First the primary silicon particles nucleate from the melt (Fig. 3a) and grow by depletion of silicon from the melt. A second effect of nucleating the silicon particles first is that the melt is depleted of impurities that would serve as nuclei for the a-aluminum phase. Also it is unlikely that a-aluminum will nucleate on the silicon particles, as the crystal lattices of both phases are quite different. So, the silicon particles continue to grow and the melt is depleted of silicon far below the eutectic composition. In the calculation shown in Fig. 3(a), the a-aluminum phase is nucleated in the corners of the simulation box that, in this calculation, represents the area of one grain. Next, a rapid dendritic growth of the aluminum phase is observed, while nearly no eutectic aluminum-silicon is formed at this stage. The solidification path is depicted schematically in Fig. 4, which deviates from the equilibrium path as a result of the suppression of nucleation of the a-aluminum phase at the eutectic point. In the melt, during the growth of the primary silicon particles, the silicon content falls significantly below the eutectic composition. Then, after nucleation of the a-aluminum phase, recalescence is observed and the concentration close to the a-aluminum phase returns to the equilibrium line. The silicon particles, on the other hand, profit from the rejection of silicon from the a-aluminum phase and continue to grow until they are engulfed by the a-dendrite. Also, tip splitting is observed (Fig. 3b). Figure 3 (c) shows the stage after nucleation of the copper silicides in the interdendritic region, and Fig. 3 (d) shows the last stage of the simulation shortly before termination of the solidification in the eutectic point with secondary silicon nucleating. This terminating eutectic phase, however, is much

438 / Modeling and Analysis of Casting Processes

(a)

(b)

spectroscopy (EDX) scan between two side branches where the relative compositions are shown for different times: immediately after the last interdendritic liquid has solidified locally, 50 s later, and 40 min later but at a temperature still above g-a transformation. All alloying elements segregate to the melt during solidification, where phosphorous shows the strongest relative segregation, while manganese and silicon are nearly indistinguishable (Fig. 6a). Directly after the last liquid has solidified at the location of the virtual EDX scan, carbon is forced to diffuse out of the interdendritic region by the cross interaction with silicon (Fig. 6b). Phosphorus follows on a longer timescale (Fig. 6c), showing an inverse segregation in the solid compared to the primary segregation during solidification, while silicon and manganese are nearly immobile in the solid. In the combined PF calculation of the solidification structure, solute partitioning, and chemical diffusion, a realistic picture of the distribution of alloying elements in a casting can be drawn.

Direct Grain Structure Simulation Using the Cellular Automaton Method

(c)

(d)

Fig. 3

Simulated solidification structure in a hypereutectic AlSiCuMg alloy in different stages. (a) Primary silicon particles and a-Al dendrite nucleated in the corners. (b) Growth of the silicon particles and the a-Al dendrite with engulfment of particles and tip splitting. (c) Precipitation of copper silicide particles. (d) Solidification structure before eutectic termination. The calculation domain is 800 by 800 mm in a two-dimensional approximation. Calculated by MICRESS

1200 TE

α

Liquid + silicon 1000

500 Aluminum + silicon

Temperature, ⬚F

Temperature, ⬚C

Liquid

Liquid + aluminum

700

800

300

0

10

Al

Silicon, wt%

Fig. 4

20

Si

Solidification path in hypereutectic Al-Si alloy

finer than the primary phases and therefore cannot be resolved in the present simulation.

Microsegregation in Steel As a second example, consider the solidification of a low-alloyed steel Fe-0.8%Mn-0.7%Si-

0.03%P-0.4%C. The emphasis of this simulation is the microsegregation profile of phosphorus after complete solidification. Phosphorus in steel today can mostly be eliminated through proper purification. However, if there is residual phosphorus left, it is known to embrittle the material and local accumulation should be avoided. Because of the chemical interaction between silicon, carbon, and phosphorus, the full diffusion matrix including cross terms has to be taken into account. The thermodynamic and diffusion data rely on the TCFE3 and MOB2 databases by Thermo-Calc Software. The calculation starts from a single seed in the corner of the calculation domain. Secondary arms of the dendritic microstructure form with a mean spacing of about 100 mm, resulting from a high cooling rate of 0.2 K/s (0.4  F/s). Figure 5 shows the microsegregation profile of the alloying elements before complete solidification. The small bar in the figure shows the position of a virtual energy-dispersive x-ray

A cellular automaton (CA) is a theoretical concept to simulate the evolution of a complex structure by cooperative action of autonomous cells. Each cell acts according to the same rules, but is dependent on the state of the neighboring cells and the local evolutions of average quantities. The rules come from microscopic descriptions developed for the evolution of a mushy zone (e.g., microsegregation) and can be taken either from a PF simulation or from analytical models (see below). For a PF simulation, a typical domain would be the CA REV cell (highlighted area in Fig. 2a shown expanded in Fig. 2b), fully paved with PF REV (small squares shown within Fig. 2b). For a CA simulation, CA REV cells are overlaid onto the domain shown in Fig. 1(a). A PF simulation using a lattice defined to cover the domain shown in Fig. 1(a) is out of reach because of computer limitations. For that reason, coupling of structure formation with a macroscopic FE method solving mass, heat, and fluid flows has been developed at the scale of the CA REV cells. Figure 7 shows schematically the superimposition of the CA lattice on the same domain used for a macroscopic finite element (FE).* Each cell n is uniquely defined by the coordinates of its center, Cn = (xn, yn), located in a finite element mesh, F. Element F has NnF nodes, labeled nFi in Fig. 7, which are locally indexed from i = 1 to NnF . In this study, NnF is equal to 3 for the triangular elements used. For the purpose of exchanging information *It is to be noticed that in Fig. 7 the FE mesh is a triangle while it is a quadrangle in Fig. 1(a). However, both types of elements are possible when coupling with the CA method.

Direct Modeling of Structure Formation / 439 from the FE nodes to the CA cells, linear internF cn 1 ,

polation coefficients, are defined between each node ni and a cell n. A variable available at the nodes, xni , can thus be used to calculate an interpolated value at cell n, xn: Nn X F

xn ¼

nF

cn 1 xni

(Eq 4)

i¼1

Similarly, when information computed at the level of the cells of the CA grid, xn, is required at the level of a FE node, xn, the following summation is performed: Nn 1 X xn ¼ cn x n i¼1 ni ni n

(Eq 5)

where Nnn is the number of cells n seen by node n. For node nF1 , it corresponds to all the cells drawn in Fig. 7. This averaging procedure is

normalized by the summation over all interpolation coefficients: Nn X n

n ¼

cnni

(Eq 6)

i¼1

Indices The state index, InS , is used to characterize the phase and the neighborhood of a cell n. It is defined as:  Cell n is liquid: InS ¼ 0.  Cell n is no longer liquid and at least one of

its nearest neighboring cell, mi (i 2 [1, Nn]), is still liquid ðImSi ¼ 0Þ: InS ¼ þ1.  Cell n is no longer liquid, and neither is its nearest neighboring cell, mi ðImSi 6¼ 0; i 2 ½1; N n Þ: InS ¼
1.

where Nn is the number of cells defined in the neighborhood of cell n. The Moore configuration considering the first and second nearest neighboring cells is used, yielding to Nn = 8. The grain index, InG , is used to track the development of the mushy zone within cell n. It is defined as:  Cell n is liquid ðInS ¼ 0Þ: InG ¼ 0.  Cell n is no longer liquid ðInS 6¼ 0Þ, and the

mushy zone is still developing: InG ¼ þ1 m with gm n <1; gn being the volume fraction of the mushy zone in cell n.  Cell n is no longer liquid ðInS 6¼ 0Þ, and its mushy zone is fully developed: InG ¼
1 for gm n ¼ 1. The CA model described hereafter is based on rules to modify the state index, InS , and the grain index, InG , associated with each cell n of a square lattice with the goal to simulate the development of the mushy zone upon solidification.

Nucleation It is assumed that nucleation of grains takes place on heterogeneous particles present in the melt as soon as a critical undercooling is reached. Because such nucleation behavior is only dependent on undercooling, it is considered as an instantaneous nucleation law. A Gaussian distribution is used to characterize the density of heterogeneous particles as a function of the critical undercooling at which they are activated (Ref 19). In order to initiate the solidification process at the microscopic scale, nucleation sites are randomly distributed among the cells of the CA square lattice. A nucleation site contained in cell n is characterized by a critical nucleation undercooling, Tnnucl , and a random crystallographic orientation, yn. There are three tests performed to nucleate a new grain in a cell n containing a nucleation site: Cross section of a dendritic microstructure in Fe-0.8%Mn-0.7%Si-0.03%P-0.4%C solidified at a cooling rate of 0.2 K/s (0.4  F/s). Calculated by MICRESS

1.6 P

Normalized field value(s)

Normalized field value(s)

4

3 C 2 Si Mn

1

0

(a)

Fig. 6

50

100 Distance, μm

150

Si

1.2 P 1.0 C 0.8 0

200

(b)

50

local liquidus temperature: Tn < TL(wn) with 1.4

Mn

1.4

 The cell is still liquid: InS ¼ 0.  The cell temperature, Tn, falls below the

Normalized field value (s)

Fig. 5

100 Distance, μm

150

200

Mn 1.2

Si

1

C P

0.8 0

(c)

50

100 Distance, μm

150

200

Element distribution in relative units, normalized by the mean concentration, in the scan position indicated in Fig. 5. (a) Before local solidification. The elements are enriched in the interdendritic region. (b) Immediately after solidification. Phosphorus is still enriched in the position of last solidification, but carbon has already diffused away due to interaction with silicon. (c) After 40 min at a temperature of approximately 1000  C (1830  F). Silicon and manganese are almost immobile in the solid and prevail in the region of last solidification, while phosphorus also was forced to diffuse into the regions of first solidification, depleted from silicon. Calculated by MICRESS

440 / Modeling and Analysis of Casting Processes y

x O

nF 1

F

cnν 1

v

yν

Cν nF 2 F xν

Fig. 7 nFi ,

nF 3

Topological links between a triangular finite element mesh, F, and a square cell of the cellular automaton grid, nF

n. Interpolation coefficients, cn1 , are defined between each cell and the NnF nodes of the finite element mesh,

with i ¼ 1; NnF and NnF ¼ 3 in this figure.

TL (wn) = TM + mwn, where TM is the melting temperature of the solvent in the binary phase diagram, m is a linear approximation of the liquidus slope of the binary phase diagram defined with the solute element, and wn is the local composition of cell n. Temperature Tn is calculated from a conversion of the average enthalpy of cell n, Hn, considering a purely liquid cell (Hn ¼ Cp Tn þ ls Hf ; Cp is the specific heat and ls Hf is the latent heat of fusion).  The temperature of cell n, Tn, deduced by interpolation from the nodes nFi defining element F in which cell n is located (Eq 4), must fall below the critical temperature defined for nucleation, Tn < TL ðwn Þ
Tnnucl . This is of course only possible providing a nucleation site was previously ascribed to cell n. These tests are applied to all the cells containing a nucleation site. If the three tests are verified, the following actions are taken to initialize the growth of the grain:  The state index of cell n, InS , is updated fol-

lowing the definitions given previously. Consequently, the state index of all cells defined in the neighborhood of cell n is also updated.  The grain index of cell n, IG n , is updated following the definitions given previously.

 The center of the growing shape of the newly

nucleated grain, Gn, is located at the center of cell n, Cn. A square shape with main diagonals corresponding to the preferential <10> directions of the dendrites stems and arms are initialized with a very small size. Angle yn defines the orientation of the [10] direction of the grain, Gn S½10 n , with respect to the Ox axis.

Figure 8 schematizes the growth propagation of the mushy zone from a cell n to a cell m. Within one time step, the four half-diagonals associated to the shape of cell n have grown from center Gn. As a result, the mushy zone has extended from the polygon delimited by the white continuous thin lines to the polygon delimited by the black continuous thin lines ½ij tips, and identified by the Sn ½ij 2 ½01; 10; 10; 01. Cell m was initially liquid ðImS ¼ 0; ImG ¼ 0Þ, while the mushy zone has developed in cell n ðInS ¼ þ1; InG ¼ þ1Þ. After the propagation of the mushy zone in cell n, Cm is now engulfed by the growing ½ij shape Sn . Capture of cell m is thus achieved by its neighboring cell n. The initial growing ½ij shape associated to a cell m, Sm (hashed area delimited by a continuous bold lines), as well as the position of the growth center of cell m, Gm, are then calculated. Two other growing shapes are schematized in Fig. 8 for cell m. Tips ½ij min (gray area delimited by dashed bold Sm lines) represent the initial shape defined at a ½01 ½10 former time when edge Sn Sn reached position Cm, thus starting to engulf cell m. Tips ½ij min thus define the minimum size associated Sm with the growth of cell m. When no movement of the grain is accounted for, the center of this first growing shape, Tm, is kept unchanged during the entire time of the simulation. If growth is maintained homothetic with respect to the velocity in all <10> directions, the maximal growing shape of cell m can be evaluated when the mushy zone has propagated to all the neighboring cells. In Fig. 8, this maximal growing ½ij max defined when shape corresponds to tips Sm the center of the southeast neighbor of cell m is ½01 ½10

reached by edge Sm Sm . All the neighboring cells of m are then captured. The current, minimal, and maximal growing ½ij ½ij min ½ij max , and Sm , shapes, respectively Sm , Sm are used to define the volume fraction of the mushy zone, gm m , associated to cell m as: "

Growth Unlike with the PF method, the development of the solid/liquid interface is not directly modeled by the CA method. Instead, only the development of the grain envelope is simulated. Since the growth kinetics required for determining the velocity of the envelope of the grains is mainly based on diffusion in the liquid at a length scale given by the radius of curvature of the solid/liquid interface, it cannot be modeled with a CA model.* Similarly, while one could also take advantage of the PF Fa(x) indicator and compute the volume fraction of solid located within the CA REV. As stated before, coupling is yet not achievable between the PF and the CA methods. Thus, the CA method makes use of growth kinetics theories and microsegregation models to approximate growth morphology and kinetics of the grain envelopes, as well as its internal fraction of solid.

gm m

# Am
Amin m ¼ Min max ;1 Am
Amin m

(Eq 7)

max where Am, Amin are the areas assom , and Am ciated with the current, initial, and maximal growing shapes, respectively. The volume fraction of solid associated to cell m, gsm , is then given by: sm gsm ¼ gm m gm

(Eq 8)

where gsm m is the volume fraction of solid ½ij located in the envelope Sm . It is to be noted that the growing shape associated with a given cell n is not a regular square as was the case *Some authors (Ref 29, 30) use a CA type of method to compute the development of microstructure at the scale of the solid/liquid interface. However, the size of the CA cells is then decreased to a size similar to that used for a PF simulation. Then realistic simulation of the development of the structure at the scale of the casting cannot be achieved.

Direct Modeling of Structure Formation / 441 [01] max Sµ

S [01] n S [01] m

min S [01] m [10] max Sµ

m

n Cm

Cn

¯ S [n10] ¯

S m[10] max

¯ 0] [1 Sm

¯

Gn

S [m10] min

Gm

[10] S [10] m = Sn min S [10] m

¯] S [01¯] min m S [01 m ¯ ] S [01 n

¯

] max S [01 m

The algorithm to model the growth of a cell n identified as part of a grain is schematized, as well as the propagation of the grain to a neighboring cell m. The growing shape associated with cell n has propagated ½ij from the white to the black continuous thin lines to reach positions Sn ; ½ij 2 ½01; 10; 10; 01. At that time, the ½ij growing shape Sn has engulfed the center of cell m, Cm. The actual size of the growing shape associated to cell ½ij m; Sm (hashed area delimited by a continuous bold line), is calculated. The initial growing center and shape ½ij min associated to cell m, respectively Gm and Sm (grey area delimited by the dashed bold lines), as well as its maximal ½ij max growing shape, Sm (delimited by the dotted bold line), used to defined the volume fraction of the mushy zone associated to cell m; gm m , are also shown.

Fig. 8

in previous procedures (Ref 31, 32). Limitation to a simple square was based on the assumption that the temperature is uniform within the cell, and all four <10> directions could consequently grow at the same velocity. Since the grain envelope is sustained by a series of neighboring cells, the temperature gradient could be taken into account by the temperature difference between cells. Because of the effect of the fluid flow on the growth kinetics of the four <10> directions, each of the directions Gn S½ij n ; ½ij 2 ½01; 10; 10; 01, makes a different angle with the fluid-flow direction. Four differ½ij ent values of the growth velocities, vn , are thus calculated. A similar extension was proposed by Takatani et al. for the study of grain texture formation in strip casting of thin steel sheets due to the effect of the relative fluid-flow velocity on grain structure formation (Ref 33). Details concerning the boundary layer correlation used to compute the supersaturation of the dendrite tips growing into an undercooled melt in the presence of fluid flow, and thus to deduce its growth velocity, are available in Ref 34. The CA model has been extended to account for the transport of the grains due to fluid flow and sedimentation. This part of the model is not presented here. More information is available in Ref 35.

Coupling of Direct Structure Simulation at Macroscopic Scale The finite element method is used to solve the average conservation equations written for the total mass, the energy, the mass of solute, and the momentum at the scale of the casting (Ref 36, 37). As outputs of a FE simulation, it is possible to predict the time evolution of the average enthalpy, the average velocity of the liquid phase (the solid being fixed), and the average solute composition. The average enthalpy and solute composition are converted into evolutions of temperature, solute composition of the liquid, and volume fraction of solid. This can be carried out by applying a simple microsegregation model, such as the lever-rule approximation deduced from the phase diagram data, at the microscopic scale of the dendrite arm spacing. In such a purely macroscopic calculation, the presence of the mushy zone is thus a direct function of the average enthalpy and the local liquidus temperature. As soon as the enthalpy falls below the value of the enthalpy that corresponds to the local liquidus temperature, the solid starts to form and the volume fraction of solid consequently increases. The final solid is formed at the eutectic temperature. The method described previously does not

consider coupling with direct simulation of structure development. In order to account for the effect of the growth undercooling of the microstructure, as well as for the transport of equiaxed grains that can freely develop in the liquid, the mesoscopic cellular automaton (CA) method has been coupled with the macroscopic FE method (Ref 38). The output of the CA model is the grain structure. Such a coupling between the mesoand macroscopic methods makes it possible to take into account nonequilibrium solidification paths due to growth undercooling. The volume fraction of solid is not only a function of the average composition and enthalpy, but also a function of the presence of the mushy zone. The presence of the mushy zone itself becomes a function of the location of the grains. The latter depends on both the undercooling of the growing dendritic grain structure and its sedimentation and transport due to the fluid flow. The coupling with the movement of the free equiaxed grains requires modifying the drag force entering the Navier-Stokes equation. This force depends on the relative velocity of the liquid with respect to the solid (i.e., velocity of the equiaxed grains), as well as on the volume fraction of the equiaxed grains in the macroscopic REV. It is computed according to the model proposed by Wang et al. (Ref 39), and it is added as a source term in the formulation of the FE method. In order to calculate the velocity of the grains that are free to move in the liquid, the model proposed by Ahuja et al. (Ref 40) has been reprogrammed. The trajectory of each individual grain is computed with a dedicated CA algorithm. The supersaturation entering the dendrite growth kinetics is computed using a correlation that accounts for the effect of the relative flow velocity with respect to the solid velocity as well as the orientation between the dendrite growth direction and the fluid-flow direction (Ref 34). It is worth noticing that the CAFE model presents similar objectives as the Eulerian model proposed by Wang and Beckermann (Ref 6). The main difference is due to the Lagrangian description used to track each individual grain with the CA method. Only the liquid velocity is calculated with the FE method. Such a simplification requires the transport of average quantities due to the movement of the solid phase to be accounted for. This is achieved by the development of coupling scheme between the CA and FE method. The experimental configuration proposed by Hebditch and Hunt (Ref 41, 42) is considered for the application of the CAFE model. It consists of a parallelepipedic cavity with dimensions 100 mm (4 in.) long, 60 mm (2 in.) high, and 13 mm (0.5 in.) thick. All faces are carefully insulated except for one of the smallest faces. The Pb-48wt%Sn binary alloy is cooled down and solidified by imposing a circulation of water in a copper chill that is maintained in contact with this face. As a consequence of this configuration, a two-

442 / Modeling and Analysis of Casting Processes

(M)

[wt%]

64

(a)

60 Sn composition [wt%]

60 56 52 48 44 -

56 52 48 44

40

40 0 (b)

0.02

(c)

0.04

0.06

0.08

0.1

Position [m]

Fig. 9

Predictions from the cellular automaton finite element model. (a) Final grain structure. (b) Segregation map of tin with its composition scale. (c) Composition profiles for a Pb-48wt%Sn alloy. Equiaxed grains nucleated in the undercooled melt are free to move due to sedimentation and buoyancy-driven flows. The FE mesh (M) is drawn together with horizontal lines indicating the location of the profiles drawn in (c). The symbols in (M) indicate the location of the measurements (Ref 41). The same styles are used in (c) and (M) for curves and symbols.

dimensional Cartesian approximation of the transport phenomena can be made by neglecting the interactions of the fluid flow with the two largest faces of the mold (Ref 43). The FE mesh is refined in both the horizontal and vertical directions as shown in Fig. 9(M). The location of the measurements of the composition in the final as-cast ingot are also shown in Fig. 9(M). A detailed table of the values of the parameters is given in Ref 38. Figure 9 presents the results of the simulations. Maps of grain structures and macrosegregation, as well as composition profiles are shown. Time evolution of the development of the grain structure shows that equiaxed grains accumulate in the bottom region of the ingot by sedimentation. The general trends of the segregation map are the same as the one predicted by a purely FE calculation (Ref 36). However, variations known as mesosegregations are also predicted at a smaller length scale. These variations are caused by the development of equiaxed grains in preferred zones, leading to instability of the segregation field that further develops with the propagation of the mushy zone. It is interesting to observe the formation of channels in the top left side of Fig. 9(b). The mechanism that led to the formation of this channel is instability of the growth front. Equiaxed grains have moved and accumulated at a position close to the actual root of the channels, forming arms of fixed equiaxed grains. These arms served as barriers to the fluid flow that then developed preferentially on their

sides. Consequently, a channel with enriched solute developed between the arms, similar to the mechanism leading to the formation of a freckle (channel segregation). The average composition profiles drawn in Fig. 9(c) also show similar trends. In the present configuration defined by the experiments performed by Hebditch and Hunt (Ref 41), comparison with a calculation that does not account for the development of the grain structure (purely macroscopic FE calculation) shows that macrosegregation is not directly influenced by the structural features. The effect of the transport of equiaxed grains on the macrosegregation is well demonstrated in the literature (Ref 44, 45), especially for larger ingot sizes. The limited size of the present Pb-48wt%Sn ingot does not favor such an effect. It should also be said that the grain structure model only considers the formation of equiaxed dendritic grains: no globular grains with high inner volume fractions of solid are taken into account. Such a limitation could be detrimental for the application of the present model to the prediction of segregation induced by grain movement. Not only the morphology of the equiaxed grain structure plays a role on segregation, but also it is expected to considerably modify the fluid flow. Despite the fact that the latter effect is accounted for in the present model, the consequence on the overall fluid flow remains limited and macrosegregation still is explained mainly by the same thermosolutal considerations that lead to the segregation map

computed with a purely macroscopic FE calculation. There are fewer studies in the literature on the effect of the grain structure on mesosegregation. The fact that instability of the growth front leads to further instability of the segregation map is well established for the formation of channels that could give rise to freckle formation. However, the proposition that accumulation of transported equiaxed grains on the growth front could lead to the formation of channels and eventually freckles is new. Only in situ diagnostics by direct visualization of the formation of the grain structure could validate this prediction. Since the grain structure is difficult to extract from Ref 41 and 42, and since the measurements were carried on a very coarse grid and for a relatively large volume (4 mm, or 0.2 in., diam rods were extracted from the as-cast structure), it is not possible to directly use these experimental results to validate the predictions of the CAFE model. Similarly, no detailed measurement of the average composition has been conducted yet with the purpose of characterizing a segregation map with a definition smaller than the grain size. It is thus difficult to validate the present predictions of mesosegregation. Furthermore, the measurements are often carried out with samples of several cubic millimeters. Such measurements do not permit validation of the simulations, unless the predicted segregation maps are also drawn by averaging the composition over the same volume as the one used

Direct Modeling of Structure Formation / 443 for the measurements. This procedure, however, has not been used here since it rubs out the effect of the grain structure.

Summary Two models used to predict direct structure formation in casting have been presented. The development of solid/liquid interfaces or of grain envelopes can be tracked using a PF method or a CA method, respectively. While growth morphology, kinetics, and segregation are direct outputs of the PF method, its application remains limited to small portions of a casting. Computer limitations are the main problem and are not likely to be solved within the next few decades (Ref 46). For that reason, the CA method has been developed with the goal of providing a direct simulation of the structure in the entire casting while not considering the development of the solid/liquid interface. As a consequence of this approximation, only the development of grain envelopes is simulated, and microscopic models that have been developed in the literature for a long time are integrated. These microscopic models include kinetics models for the growth front (grain envelope or boundary between the mushy zone and the liquid outside the envelope) or microsegregation models. The advantage of this CA mesoscopic method is the possibility of studying the influence of the development of structure on quantities characterized at the scale of the casting (e.g., macrosegregation). The development of quantitative models is very much related to the availability of experimental data. Such data are required to test the assumption of the models. Since the assumption applies at the scale of the structure, experimental data need to be collected in very well defined and characterized conditions. This is true for casting geometry, heat flow conditions, and alloy properties. Thus, the development of experimental facilities is still a need for further modeling activities. Coupling of models and numerical methods could be the solution to improving the results for engineering processing of solidification structure. However, such coupling is difficult because of the interdisciplinary nature of the field. It is recognized that such challenging coupling is not simple. REFERENCES 1. R. Trivedi and E. Sunseri, Non-Plane Front Solidification: Cellular and Dendritic Growth, and Columnar to Equiaxed Transition, Casting, S. Viswanathan, Ed., Vol 15, ASM Handbook, ASM International, 2008 2. S. Shankar, Eutectic Solidification, Casting, S. Viswanathan, Ed., Vol 15, ASM Handbook, ASM International, 2008 3. Peritectic Solidification, Casting, S. Viswanathan, Ed., Vol 15, ASM Handbook, ASM International, 2008

4. W. Kurz, Plane Front Solidification, Casting, S. Viswanathan, Ed., Vol 15, ASM Handbook, ASM International, 2008 5. M. Krane, V. Voller, and B. Li, Modeling of Transport Processes and Electromagnetics, Casting, S. Viswanathan, Ed., Vol 15, ASM Handbook, ASM International, 2008 6. C.Y. Wang and C. Beckermann, Equiaxed Dendritic Solidification with Convection: Part I: Multiscale/Multiphase Modeling, Metall. Mater. Trans., Vol 27A, 1996, p 2754–2764 7. M.A. Martorano, C. Beckermann, and Ch.-A. Gandin, Metall. Mater. Trans. A, Vol 34, 2003, p 1657 8. W. Kurz and D.J. Fisher, Fundamentals of Solidification, 3rd ed., Trans Tech Publications, Switzerland, 1992 9. A. Karma, Phase-Field Formulation for Quantitative Modeling of Alloy Solidification, Phys. Rev. Lett., Vol 87, 2001, p 115701–1–4 10. B. Echebarria, R. Folch, A. Karma, and M. Plapp, Quantitative Phase-Field Model for Alloy Solidification, Phys. Rev. E, Vol 70, 2004, p 061604–1–22 11. G. Caginalp and P. Fife, Phase-Field Methods for Interfacial Boundaries, Phys. Rev. B, Vol 33, 1986, p 7792–7794 12. L.Q. Chen, Phase-Field Models for Microstructure Evolution, Ann. Rev. Mater. Res., Vol 32, 2002, p 113–140 13. J. Eiken, B. Bo¨ttger, and I. Steinbach, Multiphase Field Approach for Multicomponent Alloys with Extrapolation Scheme for Numerical Application, Phys. Rev. E, Vol 73, 2006, p 066122–1–9 14. A.A. Wheeler, W.J. Boettinger, and G.B. McFadden, Phase-Field Model for Isothermal Phase Transitions in Binary Alloys, Phys. Rev. A, Vol 45, 1993, p 7424–7439 15. I. Steinbach, F. Pezzolla, B. Nestler, M. Seeßelberg, R. Prieler, G.J. Schmitz, and J.L.L. Rezende, A Phase Field Concept for Multiphase Systems, Phys. D, Vol 94, 1996, p 135–147 16. L. Tan and N. Zabaras, A Level Set Simulation of Dendritic Solidification with Combined Features of Front Tracking and Fixed-Domain Methods, J. Comput. Phys., Vol 211, 2006, p 36–63 17. A. Jacot and M. Rappaz, A Pseudo-Front Tracking Technique for the Modelling of Solidification Microstructures in MultiComponent Alloys, Acta Mater., Vol 50, 2002, p 1909–1926 18. D.J Browne and J.D. Hunt, A Fixed Grid Front-Tracking Model of the Growth of a Columnar Front and an Equiaxed Grain During Solidification of an Alloy, Numer. Heat Transfer, Part B: Fundamentals, Vol 45, 2004, p 395–419 19. M. Rappaz, Modeling of Microstructure Formation in Solidification Processes, Int. Mater. Rev., Vol 34, 1989, p 93–123 20. J.D. van der Waals, “The Thermodynamic Theory of Capillarity under the Hypothesis

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

of a Continuous Variation of the Density,” Amsterdam; English Trans. J.S. Robinson, J. Stat. Phys., 1979 W.J. Boettinger, J.A. Warren, and C. Beckermann, Phase-Field Simulation of Solidification, Ann. Rev. Mater. Res., Vol 32, 2002, p 163–194 L. Kaufman and H. Bernstein, Computer Calculation of Phase Diagrams with Specific Reference to Refractory Materials, Academic Press, 1970 U. Grafe, B. Bo¨ttger, J. Tiaden, and S.G. Fries, Coupling of Multicomponent Thermodynamic Databases to a PF Model: Application to Solidification and Solid State Transformations of Superalloys, Scr. Mater., Vol 42, 2000, p 1179–1186 B. Bo¨ttger, J. Eiken, and I. Steinbach, Phase-Field Simulation of Equiaxed Solidification in Technical Alloys, Acta Mater., Vol 54, 2006, p 2697–2704 R.S. Qin and E.R. Wallach, A Phase-Field Model Coupled with a Thermodynamic Database, Acta Mater., Vol 51, 2003, p 6199–6210 H. Kobayashi, M. Ode, S.G. Kim, W.T. Kim, and T. Suzuki, Phase-Field Model for Solidification of Ternary Alloys Coupled with Thermodynamic Database, Scr. Mater., Vol 48, 2004, p 689–694 L. Granacy, T. Pustai, T. Borzsonyi, J.A. Warren, and J.F. Douglas, A General Mechanism of Polycrystalline Growth, Nature Mater., Vol 3, 2004, p 645–650 A.L. Greer, P.S. Cooper, M.W. Meredith, W. Schneider, P. Schumacher, J.A. Spittle, and A. Tronche Grain Refinement of Aluminium Alloys by Inoculation, Adv. Eng. Mater., Vol 5, 2003, p 81–91 V. Pavlyk and U. Dilthey, Simulation of Weld Solidification Microstructure and Its Coupling to the Macroscopic Heat and Fluid Flow Modelling, Mod. Sim. Mater. Sci. Eng., Vol 12, 2004, p 33–45 L. Beltran-Sanchez and D.M. Stefanescu, Growth of Solutal Dendrites: A Cellular Automaton Model and Its Quantitative Capabilities, Mater. Trans. A, Vol 34, 2003, p 367–382 Ch.-A. Gandin, R.J. Schaefer, and M. Rappaz, Analytical and Numerical Predictions of Dendritic Grain Envelopes, Acta Mater., Vol 44, 1996, p 3339–3347 Ch.-A. Gandin and M. Rappaz, A 3D Cellular Automaton Algorithm for the Prediction of Dendritic Grain Growth, Acta Mater., Vol 45, 1997, p 2187–2195 H. Takatani, Ch.-A. Gandin, and M. Rappaz, EBSD Characterisation and Modelling of Columnar Dendritic Grains Growing in the Presence of Fluid Flow, Acta Mater., Vol 48, 2000, p 675–688 Ch.-A. Gandin, G. Guillemot, B. Appolaire, and N.T. Niane, Boundary Layer Correlation for Dendrite Tip Growth with Fluid Flow, Mater. Sci. Eng., Vol A342, 2003, p 44–50

444 / Modeling and Analysis of Casting Processes 35. G. Guillemot, Ch.-A. Gandin, and H. Combeau, Modeling of Macrosegregation and Solidification Grain Structures with a Coupled Cellular Automaton-Finite Element Model, ISIJ Int., Vol 46, 2006, p 880–895 36. N. Ahmad, H. Combeau, J.-L. Desbiolles, T. Jalanti, G. Lesoult, J. Rappaz, M. Rappaz, and C. Stomp, Numerical Simulation of Macrosegregation: A Comparison between Finite Volume Method and Finite Element Method Predictions and a Confrontation with Experiments, Metall. Mater. Trans., 1998, Vol 29A, p 617–630 37. M. Bellet, V.D. Fachinotti, S. Gouttebroze, W. Liu, and H. Combeau, A 3D-FEM Model Solving Thermomechanics and Macrosegregation in Binary Alloys Solidification, in Solidification Processes and Microstructures—A Symposium in Honor of Wilfried Kurz, M. Rappaz, C. Beckermann, and R. Trivedi, Ed., The Minerals, Metals and Materials Society, Warrendale, 2004, p 41–46

38. G. Guillemot, Ch.-A. Gandin, and H. Combeau, Modeling of Macrosegregation and Solidification Grain Structures with a Coupled Cellular Automaton—Finite Element Method, Solidification Processes and Microstructures: A Symposium in Honor of Prof. W. Kurz, PA, 2004, p 157–163 39. C.Y. Wang, A. Ahuja, C. Beckermann, and H.C. de Groh III, Multiparticle Interfacial Drag in Equiaxed Solidification, Metall. Mater. Trans., Vol 26B, 1995, p 111–119 40. C.Y. Wang, A. Ahuja, C. Beckermann, R. Zakhem, P.D. Weidman, and H.C. de Groh III, Drag Coefficient of an Equiaxed Dendrite Settling in an Infinite Medium, in Micro/Macro Scale Phenomena in Solidification, HTD-218, AMD-139, ASME, 1992, p 85–91 41. D.J. Hebditch and J.D. Hunt, Observations of Ingot Macrosegregation on Model Systems, Metall. Trans. Vol 5, 1974, p 1557– 1564 42. D.J. Hebditch, “Segregation in Castings,” Ph.D. Thesis, Oxford University, UK, 1973

43. J.-L. Desbiolles, Ph. The´voz, and M. Rappaz, Micro-/Macrosegregation Modeling in Casting: A Fully Coupled 3D Model, in Modeling of Casting, Welding and Advanced Solidification Processes X, D. M. Stefanescu, J. Warren, M. Jolly, and M. Krane, Ed., The Minerals, Metals and Materials Society, Warrendale, PA, 2003, p 245–252 44. C. Beckermann, Modelling of Macrosegregation: Applications and Future Needs, Int. Mater. Rev., Vol 47, 2002, p 243–261 45. G. Lesoult, V. Albert, H. Combeau, D. Daloz, A. Joly, C. Stomp, G.U. Gru¨n, and P. Jarry, Equiaxed Growth and Related Segregations in Cast Metallic Alloys, Conference IUMRS-ICAM’99, International Union of Materials Research Societies 1999, Beijing, China 46. V.R. Voller and F. Porte-Agel, Moore’s Law and Numerical Modeling, J. Comput. Phys., Vol 179, 2002, p 698–703

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 445-448 DOI: 10.1361/asmhba0005237

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Modeling of Microsegregation and Macrosegregation D.R. Poirier, The University of Arizona J.C. Heinrich, University of New Mexico

IN THE PREVIOUS SECTION OF THIS VOLUME, “Principles of Solidification,” an overview of macrosegregation can be found (Ref 1). In order to model macrosegregation, one must consider convection and the partitioning of the segregating elements at the dendritic length scale. The partitioning results in microsegregation and convection, with a small contribution from diffusion, which are the primary transport mechanisms for causing macrosegregation.

Microsegregation Most solidification processes involve alloys that solidify with a complex dendritic interface within individual grains that occupy a transient solid-plus-liquid region, called the mushy zone. There are phase-field models and interfacetracking models for simulating the development of the dendritic morphology and elemental partitioning during solidification, but these models require algorithms that are so computationally expensive that they cannot be used to predict large-scale macrosegregation. Rather, the solid-plus-liquid mixture is treated as a continuum (Ref 2, 3). First, consider a differential volume element of the solidifying alloy and assume no diffusion in the solid. Mass conservation of any solute element in the alloy, in the absence of transport across the volume element, is: ðCL  CS Þ dfS ¼ ð1  fS Þ dCL

(Eq 1)

where CL is the concentration of the element in the liquid, in weight percent; CS* is the concentration of the element at the solid-liquid interface, in weight percent; and fS is the mass fraction of solid. The left side of Eq 1 is the solute rejected by the solid in the solidification increment dfS. The concentration of the solute behind the interface does not change because of the assumption of no diffusion in the solid. The right side is the increase of solute in the interdendritic liquid. Whether CS* < CL or

CS* > CL, Eq 1 applies. The difference in concentration between the concentration of the liquid at the interface and CL is quite small in most solidification processes and neglected in this model. In the absence of transport, the overall concentration of a solute in both phases taken together remains at C0. Assuming the partition ratio, defined as k = CS*/CL, is constant, Eq 1 in its integrated form is called the Scheil (also known as Scheil-Gulliver) equation:

It is assumed that k and  are constant and that the interface advances with the square root of time rate during solidification. Equation 3 reduces to the Scheil equation when  = 0 (no diffusion), but it gives an incorrect result as the extent of diffusion in the solid increases. Clyne and Kurz (Ref 6) added a heuristic parameter to account for the extent of diffusion in the solid during solidification, 0 ; it is defined as:

CS ¼ kC0 ð1  fS Þk1

     1 1 1 0 ¼  1  exp   exp   2 2

(Eq 2)

A modification of Eq 2 can be used to estimate the amount of eutectic in a binary alloy that forms at the end of solidification. As solidification proceeds, eventually CL = CE and the amount of the liquid is (1  fS) = fE, so the mass fraction of eutectic in the as-cast microstructure can be calculated. Equations 1 and 2 are best applied to substitutional alloy elements with attendant low mass diffusivities. Interstitial elements, however, diffuse rapidly, so it is better to assume complete diffusion in the solid during solidification and apply the inverse lever rule for equilibrium (i.e., complete diffusion in the solid). Solidification software for predicting microsegregation in multicomponent alloys is commercially available; examples can be found in Ref 4 and 5.

Microsegregation with Diffusion in the Solid An approximation to account for diffusion in the solid, first devised by Brody and Flemings (Ref 2), is: CS ¼ kC0 ½1  ð1  2kÞfS ðk1Þ=ð12kÞ

(Eq 3)

where  is a Fourier diffusion number, defined as DStf / 4l2; DS is the diffusion coefficient of the solute in the solid; tf is the solidification time; and l is the dendritic length scale (often taken as the secondary dendrite arm spacing).

(Eq 4)

Equation 3 is applied with  replaced by 0 . Equation 4 accurately describes behaviors in the extremes of no diffusion and complete diffusion in the solid during solidification.

Example Simulation of Microsegregation The analytical solution of Kobayashi (Ref 7) was made convenient for digital solutions in Yeum et al. (Ref 8). As an example of the utility of the analytical model, calculated values of CS as a function of fS after complete solidification can be generated in a multicomponent alloy and compared to empirically measured microsegregation profiles. Equations 3 and 4 cannot be used for this purpose because CS* changes by diffusion as the interface advances, whereas the analytical model generates both CS* and CS*. In this example, empirical microsegregation profiles of niobium (Ref 9, 10) in a nickel-base superalloy (alloy 718) were plotted, and from many trial combinations of kNb and , the empirical coordinates were represented very well by the analytical model with kNb = 0.5 and 0.05 <  < 0.1. The wt% Cr versus fraction solid in the same alloy is also reported in Ward et al. (Ref 10). Comparing trial analytical predictions to the empirical results, reasonably good agreement was found with kCr = 1.19 and   0.1. The utility of Eq 3 and 4 is realized when one wants to estimate fraction solid as temperature

446 / Modeling and Analysis of Casting Processes for flow through dendritic equiaxed grains and columnar grains are in Ref 18 and 19. When gL = 1 (i.e., all liquid), Eq 7 reduces to the Navier-Stokes equation for a Newtonian liquid. Thus, Eq 7 allows for a natural transition in momentum transport from the mushy zone to the liquid pool. Using shrinkage in continuity, Eq 6, and including the expansion of the stress tensor, the conservation of momentum with shrinkage is:     @u u gL  ¼  rP þ gL g þur L @t gL L    @gL þ r2 u þ r @t 3  gL ½K1 u

Fig. 1

Temperature versus fraction solid during solidification of alloy 718; calculated using  = ½ for complete diffusion,  = 0 for no diffusion, and  = 0.07 and 0.10

changes during solidification. This can be done with a known relationship between the composition of the interdendritic liquid and its equilibrium temperature. Taking kNb = 0.5, kCr = 1.19, and the other partition ratios from Sung and Poirier (Ref 11), Fig. 1 was generated. There are four curves:  = 0 (no diffusion),  ¼ 1/2 (equilibrium), and  = 0.07 and 0.10 (some diffusion) for each of the six major elements in alloy 718. These calculations indicate a realistic nonequilibrium solidus of approximately 1170  C (2140  F), compared to a nonequilibrium solidus of < 1000  C (1830  F) as predicted by Scheil-type solidification. (See the article “Thermodynamics and Phase Diagrams” in this Volume for a discussion of Scheil pathway solidification.)

Continuum Model of Macrosegregation The underlying model comprises the transport equations: continuity, momentum, energy, and solute conservation. The mushy zone is treated as a porous medium. More details pertaining to the transport equations, as well as modifications for numerical implementation, can be found in Ref 12 to 17. Continuity Equation. The following assumptions are invoked: (a) only the liquid convects; and (b) the densities of the liquid and solid differ, but each is constant. Assumption (a) leads to: @ ðS gS þ L gL Þ þ r  L u ¼ 0 @t

(Eq 5)

where t is time; S and L are the densities of the solid and liquid, respectively; gS and gL are the volume fractions of solid and liquid,

respectively; and u is the superficial velocity. With solidification shrinkage defined as  = (S  L)/L, the continuity equation is: ru¼

@gL @t

(Eq 6)

Momentum Equation. A technique known as volume averaging (Ref 12) is used to derive the momentum equation. Others (Ref 13, 14) have opted for a mixture theory. Whether one chooses the former or the latter, the result is a momentum equation with the inertial terms, diffusion terms, body force, the force arising from pressure gradient, and an additional force called the Darcy term. The assumptions include: (c) negligible exchange of momentum because of shrinkage; (d) negligible momentum dispersion; (e) a Newtonian constitutive relation between stress and strain rates; and (f) first-order resistance to flow (i.e., Darcy flow). Under these assumptions and using volume averaging, the momentum equation is: L gL

  @v þ v  rv ¼  gL rP þ L gL g @t þ r  N  gL 2 ½K1 v (Eq 7)

where P is the pressure, g is gravitational acceleration, m is the viscosity of the interdendritic liquid, K is the permeability tensor, N arises from the deviatoric part of the stress tensor of the liquid, and the intrinsic velocity is related to the superficial velocity by v = u/gL. The fourth term on the right side of Eq 7 is the Darcy term. The permeability K is a conductivity for flow in the mushy zone, which depends on the volume fraction of liquid and a length scale of the dendritic solid. Models for the permeability

(Eq 8)

Equation 8 is the same as Eq 2 by Fellicelli et al. (Ref 20). The only remaining detail in the momentum equation is the Boussinesq approximation of the gravity-force term. It is assumed that the reference density of the liquid, L, is known at a reference temperature, and the effect of each element and the temperature on the density of the liquid is linear and represented by known expansivities (Ref 12, 13, 16, 17, 19, 20). Solute Conservation Equation. Assumption (a) still applies; starting with Eq 15 in Ref 21, each element in the alloy follows: r  j þ r  ðL CL uÞ þ


@ C ¼ 0 @t

(Eq 9)

where j is the diffusion flux in kg (solute) m2 s1, C is the mixture concentration in kg (solute) m3, and the second term is the advection of solute. Equation 9 applies to both the mushy zone and the all-liquid zone. The advection term can be expanded for the assumptions of constant L and constant ; then, Eq 9 becomes: 1 @gL þ u  rCL r  j þ CL  @t L   @ C ¼0 þ @t L

(Eq 10)

The flux for diffusion in the mixture is needed. This is written as: j ¼ L D0L rCL  S D0S rCS

(Eq 11)

where D0L and D0S are effective diffusivities: D0L ¼

gL DL ð1  gL ÞDS and D0S ¼ : L S

where DL and DS are the diffusion coefficients, and  L and  S are the tortuosities to account for the morphology and connectivity of the phases. Lacking data on the tortuosities, it is assumed that  L =  S = 1 (assumption f). For substitutional elements, DL >> DS, so Eq 11 reduces to: j ¼ L gL DL rCL

(Eq 12)

Modeling of Microsegregation and Macrosegregation / 447

On the other hand, interstitial elements approximate equilibrium conditions during partitioning; thus, rCS = krCL and: j ¼  Dmix rCL

(Eq 13)

with the effective diffusivity of the mixture defined as: 1 Dmix ¼ ½L gL DL þ kS ð1  gL ÞDS  

(Eq 14)

and , the mixture density, is defined as  ¼ gL L þ ð1  gL ÞS . Conservation of Energy. Equation 1 in Ref 21 can be written as:
@ H ¼ r  ðrT Þ  r  ðL HL uÞ @t

(Eq 15)

where H ¼ S ð1  gL ÞHS þ L gL HL , k is the mixture thermal conductivity, u is the superficial velocity of the liquid, and HS and HL are the intensive enthalpies of the solid and liquid (J kg1), respectively. The enthalpies are expressed in terms of the respective enthalpies at a reference temperature, TH, where the latent heat is: L ¼ HLH þ HSH

(Eq 16)

The time derivative of the volumetric enthalpy in Eq 15 is expanded as follows:
@ @gL @gL þ L HL H ¼  S HS @t @t @t @HS @HL þ L gL þ S ð1  gL Þ @t @t @gL ¼ ½L HL  S HS  @t @T þ ½S ð1  gL ÞcS þ L gL cL  @t

(Eq 17)

where cS and cL are the specific heat capacities. The advective term in Eq 15 is expanded while taking L = constant, and after expanding and combining with Eq 17, the energy equation becomes: ½S ð1  gL ÞcS þ L gL cL 

@T ¼ r  ðrT Þ @t

 L cL u  rT   @gL  S L þ ðcL  cS ÞðT  T H Þ @t

(Eq 18)

Numerically, the term with the latent heat term must be treated implicity in order for the numerical algorithm to be stable. To make this term implicit for a multicomponent alloy requires that the temperature of the mixture be related to the composition of the interdendritic liquid. Details for this procedure can be found in Ref 16 and 17. Calculation of Pressure. A penalty method can be used to impose incompressibility, which eliminates the pressure as an unknown in Eq 8. If needed, the pressure can be recovered through

a postprocessing step. To attack this problem for finite element codes, the method proposed by McBride et al. (Ref 22) is recommended.

Example Simulation of Macrosegregation The example is a simulation of macrosegregation in a vacuum arc remelted (VAR) ingot of Ti-6Al-4V. Convection in the melt pool is driven by both buoyancy and Lorentz forces. The latter are particularly important in titanium ingot processing. An “in-house” research code, MULTIA, was used to do the macrosegregation calculations, with the Lorentz forces captured by a separate code, called BAR. BAR is a proprietary code, restricted to member companies of the Specialty Metals Processing Consortium. The Lorentz forces were extracted from BAR and included as body forces in order to capture the interaction of the flows between the melt pool and the mushy zone. The simulation output comprises the temperature field and the position of the mushy zone; the concentrations of aluminum, vanodium, oxygen, and iron and their variations across the thickness of a slabshaped ingot; and velocities in the melt pool and the mushy zone. In the VAR process, consumable electrodes are used to make cylindrical ingots. Hence, in the simulation, new layers of elements are added at the top of the ingot to account for the mass, which enters the liquid pool from the electrode. The simulation, however, is for a slab-shaped ingot with two- dimensional Cartesian coordinates. Even so, the calculation is very time-consuming. The fully solidifed domain is 43 cm (17 in.) wide (half-width of the slab) and 165 cm (65 in.) high. In addition to Lorentz forces, heattransfer coefficients on each boundary were deduced from BAR and incorporated into the thermal boundary conditions. The simulation starts with a small amount of liquid alloy of the nominal composition at a uniform temperature of 1950 K (17 K > the liquidus; 3050 and 30  F above the liquidus). Figure 2 shows calculated velocities in the liquid pool at 8700 s, which is qualitatively the same as the BAR result. The maximum velocity is 40 cm/s (1.3 ft/s) and greater than the BAR result (17 cm/s, or 0.6 ft/s). In this simulation, the viscosity of the liquid equals the molecular viscosity, whereas BAR employs an eddy viscosity. Figure 3 is the simulated macrosegregation of oxygen at 8700 s. Notice that there is a band of positive segregation of oxygen near and at the midwidth of the domain. The maximum concentration of oxygen is approximately 83% greater than its nominal composition of 0.183 wt% O. Near the right side of the simulated ingot, there is a vertical band depleted in oxygen. Oxygen has a partition ratio greater than 1, whereas iron has a partition ratio less than 1. The bands therefore are enriched in iron and depleted in oxygen. The bands appeared to be weak freckles, but closer

examination and simulations of the velocity and the density field failed to reveal the characteristic freckling mechanism, in which less dense liquid in the upper part of the mushy zone flows vertically upward from a channel and into the liquid pool. When simulating macrosegregation, the sensitivity of finite element mesh spacing on simulations should be studied. A systematic sensitivity analysis of mesh spacing on simulating freckles during directional solidification (DS) (Ref 23) was consulted. Various mesh spacings applied to a DS domain of the Ti-6Al-4V alloy with the goal of finding an optimal mesh for relevant thermal conditions revealed no freckles and very little macrosegregation in the DS scenario. Hence, freckling in the VAR ingot must be discounted. Furthermore, a refined mesh used to simulate macrosegregation in the domain of the VAR slab ingot produces segregation patterns that are qualitatively similar to Fig. 3. The simulation with the finer elements shows somewhat less segregation, and the segregated bands are a little wider. The simulations were also carried out to complete solidification. Because of the Lorentz forces, the melt pool sweeps upward along the

Fig. 2

Simulated velocities in the liquid pool and outline of the mushy zone in the vacuum arc remelted slab ingot after 8700 s. Dimensions are in centimeters.

448 / Modeling and Analysis of Casting Processes ACKNOWLEDGMENTS The authors are grateful to their colleague, S.D. Felicelli of Mississippi State University, who is the primary author of MULTIA, and thank P. Sung, formerly of The University of Arizona and now at the Howmet Research Center in Whitehall, Michigan, for performing the simulations of the VAR ingot. REFERENCES

Fig. 3

Simulated macrosegregation of oxygen after 8700 s in the vacuum arc remelted slab ingot. Dimensions are in centimeters.

leading part of the mushy zone and carries oxygen-depleted melt toward the upper right corner, where there are weak secondary convection cells. Hence, a depletion of oxygen in the right side of the ingot results. The segregation band near the right side is related to the fact that the melt pool is almost stagnant where the vertical oxygen-depleted band intersects the leading part of the mushy zone. As yet, simulations of macrosegregation in full-sized ingots are very expensive in terms of computational time. The results presented here were achieved using an in-house research code, called MULTIA, which was originally designed to simulate the formation of macrosegregation (including freckles) in DS multicomponent alloys.

1. C. Beckermann, Macrosegregation, Casting, Vol 15, ASM Handbook, ASM International, 2008 2. H.D. Brody and M.C. Flemings, Solute Redistribution in Dendritic Solidification, Trans. AIME, Vol 236, 1966, p 615–624 3. M.C. Flemings, Solidification Processing, McGraw-Hill, 1974, p 31–36, 141–146, 160–164 4. W.J. Boettinger, U.R. Kattner, S.R. Coriell, Y.A. Chang, and B.A. Mueller, Development of Multicomponent Solidification Micromodels Using a Thermodynamic Phase Diagram Data Base, Modelling of Casting, Welding and Advanced Solidification Processes VII, M. Cross and J. Campbell, Ed., The Minerals, Metals and Materials Society, 1995, p 649–656 5. M.C. Schneider and C. Beckermann, Simulation of Micro-/Macrosegregation during Solidification of a Low-Alloy Steel, ISIJ Int., Vol 35, 1995, p 665–672 6. T.W. Clyne and W. Kurz, Solute Redistribution during Solidification with Rapid Solid-State Diffusion, Metall. Trans. A, Vol 12, 1981, p 965–971 7. S. Kobayashi, Solute Redistribution during Solidification with Diffusion in Solid Phase: A Theoretical Analysis, J. Cryst. Growth, Vol 88, 1988, p 87–96 8. K.S. Yeum, V. Laxmanan, and D.R. Poirier, Efficient Estimation of Diffusion during Solidification, Metall. Trans. A, Vol 20, 1989, p 2847–2856 9. A.D. Patel and Y.V. Murty, Effect of Cooling Rate on Microstructural Development in Alloy 718, Superalloys 718, 625, 706 and Various Derivatives, E.A. Loria, Ed., The Minerals, Metals and Materials Society, 2001, p 123–132 10. R.M. Ward, X. Xu, P.D. Lee, M. McLean, and M.H. Jacobs, Process/Structure Relationships during VAR of Inconel 718, Proceedings of the 2001 International Symposium on Liquid Metal Processing and Casting, A. Mitchell and J. Van Den Avyle, Ed., Sept 23–26, 2001 (Santa Fe, NM), American Vacuum Society, p 187–199 11. P.K. Sung and D.R. Poirier, Liquid-Solid Partition Ratios in Multicomponent Nickel Alloys, Metall. Mater. Trans. A, Vol 30, 1999, p 2173–2181

12. S. Ganesan and D.R. Poirier, Conservation of Mass and Momentum for the Flow of Interdendritic Liquid during Solidification, Metall. Trans. B, Vol 21, 1990, p 173–181 13. W.D. Bennon and F.P. Incropera, Continuum Model for Momentum, Heat and Species Transport in Binary Solid-Liquid Phase Change System, Part I: Model Formulation, Int. J. Heat Mass Transfer, Vol 30, 1987, p 2161–2170 14. C. Beckermann and R. Viskanta, DoubleDiffusive Convection during Dendritic Solidification of a Binary Mixture, Physichem. Hydrodyn., Vol 10, 1988, p 195–213 15. D.R. Poirier, J.C. Heinrich, and S.D. Felicelli, Simulation of Transport Phenomena in Directionally Solidified Castings, Proceedings of the Julian Szekely Memorial Symposium on Materials Processing, H.Y. Sohn, J.W. Evans, and D. Apelian, Ed., TMS, 1997, p 393–410 16. S.D. Felicelli, J.C. Heinrich, and D.R. Poirier, Finite Element Analysis of Directional Solidification of Multicomponent Alloys, Int. J. Num. Meth. Fluids, Vol 27, 1998, p 207–227 17. D.R. Poirier and P. Sung, “Simulating Macrosegregation in VAR Ingots of Titanium Alloy (Ti-6Al-4V) during Solidification,” final technical report, Contract F33615-020105219, Air Force Research Laboratory, WPAFB, Ohio, June 1, 2006 18. D.R. Poirier, J.C. Heinrich, R.G. Erdmann, and M.S. Bhat, Permeability for Flow Through Equiaxed-Dendritic Alloys, Proceedings of the Merton C. Flemings Symposium on Solidification and Materials Processing, R. Abbaschian, H. Brody, and A. Mortensen, Ed., TMS, 2001, p 287–296 19. J.C. Heinrich and D.R. Poirier, Convection Modeling in Directional Solidification, C. R. Mecanique, Vol 332, 2004, p 429–445 20. S.D. Felicelli, D.R. Poirier, and P.K. Sung, A Model for Prediction of Pressure and Redistribution of Gas Forming Elements in Multicomponent Casting Alloys, Metall. Mater. Trans. B, Vol 31, 2000, p 1283– 1292 21. D.R. Poirier, P.J. Nandapurkar, and S. Ganesan, The Energy and Solute Conservation Equations for Dendritic Solidification, Metall. Trans. B, Vol 22, 1991, p 889–900 22. E. McBride, J.C. Heinrich, and D.R. Poirier, Numerical Simulation of Incompressible Flow Driven by Density Variations during Phase Change, Int. J. Num. Meth. Fluids, Vol 31, 1999, p 787–800 23. P.K. Sung, D.R. Poirier, and S.D. Felicelli, Sensitivity of Mesh Spacing on Simulating Macrosegregation during Directional Solidification of a Superalloy, Int. J. Num. Meth. Fluids, Vol 35, 2001, p 357–370

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 449-461 DOI: 10.1361/asmhba0005238

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Modeling of Stress, Distortion, and Hot Tearing Brian G. Thomas, University of Illinois (UIUC) Michel Bellet, Centre de Mise en Forme des Mate´riaux (CEMEF)

COMPUTATIONAL MODELING of mechanical behavior during solidification is becoming more important. Thermal and microstructural simulations alone are insufficient to predict the quality of the final product that is desired by the casting industry. Accurate calculation of displacements, strains, and stresses during the casting process is needed to predict residual stress and distortion and defects such as the formation of cracks such as hot tears. It also helps predict porosity and segregation. As computing power and software tools for computational mechanics advance, it is becoming increasingly possible to perform useful mechanical analysis of castings and these important related behaviors. The thermomechanical analysis of castings presents a formidable challenge for many reasons:

on the interface between CAD design and the mechanical solvers and on computational resources.
The important length scales range from micrometers (dendrite arm shapes) to tens of meters (metallurgical length of a continuous caster), with similarly huge order-ofmagnitude range in time scales.


Many interacting physical phenomena are

Governing Equations













involved in stress-strain formation. Stress arises primarily from the mismatch of strains caused by large temperature gradients and depends on the time- and microstructuredependent inelastic flow of the material. Predicting distortions and residual stresses in cast products requires calculation of the history of the cast product and its environment over huge temperature intervals. This makes the mechanical problem highly nonlinear, involving liquid/solid interaction and complex constitutive equations. Even identifying the numerous metallurgical parameters involved in those relations is a daunting task. The coupling between the thermal and the mechanical problems is an additional difficulty. This coupling comes from the mechanical interaction between the casting and the mold components, through gap formation or the buildup of contact pressure, locally modifying the heat exchange. This adds some complexity to the nonlinear heat transfer resolution. Accounting for the mold and its interaction with the casting makes the problem multidomain, usually involving numerous deformable components with coupled interactions. Cast parts usually have very complex threedimensional shapes, which puts great demands

This article summarizes some of the issues and approaches in performing computational analyses of mechanical behavior, distortion, and hot tearing during solidification. The governing equations are presented first, followed by a brief description of the methods used to solve them, and a few examples of recent applications in shape castings and continuous casting.

«¼

The modeling of mechanical behavior requires solution of 1) the equilibrium or momentum equations relating force and stress, 2) the constitutive equations relating stress and strain, and 3) compatibility equations relating strain and displacement. This is because the boundary conditions specify either force or displacement at different boundary regions of the domain O: _

u ¼ u on @
u _ s n ¼ T on @
T

ðEq 1Þ

where _u are prescribed displacements on bound_ ary surface portion @Ou, and T are boundary surface forces or “tractions” on portion @OT. The next sections first present the equilibrium and compatibility equations and then introduce constitutive equations for the different material states during solidification. Equilibrium and Compatibility Equations. At any time and location in the solidifying material, the conservation of force (steady-state equilibrium) or momentum (transient conditions) can be expressed by: r  s þ rg  r

dv ¼0 dt

where s is the stress tensor, r is the density, g denotes gravity, v is the velocity field, and d/dt denotes the total (particular) time derivation. Stress can be further split into the deviatoric stress tensor and the pressure field. The different approaches for simplifying and solving these equations are discussed in the section “Implementation Issues.” Once the material has solidified, the internal and gravity forces dominate, so the inertia terms in Eq 2 can be neglected. Furthermore, the strains that dominate thermomechanical behavior during solidification are on the order of only a few percent, otherwise cracks will form. With small gradients of spatial displacement, ru = @u/@x, and the compatibility equations simplify to (Ref 1):

(Eq 2)

 1
ru þ ðruÞT 2

(Eq 3)

where « is the strain tensor and u is the displacement vector. This small-strain assumption simplifies the analysis considerably. The compatibility equations can also be expressed as a rate formulation, where strains become strain rates, and displacements become velocities. This formulation is more convenient for a transient computation with time integration involving fluid flow and/or large deformation. In casting analysis, the cast material may be in the liquid, mushy, or solid state. Each of these states has different constitutive behavior, as discussed in the next three sections. Liquid-State Constitutive Models. Metallic alloys generally behave as Newtonian fluids. Including thermal dilatation effects, the constitutive equation can be expressed as: «_ ¼

1 1 dr s I 2ml 3r dt

(Eq 4)

The strain-rate tensor «_ is split into two components: a mechanical part, which varies linearly with the deviatoric stress tensor s, and a thermal part. In this equation, ml is the dynamic viscosity of the liquid, r is the density, and I is the

450 / Modeling and Analysis of Casting Processes identity tensor. Taking the trace of this expression, tr«_ ¼ r  v, the mass conservation equation is recovered: dr @r þ rr  v ¼ þ r  ðrvÞ ¼ 0 dt @t

(Eq 5)

In casting processes, the liquid flow may be turbulent, even after mold filling. This may occur because of buoyancy forces or forced convection such as in jets coming out of the nozzle outlets in continuous casting processes. The most accurate approach, direct numerical simulation, generally is not feasible for industrial processes, owing to their complex-shaped domains and high turbulence. To compute just the large-scale flow features, turbulence models are used that increase the liquid viscosity according to different models of the small-scale phenomena. These models include the simple “mixing-length” models, the two-equation models such as k-e, and large eddy simulation (LES) models, which have been compared with each other and with measurements of continuous casting (Ref 2–4). Mushy-State Constitutive Models. Metallic alloys in the mushy state are two-phase liquidsolid media. Their mechanical response depends greatly on the local microstructural evolution, which involves several complex physical phenomena. An accurate description of these phenomena is useful for studying hot tearing or macrosegregation. Knowledge of the liquid flow in the mushy zone is necessary to calculate the transport of chemical species (alloying elements) (Ref 5). Knowledge of the deformation of the solid phase is important when it affects liquid flow in the mushy zone by “sponge-effects” (Ref 6). In such cases, two-phase models must be used. Starting from microscopic models describing the intrinsic behavior of the liquid phase and the solid phase, spatial averaging procedures must be developed to express the behavior of the compressible solid continuum and of the liquid phase that flows through it (Ref 7–9). If a detailed description is not really needed, such as in the analysis of residual stresses and distortions, the mushy state can be approximated as a single continuum that behaves as a non-Newtonian (i.e., viscoplastic) fluid, according to Eq 6 to 8. Thus, the liquid phase is not distinguished from the solid phase, and the individual dendrites and grain boundaries are not resolved. «_ ¼ «_ vp þ «_ th

«_ vp ¼

3 ð«_ eq Þ1m s 2K

«_ th ¼ 

1 dr I 3r dt

(Eq 6)

(Eq 7)

(Eq 8)

K is the viscoplastic consistencyp and m the strainffiffiffiffiffiffiffiffiffiffiffiffiffiffi rate sensitivity. Denoting seq ¼ 3/2sij sij the von

Mises

equivalent

stress

scalar,

and

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi _ vp e_ eq ¼ 2/3e_ vp ij e ij the von Mises equivalent

strain-rate scalar, Eq 7 yields the well-known power law: seq ¼ Kð_eeq Þm . Note that the preceding Newtonian liquid model is actually a particular case of this non-Newtonian one: Eq 4 can be derived from Eq 6, 7, and 8 taking m = 1 and K = 3ml. The solidification shrinkage is included in Eq 8: writing r = gs rS + gl rL in the solidification interval (rS, rL densities at the solidus and liquidus temperatures, respectively, gs, gl volume fractions of solid and liquid, respectively), the thermal strain rate is defined as: 1 dr 1 dgs ¼  ðrS  rL Þ dt r dt r rL  rS dgs  rL dt

tr_eth ¼ 

(Eq 9)

Solid-State Constitutive Models. In the solid state, metallic alloys can be modeled either as elastic-plastic or elastic-viscoplastic materials. In the latter class of models, one of the simpler models is expressed as follows, but it should be mentioned that a lot of models of different complexity can be found in the literature (Ref 10, 11). «_ ¼ «_ el þ «_ in þ «_ th «_ el ¼

«_ in ¼

1þn n @ _ þ T_ s_  trðsÞI E E @T @
n  T_ trðsÞI @T E 3 Dseq  s0 E1=m s K 2seq

«_ th ¼ 

1 dr I 3r dt

(Eq 10)   1þn s E ðEq 11Þ

(Eq 12)

(Eq 13)

The strain-rate tensor «_ is split into an elastic component, an inelastic (nonreversible) component, and a thermal component. Equation 11 is the hypoelastic Hooke’s law, where E is Young’s modulus, n the Poisson’s coefficient, and s_ a time derivative of the stress tensor s. Equation 12 gives the relation between the inelastic strain-rate tensor «_ in and the stress deviator, s, in which s0 denotes the scalar static yield stress, below which no inelastic deformation occurs (the expression between brackets is set to 0 when it is negative). In these equations, the temperature dependency of all the involved variables should be considered. The effect of strain hardening may appear in such a model by the increase of both the static yield stress s0 and the plastic consistency K with the accumulated inelastic strain eeq, or with another state variable that is representative of the material structure. The corresponding scalar equation relating stress and inelastic strain-rate von Mises invariants is: seq ¼ s0 þ Kð_eeq Þm

(Eq 14)

Inserting this into Eq 12 simplifies it to: «_ in ¼

3_eeq s; or; in incremental form; 2seq 3deeq d«in ¼ s 2seq

ðEq 15Þ

Although metallic alloys show a significant strain-rate sensitivity at high temperature, they are often modeled in the literature using elastic-plastic models, neglecting this important effect. In this case, Eq 15 still holds, but the flow stress is independent of the strain rate. It may depend on the accumulated plastic strain because of strain hardening. Implementation Issues. One of the major difficulties in the thermomechanical analysis of casting processes is the concurrent presence of liquid, mushy, and solid regions that move with time as solidification progresses. Several different strategies have been developed, according to the process and model objectives:
A first strategy consists in extracting the

solidified regions of the casting domain based on the thermal analysis results. Then, a small-strain thermomechanical analysis is carried out on just this solid subdomain, using a standard solid-state constitutive model. Besides difficulties with the extraction process, especially when the solidified regions have complex unconnected shapes, this method may have numerical problems with the application of the liquid hydrostatic pressure onto the new internal boundary of the solidified region. However, this simple strategy is very convenient for many practical problems, especially when the solidification front is stationary, such as the primary cooling of continuous casting of aluminum (Ref 12) and steel (Ref 13, 14). For transient problems, such as the prediction of residual stress and shape (butt-curl) during startup of the aluminum direct chill continuous casting process, the domain can be extended in time by adding layers (Ref 12).
A second strategy considers the entire casting, including the mushy, and liquid regions. The liquid, mushy, and solid regions are modeled as a continuum by adopting the constitutive equations for the solid phase (see the previous section “Solid-State Constitutive Models”) for all regions by adjusting material parameters such as K, m, E, n, s0, and r according to temperature. For example, liquid can be treated by setting the strains to 0 when the temperature is above the solidus temperature. This ensures that stress development in the liquid phase is suppressed. In the equilibrium equation, Eq 2, acceleration terms are neglected, and a small-strain analysis can be performed. The primary unknowns are the displacements, or displacement increments. This popular approach can be used with structural finite element codes, such as MARC (Ref 15) or ABAQUS (Ref 16), and with

Modeling of Stress, Distortion, and Hot Tearing / 451 commercial solidification codes or specialpurpose software, such as ALSIM (Ref 17)/ ALSPEN, (Ref 18), CASTS, (Ref 19), CON2D (Ref 20, 21), Magmasoft (Ref 22), and Procast (Ref 23, 24). It has been applied successfully to simulate deformation and residual stress in shape castings (Ref 25, 26), direct chill casting of aluminum (Ref 12, 17, 18, 27–29), and continuous casting of steel (Ref 20, 30). Despite its efficiency, this approach may suffer from several drawbacks. First, it cannot properly account for fluid flow and the volumetric shrinkage that affects flow in the liquid pool, fluid feeding into the mushy zone, and primary shrinkage depressions that affect casting shape. In addition, incompressibility of the metal in the liquid state is accounted for by increasing Poisson’s ratio to close to 0.5 which sometimes makes the solution prone to numerical instability (Ref 31, 32).
A third strategy has recently been developed that addresses the above issues. It still simulates the entire casting, but treats the mass and momentum equations of the liquid and mushy regions with a mixed velocity-pressure formulation. The primary unknowns are the velocity (time derivative of displacement) and pressure fields, which make it easier to impose the incompressibility constraint (see next section “Thermomechanical Coupling”). Indeed, the velocity-pressure formulation is also applied to the equilibrium of the solid regions to provide a single continuum framework for the global numerical solution. This strategy has been implemented into codes dedicated to casting analysis such as THERCAST (Ref 30, 33, 34) and VULCAN (Ref 35). If stress prediction is not important so that elastic strains can be ignored, then this formulation simplifies to a standard fluid-flow analysis, which is useful in the prediction of bulging and shape in large-strain processes. Example of Solid-State Constitutive Equations. Material property data are needed for the specific alloy being modeled and in a form suitable for the constitutive equations just discussed. This presents a significant challenge for quantitative mechanical analysis, because measurements are not presented in this form and only rarely supply enough information on the conditions to allow transformation to an alternate form. As an example, the following elastic-viscoplastic constitutive equation was developed for the austenite phase of steel (Ref 36) by fitting constant strain-rate tensile tests (Ref 37, 38) and constant-load creep tests (Ref 39) to the form required in Eq 10 to Eq 13.  1=m e_ eq ¼ f%C seq  s0   4:465  104 exp  T

(Eq 16)

where f%C ¼ 4:655  104 þ 7:14  10ð%CÞ þ 1:2  104 ð%CÞ2 s0 ¼ ð130:5  5:128  103 T Þeeq f2 f2 ¼ 0:6289 þ 1:114  103 T 1=m ¼ 8:132  1:54  103 T with T ½K: eq : 0 ½MPa

This equation, and a similar one for delta ferrite, have been implemented into the finite-element codes CON2D (Ref 20) and THERCAST (Ref 40) and applied to investigate several problems involving mechanical behavior during continuous casting. Elastic modulus is a crucial property that decreases with increasing temperature. It is difficult to measure at the high temperatures important to casting, owing to the susceptibility of the material to creep and thermal strain during a standard tensile test, which results in excessively low values. Higher values are obtained from high strain-rate tests, such as ultrasonic measurements (Ref 41) Elastic modulus measurements in steels near the solidus temperature range from 1 GPa (145 ksi) (Ref 42) to 44 GPa (6400 ksi) (Ref 43) with typical modulus values 10 GPa (1450 ksi) near the solidus (Ref 44–46). The density needed to compute thermal strain in Eq 4, 8, or 13 can be found from a weighted average of the values of the different solid and liquid phases, based on the local phase fractions. For the example of plain lowcarbon steel, the following equations were compiled (Ref 20) based on solid data for ferrite (a), austenite (g), and delta (d) (Ref 47, 48) and liquid (l) measurements (Ref 49). rðkg=m3 Þ ¼ ra fa þ rg fg þ rd fd þ rl fl ra ¼ 7881  0:324Tð CÞ  3  105 T ð CÞ2 rg ¼ rd ¼

100½8106  0:51T ð CÞ

½100  ð%CÞ½1 þ 0:008ð%CÞ3 100½8011  0:47T ð CÞ

½100  ð%CÞ½1 þ 0:013ð%CÞ3 rl ¼ 7100  73ð%CÞ  ½0:8  0:09ð%CÞ ½T ð CÞ  1550

ðEq 17Þ

Specialized experiments to measure mechanical properties for use in computational models will be an important trend for future research in this field.

Thermomechanical Coupling Coupling between the thermal and mechanical analyses arises from several sources. First, regarding the mechanical problem, besides the strain rate due to thermal expansion and solidification shrinkage, the material parameters of the preceding constitutive equations strongly depend on temperature and phase fractions, as shown in the previous section. Second, in the heat-transfer problem, the thermal exchange between the casting and the mold strongly

depends on local conditions such as the contact pressure or the presence of a gap between them (as a result of thermal expansion and solidification shrinkage). This is illustrated in Fig. 1 and discussed in this section. Air Gap Formation: Conductive-Radiative Modeling. In the presence of a gap between the casting and the mold, resulting from their relative deformation, the heat transfer results from concurrent conduction through the gas within the gap and from radiation. The exchanged thermal flux, qgap, can then be written: qgap ¼

kgas sðT 4  T 4 Þ ðTc  Tm Þ þ 1 c 1 m g ec þ em  1

(Eq 18)

with kgas (T) the thermal conductivity of the gas, g the gap thickness, Tc and Tm the local surface temperature of the casting and mold, respectively, ec and em their gray-body emissivities, s the Stefan-Boltzmann constant. It is to be noted that the conductive part of the flux can be written in more detail to take into account the presence of coating layers on the mold surface: conduction through a medium of thickness gcoat, of conductivity kcoat (T). It can be seen that the first term tends to infinity as the gap thickness tends to 0; this expresses a perfect contact condition, Tc and Tm tending toward a unique interface temperature. The reality is somewhat different, showing always nonperfect contact conditions. Therefore, the conductive heat-exchange coefficient hcond = kgas/g should be limited by a finite value h0, corresponding to the “no-gap” situation, and depends on the roughness of the casting surface. Specific examples of these gap heat-transfer laws are provided elsewhere for continuous casting with oil lubrication (Ref 13) and mold flux (Ref 50). Effective Contact: Heat Transfer as a Function of Contact Pressure. With effective contact, the conductive heat flux increases with the contact pressure according to a power law (Ref 51). Still denoting h0 as the heat-exchange coefficient corresponding to no gap and no contact pressure, the interfacial heat flux is: qcontact ¼ ðh0 þ ApB c ÞðTc  Tm Þ

(Eq 19)

with pc the contact pressure, A and B two parameters that depend on the materials, the presence of coating or lubricating agent, the surface roughness, and the temperature. The parameters and possibly the laws governing their evolution need to be determined experimentally.

Numerical Solution The thermomechanical modeling equations just presented must be solved numerically, owing to the complex shape of the casting process domain, and the highly nonlinear material properties. The calculation depends greatly on the numerical resolution of time and space. Although finite-difference approaches are

452 / Modeling and Analysis of Casting Processes

Heat-exchange coefficient h (W/(m2•K))

B h = h0 + Apc

h= h0

kgas g

+

2 )(T + T ) s (T2c + Tm c m

ec

Finite Element Formulation and Numerical Implementation In the framework of the small-strain approach presented previously (see the section “Implementation Issues”), having displacements for primitive unknowns, the weak form of the equilibrium equation, Eq 2, neglecting inertia terms, is written as:




-1

Boundary Conditions: Modeling of Contact Conditions. Multidomain Approaches

Modeling of the local heat transfer coefficient in the gap and effective contact situations

popular for heat-transfer, solidification, and fluid-flow analyses, the finite element formulation is usually preferred for the mechanical analysis, owing to its historical advantages with unstructured meshes and accurate implicit solution of the resulting simultaneous algebraic equations. The latter are discussed below.

ð

1 em

Effective gap

Effective contact

8u

+

Gap width g (m)

Contact pressure pc (Pa)

Fig. 1

1

ð

s : d«ðu ÞdV  ð

@


 rg  u dV ¼ 0

ðEq 20Þ




where T is the external stress vector. The vector test functions u* can be seen as virtual displacements in a statement of virtual work. If the third strategy described in the section “Implementation Issues” is adopted, with velocity and pressure as primary unknown variables, the weak form of the momentum equation (Eq 2) is written as (Ref 52): ð

ð _ ÞdV  pr  v dV s : «ðv

8

> > > 8v > > > > > > > > > > > > > <




> > > > > > > > > > > > > > 8p

> > :

dv þ r  v dV ¼ 0 dt ð
p trð«_ in ÞdV ¼ 0

ð

 @


ð




The first equation contains vector test functions v*, which can be seen as virtual velocities in a statement of virtual power. Unlike Eq 20, the pressure p is a primary variable, and only the deviatoric part of the constitutive equations is involved (to determine the stress deviator s). This is why the second equation is needed, which consists of a weak form of the incompressibility of inelastic deformations. Equations 20 and 21 are spatially discretized using the standard finite-element method, as explained in detail in many references (Ref 52). Combined with time discretization using finite differences, this leads to a set of nonlinear equations to be solved at each time increment. In the context of the displacement strategy, Eq 20, this leads to: RðUÞ ¼ 0

T  u dS




ð T  v dS  rg  v dV




(Eq 21)

(Eq 22)

where R is the vector of the nodal equilibrium residues (number of components: 3  number of nodes, in three dimensions), and U is the vector of nodal incremental displacements (same size). Adopting the velocity-pressure strategy, Eq 21 leads to a set of nonlinear equations: R0 ðV; PÞ ¼ 0

constitutive equations, when they depend on strain rate or strain. When the constitutive equations are highly nonlinear, an implicit algorithm is useful to perform time integration at each Gauss point to provide better estimates of inelastic strain at the local level (Ref 53–55).

(Eq 23)

where R0 is the vector of the nodal residues (number of components: 4  number of nodes, in dimension 3), V is the vector of nodal velocities (size: 3  number of nodes), and P is the vector of nodal pressures (size: number of nodes). The global finite-element systems Eq 22 or Eq 23 are usually solved using a full or modified Newton-Raphson method (Ref 16, 31), which iterates to minimize the norm of the residue vectors R or R0 . Alternatively, explicit methods may be employed at this global level. At the local (finite element) level, an algorithm is also required to integrate the

At the interface between the solidifying material and the mold, a contact condition is required to prevent penetration of the shell into the mold, while allowing shrinkage of the shell away from the mold to create an interfacial gap: 8 < sn  n  0 g 0 : ðs n  nÞg ¼ 0

(Eq 24)

where g is the local interface gap width (positive when air gap exists effectively, as in section 0) and n is the local outward unit normal to the part. Equation 24 can be satisfied with a penalty condition, which consists of applying a normal stress vector T proportional to the penetration depth (if any) via a penalty constant wp: T ¼ s n ¼ p hgin

(Eq 25)

Here again, the brackets denote the positive part; a repulsive stress is applied only if g is negative (penetration). Different methods of local adaptation of the penalty coefficient wp have been developed, including the augmented Lagrangian method (Ref 56). More complex and computationally expensive methods, such as the use of Lagrange multipliers may also be used (Ref 57). The possible tangential friction effects between part and mold can be taken into account by a friction law, such as a Coulomb model for instance. In this case, the previous stress vector has a tangential component, Tt, given by: Tt ¼ mf pc

1 ðv  vmold Þ kv  vmold k

(Eq 26)

where pc ¼ sn ¼ sn  n is the contact pressure, and mf the friction coefficient. The previous approach can be extended to the multidomain context to account for the deformation of mold components. The local stress vectors calculated by Eq 25 can be applied onto the surface of the mold, contributing then to its deformation. For most casting processes, the mechanical interaction between the cast product and the mold is sufficiently slow (i.e., its characteristic time remains significant with respect to the process time) to permit a staggered scheme within each time increment: the mechanical problem is successively solved in the cast product and in the different mold components. A global updating of the different configurations is then performed at the end of the time

Modeling of Stress, Distortion, and Hot Tearing / 453 increment. This simple approach gives access to a prediction of the local air gap size g, or alternatively of the local contact pressure pc, that is used in the expressions of the heat-transfer coefficient, according to Eq 18 and 19 (Ref 58).

Treatment of the Regions in the Solid, Mushy, and Liquid States Solidified Regions: Lagrangian Formulation. In casting processes, the solidified regions generally encounter small deformations. It is thus natural to embed the finite element domain into the material, with each node of the computational grid corresponding with the same solid particle during its displacement. The boundary of the mesh corresponds then to the surface of the casting. This method, called Lagrangian formulation, provides the best accuracy when computing the gap forming between the solidified material and the mold. It is also the more reliable and convenient method for time integration of highly nonlinear constitutive equations, such as elastic-(visco)plastic laws presented in the section “Solid-State Constitutive Models.” Mushy and liquid regions: ALE modeling. When the mushy and liquid regions are modeled in the same domain as the solid (see the discussion in the section “Implementation Issues”), they are often subjected to large displacements and strains arising from solidification shrinkage, buoyancy, or forced convection. Similar difficulties are generated in casting processes such as squeeze casting, where the entire domain is highly deformed. In these cases, a Lagrangian formulation would demand frequent remeshings to avoid mesh degeneracy, which is both computationally costly and detrimental to the accuracy of the modeling. It is then preferable to use a so-called Arbitrary Lagrangian Eulerian formulation (ALE). In a Eulerian formulation, material moves through the computational grid, which remains stationary in the “laboratory” frame of reference. In the ALE formulation, the updating of the mesh is partially independent of the velocity of the material particles to maintain the quality of the computational grid. Several methods can be used, including the popular “barycentering” technique, which keeps each node at the geometrical centroid of a set of its

neighbors. This method involves significant extra complexity to account for the advection of material through the domain, and the state variables such as temperature and inelastic strain must be updated according to the relative velocity between the mesh and the particles. In doing this, some surface constraints must be enforced to ensure mass conservation, expressing that the fluxes of mesh velocity and of fluid particle velocity through the surface of the mesh should remain identical. A review on the ALE method in solidification modeling is available, together with some details on its application (Ref 33).

Thermomechanical Coupling Because of the interdependency of the thermal and mechanical analyses, as presented in the section “Thermomechanical Coupling,” their coupling should be taken into account all throughout the cooling process. In practice, the cooling time is decomposed into time increments, each increment requiring the solution of two problems: the energy conservation and the momentum conservation. With the highly nonlinear elastic-viscoplastic constitutive equations typical of solidifying metals, the incremental steps required for the mechanical analysis to converge are generally much smaller than those for the thermal analysis. Thus, these two analyses are generally performed in succession and only once per time increment. However, in the case of very rapid cooling, these solutions might preferably be performed together (including thermal and mechanical unknowns in a single set of nonlinear equations), or else separately but iteratively until convergence at each time increment, otherwise the time step has to be dramatically reduced.

Model Validation Model validation with both analytical solutions and experiments is a crucial step in any computational analysis, and thermomechanical modeling is no exception. Weiner and Boley (Ref 59) derived an analytical solution for unidirectional solidification of an unconstrained plate with a unique solidification temperature, an elastic perfectly plastic constitutive law and constant properties. The plate is subjected to sudden surface quench from a uniform initial temperature to a constant mold temperature. This benchmark problem is ideal for estimating the discretization errors of computational thermal-stress models, as it can be solved with a simple mesh consisting of one row elements, as shown in Fig. 2. Numerical predictions should match with acceptable precision using the same element type and mesh refinement planned for the real problem. For example, the solidification stress analysis code, CON2D (Ref 20) and the commercial code ABAQUS were applied for typical conditions of steel casting (Ref 21). Figures 3 and 4 compare the temperature and stress profiles in the plate at two times. The temperature profile through the solidifying shell is almost linear. Because the interior cools relative to the fixed surface temperature, its shrinkage generates internal tensile stress, which induces compressive stress at the surface. With no applied external pressure, the average stress through the thickness must naturally equal 0, and stress must decrease to 0 in the liquid. Stresses and strains in both transverse directions are equal for this symmetrical problem. The close agreement demonstrates that both computational models are numerically

y

x z

Molten metal

Fig. 2

One-dimensional slice domain for modeling solidifying plate

Fig. 3

Temperatures through solidifying plate at different times comparing analytical solution and numerical predictions

454 / Modeling and Analysis of Casting Processes consistent and have an acceptable mesh resolution. Comparison with experimental measurements is also required to validate that the modeling assumptions and input data are reasonable.

Example Applications Sand Casting of Braking Disks. The finite element software THERCAST for thermomechanical analysis of solidification (Ref 34) has been used in the automotive industry to predict distortion of gray-iron braking discs cast in sand molds

(Ref 60). Particular attention has been paid to the interaction between the deformation of internal sand cores and the cast parts. This demands a global coupled thermomechanical simulation, as presented previously. Figure 5 illustrates the discretization of the different domains involved in the calculation. The actual cooling scenario has been simulated: cooling in mold for 45 min, shakeout, and air cooling for 15 min. Figure 6 shows the temperature evolution at different points in a horizontal cross section at midheight in the disc, revealing: solidification after 2 min, and solidstate phase change after 20 min. The calculated deformation of the core blades shows thermal

Fig. 4

Transverse (Y and Z ) stress through solidifying plate at different times comparing analytical solution and numerical predictions

Fig. 5

Finite element meshes of the different domains: part, core, and two half molds

buckling caused by the very high temperature, and constraint of their dilatation, as shown in Fig. 7. This deformation causes a difference in thickness between the two braking tracks of the disc. Such a defect needs heavy and costly machining operations to produce quality parts. Instead, process simulation allows the manufacturer to test alternative geometries and process conditions in order to minimize the defect. Similar thermomechanical calculations have been made for plain discs, leading to comparisons with residual stress measurements by means of neutrons and x-ray diffraction (Ref 61). As shown in Fig. 8, calculations are consistent with measurements to within 10 MPa (1.5 ksi). Continuous Casting of Steel Slabs. Thermomechanical simulations are used by steelmakers to analyze stresses and strains both in the mold and in the secondary cooling zone below. One goal is to quantify the bulging of the solidified crust between the supporting rolls that is responsible for the tensile stress state in the mushy core, which in turn induces internal cracks and macrosegregation (Ref 62, 63). Two- and three-dimensional finite element models have been recently developed, for the entire length of the caster using THERCAST, as described elsewhere (Ref 40, 64). The constitutive models were presented in the section “Governing Equations.” Contact with supporting rolls is simulated with the penalty formulation discussed in the section “Boundary Conditions: Modeling of Contact Conditions. Multidomain Approaches” adapting penalty coefficients for the different rolls continuously to control numerical penetration of the strand. Figure 9 shows results for a vertical-curved machine (strand thickness 0.22 m, or 0.75 ft, casting speed 0.9 m/min, or 3 ft/min, material Fe-0.06wt%C) at around 11 m (36 ft) below the meniscus. The pressure distribution reveals a double alternation of compressive and

Modeling of Stress, Distortion, and Hot Tearing / 455

Fig. 6

Temperature evolution in the part at different points located in the indicated section

Fig. 7

Deformation of core blades in a radial section, after a few seconds of cooling. On the left, displacements are magnified (100). The temperature distribution is superimposed. On the right, the difference in thickness between the two braking tracks is shown.

70 MPa (10 ksi)

-30 MPa (-4 ksi)

60 ⫾ 8 MPa -40 ⫾ 20 MPa (9 ⫾ 1 ksi) (-6 ⫾ 3 ksi)

Fig. 8

Calculated sqq

Measured sqq

-10 MPa (-1.5 ksi)

Calculated srr

20 ⫾ 22 MPa -25 ⫾ 20 MPa (3 ⫾ 3 ksi) (-4 ⫾ 3 ksi)

Measured srr

10 MPa (1.5 ksi)

Residual hoop stresses (left) and radial stresses (right) in a radial section on as-cast plain discs made of gray iron. Top line, calculated values; bottom line, measured values

456 / Modeling and Analysis of Casting Processes

Fig. 9

Predictions in the middle of the secondary cooling zone, about 11 m (36 ft) below the meniscus. The finite element mesh, (top left) features a fine band of 20 mm (0.8 in.). The pressure distribution (right) reveals alternating stress, including tension near the solidification front (the mushy zone is materialized by 20 lines separated by an interval Dgl = 0.05).

depressive zones. First, the strand surface is in a compressive state under the rolls where the pressure reaches its maximum, 36 MPa (5 ksi). Conversely, it is in a depressive (tensile) state between rolls, where the pressure is minimum (9 MPa, or 1 ksi). Near the solidification front (i.e., close to the solidus isotherm), the stress alternates between tension (negative pressure of about 2 MPa, or 0.3 ksi) beneath the rolls and compression in between (2–3 MPa, or 0.3–0.4 ksi). These results agree with previous structural analyses of the deformation of the solidified shell between rolls, such as those carried out in static conditions by Wu¨nnenberg and Huchingen (Ref 65), Miyazawa and Schwerdtfeger (Ref 62), or by Kajitani et al. (Ref 66) on small slab sections moving downstream between rolls and submitted to the metallurgical pressure onto the solidification front. The influence of process parameters on the thermomechanical state of the strand can then be studied using such numerical models. An example is given in Fig. 10, presenting the sensitivity of bulging to the casting speed. It can

also be seen that bulging predictions are sensitive to the roll pitch, a larger pitch between two sets of rolls inducing an increased bulging. These numerical simulations can then be used to study possible modifications in the design of continuous casters, such as the replacement of large rolls by smaller ones to reduce the pitch and the associated bulging (Ref 67).

Hot-Tearing Analysis Hot-tear crack formation is one of the most important consequences of stress during solidification. Hot tearing is caused by a combination of tensile stress and metallurgical embrittlement. It occurs at temperatures near the solidus when strain concentrates within the interdendritic liquid films, causing separation of the dendrites and intergranular cracks at very small strains (on the order of 1%). This complex phenomenon depends on the ability of liquid to flow through the dendritic structure to feed the volumetric shrinkage, the strength of the surrounding

dendritic skeleton, the grain size and shape, the nucleation of supersaturated gas into pores or crack surfaces, the segregation of solute impurities, and the formation of interfering solid precipitates. The subsequent refilling of hot tears with segregated liquid alloy can cause internal defects that are just as serious as exposed surface cracks. The hot tearing of aluminum alloys is reviewed elsewhere (Ref 68). Hot-tearing phenomena are too complex, too small-scale, and insufficiently understood to model in detail, so several different criteria have been developed to predict hot tears from the results of a thermomechanical analysis. Thermal Analysis Criteria. Casting conditions that produce faster solidification and alloys with wider freezing ranges are more prone to hot tears. Thus, many criteria are solely based on thermal analysis. Clyne and Davies simply compare the local time spent between two critical solid fractions gs1 and gs2 (typically 0.9 and 0.99, respectively), with the total local solidification time (or a reference solidification time) (Ref 69). The “hot-cracking susceptibility” is defined as:

Modeling of Stress, Distortion, and Hot Tearing / 457

Fig. 10

Slab bulging calculated at two different casting speeds: 0.9 and 1.2 m/min (3 and 4 ft/min). The slab bulging increases with the casting speed. Source: Ref 67

HCSClyne ¼

t0:99  t0:90 t0:90  t0:40

(Eq 27)

Classical Mechanics Criteria. Criteria based on classical mechanics often assume cracks will form when a critical stress is exceeded, and they are popular for predicting cracks at lower temperatures (Ref 70–73). This critical stress depends greatly on the local temperature and strain rate. Its accuracy relies on measurements, such as the submerged split-chill tensile test for hot tearing (Ref 74–76). Measurements often correlate hot-tear formation with the accumulation of a critical level of mechanical strain while applying tensile loading within a critical solid fraction where liquid feeding is difficult. This has formed the basis for many hot-tearing criteria. That of Yamanaka et al. (Ref 77) accumulates inelastic deformation over a brittleness temperature range, which is defined, for example as gs 2 ½0:85; 0:99 for a Fe-0.15wt%C steel grade. The local condition for fracture initiation is then: Xgs2 gs1

ein ecr

(Eq 28)

in which the critical strain ecr is 1.6% at a typical strain rate of 3  104 s1. Careful measurements during bending of solidifying steel ingots have revealed critical strains ranging from 1 to 3.8% (Ref 77, 78). The lowest values were found at high strain rate and in crack-sensitive grades (e.g., high-sulfur peritectic steel) (Ref 77). In aluminum-rich Al-Cu alloys, critical strains were reported from 0.09 to 1.6% and were relatively independent of strain rate (Ref 79). Tensile stress is also a requirement

for hot-tear formation (Ref 77). The maximum tensile stress occurs just before formation of a critical flaw (Ref 79). The critical strain decreases with increasing strain rate, presumably because less time is available for liquid feeding, and also decreases for alloys with wider freezing ranges. Won et al. (Ref 80) suggested the following empirical equation for the critical strain in steel, based on fitting measurements from many bend tests: ecr ¼

0:02821 e_ 0:3131 TB0:8638

(Eq 29)

where e_ is the strain rate and DTB is the brittle temperature range, defined between the temperatures corresponding to solid fractions of 0.9 and 0.99. Mechanistically Based Criteria. More mechanistically based hot-tearing criteria include more of the local physical phenomena that give rise to hot tears. Feurer (Ref 81) and more recently Rappaz et al. (Ref 82) have proposed that hot tears form when the local interdendritic liquid feeding rate is not sufficient to balance the rate of tensile strain increase across the mushy zone. The criterion of Rappaz et al. predicts fracture when the strain rate exceeds a limit value that allows pore cavitation to separate the residual liquid film between the dendrites: 1 22 krT k rL ðpm  pC Þ R 180ml rS

r  rL H vT S rS

e_

ðEq 30Þ

in which ml is the dynamic viscosity of the liquid phase, l2 is the secondary dendrite arm

Fig. 11

Model domain

spacing, pm is the local pressure in the liquid ahead of the mushy zone, pC is the cavitation pressure, and vT is the velocity of the solidification front. The quantities R and H depend on the solidification path of the alloy: R¼

ð T1 T2

ð T1

g2s F ðT Þ dT gl 3

g2s dT 2 T2 gl ðT 1 F ðT Þ ¼ gs dT krT k T2 H¼

ðEq 31Þ

where the integration limits are calibration parameters that also have physical meaning (Ref 83). The upper limit T1 may be the liquidus or the coherency temperature, while the lower limit T2 typically is within the solid fraction range of 0.95 to 0.99 (Ref 84). Case Study: Billet Casting Speed Optimization. A Lagrangian model of temperature, distortion, strain, stress, hot tearing has been applied to predict the maximum speed for continuous casting of steel billets without forming off-corner internal cracks. The two-dimensional transient finite-element thermomechanical model, CON2D (Ref 20, 21), has been used to track a transverse slice through the solidifying steel strand as it moves downward at the casting speed to reveal the entire three-dimensional stress state. The two-dimensional assumption produces reasonable temperature predictions because axial (z-direction) conduction is negligible relative to axial advection (Ref 50). In-plane mechanical predictions are also reasonable because bulging effects are small and the undiscretized casting direction is modeled with the appropriate condition of

458 / Modeling and Analysis of Casting Processes

 qðMW=m2 Þ ¼

5  0:2444tðsÞ t  1:0 s (Eq 32) 4:7556tðsÞ0:504 t > 1:0 s

Sample results are presented here for onequarter of a 120 mm2 (0.2 in.2) billet cast at speeds of 2.0 and 5.0 m/min (6.5 to 16.5 ft/s). The latter is the critical speed at which hot-tear crack failure of the shell is just predicted to occur. The temperature and axial (z) stress distributions in a typical section through the wide face of the steel shell cast at 2.0 m/min (6.5 ft/s) are shown in Fig. 12 and 13 at four different times during cooling in the mold. Unlike the analytical solution in Fig. 3, the surface temperature drops as time progresses. The corresponding stress distributions are qualitatively similar to the analytical solution (Fig. 4). The stresses increase with time, however, as solidification progresses. The realistic constitutive equations produce a large region of tension near the solidification front. The magnitude of these stresses (as well as the corresponding strains) is not predicted to be enough to cause hot tearing in the mold, however. The results from two different codes reasonably match, demonstrating that the formulations are accurately implemented, convergence has been achieved, and the mesh and time-step refinement are sufficient.

Distance to chilled surface, in. 0.2 0.4

Distance to chilled surface, in. 0.2

1550

0.4

0.6 2800

0.6

4 0.5

1500 2

1450

2600 0

1350

−2

Abaqus 1 s

1250

CON2D 1 s Abaqus 5 s

1200

2200

CON2D 5 s 1150

Abaqus 10 s

1100

CON2D 10 s Abaqus 21 s

1050 1000

0

−0.5 Abaqus 1 s

−4

CON2D 1 s −6

Abaqus 5 s

−1

Stress, ksi

2400

1300

Stress, MPa

1400 Temperature, ⬚F

Temperature, ⬚C

generalized plain strain. Other applications with this model include the prediction of ideal taper of the mold walls (Ref 85) and quantifying the effect of steel grade on oscillation mark severity during level fluctuations (Ref 86). The model domain is an L-shaped region of a two-dimensional transverse section, shown in Fig. 11. Removing the central liquid region saves computation and lessens stability problems related to element “locking.” Physically, this “trick” is important in two-dimensional domains because it allows the liquid volume to change without generating stress, which mimics the effect of fluid flow into and out of the domain that occurs in the actual opentopped casting process. Simulations start at the meniscus, 100 mm (4 in.) below the mold top, and extend through the 800 mm (32 in.) long mold and below, for a caster with no submold support. The instantaneous heat flux, given in Eq 32, was based on plant measurements (Ref 45). It was assumed to be uniform around the perimeter of the billet surface in order to simulate ideal taper and perfect contact between the shell and mold. Below the mold, the billet surface temperature was kept constant at its circumferential profile at mold exit. This eliminates the effect of spray cooling practice imperfections on submold reheating or cooling and the associated complication for the stress/ strain development. A typical plain carbon steel was studied (0.27% C, 1.52% Mn, 0.34% Si) with 1500.7 C (2733 F) liquidus temperature, and 1411.8 C (2573 F) solidus temperature. Constitutive equation and properties are given in the sections “Solid-State Constitutive Models” and “Example of Solid-State Constitutive Equations.”

CON2D 5 s

−8

Abaqus 10 s

2000

CON2D 21 s

−10

CON2D 10 s

−1.5

Abaqus 21 s

−12

CON2D 21 s

0

Fig. 12

1

2

3 4 5 6 7 8 9 10 11 12 13 14 15 Distance to chilled surface, mm

Temperature distribution along the solidifying slice in continuous casting mold

Figure 14(a) shows the distorted temperature contours near the strand corner at 200 mm (8 in.) below the mold exit, for a casting speed of 5.0 m/min (16.5 ft/min). The corner region is coldest, owing to two-dimensional cooling. The shell becomes hotter and thinner with increasing casting speed, owing to less time in the mold. This weakens the shell, allowing it to bulge more under the ferrostatic pressure below the mold. Figure 14(b) shows contours of “hoop” stress constructed by taking the stress component acting perpendicular to the dendrite growth direction, which simplifies to sx in the lower right portion of the domain and sy in the upper left portion. High values appear at the off-corner subsurface region, due to a hinging effect that the ferrostatic pressure over the entire face exerts around the corner. This bends the shell around the corner and generates high subsurface tensile stress at the weak solidification front in the off-corner subsurface location. This tensile stress peak increases slightly and moves toward the surface at higher casting speed. Stress concentration is less, and the surface hoop stress is compressive at the lower casting speed. This indicates no possibility of surface cracking. However, tensile surface hoop stress is generated below the mold at high speed in Fig. 14(b) at the face center due to excessive bulging. This tensile stress, and the accompanying hot-tear strain, might contribute to longitudinal cracks that penetrate the surface. Hot tearing was predicted using the criterion in Eq 28 with the critical strain given in Eq 29. Inelastic strain was accumulated for the component oriented normal to the dendrite growth direction, because that is the weakest direction and corresponds to the measurements used to obtain Eq 29. Figure 14(c) shows contours of hot-tear strain in the hoop direction. The highest values appear at the off-corner subsurface region in the hoop direction. Moreover, significantly higher values are found at higher casting speeds. For this particular example, hot-tear strain exceeds the threshold at 12 nodes, all located near the off-corner subsurface region.

−14

−2 0

1

Fig. 13

2

3

4 5 6 7 8 9 10 11 12 13 14 15 Distance to chilled surface, mm

Lateral (y and z) stress distribution along the solidifying slice in continuous casting mold

This is caused by the hinging mechanism around the corner. No nodes fail at the center surface, in spite of the high tensile stress there. The predicted hot-tearing region matches the location of off-corner longitudinal cracks observed in sections through real solidifying shells, such as the one pictured in Fig. 15. The bulged shape is also similar. Results from many computations were used to find the critical speed to avoid hot-tear cracks as a function of section size and working mold length, presented in Fig. 16 (Ref 46). These predictions slightly exceed plant practice, which is generally chosen by empirical trial and error (Ref 87). This suggests that plant conditions such as mold taper are less than ideal, that other factors limit casting speed, or those speeds in practice could be increased. The qualitative trends are the same. This quantitative model of hot tearing provides many useful insights into the continuous casting process. Larger section sizes are more susceptible to bending around the corner and so have a lower critical casting speed, resulting in less productivity increase than expected. The trend toward longer molds over the past three decades enables a higher casting speed without cracks by producing a thicker, stronger shell at mold exit.

Conclusions Mechanical analysis of casting processes is growing in sophistication, accuracy, and phenomena incorporated. Quantitative predictions of temperature, deformation, strain, stress, and hot tearing in real casting processes are becoming possible. Computations are still hampered by the computational speed and limits of mesh resolution, especially for realistic three-dimensional geometries and defect analysis. ACKNOWLEDGMENT The authors wish to thank the Continuous Casting Consortium and the National Center

Modeling of Stress, Distortion, and Hot Tearing / 459

Fig. 14

Distorted contours at 200 mm (8 in.) below mold exit. (a) Temperature. (b) Hoop stress. (c) Hot-tear strain Section size, in.2 0.15 6

0.20

0.25

0.30

0.35

0.40 20

15 4

3

10

2 CON2D

0 100

Fig. 15

Off-corner internal crack in breakout shell from a 175 mm (7 in.) square bloom

for Supercomputing Applications at the University of Illinois, the French Ministry of Industry, the French Technical Center of Casting Industries (CTIF), and the companies ArcelorMittal, Ascometal, Aubert et Duval, Industeel, and PSA Peugeot-Citroe¨n, for support of this work.

1. G.E. Mase and G.T. Mase, Continuum Mechanics for Engineers, 2nd ed., CRC Press, 1999 2. B.G. Thomas, Q. Yuan, S. Sivaramakrishnan, T. Shi, S.P. Vanka, and M.B. Assar, Comparison of Four Methods to Evaluate Fluid Velocities in a Continuous Casting Mold, ISIJ Int., Vol 41 (No. 10), 2001, p 1266–1276 3. Q. Yuan, B. Zhao, S.P. Vanka, and B.G. Thomas, Study of Computational Issues in Simulation of Transient Flow in Continuous Casting, Steel Res. Int., Vol 76 (No. 1, Special Issue: Simulation of Fluid Flow in Metallurgy), 2005, p 33–43 4. Q. Yuan, S. Sivaramakrishnan, S.P. Vanka, and B.G. Thomas, Computational and

120

140

160

5

Plant

10.

11.

2000 data 1990 data 1970 data 180

200

220

240

260

Section size, mm2

12.

Fig. 16

Comparison of critical casting speeds, based on hot-tearing criterion (Ref 46), and typical plant practice (Ref 87)

5.

REFERENCES

Working mold length 1000 mm (40 in.) 700 mm (28 in.) 500 mm (20 in.)

1

Critical casting speed, ft/min

Critical casting speed, m/min

5

6.

7.

8.

9.

Experimental Study of Turbulent Flow in a 0.4-Scale Water Model of a Continuous Steel Caster, Metall. Mater. Trans., Vol 35B (No. 5), 2004, p 967–982 C. Beckermann, Macrosegregation, in Encyclopedia of Materials: Science and Technology, Elsevier, 2001, p 4733–4739 G. Lesoult, C.-A. Gandin, and N.T. Niane, Segregation During Solidification with Spongy Deformation of the Mushy Zone, Acta Mater., Vol 51, 2003, p 5263–5283 L.C. Nicolli, A. Mo, and M. M’Hamdi, Modeling of Macrosegregation Caused by Volumetric Deformation in a Coherent Mushy Zone, Metall. Mater. Trans. A, Vol 36, 2005, p 433–442 V.D. Fachinotti, S. Le Corre, N. Triolet, M. Bobadilla, and M. Bellet, Two-Phase Thermo-mechanical and Macrosegregation Modelling of Binary Alloys Solidification with Emphasis on the Secondary Cooling Stage of Steel Slab Continuous Casting, Int. J. Num. Meth. Eng., Vol 67 (No. 10), 2006, p 1341–1384 O. Ludwig, C.-L. Martin, J.-M. Drezet, and M. Sue´ry, Rheological Behaviour of Al-Cu

13.

14.

15. 16. 17.

18.

Alloys During Solidification: Constitutive Modelling, Experimental Identification and Numerical Study, Met. Mater. Trans., Vol 36A, 2005, p 1525–1535 Y. Estrin, A Versatile Unified Constitutive Model Based on Dislocation Density Evolution, in Constitutive Modelling—Theory and Application, Vol MD-vol26/AMDvol121, ASME, 1991, p 65–75 C. Agelet de Saracibar, M. Cervera, and M. Chiumenti, On the Constitutive Modeling of Coupled Thermomechanical PhaseChange Problems, Int. J. Plast., Vol 17, 2001, p 1565–1622 J. Sengupta, S.L. Cockcroft, D.M. Maijer, and A. Larouche, Quantification of Temperature, Stress, and Strain Fields During the Start-up Phase of Direct Chill Casting Process by Using a 3D Fully Coupled Thermal and Stress Model for AA5182 Ingots, Mater. Sci. Eng. A, Vol 397, 2005, p 157–177 J.E. Kelly, K.P. Michalek, T.G. O’Connor, B.G. Thomas, and J.A. Dantzig, Initial Development of Thermal and Stress Fields in Continuously Cast Steel Billets, Metall. Trans. A, Vol 19A (No. 10), 1988, p 2589–2602 A.E. Huespe, A. Cardona, and V. Fachinotti, Thermomechanical Model of a Continuous Casting Process, Comput. Meth. Appl. Mech. Eng., Vol 182 (No. 3), 2000, p 439–455 MARC Users Manual, MARC Analysis Research Corp., Palo Alto, CA, 1991 “ABAQUS Theory Manual v6.0,” Abaqus, Inc., Pawtucket, RI, 2004 A. Hakonsen and D. Mortensen, Modeling of Casting, Welding and Advanced Solidification Processes VII, TMS, Warrendale, PA, 1995, p 963–970 H.G. Fjaer and A. Mo, ALSPEN—A Mathematical Model for Thermal Stresses in DC-Cast Al Billets, Metall. Trans., Vol 21B (No. 6), 1990, p 1049–1061

460 / Modeling and Analysis of Casting Processes 19. G. Laschet, J. Jakumeit, and S. Benke, Thermo-mechanical Analysis of Cast/ Mould Interaction in Casting Processes, Z. Metallkd., Vol 95, 2004, p 1087–1096 20. C. Li and B.G. Thomas, Thermo-Mechanical Finite-Element Model of Shell Behavior in Continuous Casting of Steel, Metall. Mater. Trans. B, Vol. 35B (No. 6), 2004, p 1151– 1172 21. S. Koric and B.G. Thomas, Efficient Thermo-Mechanical Model for Solidification Processes, Int. J. Num. Meth. Eng., Vol 66 (No. 12), 2006, p 1955–1989 22. MAGMASOFT, http://www.magmasoft. com, 2007 23. PROCAST, Procast, http://www.esi-group. com/SimulationSoftware/Die_Casting_Solution/Products/ Casting_simulation/, 2007 24. M. Samonds and J.Z. Zhu, Coupled Thermal-Fluid-Stress Analysis of Castings, Proc. MCWASP IX, Ninth Int. Conf. on Modeling of Casting, Welding and Advanced Solidification Processes, P.R. Sahm, P.N. Hansen, and J.G. Conley, Ed. (Aachen, Germany), Aug 20–25, 2000, Shaker Verlag, Aachen, 2000, p 80–87 25. J.M. Drezet, A. Bughardt, H.G. Fjaer, and B. Magnin, Thermomechanical Effects in D.C. Casting of Aluminum Alloy: A Numerical Benchmark Study, Mater. Sci. Forum, Vol 329–330, 2000, p 493–500 26. H.G. Fjaer and A. Hakonsen, The Mechanism of Pull-in During DC-Casting of Aluminum Sheet Ingots, Light Metals 1997, TMS, Warrendale, PA, 1997, p 683–690 27. A. Mo, M. Rappaz, and L.L. Martin, Aluminum, Vol 78 (No. 10), 2002, p 856–864 28. J.M. Drezet, A. Bughardt, H.G. Fjaer, and B. Magnin, Mater. Sci. Forum, Vol 329– 330, 2000, p 493–500 29. A. Henry, Light Metal Age, Vol 58 (No. 7– 8), 2000, p 66–67 30. M. Bellet, O. Jaouen, and I. Poitrault, An ALE-FEM Approach to the Thermomechanics of Solidification Processes with Application to the Prediction of Pipe Shrinkage, Int. J. Num. Meth. Heat Fluid Flow, Vol 15, 2005, p 120–142 31. O.C. Zienkiewicz and R.L. Taylor, The Finite Element Method, 4th ed., McGraw Hill, 1988 32. J.-M. Drezet, B. Commet, H.G. Fjaer, and B. Magnin, Stress-Strain Computations of the DC Casting Process of Aluminum Alloy: A Sensitivity Study on Material Properties, Proc. MCWASP IX, Ninth Int. Conf. on Modeling of Casting, Welding and Advanced Solidification Processes, P.R. Sahm, P.N. Hansen, and J.G. Conley, Ed., (Aachen, Germany), Aug 20–25, 2000, Shaker Verlag, Aachen, 2000, p 33–40 33. M. Bellet and V.D. Fachinotti, ALE Method for Solidification Modelling, Comput. Meth. Appl. Mech. Eng., Vol 193, 2004, p 4355–4381 34. Thercast, presentation, www.transvalor. com and www.scconsultants.com, 2007

35. M. Chiumenti, M. Cervera, and C.A.d. Saracibar, Coupled Thermomechanical Simulation of Solidification and Cooling Phases in Casting Processes, Proc. MCWASP XI, 11th Int. Conf. on Modeling of Casting, Welding and Advanced Solidification Processes, C.-A. Gandin and M. Bellet, Ed., TMS, Warrendale, PA, 2006, p 201–208 36. P. Kozlowski, B.G. Thomas, J. Azzi, and H. Wang, Simple Constitutive Equations for Steel at High Temperature, Metall. Trans. A, Vol 23A (No. 3), 1992, p 903–918 37. P.J. Wray, Plastic Deformation of DeltaFerritic Iron at Intermediate Strain Rates, Metall. Trans. A, Vol 7A, Nov 1976, p 1621–1627 38. P.J. Wray, Effect of Carbon Content on the Plastic Flow of Plain Carbon Steels at Elevated Temperatures, Metall. Trans. A, Vol 13A (No. 1), 1982, p 125–134 39. T. Suzuki, K.H. Tacke, K. Wunnenberg, and K. Schwerdtfeger, Creep Properties of Steel at Continuous Casting Temperatures, Ironmaking Steelmaking, Vol 15 (No. 2), 1988, p 90–100 40. F. Costes, A. Heinrich, and M. Bellet, 3D Thermomechanical Simulation of the Secondary Cooling Zone of Steel Continuous Casting, Proc. MCWASP X, Tenth Int. Conf. on Modeling of Casting, Welding and Advanced Solidification Processes, D.M. Stefanescu, J.A. Warren, M.R. Jolly, and M.J.M. Krane, Ed., TMS, Warrendale, PA, 2003, p 393–400 41. D.L. Donsbach and M.W. Moyer, Ultrasonic Measurement of Elastic Constants at Temperatures from 20 to 1100 C, Ultrasonic Materials Characterization, Special Pub. 596, H. Berger and M. Linzer, Ed., National Bureau of Standards, 1980 42. O.M. Puhringer, Strand Mechanics for Continuous Slab Casting Plants, Stahl Eisen, Vol 96 (No. 6), 1976, p 279–284 43. D.R. Hub, Measurement of Velocity and Attenuation of Sound in Iron up to the Melting Point, Proc. IVth Int. Vong. Acoustics (Copenhagen), 1962, paper 551 44. H. Mizukami, K. Murakami, and Y. Miyashita, Mechanical Properties of Continuously Cast Steels at High Temperatures, Tetsu-to-Hagane, Vol 63 (No. 146), 1977, p S652 45. C. Li, and B.G. Thomas, Maximum Casting Speed for Continuous Cast Steel Billets Based on Sub-Mold Bulging Computation, Steelmaking Conf. Proc., Vol 85 (Nashville, TN), March 10–13, 2002, ISS, Warrendale, PA, 2002, p 109–130 46. C. Li and B.G. Thomas, Thermo-Mechanical Finite Element Model of Bulging and Hot Tearing During Continuous Casting of Steel Billets, Modeling of Casting, Welding, and Advanced Solidification Processes, Vol X, D. Stefanescu, J. Warren, M. Jolly, and M. Krane, Ed. (San Destin, FL), May 25–30, 2003, TMS, Warrendale, PA, 2003, p 385–392

47. K. Harste, A. Jablonka, and K. Schwerdtfeger, Shrinkage and Formation of Mechanical Stresses During Solidification of Round Steel Strands, Fourth Int. Conf. on Continuous Casting (Centres de Recherches Metallurgiques and Verein Deutscher Eisenhuttenleute), Stahl und Eisen, Brussels, 1988, p 633–644 48. K. Harste, “Investigation of the Shrinkage and the Origin of Mechanical Tension During the Solidification and Successive Cooling of Cylindrical Bars of Fe-C Alloys,” Ph.D. dissertation thesis, Technical University of Clausthal, 1989 49. I. Jimbo and A. Cramb, The Density of Liquid Iron-Carbon Alloys, Metall. Trans. B, Vol 24B, 1993, p 5–10 50. Y. Meng and B.G. Thomas, Heat Transfer and Solidification Model of Continuous Slab Casting: CON1D, Met. Mater. Trans., Vol 34B (No. 5), 2003, p 685–705 51. C.V. Madhusudana and L.S. Fletcher, Contact Heat Transfer—The Last Decade, AIAA J., Vol 24, 1985, p 510–523 52. M. Rappaz, M. Bellet, and M. Deville, Numerical Modeling in Materials Science and Engineering, in Springer Series in Computational Mathematics, Springer Verlag, Berlin, 2003 53. S. Nemat-Nasser, and Y.F. Li, An Explicit Algorithm for Large-Strain, Large-Strain Rate Elastic-Viscoplasticity, Comput. Meth. Appl. Mech. Eng., Vol 48, 1992, p 205–219 54. A.M. Lush, G. Weber, and L. Anand, An Implicit Time-integration Procedure for A Set of Internal Variable Constitutive Equations for Isotropic Elastic-Viscoplasticity, Int. J. Plast., Vol 5, 1989, p 521–549 55. H. Zhu, “Coupled Thermal-Mechanical Finite-Element Model with Application to Initial Solidification,” Thesis, University of Illinois, 1993 56. R. Glowinski and P.L. Tallee, Augmented Lagrangian and Operator-Splitting Methods in Non-linear Mechanics, Studies in Applied Mathematics, SIAM, Vol 9, 1989, p 295 57. P. Wriggers and G. Zavarise, On Contact between Three-Dimensional Beams Undergoing Large Deflections, Commun. Num. Meth. Eng., Vol 13, 1997, p 429–438 58. O. Jaouen and M. Bellet, A Numerical Mechanical Coupling Algorithm for Deformable Bodies: Application to Part/ Mold Interaction in Casting Process, Proc. Eighth Int. Conf. on Modelling of Casting, Welding and Advanced Solidification Processes, B.G. Thomas and C. Beckermann, Ed. (San Diego, CA), June 7–12, 1998, TMS, Warrendale, PA, 1998, p 739–746 59. J.H. Weiner and B.A. Boley, Elasto-plastic Thermal Stresses in a Solidifying Body, J. Mech. Phys. Solids, Vol 11, 1963, p 145– 154 60. M. Bellet, C. Aliaga, and O. Jaouen, Finite Elements for a Thermomechanical

Modeling of Stress, Distortion, and Hot Tearing / 461

61.

62.

63.

64.

65.

66.

67.

68.

69.

Analysis of Solidification Processes, Modeling of Casting, Welding, and Advanced Solidification Processes IX, Shaker Verlag GmbH, Aachen, 2000, p 10–17 S. David and P. Auburtin, Numerical Simulation of Casting Processes. Benefits of Thermomechanical Simulation in Automotive Industry, Conference Mate´riaux 2002, Tours, France, Proc., Vol on CD, Universite´ Technologique de Belfort-Montbe´liard, 2002, p 5 (in French) K. Miyazawa and K. Schwerdtfeger, Macrosegregation in Continuously Cast Steel Slabs: Preliminary Theoretical Investigation on the Effect of Steady State Bulging, Arch. Eisenh€ utten., Vol 52 (No. 11), 1981, p 415–422 G. Lesoult and S. Sella, Analysis and Prevention of Centreline Segregation During Continuous Casting of Steel Related to Deformation of the Solid Phase, Solid State Phenom., Vol 3, 1988, p 167–178 M. Bellet and A. Heinrich, A Two-Dimensional Finite Element Thermomechanical Approach to a Global Stress-Strain Analysis of Steel Continuous Casting, ISIJ Int., Vol 44, 2004, p 1686–1695 K. Wu¨nnenberg and D. Huchingen, Strand Bulging between Supporting Rollers During Continuous Slab Casting, Stahl Eisen, Vol 98 (No. 6), 1978, p 254–259 T. Kajitani, J.-M. Drezet, and M. Rappaz, Numerical Simulation of DeformationInduced Segregation in Continuous Casting of Steel, Metall. Mater. Trans. A, Vol 32, 2001, p 1479–1491 N. Triolet and M. Bobadilla, Mastering Steel Slab Internal Soundness and Surface Quality Issues through Thermomechanical Modelling of Continuous Casting, Proc. MCWASP-XI, 11th Int. Conf. on Modelling of Casting, Welding, and Advanced Solidification Processes, C.A. Gandin and M. Bellet, Ed. (Opio, France), May 28 to June 2, 2006), TMS, Warrendale, PA, 2006, p 753–760 D.G. Eskin, Suyitno, and L. Katgerman, Mechanical Properties in the Semi-Solid State and Hot Tearing of Aluminum Alloys, Prog. Mater. Sci., Vol 49, 2004, p 629–711 T.W. Clyne and G.J. Davies, Comparison between Experimental Data and

70.

71. 72.

73.

74.

75.

76.

77.

78.

Theoretical Predictions Relating to Dependence of Solidification Cracking on Composition, Solidification and Casting of Metals, The Metals Society, London, 1979, p 275–278 K. Kinoshita, T. Emi, and M. Kasai, Thermal Elasto-plastic Stress Analysis of Solidifying Shell in Continuous Casting Mold, Tetsu-to-Hagane, Vol 65 (No. 14), 1979, p 2022–2031 J.O. Kristiansson, Thermal Stresses in the Early Stage of Solidification of Steel, J. Thermal Stresses, Vol 5, 1982, p 315–330 B.G. Thomas, I.V. Samarasekera, and J.K. Brimacombe, Mathematical Model of the Thermal Processing of Steel Ingots, Part II: Stress Model, Metall. Trans. B, Vol 18B (No. 1), 1987, p 131–147 K. Okamura and H. Kawashima, Calculation of Bulging Strain and its Application to Prediction of Internal Cracks in Continuously Cast Slabs, Proc. Int. Conf. Comp. Ass. Mat. Design Proc. Simul., ISIJ, Tokyo, 1993, p 129–134 P. Ackermann, W. Kurz, and W. Heinemann, In Situ Tensile Testing of Solidifying Aluminum and Al-Mg Shells, Mater. Sci. Eng. Vol 75, 1985, p 79–86 C. Bernhard, H. Hiebert, and M.M. Wolf, Simulation of Shell Strength Properties by the SSCT Test, ISIJ Int. (Japan), Vol 36 (Suppl. Science and Technology of Steelmaking), 1996, p S163–S166 M. Suzuki, C. Yu, and T. Emi, In-Situ Measurement of Tensile Strength of Solidifying Steel Shells to Predict Upper Limit of Casting Speed in Continuous Caster with Oscillating Mold, ISIJ Int., Iron Steel Inst. Jpn., Vol 37 (No. 4), 1997, p 375–382 A. Yamanaka, K. Nakajima, K. Yasumoto, H. Kawashima, and K. Nakai, Measurement of Critical Strain for Solidification Cracking, Modelling of Casting, Welding, and Advanced Solidification Processes, Vol V, M. Rappaz, M.R. Ozgu, and K.W. Mahin, Ed., (Davos, Switzerland), TMS, Warrendale, PA, 1990, p 279–284 A. Yamanaka, K. Nakajima, and K. Okamura, Critical Strain for Internal Crack Formation in Continuous Casting, Ironmaking Steelmaking, Vol 22 (No. 6), 1995, p 508–512

79. P. Wisniewski and H.D. Brody, Tensile Behavior of Solidifying Aluminum Alloys, Modelling of Casting, Welding, and Advanced Solidification Processes, Vol V, M. Rappaz, M.R. Ozgu, and K.W. Mahin, Ed., (Davos, Switzerland), TMS, Warrendale, PA, 1990, p 273–278 80. Y.-M. Won, T.J. Yeo, D.J. Seol, and K.H. Oh, A New Criterion for Internal Crack Formation in Continuously Cast Steels, Metall. Mater. Trans. B, Vol 31B, 2000, p 779–794 81. U. Feurer, Mathematisches modell der Warmrissneigung von bina¨ren aluminium legierungen, Giessereiforschung, Vol 28, 1976, p 75–80 82. M. Rappaz, J.-M. Drezet, and M. Gremaud, A New Hot-Tearing Criterion, Metall. Mater. Trans. A, Vol 30A (No. 2), 1999, p 449–455 83. J.M. Drezet and M. Rappaz, Prediction of Hot Tears in DC-Cast Aluminum Billets, Light Metals, J.L. Anjier, Ed., TMS, Warrendale, PA, 2001, p 887–893 84. M. M’Hamdi, S. Benum, D. Mortensen, H. G. Fjaer, and J.M. Drezet, The Importance of Viscoplastic Strain Rate in the Formation of Center Cracks During the Start-up Phase of Direct-Chill Cast Aluminium Extrusion Ingots, Metall. Mater. Trans. A, Vol 34, 2003, p 1941–1952 85. B.G. Thomas and C. Ojeda, Ideal Taper Prediction for Slab Casting, ISSTech Steelmaking Conference (Indianapolis, IN), April 27–30, 2003, Vol 86, 2003, p 396– 308 86. J. Sengupta and B.G. Thomas, Effect of a Sudden Level Fluctuation on Hook Formation During Continuous Casting of UltraLow Carbon Steel Slabs, Modeling of Casting, Welding, and Advanced Solidification Processes XI (MCWASP XI) Conference, C.Z. Gandin, M. Bellet, and J.E. Allison, Ed. (Opio, France, May 28 to June 2, 2006), TMS, Warrendale, PA, 2006, p 727–736 87. E. Howard and D. Lorento, Development of High Speed Casting, 1996 Electric Furnace Conference Proceedings, (Dallas, TX), Dec 9–12, 1996, ISS, Warrendale, PA, 1996

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 462-467 DOI: 10.1361/asmhba0005239

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Practical Issues in Computer Simulation of Casting Processes Itsuo Ohnaka, Osaka Sangyo University, Japan

COMPUTER SIMULATIONS usually follow the procedure shown in Fig. 1. This section describes important elements and points of the simulations, namely, setting clear simulation objectives, selection of proper simulation code, hints in modeling of shape and phenomena, initial and boundary conditions, physical properties, enmeshing, and evaluation of simulation results, providing foundry process engineers some insights into the application of models to real world problems.

Setting Simulation Objectives, Modeling, and Selection of Simulation Code

it is better to check and correct them with a commercial code (such as MagicsRP). Further, it should be noted that often the 3D solid models are of the finished part rather than the cast shape. Actual final part shapes often have details that need to be altered or eliminated to allow for enmeshing and practical computer simulation. Often, the part geometry data are only available for cast products. For casting simulation one needs a description of the mold geometry, which is different from the product geometry, because a proper mold design must consider thermal shrinkage, draft, cutting stock, ingate, riser, and so forth. Therefore, if the mold geometry is not provided, the software must allow

It is necessary to clarify the objectives and simulation conditions, based on a careful understanding of real situations and problems. The objectives include estimating casting defects to set proper gating and rigging, investigating and reducing casting defects, cost reduction, control of microstructure and properties, and so forth. The simulation conditions include adequate time, cost, human and computer resources, and so forth. Based on these objectives and conditions, the casting process concerned is modeled, leading to selection of the proper simulation code. Modeling is used to determine what kind of shape and parameters should be considered, including the necessary accuracy of the simulation. Although users usually cannot afford to use multiple codes, at least the proper function provided by a code should be selected based on the conditions and modeling described previously.

Start

Setting objectives Modeling and selecting a code to use Modeling of shape and size Preprocessor

Input of materials data and initial and boundary conditions Enmeshing

Numerical calculation

Solver

Mold filling Solidification Defects Structure Properties, etc.

Output and display

Modeling of Shape and Phenomena When three-dimensional (3D) digital data are available, the user can use them alone, making standard triangulation language (STL) data. However, because STL data can have defects,

for its creation. Care should be taken to define the initial mold features since the level of effort required by the user to reach a sound process design is directly related to how well the initial design represents an optimized solution. Further, the real mold size differs from its pattern or computer-aided design (CAD) data, because of mold coating, production error, and other factors such as springback in sand molding. Measuring important places of the mold, in particular, thin sections, is recommended. Computing cost can be greatly reduced if some approximation is possible, as demonstrated in Fig. 2. Figure 2(a) and (b) show how onedimensional (1D) or two-dimensional (2D)

Evaluation

End

Fig. 1

Simulation procedure

Postprocessor

Practical Issues in Computer Simulation of Casting Processes / 463 D W1

y

x

x

W2

W1/D, W2/D > 15

D2

y

D1

L

simulation, the metal delivery system prior to the chork (gate, runner, sprue) can be ignored through intelligent selection of pressure and/or inflow velocity, which can be approximated through hand calculations. Further, it is possible to analyze only a part of the mold cavity, for example, the part near the gates, if the boundary condition can be properly set.

x

L/D1, L /D2 > 15 x

y

Cooling

x

r

y

L

Initial and Boundary Conditions and Physical Properties

x

D L /D > 15

Initial Conditions

(b)

(a)

z

y

D r

r

x

r/D >> 1 (c)

(d) z

y

x

r (f)

(e)

Fig. 2

Modeling approximations of shapes (a) 1D approximation. (b) 2D approximation. (c) 2D approximation where curvature radius is much greater than the thickness. (d) Rotation symmetry approximation neglecting small irregularities. (e) Rotation symmetry approximation. (f) Half cut body of a symmetrical body. Source: Ref 1

h

Fig. 3

Cross-sectional area, A Pressure P = rgh

Boundary condition for a gravity casting. r, melt density; g, acceleration of gravity

approximations are possible for thin plates, long bars, long plates, and cases where the main heat flow direction is one or two dimensional, such as when the object is cooled from specific sides. Figure 2(c) illustrates a 2D approximation for the case where the curvature radius of a rotation symmetry body is much bigger than its thickness, and Fig. 2(d) shows that neglecting small irregularities has little effect on the results. Applying a rotation symmetry approximation to modify the shape a bit is shown in Fig. 2(e), and Fig. 2(f) demonstrates use of a cut body for symmetrical shapes. It should be noted that while one can take advantage of symmetry for all types of analyses,

deviations from actual geometry may provide erroneous results for, in particular, fluid-flow analyses. It is also important to properly determine the simulation domain. It is useful to consider how far the change in a particular parameter propagates. For example, the region within the diffusion length defined by (at)1/2 (where t is time and a is thermal or mass diffusion coefficient) is the sufficient domain for thermal diffusion or mass diffusion problems. In the case of thermal radiation or deformation problems, the domain becomes very large, because radiation and stress propagate at the speeds of light and sound, respectively. In the case of mold-filling

To define the starting or initial conditions of the simulation, measured or estimated values for temperature, liquid metal pressure and velocity, stress, and so forth must be used. For example, simulation results of mold filling can be used to define the initial temperature distribution for subsequent solidification simulation. In many cases, a uniform initial temperature can be used to simulate solidification phenomena. There are instances where the uniform initial temperature concept fails, as with metallic molds such as die or permanent mold casting, where the mold is quickly cycled through hundreds or thousands of casting cycles. For these types of molds, one typically must simulate the casting of at least 5 to 10 casting cycles to ensure a good initial temperature distribution in the mold. Some judgment must be used when defining some inflow conditions. For example, when many molds or a large mold are poured with a ladle, taking a prolonged pouring time, the pouring temperature may vary with time because of cooling within the ladle. Properly accounting for the variation in metal inflow temperature will often enhance the prediction of some casting defects, such as size and location of solidification shrinkage.

Boundary Conditions Flow Simulation. The inlet boundary conditions in fluid-flow simulations greatly affect the results, requiring careful definition of these conditions. In the case of high-pressure die casting, the inlet melt velocity is typically defined by the plunger velocity. However, it is sometimes necessary to consider the flow in the plunger sleeve, requiring a simulation of flow in the plunger sleeve because too much air in the plunger sleeve can affect the melt flow. Furthermore, not only the gas but also the surface oxide and solidified pieces in the plunger sleeve are sources of many casting defects. In the case of gravity casting the inlet melt velocity can be complicated. For example, as seen in Fig. 3, the most accurate simulation would consider the melt flow resulting from tilting the ladle. However, this phenomenon is not usually supported by simulation codes or it requires a large amount of computer

464 / Modeling and Analysis of Casting Processes resources, especially central processing unit (CPU) time. Therefore, it is recommended to use a pressure-boundary condition and melt inlet area, A, that is measured. Although the correct boundary condition on the mold wall is a nonslip condition ( i.e., the fluid velocity at a solid boundary is zero), even slip conditions usually result in similar simulation results. It should be noted that venting and mold permeability (including mold coating effects) often affect the mold-filling behavior, especially when isolated pockets of gas cannot quickly diffuse through the mold wall or vents. Solidification and Heat Treatment Simulation. Heat is transferred in any of three modes: conduction, convection, and radiation. One must consider which of these to include in the simulation. Conduction (except for some approximation methods) is always included. Convection and radiation effects are often approximated as a single convective boundary condition—convection between the casting and mold, and between the mold and surroundings. The thermal resistance between the mold and casting greatly affects the simulation results for metallic mold casting simulations, including high-pressure die casting. It is also true that one must use an accurate convective heat-transfer coefficient between castings and metallic chills. Proper values are measured or estimated before the simulation. Although literature values (Ref 3) are useful, the generalized heat-transfer methods must be carefully applied to specific casting designs. Stress and Deformation Simulation. Estimation of stress-deformation-related defects such as distortion and hot tears requires stress and deformation simulation. The element shape and position of nodal points where displacement is designated greatly affects the stress and strain at the boundary. Namely, when the nodal points designated at element vertices of rectangular solid elements are applied to curved bodies, a large error is generated, because of stress concentration. Therefore, the use of isoparametric or tetrahedron elements for the finite element method (FEM) is recommended. The finite volume method (FVM) with inner nodal points, where the nodal points are designated at element centers, results in reasonable simulation accuracy even for rectangular solid elements (Ref 4, 5). The difficulty with casting stress and deformation analyses is the issue of contact, namely the effect of the mold. However, the analyses can be solved under the free-boundary condition, where the casting is taken out of the mold at high temperatures and cooled. It is also possible to use the surface element method (Ref 6) where a stiffness of the mold obtained with application of unit displacement to the mold, or to approximate the mold as a rigid body with introduction of spring elements between the rigid mold and casting, setting a proper spring constant depending on the strength of the mold. Simulation of hot tear problems of steel often requires

a proper setting of friction boundary conditions between the casting and mold.

Generally speaking, the more accurate the modeling is, the more accurate input data must be.

Physical Properties

Enmeshing

The values of physical properties such as thermal conductivity and specific heat are obtained from the literature (Ref 7,8) or can be found in the database provided by commercial codes. However, it should be noted that the sand properties (especially thermal conductivity) vary greatly with density (Ref 9); therefore, the properties of the sand mold are greatly affected by the degree of compaction, which is affected by the molding method and even the location in the mold. When proper values of physical properties or boundary conditions are not available, a sensitivity analysis where simulation results with different values are compared is recommended. A value can be safely used if the simulation results change little. If they change a lot, the value should be selected carefully. In some cases the values should be measured directly. Because it is not easy to accurately measure the physical properties or boundary conditions, they are often determined through comparison of simulated results and easy-to-measure values such as temperature change and filling time.

Element Shape Although FEM codes can use various element shapes, codes using the FVM usually limit the shape to rectangular solids. In any case, it is better to avoid highly skewed and/or very elongated (high-aspect-ratio) elements. Selection of element shape is important, in particular, in fluid-flow simulations as shown in Fig. 4. Isoparametric elements are the best for the cavity with curved mold wall, while CPU time is usually longer and more memory is required. It should be noted that when rectangular solid elements are used for curved surfaces of a casting, the contact area between the mold and casting becomes larger than the real value, even when very small elements are used, affecting the accuracy of the solidification simulation. Further, there can be increased errors for stress analyses as mentioned before (see the paragraph “Solidification and Heat Treatment Simulation” in the section “Boundary Conditions” in this article.

(a)

0.13 s

0.60 s

0.57 s

1.60 s

(b)

0.15 s

0.60 s

0.55 s

1.60 s

(c)

0.15 s

0.60 s

0.55 s

1.60 s

(d)

0.15 s

0.60 s

0.55 s

1.60 s

Fig. 4

Effect of enmeshing on mold-filling simulation and comparison with water model (a). Results with (b) regular/ irregular elements, (c) regular elements, and (d) isoparametric elements. Source: Ref 10

Practical Issues in Computer Simulation of Casting Processes / 465 Element Size Although, generally speaking, a smaller element size results in a higher simulation accuracy, the CPU time becomes longer and the required memory size larger. Therefore, although challenging, it is important to select the proper element size. It is usually recommended to use fine meshes in the region where temperature or velocity gradients are steep and/or in critical regions of the casting. Sections should not be enmeshed with only one or two elements.

Evaluation of Simulation Results

Table 1

Comparison of direct observation methods of mold filling X-ray method

Apparatus cost Mold cost

Expensive Low

Low Rather low to high

Operation

Rather easy but sometimes remote handling and an authorized x-ray protection supervisor are required Real phenomena

Easy

Resolution of bubbles or inclusion Advantage

Depending on thickness and density difference Usually 2–3 mm for air bubble Real phenomena

Difficult to observe

Problems

High apparatus cost Limitation of observation position Accurate information of inside phenomena

Not able to see the inside phenomena Reaction of melt with the mold Difficult to observe the effect of permeability of mold

Phenomena observed

Validation and Accuracy of Simulation When the numerical calculations are finished, the results are evaluated. In the case of iterated simulations for conditions similar to those where the simulation accuracy is known, the results can be used as they are. In the case of an initial simulation, however, it is necessary to verify that the results are reasonable. Usually the results are compared with simulations for similar castings or, if possible, with measured values. Even comparison with estimated results is valuable. When simulation results do not agree sufficiently well with independent estimates, the input data such as shape, physical properties, and boundary and initial conditions should be checked for accuracy. Further, it is important to recognize the limitation of modeling, in particular, for estimating casting defects from simulation results (see the next section of this article for estimation of casting defects). For mold-filling simulation, the easiest way to establish simulation accuracy is to compare it with short-shot results. A short-shot test involves pouring or ejecting a small amount of melt that is insufficient to fill the mold into the mold. The filling pattern is compared with the simulation results. This technique is usually applied to metallic mold casting, in particular with lowered mold temperatures. The short-shot method is difficult to apply to sand mold casting, because it is difficult to stop the flow quickly. However, recent high-pressure die-casting machines have a function to stop the plunger during mold filling, resulting in a short shot. Although mold-filling analysis devices using electric contacts can provide more accurate data to validate the simulation, direct observation provides the most detailed information (Ref 11). There are three direct observation methods: observation with x-rays (Ref 11–15), observation through transparent molds (Ref 15), and water or low melting liquid model simulation. Table 1 compares these methods, and examples are demonstrated in Fig. 4 and 5 (Ref 12). The x-ray and transparent mold methods can also be applied to the sand molding process. For solidification simulation the measured temperature curves are good validation. However, it is not easy to measure the temperature

Water or low melting temperature liquid method

Transparent mold method

Rather low High for complicated or large mold Not so easy

Surface phenomena

Surface and inside but not able to observe surface tension-related phenomena (water model) 01 to 0.5mm

Easy Rather low observation cost

Easy Rather low observation cost Possible to observe the entrapment of rather small bubbles Not able to observe surface tension-related phenomena (water model) High cost for complicated or large mold

outlet to vacuum chamber

0.85 s

1.50 s 1.75 s (a) simulation results

2.20 s

0.85 s

1.45 s

1.89 s

B A 1.71 s

(b) x-ray observation results

Fig. 5

Comparison of simulated mold-filling patterns with x-ray observation at reduced pressure of 8 kPa and reducing pressure rate of 4.2 kPa/s. A and B, lower and upper part of the casting. Graphite mold and Al-7Si-0.4Mg melt. Source: Ref 11

within 5% error. Sometimes it is more convenient to estimate the hot spot from microstructure observation. In the case of steel, aluminum alloys, magnesium alloys and others, since the secondary dendrite arm spacing is a function of solidification time, it is a useful scale to estimate the cooling rate [see the articles in the Section “Principles of Solidification” in this Volume. Although validation of microstructure simulation including macrosegregation is usually made by comparing a simulated microstructure with a solidified one, it is possible also to use direct observation with x-ray for macrosegregation (Ref 17, 18) and with the synchrotron x-ray direct beam x-ray imaging technique (Ref 19) for microstructure simulation.

Estimation of Casting Defects Phenomenon While researchers have continued to provide an improved understanding of the relevant

phenomena associated with predicting casting defects, no single simulation code contains all the mathematical tools necessary to predict the wide range of defects associated with castings. Selection of the proper simulation code and proper use of the defect prediction tools must be made to properly predict specific types of defects. Entrapment of Gas and Inclusions. There are not many commercial codes providing direct simulation of gas entrapment. However, it can be estimated it from the flow simulation to some extent. Estimation from Flow Field. Most commercial codes predict the mold-filling sequence. If the final positions to fill are not adjacent to the sand mold or vents, the possibility of gas entrapment is high. Furthermore, if the flow field shows a vortex or swirls, their centers are possible locations for entrapment. Very fast filling velocities should be avoided. The simplest way to characterize fill velocity is

466 / Modeling and Analysis of Casting Processes with a dimensionless parameter, known as Reynolds number (Re), which is defined as Re = rVD/m where V is the flow velocity, D is the pipe diameter (or 4 cross-sectional area/wetted perimeter), r is the fluid density, and m is the fluid viscosity. Fill rates having Re > 10,000 should be avoided. It should be noted that most commercial codes assume laminar flow or a twoequation turbulent model such as a k-e model. Different free surface flow behavior occurs with each of these models. Further, when the melt collides with mold above a critical speed, a lot of gas bubbles or inclusions are entrapped. Larger bubbles float up and/or escape to the mold. Estimation Form Filling Behavior of Elements. It can be assumed that the gas or inclusion is entrapped in the elements that are filled with all inward flow. Based on this assumption, a validation method is applied with an entrapment criteria function that takes filling behavior into account (Ref 20). Estimation from Vortex Intensity. As mentioned previously, gas bubbles or inclusions can be entrapped in vortexes during mold filling. By assuming that the degree of entrapment is proportional to vortex intensity, the vortex intensity can be used to estimate entrapment (Ref 21). Direct Simulation of Entrapment. When the melt-free surface collides with another free surface or with the mold, gas or surface oxides can be entrapped. This defect can be simulated if the free surface shape is well simulated (Ref 22). However, it is not easy to completely model the gas and inclusion entrapment during mold filling, because the accurate simulation of free-surface phenomena is not easy and the wettability of the melt with the mold is usually not considered in the fluid-flow simulation. The users should carefully evaluate the results if a code provides the entrapment simulation. Porosity Estimation. There are various kinds of porosity defects and their causes are different. However, many commercial codes predict the defects using criteria considering temperature field or fraction solid distribution, and so forth. They are generally useful in predicting trends associated with process and/or mold design changes. In the real world, the shape, size, and distribution of pores differ even when the melt and mold cavity are said to be the same. One of the reasons is that porosity defects are affected by the distribution of heterogeneous nucleation sites and gas content, which is affected by melt cleanliness and gas content in the melt. The gas content is affected by not only the melting operation but also gas entrapment and absorption during mold filling. In particular, inclusions, including surface oxides, greatly affect the porosity defects for aluminum and magnesium alloys; however, most commercial codes do not consider these effects, which limits their accuracy in porosity defect estimation. In the case of spheroidal graphite cast iron, accurate porosity estimation requires at least simulation of the expansion of the solidified

casting due to graphite formation (Ref 23). Criteria methods considering just the effect of graphite are limited in accuracy because such criteria vary with casting shape and size. It is most difficult to estimate porosity defects that appear sporadically, for example, once in 100 castings or once in 1000 castings. Although models have not been established for such cases, some simulation might be possible with the following considerations:  Change of melt quality in terms of gas and

inclusion content

 Change in pouring conditions such as pour-

ing temperature (usually not controlled), intermittent pouring or melt spilling from the previously poured mold, dimensional change of pouring basin caused by wear or change of refractory materials, and so forth  Change in sand conditions such as water content or generation of clots or globs containing more clay and water, and so forth, which affects thermal conductivity of the mold and gas porosity  Change in pattern dimensions caused by wear or breakout affecting mold dimensions  Change in mold coating affecting mold dimensions and flow resistance for the gas escape Cold Shut, Misrun, Flow Marks or Surface Folds. Misruns can be estimated from the temperature decrease or solid fraction increase of the melt-free surface obtained from the moldfilling simulation. When the melt pressure affects solid movement as in the case of highpressure die casting, it is necessary to use a proper empirical value of the critical fraction solid. Cold shuts can be estimated at the place where the melt front collides with the part having a higher fraction solid than some critical value.

Estimation of Structure and Properties The casting properties are affected by not only the structure but also defects such as porosity and inclusions, surface roughness, and so forth. The microstructure is affected by the distribution of heterogeneous nuclei. Therefore, the simulation results should be used carefully, considering the assumptions made and limitations. Sometimes it is more accurate to estimate the properties directly from experimental data correlating the calculated cooling rate and measured properties.

Optimization Although it is often claimed that simulation codes can determine optimum casting conditions or size and dimensions, in reality, optimization is limited to rather special cases. This is because the real optimum condition is very

difficult to determine. For example, the conventional methods use iterative methods to take maximum or minimum of an evaluation function; however, the evaluation function is rather simple and neglects cost. REFERENCES 1. I. Ohnaka, J. Jpn. Inst. Light Me., Vol 54, 2004, p 304–403 2. D.R. Poirier and G.H. Geiger, Transport Phenomena in Materials Processing, TMS, 1994 3. A. Bejan and A.D. Krans, Heat Transfer Handbook, John Wiley & Sons, 2003; and other heat-transfer handbooks 4. I. Demirdzic and D. Martinovic, Comput. Methods Appl. Mech. Eng., Vol 109, 1993, p 331 5. A. Sato, I. Ohnaka, and J. Iwane: J. Jpn. Foundry Eng. Soc., Vol 78, 2006, p 231 6. D. Metzger, K.J. New, and J. Dantzig, Appl. Math. Model., Vol 25 2001, p 825 7. Y.S. Touloukian, Ed., Thermophysical Properties of Matter, TPRC Data Series, Plenum Publishing, 1970 8. R.D. Pehlke, A. Jeyarajan, and H. Wada, Summary of Thermal Properties for Casting Alloys and Mold Materials, University of Michigan Press, 1982 9. M. Kubo, I. Ohnaka, and T. Fukusako, Measurement of Thermal Conductivity of Sand Molds by Parameter Optimization Technique, Trans. Jpn. Foundrymen’s Soc., Vol 2, 1983, p 41–45 10. J.D. Zhu and I. Ohnaka, A New Simulation Method of Mold Filling by Using Regular and Irregular-Mixed Elements, Proc. of Modeling of Casting and Solidification Processes IX, P. R. Sahm, P. N. Hansen, and J. G. Conley, Ed., Shaker Verlag, Aachen, Germany, 2000, p 428–435 11. I. Ohnaka, H. Yasuda, A. Sugiyama, and T. Ohmichi, Direct Observation of Casting and Solidification, Trans. Indian Inst. Met., Vol. 58 (No. 4), 2005, p 603–610 12. S. Kashiwai, I. Ohnaka, A. Kimatsuka, T. Kaneyoshi, T. Ohmichi, and J.D. Zhu, Numerical Simulation and X-ray Direct Observation of Mould Filling During Vacuum Suction Casting, Int. J. Cast Met. Res., Vol 18 (3), 2005, p 144–148 13. B. Sirrell, M. Holliday, and J. Campbell, Modeling of Casting, Welding and Advanced Solidification Processes VII, M. Cross and J. Campbell, Ed., TMS, 1995, p 915 14. X. Yang, D. Din, and J. Campbell, Int. J. Cast Metals Res., Vol 11, 1998, p 1 15. C. Bates, http://www.eng.uab.edu/mse/ labs/cel/ 16. P. Larsen and N. Tiedje, Proc. 66th World Foundry Congress, Istanbul, 2004, p 223 17. M.P. Stephenson and J. Beech, Solidification and Casting of Metals, The Metals Society, London, 1979, p 34 18. M.R. Bridge and J. Beech Solidification Technology in the Foundry and Casthouse, The Metals Society, London, 1983, p 478

Practical Issues in Computer Simulation of Casting Processes / 467 19. H. Yasuda, I. Ohnaka, K. Kawasaki, A. Sugiyama, T. Ohmichi, J. Iwane, and K. Umetani, J. Cryst. Growth, Vol 262, 2004, p 645 20. Y. Ohtsuka, T. Ono, K. Mizuno, and E. Matsubara, Imono (J. Jpn. Foundrymen’s Associ.), Vol 60, 1988, p 757

21. T. Sakurai, J. Jpn. Foundry Eng., Vol 73, 2001, p 293 22. I. Ohnaka, Recent Development of Numerical Modeling of Casting and Solidification, Modeling of Casting, Welding and Solidification Processes X, TMS, 2003, p 403–414

23. I. Ohnaka, J. Iwane, H. Yasuda, and J.D. Zhu, Prediction of Porosity Defect in Spheroidal Graphite Iron Castings, Int. J. Cast Met. Res., Vol 16 (No. 1–3), 2003, p 293–299

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 468-481 DOI: 10.1361/asmhba0005240

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Thermophysical Properties Juan J. Valencia, Concurrent Technologies Corporation Peter N. Quested, National Physical Laboratory

ADVANCED COMPUTER SIMULATION TECHNOLOGY is a powerful tool used to understand the critical aspects of heat transfer and fluid transport phenomena and their relationships to metallurgical structures and defect formation in metal casting processes. Computational models are enabling the design and production of more economical and higher-quality castings. In order to produce accurate and reliable simulation of the complex solidification processes, accurate, self-consistent, and realistic thermophysical properties input data are necessary. Unfortunately, reliable data for many alloys of industrial interest are very limited. Sand, ceramic, and metal molds are extensively used to cast most metals. During the solidification process, the predominant resistance to heat flow is within the mold/metal interface and the mold itself; thus, the primary interest is not the mold thermal history but rather the rate at which the heat is extracted from the solidifying metal. Therefore, heat transfer is the governing phenomenon in any casting process. Heat transfer is fundamentally described by the heat-transfer coefficient, the temperature gradient, the geometry of the system, and the thermophysical properties of both metal and mold material. Table 1 shows the required thermophysical properties that must be available for input before reliable numerical simulations of a casting process can be performed, as well as their influence in the prediction of defects. Current commercial software requires that the thermal conductivity, specific heat capacity, latent heat, solidus and liquidus temperatures, and density must be known for heat-transfer operations. Viscosity, density, wetting angle, and surface tension of the molten alloy are required for fluid flow operations. In addition to the metal properties, mold materials properties are also needed to conduct an effective simulation.

Sources and Availability of Reliable Data The thermophysical property data found in the literature for engineering and design calculations

of casting processes must not be used indiscriminately without knowing their source and reliability. Prior to using a given set of data, it is very important to critically evaluate and analyze the available thermophysical property data, to give judgment on their reliability and accuracy. There are several main sources of thermophysical property data that provide the most authoritative and comprehensive compilations of critically and systematically evaluated data that are presently available. The challenge of finding data is discussed in Ref 2. The data have been published and can be found in the following resources:
The Center for Information and Numerical











Data Analysis and Synthesis (CINDAS) generated and recommended reference values for diverse materials (Ref 3–12). Smithells’ Metals Reference Book provides an extensive compilation of thermochemical data for metals, alloys, and compounds of metallurgical importance (Ref 13). Summary of Thermal Properties for Casting Alloys and Mold Materials by R.D. Pehlke and co-workers (Ref 14). The ASM International Materials Properties Database Committee publishes a comprehensive thermal properties database of most commercially available metals (Ref 15). Recommended Values of Thermophysical Properties for Selected Commercial Alloys by K.C. Mills. Experimental determination, estimation, and validation of the thermophysical properties in the solid and liquid states (Ref 16)

Computer models based on first principles of thermodynamics and kinetics of phase transformations have been developed to calculate thermophysical properties for various materials in the solid and liquid states (Ref 17–22). However, their use is still limited due to the lack of thermodynamic data and accurate measurements of thermophysical properties for materials of industrial interest. Also, sensitivity studies (Ref 23) are necessary to truly evaluate the reliability of calculated thermophysical property data from these models in actual casting processes.

Limitations and Warning on the Use of Data The thermophysical properties data presented in this article are provided to assist in the materials properties selection for the simulation of casting processes. Great effort has been exercised in the compilation and analysis of the data, and careful attention has been taken to faithfully duplicate the data and their sources found in the literature. The thermophysical properties data provided here shall bear the warning “not for design purposes.” It is the full responsibility of the reader to further investigate the sources of information and follow all necessary engineering steps to make sure the validity and quality of the data meet the requirements of the intended application.

Methods to Determine Thermophysical Properties Experimental determinations of reliable thermophysical properties are difficult. In the solid state, the properties recorded in the technical literature are often widely diverging, conflicting, and subject to large uncertainties. This problem is particularly acute for materials in the mushy and liquid state. Also, accurate, consistent, and reliable thermophysical property measurements are experimentally difficult. Convection effects in molten samples and their interactions and reactivity with their containers and environment often exacerbate the difficulties. The measurements are difficult because of high temperatures and the reactivity of some alloys. Strategies adopted to minimize these effects are:
Perform the experiments quickly using tran-

sient methods


Choose crucible materials that contact the

sample to minimize reactions


Eliminate container levitation
Measure properties in microgravity
Control the composition of the atmosphere

Mills et al. (Ref 24) describe the necessity to exercise care when analyzing the experimental results. The following cases are discussed:

Thermophysical Properties / 469
Thermal diffusivity measurements for the

Table 1 Thermophysical property data required for metal casting Casting process component

Furnace metal

Transport phenomena for casting

S

Heat transfer
Conduction
Convection O
Radiation L

I Mold core chill

D Mass transfer (fluid flow) I

Thermophysical data required

Heat-transfer coefficient
Metal/mold
Metal/core
Metal/chill
Mold/chill
Mold/environment Emissivity—Metal/mold/furnace wall Temperature-dependent parameters
Density
Heat capacity
Conductivity Latent heat of fusion Liquidus and solidus

Effective design for:
Riser
Chill
Insulation Solidification direction Solidification shrinkage Porosity Hot spots

Temperature dependent
Viscosity
Surface tension
Density

Effective design for:
Ingate
Runner
Vents Pouring parameters
Temperature
Pouring rate Mold filling time Cold shut Missruns

Phase diagram Phase chemical composition Capillarity effect (Gibbs-Thompson coefficient) Nucleation and growth parameters Solid fraction vs, temperature Diffusivity
Solubility Temperature-dependent parameters
Coefficient of thermal expansion
Stress/strain

Microsegregation Macrosegreation Grain size Grain orientation Phase morphology Mechanical properties

F Insulation

I

Microstructural evolution

C

A Stress analysis T I

Computer modeling for process, part design, and defect prediction

Casting design for:
Dimension and distortion Internal stresses Hot tears and hot cracks

O N Source: Ref 1

Table 2 Thermophysical and mechanical properties needed for casting process simulation and common measurement techniques Thermophysical property

Thermal conductivity Heat capacity Density Thermal diffusivity Heat of fusion Transformation temperatures Fraction solid Electrical resistivity Hemispherical emissivity Viscosity Surface tension Young’s modulus, Poisson’s ratio Thermal expansion Yield strength

Measurement technique

Comparative stationary (solid), indirect (liquid) Differential scanning calorimetry, pulse heating; drop calorimetry Archimedian balance, push-rod dilatometry; levitation Laser flash Differential scanning calorimetry, pulse heating Differential scanning calorimetry, thermal analysis Differential scanning calorimetry, thermal analysis Pulse heating, four-point probing Pulse heating Levitation, viscometer Levitation, sessile drop Tensile test, sound speed measurements Push-rod dilatometry Tensile test

Source: Ref 26


The

specific heat capacity (Cp) peaks recorded by differential scanning calorimetry (DSC) for fusion (or solidification) are only apparent Cp values, since they contain enthalpy contributions.
Use of these apparent Cp values for the conversion of thermal diffusivities to conductivities can lead to massive overestimation of thermal conductivities for temperatures in the transition range.


There is a temperature lag between the sam-

ple and reference pans in differential thermal analysis (DTA)-type DSC instruments, which causes some uncertainty in the temperature scale for fraction solid/temperature relations. The temperature difference between the sample and reference pans results in heating/cooling rate-dependent variations in transition temperatures (see also Ref 25 for a detailed critique).

mushy region are prone to error because some of the energy supplied may be used for further melting of the alloy.
Oxide films on the surface of the molten alloy can affect measurements of physical properties; wettability of the metal on the crucible and/or rotor can affect viscosity measurements and nonwetting, leading to low values for the viscosity. Table 2 lists some common techniques used for the measurement of relevant thermophysical properties. Numerous methods exist for the measurement of thermophysical properties of metallic materials and are cited in the literature (Ref 16, 27–47). However, only a few of them have been standardized, and most of them are limited to the solid-state properties. The current ASTM International standards and selected CEN and ISO standards include:
Specific heat capacity: Differential scanning

calorimetry, ASTM E 1269, E 967, E 968, E 2253, E 793, and D 2766 (Ref 48–53). Ceramics, EN 821–3 drop and DSC (Ref 54)
Thermal expansion: Dilatometry, ASTM E 228; interferometry, ASTM E 289; and thermomechanical analysis, ASTM E 831 (Ref 55–57). Ceramics, EN 821–1 and ISO 17562, both dilatometry (Ref 58, 59)
Thermal conductivity: Modulated-temperature scanning calorimetry, ASTM E 1952; thermal diffusivity of solids by the laser flash method, ASTM E 1461; and the steady-state heat flow, ASTM C 518 (Ref 60–62). Ceramics, EN 821–2 and ISO 18755, both laser flash (Ref 63, 64)
Thermal emittance: Radiometric techniques, ASTM E 307 and E 408 (Ref 65, 66) Best practices for the measurement of alloy freezing and melting temperatures by DTA and DSC measurements are found in Boettinger et al. (Ref 25). Standardized methods to determine the thermal properties of liquid metals are practically nonexistent. Several methods, such as the oscillating viscometer (Ref 67); the levitated drop apparatus to measure surface tension, density, and viscosity (Ref 34, 67–69); and the laser flash to measure thermal diffusivity and specific heat capacity (Ref 37–43), are successfully being used. Reference 33 provides a comprehensive critical survey of the microstructural characteristics of liquid metals, which determine properties of viscosity, surface tension, density, specific heat capacity, thermal conductivity, electrical resistivity, diffusion, and velocity of sound transmission. The experimental techniques used to obtain these data are also reviewed, with correlations and reference data. In this article, the methods to measure thermophysical properties are not discussed further; instead, available thermophysical properties for pure metals and some commercial alloys are presented.

470 / Modeling and Analysis of Casting Processes

Specific Heat Capacity and Enthalpy of Transformation

Table 3 Specific heat capacity values and enthalpy of fusion for pure metals Element

Specific heat capacity of a material is the amount of thermal energy needed to change the temperature of a unit mass (m) of a substance by one degree Kelvin. The specific heat capacity is an extensive property of matter that depends on the amount of species in the system and is sensitive to phase changes. Specific heat capacity can be defined for a constant volume (Cv) or for a constant pressure (Cp). The specific heat capacity in SI units is expressed in J/kg  K. Also, the specific heat term is often used interchangeably with heat capacity. While this is not precisely correct, it is not a cause of misunderstanding. The total amount of thermal energy or enthalpy, DH, associated with the specific heat capacity and a temperature change (T1 to T2) is given by: Tð2


H ¼

Cp dT

(Eq 1)

T1

A good approximation to estimate the specific heat capacity as a function of temperature is given by: Cp ¼ a þ bT þ cT 2 þ . . .

(Eq 2)

where Cp is the molar heat capacity; a, b, and c, are constants; and T is the temperature in degrees Kelvin. Table 3 shows the specific heat capacity of solids as a function of temperature, the specific heat capacity of liquids at the melting point (Tm), and the enthalpy of fusion for most common elements found in cast metals. Changes in specific heat capacity and most thermophysical properties with changing temperature in liquid metals may be gradual and continuous, rather than showing the abrupt effects of phase transitions that take place in the mushy and solid state. Thus, a reasonable estimate of the specific heat capacity for a liquid alloy (CpL) can be calculated from the elemental heat capacities of the components in the alloy by using the commonly known Kopp-Neuman rule of mixtures (Ref 72, 73): CpL ¼

n X

Yi ðCp ÞLi in cal=ðmol  KÞ

(Eq 3)

i¼1

where Yi is the atomic fraction of element i in the alloy. The small changes in the specific heat capacity of the liquid with temperature allows for a reasonable estimate (þ 3%) using Eq 3. Equation 3 can be expressed in SI units by multiplying CpL by 4.184 J/cal and dividing by the atomic mass of the element i in kg/mol. Major changes in the specific heat capacity with heating or cooling rates are observed in the solid-liquid range (mushy region). Macroand microsegregation, as well as the presence of eutectics, peritectics, and other phase transformations that occur during solidification, cannot be easily described by Eq 1. Instead, the

Al(s) Al(l) C(graphite) Co(b) Co(g) Co(l) Cr(s) Cr(l) Cu(s) Cu(l) Fe(a, d) Fe(g) Fe(l) Hf(s) Li(s) Li(l) Mg(s) Mg(l) Mn(a) Mn(b) Mn(g) Mn(d) Mn(l) Mo(s) Nb(s) Ni(a) Ni(b) Ni(l) Pb(s) Pb(l) Si(s) Si(l) Ta Ti(a) Ti(b) Ti(l) V(s) W(s) Y(a) Y(b) Y(l) Zn(s) Zn(l) Zr(a) Zr(b)

Cp = a + bT + cT2 +. . .(a), J/K  mol

4.94 + 2.96  103T 7.0 4.1 + 1.02  103T  2.10  105T2 3.3 + 5.86  103T 9.60 9.65 5.84 + 2.35  103T  0.88  105T2 9.4 5.41 + 1.4  103T 7.50 8.873 + 1.474  103T  56.92T1/2 5.85 + 2.02  103T 9.74 + 0.4  103T 5.61 + 1.82  103T 3.33 + 8.21  103T 5.85 + 1.31  103T + 2.07  105T2  467  106T2 5.33 + 2.45  103T  0.103  105T2 7.68 5.70 + 3.38  103T  0.375  105T2 8.33 + 0.66  103T 6.03 + 3.56  103T  0.443  105T2 11.10 11.0 5.77 + 0.28  103T + 2.26  106T2 5.66 + 0.96  103T 7.80  0.47  103T  1.335  105T2 7.10 + 1.0  103T  2.23  105T2 9.20 5.63 + 2.33  103T 7.75  0.74  103T 5.72 + 0.59  103T  0.99  105T2 6.498 6.65  0.52  103T  0.45  105T2 + 0.47  106T2 5.28 + 2.4  103T 4.74 + 1.90  103T 11.042 4.90 + 2.58  103T + 0.2  105T2 5.74 + 0.76  103T 5.72 + 1.805  103T + 0.08  105T2 8.37 9.51 5.35 + 2.4  103T 7.5 5.25 + 2.78  103T  0.91  105T2 5.55 + 1.11  103T

Temperature range, K

298–932 932–1273 298–2300 715–1400 1400–Tm Tm–1900 298–Tm Tm 298–Tm Tm–1600 298–Tm 1187–1664 Tm–2000 298–1346 273–Tm Tm–580 298–Tm Tm–1100 298–1000 1108–1317 1374–1410 1410–1450 Tm–Tbp(c) 298–2500 298–1900 298–630 630–Tm Tm–2200 298–Tm Tm–1300 298–1200 Tm–1873 298–2300 298–1155 1155–1350 Tm–2073 298–1900 298–2000 298–1758 1758–Tm Tm–1950 298–Tm Tm–Tbp 298–1135 1135–Tm

Tm(b), K

CpLm ; J=g  K

933

1.18

397

16, 70

4073 1768

... 0.59

... 275

70 13, 16, 70

2130

0.78

401.95

13

1356

0.495

208.7

13, 16

1809

0.762

247

13, 16, 71

2500 454

... ...

134.85 422.1

13 13, 70

922

1.32

349

13, 16

1517

0.838

267.5

13

2893 2740 1726

0.57 0.334 0.63

371 315.4 292.4

13 13 13, 16, 70

600

0.142

23.2

13

1685

0.968

1877

13, 16

3288

...

136.5

13, 16

1940

0.965

295

13, 16

2175 3673 1803

... ... 0.394

328.6 176.8 128.6

13 ... 13

0.481

112

13, 16

... ...

211.6 ...

13 ...

692.5 2125 ...

DHf, J/g

Reference

(a) Cp in SI units (J/kg  K) when multiplied by 4.184 J/cal and divided by the corresponding element atomic mass (kg/mol). (b) Tm = melting point. (c) Tbp = boiling point

behavior in the solid or mushy state is more complex, since the phase transformations are dependent on the heating and cooling rates and on the chemistry of the alloy system. The dynamic characteristics of a given casting process require the input of all liquid-to-solid changes to understand the behavior of the solidifying metal. Therefore, in casting processes, determination of the specific heat capacity must be conducted on cooling for the solidifying metal and on heating for the mold and core materials, since the latter absorb most of the superheat and latent heat of solidification.

Enthalpy of Melting, Solidus, and Liquidus Temperatures The enthalpy or latent heat of melting (DHf) is the heat that is required during solid-to-liquid

transformations, and the latent heat of solidification is the heat released during liquid-to-solid transformations. The latent heats of melting and solidification in a multicomponent alloy system occur over a temperature range. The temperature at which the alloy starts to melt is called the solidus temperature, and the temperature at which the melting is completed is called the liquidus temperature. In actual melting and casting processes, equilibrium conditions do not exist, since melting and solidification processes are ruled by the rate of phase transformations and by the heat and mass transfer phenomena. On melting, high heating rates may displace the solidus and liquidus temperatures to higher values, while on cooling and prior to the nucleation of the solid phase, the molten alloy is usually undercooled. High undercooling generally decreases the liquidus and solidus temperatures. Also, the

Thermophysical Properties / 471 degree of undercooling directly affects the kinetics of the liquid-solid phase transformation and the type of second phases that evolve during solidification. A more detailed discussion of nonequilibrium structures can be found in Ref 74. In actual casting processes, depending on the process and degree of inhomogeneity and impurities in a cast material, the liquidus and solidus temperatures can differ by several or even tens of degrees from equilibrium. The latent heat, solidus, liquidus, and other phase transformation temperatures are determined using the same techniques as for heat capacity and are described elsewhere in the literature (Ref 27). The enthalpy of fusion and the solidus and liquidus temperatures for various alloys of commercial interest are shown in Table 4. The heat capacity, along with the thermal conductivity and density, for some commonly used sand molds is shown in Table 11.

Coefficient of Thermal Expansion The coefficient of linear thermal expansion (a) is a material property that indicates the extent to which the material expands or contracts with temperature changes. At a constant pressure, the true coefficient of volumetric thermal expansion (aV, or commonly b) is defined by the changes that occur by a differential temperature change (dT). This is usually expressed by the relationship: aV ¼

1 V



@V @T

 (Eq 4) P

where V is the volume at a temperature, T, at a constant pressure, P. The corresponding definition for the linear coefficient of expansion can be represented by the relationship: al ¼

  1 @l l @T P

(Eq 5)

Usually, the coefficient of thermal expansion is not measured directly but is calculated by the derivative of the equation that represents the expansion. Also, the instantaneous coefficient of linear thermal expansion is frequently defined as the fractional increase of length per unit rise in temperature. Further analyses and the theory of thermal expansion can be found in the literature (Ref 6). The temperature dependence of the coefficient of thermal expansion for solids is very complex, but it has been shown that it varies inversely with the melting point (Tm) of the material and is expressed by: a ¼ gG =100Tm

(Eq 6)

where gG is the Gru¨neisen parameter and, for most solid materials, is close to 1. Semiempirical analyses (Ref 6) of the thermal expansion of crystalline materials, such as close-packed

metals, have shown that the mean coefficient of linear thermal expansion (am) can also be related to their melting temperature by the relationship: ðLm  L0 Þ=L0 ¼ Tm am  0:0222

(Eq 7)

where Lm is the length at Tm. It has been found that most metals with melting points above 900 K linearly expand approximately 2% on heating from 298 K to Tm (Ref 6). Many metals exhibit a thermal expansion am  104 K1 in the liquid phase just above the liquidus temperature. Table 5 shows the linear expansion (DL/L0) and the coefficient of linear thermal expansion (a) for some pure metals at temperatures closer to their melting points (Ref 6). Unfortunately, few data at the solidus and/or liquidus are available for alloys of commercial interest. Data have been determined in the liquid by levitation, and recent developments with piston dilatometry enable measurements across the liquid/solid region. Using this technique, Blumm and Henderson (Ref 75) and Morrell and Quested (Ref 76) have performed measurements on nickel, aluminum, and cast irons, and typical data for alloys are included in Tables 14 to 17.

Density Density (r) is defined as the mass per unit volume of a material. The reciprocal of the density (1/r) is the specific volume. Room temperature and liquid density for pure metals and some materials of commercial interest are available in the literature (Ref 13, 15, 16, 28, 77). Accurate density values are highly desirable since it is a variable in the calculation of thermal conductivity, surface tension, and viscosity. Further, the evaluation of fluid flow phenomena in a solidifying metal is dominated by the changes of density. Density of liquid metals is also useful to calculate volume changes during melting, solidification, and alloying. As a general rule, the average density change of nonclose-packed metals on fusion is approximately 3% while the average volume change of a close-packed metal does not exceed 5% (Ref 71). Bismuth, gallium, antimony, germanium, silicon, cerium, and plutonium are the exceptions to the general rule, since these elements contract on melting. Density of solid alloys as a function of temperature can be calculated from thermal expansion data using the following relationship: rT ¼ rRT =ð1 þ
Lexp Þ3

(Eq 8)

where DLexp is the linear expansion (DLexp = (LT  L0)/L0 = aT) at a temperature, T. Alternatively, the density of a heterogeneous phase mixture (rm) containing a number, n, of phases can be roughly estimated using the empirical relationship:

rm ¼ P n

1

(Eq 9)

ðXp =rp Þ

i¼1

where Xp and rp are the fraction and density of the phase, respectively, at a given temperature. The density of pure liquid metals as a function of temperature can be reasonably estimated from the following empirical equation (Ref 28): rL ¼ a  bðT  Tm Þ

(Eq 10)

where rL is in g/cm3, a (the density at the liquidus temperature, rm) and b are dimensionless constants, and T is the temperature above the melting point (Tm). Both temperatures are in degrees Kelvin. Table 6 shows the density at the melting point and the values for the parameters a and b for estimation of the liquid density as a function of temperature for various elements. Table 6 also shows the viscosity and activation energy for viscous flow for some pure metals.

Surface Tension Knowledge of the surface tension phenomena of metals is essential in the understanding of solidification during casting. In solidification phenomena, the transformation of liquid into solid requires the creation of curved solid/liquid interfaces that lead to capillarity, microscopic heat flow, and solute diffusion effects. The interplay of the heat flow and solute diffusion effects determines the solidification morphologies. Solute diffusion effects have the tendency to minimize the scale of the morphology, while the capillarity effects tend to maximize the scale. A compromise between these two tendencies has a profound effect on the crystal morphologies with respect to nucleation, interface instability, and dendritic and eutectic growth (Ref 80). The surface energy has an important role because of the creation of the solid/liquid interface area. Also, during solidification, a decrease in the equilibrium melting point produces a positive undercooling and is associated to the solid/liquid interface curvature, which is usually convex toward the liquid phase. This curvature effect is often called the Gibbs-Thomson effect (G = gs/l/DSf), which is a function of the solid/liquid interfacial energy (gs/l) and the volume entropy of fusion (DSf). The Gibbs-Thompson effect is of the order of 107 mK for most metals. This indicates that the effect of the surface energy becomes important only when a given morphology such as nuclei, interface perturbations, dendrites, and eutectic phases have a radius less than 10 mm (Ref 80). Unfortunately, data on the surface energy or surface tension (J/m2) for most metals and their alloys are very limited or not available. Surface tension of liquid metals can be measured using various contact and noncontact techniques (Ref 33–36, 68, 69). Semitheoretical models have been developed to calculate the surface tension of pure metals.

472 / Modeling and Analysis of Casting Processes Table 4

Specific heat capacity data, latent heat of fusion, and solidus and liquidus temperatures for some alloys of commercial interest

Material

Aluminum alloys A319 (LM4; Al-5Si-3Cu) A356 (LM25; Al-7Si-0.5 Mg) 2003 3004 2024-T4 6061-T6 7075-T6

Nominal composition, wt%

Al-3Cu-5Si-1Zn-0.35Ni-0.4Mn-0.1Mg Al-7Si-0.5Fe-0.4Mg-0.3Mn-0.2 Cu-0.1Ni-0.2Ti Al-4.5Cu Al-0.2Cu-1Mg-1Mn-0.43Fe-0.14 Si-0.25Zn Al-4.4Cu-1.5Mg-0.6Mn-0.5Fe-0.5 Si-0.25Zn-0.1Cr-0.15Ti Al-0.3Cu-1Mg-0.15Mn-0.7Fe-0.6 Si-0.25Zn-0.04Cr-0.15Ti Al-1.6Cu-2.5Mg-0.3Mn-0.5Fe-0.4 Si-5.6Zn-0.2Cr-0.2Ti

Copper alloys Cu-Al (Albronze)

Cu-9.7Al-4.6Fe-0.64Mn-4.6Ni

Brass

70Cu-30Zn

Brass

60Cu-40Zn

Copper-nickel

70Cu-30Ni-Fe-Mn

Iron and steel alloys Carbon steel AISI 1008 Carbon steel AISI 1026 1% Cr 304 stainless steel 316 stainless steel 420 stainless steel Ductile iron Gray cast iron

Fe-0.08C-0.31Mn-0.08Si0.45Cr-0.03P-0.05S Fe-0.23C-0.63Mn-0.11Si-0.07Ni0.03P-0.03S Fe-0.3C-0.69Mn-0.2Si-1.1 Cr-0.7Ni-0.012Mo-0.039P-0.036S Fe-0.08C-19Cr-0.3Cu-2Mn-9.5Ni Fe-0.08C-17Cr-0.3Cu-2Mn-2.5 Mo-12Ni-1Si Fe-0.3C-13Cr-0.12Cu-0.5Mn-0.06 Mo-0.5Ni-0.4Si Fe-3.61C-2.91Si -0.08Cr-0.12 Cu-0.65Mn-0.02Mo-0.13Ni0.002Mg Fe-3.72C-1.89Si-0.95Cr-0.66Mn0.59Mo-0.19Ni0.002Mg

Magnesium alloys AZ31B

Mg-3Al-1Zn-0.5Mn

AZ91B

Mg-9Al-0.6Zn-0.2Mn

KIA

Mg-0.7Zr

ZK51A

Mg-4.6Zn-0.7Zr

HM11A

Mg-1.2Mn-1.2Th

EZ33

Mg-3Ce-2Zn-0.6Zr

Nickel alloys Single-crystal CMSX-4 Hastelloy C Hastelloy X Inconel 718

Ni-10Co-6.5Cr-6.5Ta-6.4W-3Re1Ti-0.15Fe-0.04Si-0.006C Ni-16Mo-16Cr Ni-22Cr-18.5Fe-9Mo-1.5Co-0.6W0.5Si-0.1C Ni-19Cr-16.7Fe-5.2Nb-3.1Mo-1Co0.9Ti-0.35Mn-0.35Si-0.08C

Titanium alloys Ti-6Al-4V Ti-5.5–6.7Al-3.5–4.5V-0.25(O2 + N2)0.03Fe-0.0125H2 Zinc alloys Zn-Al

Zn-4.5Al-0.05Mg

Specific heat capacity, J/g  K

Temperature range, K

Cp at 25  C

Cp at TL

DHf, J/g

TS, K

TL, K

Reference

Cp = 0.7473 + 2  104T + 5  107T2 Cp = 0.7284 + 5  104T  8  107T2

298–780

0.87

1.17

393

798

898

16

298–840 Trans. T 653

0.88

1.16

425

840

887

16

Cp = 0.749 + 4.44  104T Cp = 1.287  2.5  104T Cp = 0.7989 + 3  104T  9  108T2 Cp = 0.7688 + 3  104T  2  107T2 Cp = 0.7067 + 6  104T  1  107T2 Cp = 0.7148 + 5  104T + 4  1010T2

T < 775 775–991 298–873

0.882

1.059

...

775

911

14

0.90

1.22

383

890

929

16

298–811

0.87

1.14

297

811

905

16

298–873

0.87

1.17

380

873

915

16

298–805

0.86

1.13

358

805

901

16

Cp = 0.353 + 3  104T  1  107T2 Cp = 0.582 Cp = 0.355 + 1.36  104T Cp = 1.32 + 6.75  104T Cp = 0.49 Cp = 0.354 + 1.11  104T Cp = 0.689 + 1.0  103T Cp = 0.489 Cp = 0.37 + 1.13  104T Cp = 0.348 + 1.28  104T Cp = 0.543

298–1313

0.442

0.582

240

1313

1350

16

0.396

0.49

164.8

1188

1228

14

0.387

0.489

160.1

1173

1178

14

0.404

0.543

...

1443

1513

14

Cp = 0.593 + 4.8  105T

1273–1550

0.469

...

...

1768

1808

14

1379–1550

0.469

...

...

1768

1798

14

1173–1683

0.477

0.856

251

1693

1793

14

Cp = 0.443 + 2  10 T  8  1010T2 Cp = 0.412 + 2  104T  2  108T2 Cp = 1.92  1.587  102T Cp = 0.569 Cp = 0.80

298–1727

0.49

0.80

290

1673

1727

14, 16

298–1658

0.45

0.79

260

1658

1723

14, 16

1150–1173 T > 1173 1373

0.477

...

304

1727

1783

14

0.48

0.83

220

1413

1451

16

Cp = 0.66

1353

0.49

0.95

240

1353

1463

16

 905

1.01

1.415

...

839

905

14

 869

0.98

1.428

...

742

869

14

1.005

1.428

...

922

923

14

 914

1.022

1.371

...

822

914

14

 923

0.946

1.411

...

905

923

14

0.98

1.336

343

818

913

16

0.397

0.636

240

1593

1653

16

Cp = 0.281 + 0.283  10 T T  1534 Cp = 0.4384 + 1  104T 1073–1533 + 5  108T2 Cp = 0.65 1443

0.283 0.439

0.677

... 276

1534 1533

1578 1628

14 16

0.435

0.72

210

1533

1609

16

Cp = 0.4115 + 2  104T + 5  1010T2 Cp = 0.83

1268–1923 T 1923

0.546

0.83

286

...

1923

16

Cp = 0.50 Cp = 0.52  6  105 (T  387  C)

630 Liquid T 660

0.41

0.51

114

630

660

16

4

Cp = 0.354 + 2.1  10 T 3

Cp = 0.436 + 1.22  10 T 4

Cp Cp Cp Cp Cp Cp Cp Cp Cp Cp Cp Cp

= = = = = = = = = = = =

1.88  5.22  104T 0.979 + 4.73  104T 0.251 + 1.354  103T 1.43 0.873 + 0.463  103T 1.43 1.945  6.28  104T 0.90 + 0.516  103T 0.65 + 8.75  103T 1.482 1.21 1.336

1313–1773 298–1188 1188–1228 1228  T T  1173 1173–1178 1178  T T  1443 1443–1513 1513  T

839  T 905  T 742  T 869  T T  922 923  T 822  T 914  T 903  T 923  T 818 913  T

1653  T

Cp = 0.675 3

Thermophysical Properties / 473 Table 5 Thermal expansion of selected pure metals at temperatures close to melting Element

Melting point (Tm), K

LE%(a), DL/L0

CTE (a)(b), 106/K

T(c), K

Ag Al Au Be Bi Cd Ce Cr Co Cu a-Fe Fe (a-g) g-Fe Hf La Li Mg Mn Mo Nb Ni Pb Pd Pt Re Rh Sb Si Sn Ta a-Ti Ti (a-b) Ti-b V W Zn Zr

1233.7 933.5 1336 1562 544.6 594 1072 2148 1766 1356 1185 1185 1811 2216 1195 453.5 924 1525 2880 2741 1727 660.6 1825 2042 3431 2236 904 1683 505 3250 1156 1156 1958 2185 3650 692 2123

2.11 1.764 1.757 2.315 0.307 1.028 0.512 2.02 1.5 2.095 1.37 0.993 2.077 0.712 0.497 0.804 1.886 6.604 2.15 1.788 2.06 0.988 1.302 1.837 1.941 1.526 0.588 ... 0.516 3.126 0.918 0.868 1.411 2.16 2.263 1.291 1.139

28.4 37.4 22.1 23.7 12.4 40 9.4 19 17.7 25.8 16.8 23.3 23.3 8.4 11.3 56 37.6 ... 16.5 10.1 20.3 36.7 16.9 14.9 9.8 15.4 11.7 3.8 27.2 24.4 11.8 11 13.5 17.2 11.6 34 11.3

1200 900 1300 1500 525 590 1000 1900 1200 1300 1185 1185 1650 1300 1000 450 900 1500 2800 2300 1500 600 1200 1900 2800 1600 800 ... 500 3200 1156 1156 1600 2000 3600 690 1800

(a) LE% = percent of linear expansion. (b) CTE = coefficient of linear expansion. (c) T = temperature. Source: Ref 6

The rigid sphere model (Ref 81) assumed a structure of the liquid metal where the collision diameter may be estimated by the molar volume (V) at the melting temperature. Then, the surface tension can be expressed by: g ¼ ð3:6Tm V 2=3 Þ  103 J=m2

(Eq 11)

Alternatively, the surface tension has been correlated to the heat vaporization (DHv) caused by the breaking of the interatomic bonds during evaporation in the liquid state (Ref 82). Usually, metals with large atomic volume have low energies of vaporization: g ¼ 1:8  109 ð
Hv =V 2=3 Þ

(Eq 12)

The surface tension for pure metals as a function of temperature (g f(T)) can also be calculated using Eq 13 and Table 7: gfðT Þ ¼ g0i þ cðT  Tm Þ

(Eq 13)

where g0 is the surface tension at the melting point, and c is dg/dT, Tm, and T > Tm; T and Tm are in degrees Kelvin. There are two definitive compendia of surface tension values by Keene: the first for pure metals (Ref 83) and the other for iron and its binary alloys (Ref 84). The review of elements

Table 6 Density at the melting point, dimensionless values for the parameters a and b, viscosity, and activation energy for viscous flow for selected elements Density, g/cm3 rL = a  b (T  Tm) (Ref 28) Element

Melting point, K

Ag Al Au B Be Bi Cd Ce Cr Co Cu Fe Hf La Li Mg Mn Mo Nb Ni Pb Pd Pt Re Sb Si Sn Ta Ti V W Zn Zr

1233.7 933.5 1336 2448 1550 544 593 1060 2148 1766 1356 1811 2216 1203 453.5 924 1525 2880 2741 1727 660.6 1825 2042 3431 904 1683 505 3250 1958 2185 3650 692 2123

Viscosity (Ref 13, 78, 79)

Measured (Ref 13) at Tm

a

b  104

h at Tm, mN  s/m2

ho mN  s/m2

E, kJ/mol

9.346 2.385 17.36 2.08 at 2346 K 1.690 10.068 8.020 6.685 6.28 7.760 8.000 7.015 11.10 5.955 0.525 1.590 5.730 9.35 7.830 7.905 10.678 10.490 19.00 18.80 6.483 2.524 7.000 15.00 4.110 5.700 17.60 6.575 5.800

9.329 2.378 17.346 ... 1.690 10.031 7.997 6.689 6.280 7.740 8.033 7.035 ... 5.950 0.5150 1.589 5.750 ... ... 7.890 10.587 18.909 18.909 ... 6.077 2.524 6.973 ... 4.140 5.36 ... 6.552 ...

10.51 3.111 17.020 ... 1.165 12.367 12.205 2.270 7.230 9.500 7.953 9.26 ... 2.370 1.201 2.658 9.300 ... ... 9.910 12.220 28.826 28.826 ... 6.486 3.487 7.125 ... 2.260 3.20 ... 9.502 ...

3.88 1.34 5.38 ... ... 1.85 2.28 2.88 ... 4.49 4.10 5.85 ... 2.45 0.55 1.32 ... ... ... 4.60 2.61 ... ... ... 1.48 0.94 2.00 ... 5.20 ... ... 3.85 8.0

0.4532 0.185 1.132 ... ... 0.4458 0.3001 ... ... 0.2550 0.3009 0.191 ... ... 0.1456 0.0245 ... ... ... 0.1663 0.4636 ... ... ... 0.0812 ... 0.5382 ... ... ... ... 0.4131 ...

22.2 15.4 15.9 ... ... 6.45 10.9 ... ... 44.4 30.5 51.5 ... ... 5.56 30.5 ... ... ... 50.2 8.61 ... ... ... 22 ... 5.44 ... ... ... ... 12.7 ...

Source: Ref 129

has recently been updated (Ref 85). Variations on reported surface tension of pure elements and alloys are expected because of the experimental techniques and the strong effect of surface-active impurities such as soluble oxygen, sulfur, and tellurium in liquid metals on their surface tension. Therefore, the levels of impurities in solution should be carefully controlled and taken into account in the determination and assessment of the surface tension of metallic alloys. It should be noted that in practice it is difficult to control soluble oxygen levels. Marangoni flows are those driven by surface tension gradients. In general, surface tension depends on both the temperature and chemical composition at the interface; consequently, Marangoni flows may be generated by gradients in either temperature or chemical concentration at an interface. Because the temperature coefficient for surface tension, dg/dT, can change sign from a negative to positive as the impurity concentration increases, the direction of flow in shallow pools of liquid (Marangoni flow) with uneven temperature distributions can be reversed with different materials with different surface-active element concentrations. These effects are established to have implications in melt processes such as the weld quality and flow during melt refining. Fifty ppm of either oxygen or sulfur cause a decrease of 25% in g and a change from negative (dg/dT) to positive

(dg/dT), which will reverse the direction of any Marangoni convection. Brooks and Quested (Ref 86) have recently reviewed both the surface tension (g) and dg/dT for a variety of ferritic and austenitic steels as a function of sulfur content. In general, (g) decreases with increasing sulfur levels, and dg/dT increases from negative to positive with increasing sulfur levels. For detailed compositions of the steels, the reader is referred to Ref 86. Thermodynamic models to calculate the surface tensions of liquid alloys have been developed (Ref 46, 87–90). These models were based on earlier work on surface tension prediction of binary (Ref 89) and ternary (Ref 90) solutions and on the Buttler equation (Ref 91). In order to calculate the surface tension of a liquid metal using the Buttler equation, the surface tensions and surface areas of the pure constituent elements and the excess Gibbs energy of the liquid metal must be known. The excess Gibbs energy is the same as that used for calculating the phase diagram and thermodynamic properties. In this thermodynamic approach, the description of the alloy system must be established before any property of the multicomponent alloy can be calculated. Model parameters for several alloy systems have been determined (Ref 92–103). Table 8 shows the surface tension values for some alloys of industrial interest (Ref 16).

474 / Modeling and Analysis of Casting Processes Table 7 Surface tension for pure liquid metals Element

Ag Al Au B Be Bi Cd Ce Cr Co Cu Fe Hf La Li Mg Mn Mo Nb Ni Pb Pd Pt Re Sb Si Sn Ta Ti V W Zn Zr

i0 at Tm mN/m

dg/dT, mN/m  K

903 914 1140 1070 1390 378 570 740 1700 1873 1285 1872 1630 720 395 559 1090 2250 1900 1778 468 1500 1800 2700 367 865 544 2150 1650 1950 2500 782 1480

0.16 0.35 0.52 ... 0.29 0.07 0.26 0.33 0.32 0.49 0.13 0.49 0.21 0.32 0.15 0.35 0.2 0.30 0.24 0.38 0.13 0.22 0.17 0.34 0.05 0.13 0.07 0.25 0.26 0.31 0.29 0.17 0.20

Source: Ref 13

Table 8 Surface tension for some industrial alloys Material(a)

Cu-Al (Al-bronze) Single-crystal CMSX-4 Hastelloy X Inconel 718 Zn-Al

Surface tension(b), mN/m

1240 1215 1850 1850 1880 1865 1882 1866 830 807

are identical to the units for diffusion coefficients and thermal diffusivity. Several methods exist to measure the viscosity of liquid metals (Ref 67, 104, 105). However, accurate and suitable methods to measure the viscosity of liquid metals and their alloys are restricted due to the relatively low viscosity, high liquidus temperatures, and chemical reactivity of melts. The reciprocal of the viscosity is known as the fluidity. The kinematic viscosity is the ratio of the viscosity to density (n = Z/r). This is an important parameter in fluid mechanics. The kinematic viscosity represents the transverse diffusion of momentum down a velocity gradient that is necessary to describe mold filling in a casting process. The Arrhenius equation is the most common form of representing the temperature dependence (Ref 104) of viscosity: Z ¼ A expðEv =RT Þ

(Eq 14)

where Ev is the activation energy for viscous flow, and R is the ideal gas constant (8.3144 J/K). Andrade (Ref 106) derived a semiempirical relation to determine the viscosity of elemental liquid metals at their melting temperatures. Andrade’s relationship is based on the quasicrystalline theory that assumes that the atoms in the liquid at the melting point are vibrating in random directions and periods, just as in the solid state. Modification to Andrade’s equation, based on the characteristics of atomic vibration frequency at the melting point, gives (Ref 104): Zm ¼ 1:7  107 r2=3 Tm 1=2 M 1=6 ðPa  sÞ

Temperature, K

1350 1473 1653 1873 1628 1773 1609 1773 660 773

(a) Chemistries given in Table 4. (b) Values depend on oxygen and sulfur contents. Source: Ref 16

Viscosity The viscosity is the resistance of the fluid to flow when subjected to an external shear force. The shear stress (t), or the force per unit area, causing a relative motion of two adjacent layers in a liquid is proportional to the velocity gradient (du/dy), which is normal to the direction of the applied force (t = Z du/dy), where the proportionality factor, Z, is termed the viscosity. This concept is known as Newton’s law of viscosity. Most liquid metals are believed to follow a Newtonian behavior. The unit of viscosity is called Poise (P) (1P = 1 dyne  s/cm2 = 1 g/cm  s = 1 mPa  s). The parameter (Z/r) is referred to as kinematic viscosity and has units (m2/s), which

(Eq 15)

where r is the density, M is the atomic mass, and Tm is the melting point in degrees Kelvin. The temperature dependence of the viscosity for most pure liquid metals can be expressed by: Z ¼ Zo expð2:65T 1:27 =RT Þ

ðmPa  sÞ

(Eq 16)

At T = Tm and Z = Zm, then the value for Zo is determined as follows: Zo ¼ Zm =½expð2:65T 1:27 =RT Þ

(Eq 17)

With the exception of silicon, manganese, chromium, hafnium, palladium, and vanadium, these equations allow a reasonable prediction of the viscosity as a function of temperature for most pure liquid metals. Further modifications of Andrade’s equation have been developed in an attempt to estimate more accurately the viscosity of liquid metals (Ref 107–111). An empirical relationship between viscosity and surface tension for pure liquid metals has been developed (Ref 111): Zm ¼ 2:81  104 ½ðMgm Þ1=2 =V 1=3

(Eq 18)

where M is the atomic mass, V is the atomic volume, and gm is the surface tension at the melting point.

Table 6 shows the viscosity of liquid metals at their melting temperature and their activation energy for viscous flow (Ref 78, 79), and Table 9 shows the viscosities of some metals and some alloys of commercial interest (Ref 16).

Electrical and Thermal Conductivity Thermal and electrical conductivities are intrinsic properties of materials, and they reflect the relative ease or difficulty of energy transfer through the material. Metals are well known for their high electrical conductivity, which arises from the easy migration of electrons through the crystal lattice. The conductivity on melting for most metals decreases markedly due to the exceptional disorder of the liquid state. Generally, the electrical resistivity of some liquid metals just above their melting points is approximately 1.5 to 2.3 greater than that of the solids just below their melting temperature (Ref 33). Examples of exceptions are iron, cobalt, and nickel. Mott (Ref 112) derived an empirical equation to estimate the ratio of liquid/solid electrical conductivity (se,l/se,s) at the melting point of the pure metal. The Mott equation is expressed as follows: lnðssol =sliq Þ ¼ Cð
Hm =Tm Þ ¼ Cð
Sm Þ

(Eq 19)

where DHm is the enthalpy of fusion in KJ/mol, Tm is the melting temperature in degrees Kelvin, DSm is the entropy of fusion, and C is a constant. With the exception of a few metals (e.g., antimony, bismuth, gallium, mercury, and tin), the simple relationship proposed by Mott is in good agreement with experimental measurements. Reviews of the electrical and thermal conductivity theories can be found in the literature (Ref 113–119). Of particular interest to high-temperature technology, such as casting processes, is the theoretical relation between the thermal (k) and electrical (s) conductivities known as the Wiedman-Franz law (WFL) and the constant of proportionality known as the Lorentz number, L (Ref 120). The WFL relationship (Eq 20) appears to hold reasonably well for pure metals around the melting point, but large departures can occur at lower temperatures in the solid: L ¼ ðk=sT Þ ¼ 2:445  108

W  =K2

(Eq 20)

Convective flow that usually occurs in the liquid should not have any effect on the electrical conductivity. Therefore, it should be possible to estimate the thermal conductivity of liquid alloys from the electrical conductivity values. The electrical conductivity for molten binary alloy has been estimated using Eq 21 (Ref 33): sT 1 ¼ s1 x1 þ s2 x2 þ s3 x3 þ . . .

(Eq 21)

A negative departure of less than 10% from linearity has been observed for most alloys. Similarly, the temperature dependence of the

Thermophysical Properties / 475 Table 9

Density, thermal conductivity, surface tension, and viscosity for some commercial alloys

Aluminum alloys A319

273 K 2.75

LM25 (A356)(d)

2.68

2003(e)

...

3004

2.72

2024-T4

2.785

6061-T6

2.705

7075-T6

2.805

Copper alloys Cu-Al (Al-bronze)

7.262

Brass(e) 70Cu-30Zn

...

Brass(e) 60Cu-40Zn

...

Copper-nickel(e) 70Cu-30Ni-Fe-Mn

...

Iron and steel alloys Steel AISI 1008(e)

Thermal conductivity(b), W/m  K (temp. range in K)

Density, g/cm3 T in K

Material(a)

7.86 ...

Estimated viscosity(c), mPa  s

rs = 2.753  22.3  102 (T  298) ks = 76.64 + 0.2633T  2  104T2 rl = 2.492  27.0  102 (T  894) kl = 70 kl = 71 rs = 2.68  21.2  102 (T  298) ks = 149.7 + 0.0809T  1  104T2 rl = 2.401  26.4  102 (T  887) kl = 65.8 kl = 70 ... ks = 192.5 kl = 818.7  0.808T kl = 86.3 rs = 2.72  23.4  102 (T  298) ks = 124.7 + 0.56T + 1  105T2 rl = 2.4  27.0  102 (T  929) kl = 61 rs = 2.785  21.3  102 (T  298) ks = 188 rl = 2.50  28.0  102 (T  905) kl = 85.5 rs = 2.705  20.1  102 (T  298) ks = 7.62 + 0.995T  17  104T2 + 1  106T3 rl = 2.415  28.0  102 (T  915) ks = 66.5 kl = 90 rs = 2.805  22.4  102 (T  273) ks = 196 rl = 2.50  28.0  102 (T  901) ks = 193 kl = 85

(T: 298–773) (T: 894) (T: 1073) (T: 298–840) (T: 887) (T: 1073) (T: 573–775) (T: 775–911) (T: 1023) (T: 298–890) (T 929) (T: 473–573) (T: 811–905) (T: 298–773) (T: 873) (T: 915) (T: 673–773) (T: 805) (T: 901)

1.3 at 894 K 1.1 at 1073 K

rs = 7.262  48.6  102 (T  298) ks = 7.925 + 0.1375T  6  105T2 rl = 6.425  65.0  102 (T  1350) ks = 42 kl = 27 ... ks = 140.62 + 112.14  104T ks/l = 2430.3  191.61  102T kl = 45.43 + 26  103T ... ks = 182.95 + 366.1  104T ks/l = 16479.5  13.93 T kl = 39.724 + 26  103T ... ks = 16.041 + 438.9  104T ks/l = 796.018  502.63  103T kl = 3.8 + 26  103T

(T: 373–773) (T: 1313) (T: 1373) (T: 460–1188) (T: 1188–1228) (T 1228) (T: 620–1173) (T: 1173–1178) (T 1178) (T  1443) (T: 1443–1513) (T 1513)

6.3 at 1350 K 5.2 at 1473 K

(T: 1122–1768) (T 1768) (T: 1073–1693) (T: 1693–1793) (T 1793) (T: 298–1633) (T: 1644–1672) (T 1793) (T: 298–1573) (T: 1644–1672) (T 1672) (T  1727) (T: 1727–1783) (T 1783) (T: 1373) (T: 1451) (T: 1673) (T: 1353) (T: 1463) (T: 1673)

...

rl = 7.0 at 1823 K ...

1% Cr steel(e) Fe-0.3C-0.69Mn-0.2Si-1.1Cr0.7Ni 304 stainless steel

8.02

rs = 8.02  50.1  102 (T  298) rl = 6.90  80.0  102 (T  Tm)

316 stainless steel

7.95

rs = 7.95  50.1  102 (T  298) rl = 6.881  77.0  102 (T  Tm)

420 stainless steel(e)

7.7

rl = 7.0 at 1823 K

Ductile iron

7.3

Gray cast iron

7.2

rs = 7.06 at 1373 K rl = 6.62 at 1451 K rl = 6.586 at 1573 C rs = 6.992 at 1273 K r = 6.964 at 1353 K rl = 7.495  77.0  102T

Magnesium alloys AZ31B(e)

...

AZ91B(e)

...

KIA(e)

...

ZK51A(e)

...

HM11A(e)

...

EZ33

1.8

Nickel alloys Single crystal CMSX-4

8.7

ks = 13.58 + 11.3  103T kl = 280.72  14.0  103T ks = 14.53 + 105.0  104T ks/l = 91.74  351.0  104T kl = 7.85 + 116.8  104T ks = 10.33 + 15.4  103T  7.0  107T2 ks/l = 355.93  196.8  103T kl = 6.6 + 12.14  103T ks = 6.31 + 27.2  103T  7.0  106T2 ks/l = 355.93  196.8  103T kl = 6.6 + 121.4  104T ks = 20 + 61.5  104T ks/l = 133.4  594.9  104T kl = 6.5 + 116.8  104T ks = 31 ks = 28 kl = 29 ks = 29 ks = 26 kl = 28

1.38 at 887 K 1.1 at 1073 K ... 1.15 at 929 1.05 at 973 1.30 at 905 1.1 at 1073 1.15 at 915 1.0 at 1073

K K K K K K

1.3 at 901 K 1.1 at 1073 K

... ... ...

... 8.0 at 1727 K

8.0 at 1723 K ... 14.0 at 1451 K 11.5 at 1573 K 9.0 at 1673 K 14.3 at 1463 K 14.0 at 1473 K 10.5 at 1573 K

ks = 67.12 + 655.7  104T ks/l = 830.9  844.8  103T kl = 3.05 + 70.0  103T ... ks = 18.27 + 112.11  103T ks/l = 372  364.6  103T kl =  5.63 + 7.0  102T ... ks = 127.16 + 142.9  104T kl = 11.66 + 7.0  102T ... ks = 71.96 + 154.35  103T  93.8  106T2 ks/l = 688  672.2  103T kl = 9.62 + 70.0  103T ... ks = 73.35 + 133.43  103T  73.9  106T2 ks/l = 2887  305.0  102T kl = 8.0 + 70.0  103T rs = 1.8  14.3  102 (T  298) ks = 156 rl = 1.663  27.0  102 (T  913) kl = 91

(T: 499–839) (T: 839–905) (T 905) (T  742) (T: 742–869) (T 869) (T: 520–922) (T 923) (T  822) (T: 822–914) (T 914) (T  903) (T: 903–923) (T 923) (T: 818) (T: 913)

1.5 at 913 K 1.4 at 973 K

rs = 8.7  45.8  102 (T  298) ks = 27.2 rl = 7.754  90.0  102 (T  1653) kl = 25.6

(T: 1573)(f) (T: 1653–1673)

6.7 at 1653 K 5.3 at 1773 K

...

... ... ... ... ...

(continued) (a) Chemistry of the alloys is given in Table 4. (b) The polynomial equations that represent the thermal conductivity for most of the alloy systems shown in this table are estimated from the data recommended in Ref 16. Other data indicated as (e) have been obtained from Ref 14. For further analysis of the data it is recommended to consult the original source cited in the given literature. (c) Estimated values of viscosity may vary from þ10 to þ30%   (Ref 16). (d) Aluminum alloy with the British designation LM25 is equivalent to the Aluminum Association designation of A356. (e) Data from Ref 14. (f) Based on estimated Cp value from lS = arCp (Ref 16)

476 / Modeling and Analysis of Casting Processes Table 9

(continued) Thermal conductivity(b), W/m  K (temp. range in K)

Density, g/cm3 T in K

Material(a)

rs = 8.24  38.1  102 (T  298) rl = 7.42  83.0  102 (T  1628) rs = 8.19  39.2  102 (T  298) rl = 7.40  88.0  102 (T  1609)

ks = 3.36 + 17.3  103T + 2.0  106T2 kl = 29.0 ks = 39.73 + 32.4  103T + 2.0  105T2 kl = 29.6

Estimated viscosity(c), mPa  s

Hastelloy X

8.24

Inconel 718

8.19

Titanium alloys Ti-6Al-4V

4.42

rs = 4.42  15.4  102 (T  298) ks = 0.797 + 18.2  103T  2.0  106T2 rl = 3.92  68.0  102 (T  1923) kl = 33.4 kl = 34.6

(T: 1268–1923) (T: 1923) (T: 1973)

3.25 at 1923 K 2.66 at 2073 K

Zinc alloys Zn-Al

6.7

rs = 6.7  60.3  102 (T  298) ks = 98.0 rl = 6.142  97.7  102 (T  660) kl = 40.0

(T: 630) (T: 660)

3.5 at 673 K 2.6 at 773 K

(T: (T: (T: (T:

1073–1533) 1428–1773) 1173–1443) 1609–1873)

7.5 at 1628 K 5.5 at 1773 K 7.2 at 1609 K 5.31 at 1773 K

(a) Chemistry of the alloys is given in Table 4. (b) The polynomial equations that represent the thermal conductivity for most of the alloy systems shown in this table are estimated from the data recommended in Ref 16. Other data indicated as (e) have been obtained from Ref 14. For further analysis of the data, it is recommended to consult the original source cited in the given literature. (c) Estimated values of viscosity may vary from þ10 to þ30%   (Ref 16). (d) Aluminum alloy with the British designation LM25 is equivalent to the Aluminum Association designation of A356. (e) Data from Ref 14. (f) Based on estimated Cp value from lS = arCp (Ref 16)

Table 10

Electrical resistivity (re) and thermal conductivity of solid and liquid metals Electrical resistivity (Ref 13, 33)

Element

Ag Al Au B Be Bi Cd Ce Cr Co Cu Fe Hf La Li Mg Mn Mo Nb Ni Pb Pd Pt Re Sb Si Sn Ta Ti V W Zn Zr

T m, K

1233.7 931 1336 2448 1550 544 593 1060 2148 1766 1356 1809 2216 1203 453.5 924 1525 2880 2741 1727 660.6 1825 2042 3431 904 1683 505 3250 1958 2185 3650 692 2123

re,s, mV  cm

8.2 10.9 13.68 ... ... ... 17.1 ... ... 97 9.4 122 ... ... ... 15.4 66 ... ... 65.4 49.0 ... ... ... 183 ... 22.8 ... ... ... ... 16.7 ...

re,l, mV  cm

17.2 24.2 31.2 210.0 45.0 129.0 33.7 126.8 31.6 102 20.0 110 218.0 138 240.0 27.4 40 60.5 105.0 85.0 95.0 ... 73.0 145 113.5 75 48.0 118.0 172 71.0 127 37.4 153

electrical conductivity can be estimated using Eq 22: sT ¼ sT 1 f1 þ ðds=dT Þalloy g

(Eq 22)

Thermal conductivity (Ref 122)

re,l = aT + b

ksm

re,l/re,s

a, mV  cm/K

b, mV  cm

T range Tm to T, K

2.09 2.20 2.28 ... ... ... 1.97 ... ... 1.05 2.1 0.9 ... ... ... 1.78 0.61 ... ... 1.3 1.94 ... ... ... 0.61 ... 2.10 ... ... ... ... 2.24 ...

0.0090 0.0145 0.0140 ... ... ... Not linear ... ... 0.0612 0.0102 0.033 ... ... ... 0.005 No data ... ... 0.0127 0.0479 ... ... ... 0.270 ... 0.0249 ... ... ... ... Not linear ...

6.2 10.7 12.5 ... ... ... ... ... ... 6.0 6.2 50 ... ... ... 22.9 ... ... ... 63 66.6 ... ... ... 87.9 ... 35.4 ... ... ... ... ... ...

1473 1473 1473 ... ... ... ... ... ... 1973 1873 1973 ... ... ... 1173 ... ... ... 1973 1273 ... ... ... 1273 ... 1473 ... ... ... ... ... ...

þ x3 ðds3 =dT Þ þ . . . (Eq 23)

Once sT is calculated at a given T, then the thermal conductivity is calculated using the WLF equation (Eq 20). Excellent compilations of conductivity properties of pure elements can be found in the literature (Ref 13, 121, 122). Table 10 shows the

362 211 247 ... ... 7.6 37  þ3 21 45 45 330 34 39 ... 71 145 24 87 78 70 30 99 80 65  þ5 17 25 59.5 70 31 51 95 90 38

klm

175 91 105 ... ... 12 90 22 35 36 163 33 ... 17 43 79 22 72 66 60 15 87 53 55 25 56 27 58 31 43.5 63 50 36.5

D Sm, J/mol  K

9.15 10.71 9.39 ... ... 20.75 10.42 2.99 9.63 9.16 9.77 7.62 10.9 5.19 6.6 9.18 8.3 12.95 10.90 10.11 7.95 9.15 10.86 17.5 22 29.8 13.9 11.1 7.28 9.85 14.2 10.6 9.87

K

0.0794 0.0785 0.0911 ... ... 0.022 ... ... 0.0261 0.0243 0.0722 0.0039 ... ... 0.076 0.066 0.0102 ... ... 0.0152 0.0872 ... 0.0392 ... ... ... 0.0567 0.0169 0.0 0.0161 0.029 0.055 0.0041

electrical resistivity for some solid and liquid metals at the melting point. The resistivity data for the liquid, re,l, from the melting point to a given temperature in Table 9 can be calculated by the expression:

of the solid, ksm, and liquid, klm, at the melting point:

re;l ¼ aT þ b

where K is a constant. This equation would be useful to estimate the thermal conductivity of a liquid alloy by determining the thermal conductivity (or electrical conductivity) at the melting point. The only limitation is the constant K, which would not have a uniform value for all metals and alloys (Ref 122). Table 10 shows the thermal conductivity for pure metals in the solid and liquid state at the melting point as well as the estimation of the

where: ds=dT ¼ x1 ðds1 =dT Þ þ x2 ðds2 =dT Þ

W/m  K

(Eq 24)

The values for constants a and b for the various metals are given in Table 10. Note also that the electrical resistivity data given in Table 10 are for bulk metals and may not be applicable to thin films. Mills and co-workers (Ref 122) combined Mott (Eq 19) and WFL (Eq 20) to establish a relationship between the thermal conductivity

  ksm ¼ K
Sm ln klm

(Eq 25)

Thermophysical Properties / 477 Table 11 Mold material

Heat capacity and thermal conductivity of some mold materials Dry density, g/cm3 Specific heat capacity, kJ/Kg  K

Silica molding sands 20–30 mesh 50–70 mesh 80–120 mesh Silica sand (22 mesh) + 7% bentonite + 4% bentonite Olivine sand + 4% bentonite Zircon sand + 4% bentonite Chromite sand 3.9% bentonite Graphite (chill foundry grade)

Investment casting Zircon-30% alumina-20% silica

Thermal conductivity, W/m1  K

1.730 1.634 1.458

Cp = 0.5472  1.147  103T  5.401  107T Cp = 1.066 + 8.676  105T

(T < 1033) (T > 1033)

k = 0.604  0.767  103T + 0.795  106T2 k = 0.676  0.793  103T + 0.556  106T2

(T < 1033 (T > 1033)

1.520 1.60 1.83 2.125 2.780 2.96

Cp = 0.4071(T  273)0.154

(T  1600)

Cp = 0.3891(T  273)

0.162

(T  1300)

Cp = 0.2519(T  273)

0.946  0.903  103T + 0.564  106T2 1.26  0.169  102T + 0.105  105T2 0.713 + 0.349  104T 1.82  1.88  102T + 0.10  105T2 1.19  0.948  103T + 0.608  106T2 1.82  0.176  102T + 0.984  106T2

(T < 1500) (T < 1500)

0.170

k k k k k k

2.75 1.922

Cp Cp Cp Cp

= = = =

0.318(T  273) 0.11511 + 2.8168  103T 0.6484 + 1.305  103T 1.3596 + 4.2797  104T 0.158

...

2.48–2.54

(T  1300) (T  1300) (T < 505) (505–811) (T > 811)

= = = = = =

3

(T < 1500)

6 2

k = 941  0.753  10 T + 0.561  10 T k = 135.99  8.378  102T 2

k = 103.415  4.647  10 T 4

(T < 1300)

k = 3.03  3.98  10 T + 508 T

(T < 873) (T > 873)

1

(T: 375–1825)

Source: Ref 14

constant K based on the WFL (Ref 122). The experimental errors in the measurement of the thermal conductivity of the liquid can be larger than 5% for the thermal conductivity and  þ3% for electrical conductivity (Ref 122). Calculating thermal conductivity from measured thermal diffusivity values may reduce some of the experimental errors. Thermal diffusivity is the ability of a material to self-diffuse thermal energy. This is determined by combining the material ability to conduct heat and its specific heat capacity. The density is involved due to the given specific heat capacity in units of heat per unit mass, while conductivity relates to the volume of material. Thus, the thermal conductivity (k) and thermal diffusivity (a) that measure the heat flow within materials are related by their specific heat capacity (Cp) and density (r) by the relationship: k ¼ aCp r

(Eq 26)

Table 9 shows some of the data available in the literature for ferrous and nonferrous alloys (Ref 14). In casting processes, the need for thermal conductivity data for mold materials becomes more crucial. Mold materials are usually bulk, porous, complex sand-polymer mixtures and ceramic materials, and their thermal conductivity is certainly different from the intrinsic thermal conductivity of the base material. Methods to estimate the thermal conductivity of these materials can be found in the literature (Ref 123). Table 11 shows the thermal conductivity for some mold materials (Ref 14).

Emissivity Thermal radiation is an important heat-transfer phenomenon in casting processes, since heat losses during pouring of the molten metal and heat radiation from the mold contribute to the overall heat balance during solidification of the cast product. The thermal radiation

is defined by Plank’s law. The integrated form of Plank’s equation gives the total emissive power of a body (e). This is known as the Stefan-Boltzmann equation and is represented by: e ¼ esT 4

(Eq 27)

where e is the emissivity, and s is the StefanBoltzmann constant. The Stefan-Boltzmann constant is given by: s¼

25 4 ¼ 5:67  108 J=m2  s  K4 15c2 h

(Eq 28)

where k and h are the Boltzmann’s and Plank’s constants, respectively, and c is the speed of light. Because emissivity data throughout the wavelength spectrum are not available for most metallic materials, the following empirical equation has been employed to represent the total emissivity as a function of temperature (Ref 124): et ¼ K1

pffiffiffiffiffiffiffiffiffiffi ðrT Þ  K2 rT

(Eq 29)

where K1 and K2 are constants, K1 = 5.736, K2 = 1.769, and r is the electrical resistivity in O  m. Equation 29 is in reasonable agreement with experimental data, and it shows an increase of et with T. However, deviations for some materials at high temperatures can be expected because of the spectral emissivity that changes with the wavelength and direction of emission. Nevertheless, in most practical situations, including in casting processes, an average emissivity for all directions and wavelengths is used. The emissivity values of metals or other nonmetallic materials depend on the nature of the surfaces, such as the degree of oxidation, surface finish, and the grain size. Table 12 gives some total normal emissivity for some pristine and oxidized metals, while Table 13 gives the total emissivity for some alloys and refractory materials (Ref 125).

Additional emissivity data for various materials have been compiled and can be found in the literature (Ref 5, 124, 126–128). The data quoted in Tables 12 and 13 provide a guideline and must be used with discretion since, in practical applications, the values of emissivity may change considerably with oxidation and roughening of the surfaces. Therefore, it is important that the total emissivity should be determined for the actual surface conditions of the materials in question.

Typical Thermophysical Properties Ranges of Some Cast Alloys Tables 14 to 17 show the typical range of thermophysical properties for aluminum, magnesium, cast iron, and nickel alloys. Unfortunately, numerical values of those thermophysical properties in the mushy zone or even in the liquid state are not available. However, one could infer the expected trends as noted in the tables.

Summary This article introduces the methods, concepts, and typical values for the main types of thermophysical data needed for processing modeling of castings. The following conclusions can be drawn:
Throughout this survey, values have been

quoted. These should be used with discretion and taken as representative values for materials and not necessarily values for a particular system.
The properties considered are specific heat capacity, enthalpy of melting, solidus and liquidus temperatures, coefficient of thermal expansion, density, surface tension, viscosity, electrical and thermal conductivity, and emissivity appropriate to metals, alloys, and molds.
The measurement of these properties presents the experimentalist with several challenges

478 / Modeling and Analysis of Casting Processes Table 12 Metal

Ag Al Be Cr Co Cu: solid Cu: liquid Hf Fe Mo Ni Nb Pd Pt Rh Ta Ti(a) W

Alloys Cast iron

Total normal emissivity of pristine and oxidized metals Pristine

Temperature, K

0.02–0.03 0.064 0.87 0.11–0.14 0.34–0.46 0.02 0.12 0.32 0.24 0.27 0.14–0.22 0.18 0.15 0.16 0.09 0.18 0.47 0.17 0.18 0.23 ... 0.29

773 773 1473 773 773 773 1473 1873 1273 1873 1273 1873 1473 1473 1673 1873 1673 1673 1873 2273 ... 1873

Oxidized

Temperature, K

... 0.19 ... 0.14–0.34 ... 0.24

... 873 ... 873 ... 1073

... 0.57 0.84 0.49–0.71 0.74 0.124 ... ... 0.42 ... ...

... 873 673 1073 1073 1273 ... ... 873 ... ...

... 0.78

... 873

(a) Spectral normal emissivity at 65 mm wavelength. Source: Ref 124

Table 13

Total normal emissivity of various alloys and refractory materials

Alloys

Commercial aluminum Oxidized Cast iron: Solid Liquid Commercial copper: heated at 872 K Liquid Steel: plate rough Steel liquid Refractory materials Alumina: 85–99.5Al2O3-0–12SiO2-0–1Fe2O3 Alumina-silica: 58–80Al2O3-16–18SiO2-0.4Fe2O3 Fireclay brick Carbon rough plate Magnesite brick Quartz, fused Zirconium silicate

«

Temperature, K

0.09 0.11–0.19 0.60–0.70 0.29 0.57 0.16–0.13 0.94–0.97 0.42–0.53

373 472–872 1155–1261 1873 472–872 1349–1550 273–644 1772–1922

0.5–0.18 0.61–0.43 0.75 0.77–0.72 0.38 0.93 0.92–0.80 0.80–0.52

1283–1839 1283–1839 1273 373–773 1273 294 510–772 772–1105

Note: A linear interpolation of the emissivity values with temperature can be done when the emissivity values are separated by a dash. It should be noted that some materials (such as alumina) are semitransparent, and their measured emissivity depends on the thickness of the sample. Source: Ref 125

Table 14

Range of thermophysical and mechanical properties of cast aluminum alloys

Property

Thermal conductivity, W/K  m Heat capacity, J/kg  K Density, kg/m3 Viscosity, mm2/s Young’s modulus, GPa Thermal expansion, mm/m   C Heat of fusion, kJ/kg

At room temperature

At solidus temperature

100–180 880–920 2600–2800

150–210 1100–1200 2400–2600

68–75 20–24

40–50

In the mushy zone

In the liquid range

Dropping 60–80 Almost constant 1100–1200 Dropping 2200–2400 Sharply decreasing 0.4–0.5 Dropping to zero 25–30 Sharply increasing 400–500

Source: Ref 26

associated with high temperatures and the reactivity of the materials with containers and atmospheres.
The prediction of properties has advanced in recent years, but some care is needed in validating the data.
Before commissioning expensive work to measure properties, the modeler is advised to check the sensitivity of the

required predictions to changes in the input thermophysical data. This will enable measurement efforts to concentrate on the critical data.
Among properties not considered are: a. Chemical diffusion coefficients and partition coefficients, which can strongly affect the chemistry at the microstructural scale

b. Mechanical properties of the metal close to the solidus temperature where it is very weak, with implications to the strength such as susceptibility to hot tearing and cracking c. Mechanical properties of molds at appropriate temperatures during casting REFERENCES 1. J.J. Valencia, Symposium on Thermophysical Properties: Metalworking Industry Needs and Resource, Concurrent Technologies Corporation, Oct 22–23, 1996 2. A. Ludwig, P. Quested, and G. Neuer, How to Find Thermophysical Material Property Data for Casting Simulations, Adv. Eng. Mater., Vol 3, 2001, p 11–14 3. Y.S. Touloukian, C.Y. Ho, et al., Thermophysical Properties Research Center Data Book, Purdue University, 1960–1966 4. Y.S. Touloukian, Ed., Thermophysical Properties of High Temperature Solid Materials, MacMillan Co., 1967 5. Y.S. Touloukian, J.K. Gerritsen, and N.Y. Moore, Ed., Thermophysical Properties Literature Retrieval Guide, Basic Edition, Plenum Press, 1967 6. Y.S. Touloukian and C.Y. Ho, Ed., Thermophysical Properties of Matter—The TPRC Data Series, IFA/Plenum Data Co., 1970–1979 7. Y.S. Touloukian, J.K. Gerritsen, and W.H. Shafer, Ed., Thermophysical Properties Literature Retrieval Guide, Supplement I (1964–1970), IFI/Plenum Data Co., 1973 8. Y.S. Touloukian and C.Y. Ho, Ed., Thermophysical Properties of Selected Aerospace Materials, Part I: Thermal Radiative Properties; Part II: Thermophysical Properties of Seven Materials, Purdue University, TEPIAC/CINDAS, Part I, 1976; Part II, 1977 9. J.K. Gerritsen, V. Ramdas, and T.M. Putnam, Ed., Thermophysical Properties Literature Retrieval Guide, Supplement II (1971–1977), IFI/Plenum Data Co., 1979 10. J.F. Chaney, T.M. Putnam, C.R. Rodriguez, and M.H. Wu, Ed., Thermophysical Properties Literature Retrieval Guide (1900–1980), IFI/Plenum Data Co., 1981 11. Y.S. Touloukian and C.Y. Ho, Ed., McGraw-Hill/CINDAS Data Series on Materials Properties, McGraw-Hill Book Co., 1981 12. Thermophysical Properties of Materials: Computer-Readable Bibliographic Files, Computer Magnetic Tapes, TEPIAC/CINDAS, 1981 13. C.J. Smithells, General Physical Properties, Metals Reference Book, 7th ed., E.A. Brandes and G.B. Brook, Ed., Butterworth-Heinemann, 1992, p 14–1 14. R.D. Pehlke, A. Jeyarajan, and H. Wada, Summary of Thermal Properties for Casting Alloys and Mold Materials, Grant DAR78–26171, The University of

Thermophysical Properties / 479 Table 15

Range of thermophysical and mechanical properties of cast magnesium alloys

Property

Thermal conductivity, W/K  m Heat capacity, J/kg  K Density, kg/m3 Viscosity, mm2/s Young’s modulus, GPa Thermal expansion, mm/m   C Heat of fusion, kJ/kg

At room temperature

At solidus temperature

50–85 1000–1050 1750–1850

80–120 1150–1250 1650–1750

In the mushy zone

In the liquid range

Dropping 50–70 Almost constant 1200–1350 Dropping 1550–1650 Sharply decreasing 0.6–0.7 30–35 Dropping to zero 30–34 Sharply increasing 280–380

42–47 24–26

26.

Source: Ref 26

27. Table 16

Range of thermophysical and mechanical properties of cast iron

Property

Thermal conductivity, W/K  m Heat capacity, J/kg  K Density, kg/m3 Viscosity, mm2/s Young’s modulus, GPa Thermal expansion, mm/m   C Heat of fusion, kJ/kg

At room temperature

At solidus temperature

In the mushy zone

In the liquid range

30–50 460–700 6900–7400

25–30 850–1050 6750–7350 ... 60–100 17–22 ...

Almost constant Dropping Shrinking/expanding Sharply decreasing Dropping to zero Sharply increasing 200–250

25–35 800–950 6700–7300 0.5–0.8 ... ... ...

80–160 11–14

28. 29.

Source: Ref 26

Table 17

Range of thermophysical and mechanical properties of nickel-base superalloys

Property

Thermal conductivity, W/K  m Heat capacity, J/kg  K Density, kg/m3 Viscosity, mm2/s Young’s modulus, GPa Thermal expansion, mm/m   C

At room temperature

At solidus temperature

8–14 420–470 8200–8700

28–32 620–700 7500–8000

180–220 12–13

100–120 12–13

In the mushy zone

In the liquid range

Slightly dropping Rising Dropping Sharply decreasing

25–35 700–800 7000–7500 0.6–1.0

Dropping to zero 20–23 Sharply 200–220 increasing

...

30.

31.

Heat of fusion, kJ/kg Source: Ref 26

32.

15. 16.

17. 18.

19.

20.

Michigan—National Science Foundation, Applied Research Division, Dec 1982 Thermal Properties of Metals, ASM Ready Reference, F. Cverna, Ed., ASM International, 2002 K.C. Mills, Recommended Values of Thermophysical Properties for Selected Commercial Alloys, National Physical Laboratory and ASM International, Woodhead Publishing Limited, Cambridge, England, 2002 J. Miettien, Metall. Trans. A, Vol 23, 1992, p 1155–1170 J. Miettien and S. Louhenkilpi, Calculation of Thermophysical Properties of Carbon and Low Alloy Steels for Modeling of Solidification Processes, Metall. Trans. B, Vol 25, 1994, p 909–916 Y.A. Chang, A Thermodynamic Approach to Obtain Materials Properties for Engineering Applications, Proceedings of a Workshop on the Thermophysical Properties of Molten Materials, Oct 20–23, 1992 (Cleveland, OH), NASA Lewis Research Center, p 177–201 K.C. Mills, A.P. Day, and P.N. Quested, Estimating the Thermophysical Properties of Commercial Alloys, Proceedings

21.

22.

23.

24.

25.

of Nottingham Univ.-Osaka Univ. Joint Symposium, Sept 1995 (Nottingham) N. Saunders, A.P. Midwnik, and J.-Ph. Schille´, Modelling of the Thermophysical and Physical Properties for Solidification of Ni-Based Superalloys, Proceedings of the 2003 International Symposium on Liquid Metal Processing and Casting, P. Lee et al., Ed., p 253–260 J.A.J. Robinson, A.T. Dinsdale, and P.N. Quested, The Prediction of Thermophysical Properties of Steels and Slags, Second International Congress on the Science and Technology of Steelmaking, University of Wales, Swansea, April 2001 X.L. Yang, P.D. Lee, R.F. Brooks, and R. Wunderlich, The Sensitivity of Investment Casting Simulations to the Accuracy of Thermophysical Property Values, Superalloys 2004, K.A. Green et al., Ed., TMS, 2004, p 951–958 K.C. Mills, B.J. Monaghan, R.F. Brooks, and P.N. Quested, Some Cautionary Notes on the Measurement and Use of Thermophysical Property Data, High Temp.— High Press., Vol 34, 2002, p 253–264 W.J. Boettinger, U.R. Kattner, K.-W. Moon, and J.H. Perepezko, “DTA and

33. 34.

35.

36.

37.

38.

Heat-Flux DSC Measurements of Alloy Melting and Freezing,” NIST Special Publication 960–15, 2006 E. Kaschnitz, A. Bu¨hrig-Polaczek, and G. Pottlacher, “Thermophysical Properties of Casting Materials,” World Foundrymen Organization—International Commission 3.3, Casting Simulation—Background and Examples from Europe and USA, WFO, 2002 K.D. Maglic, A. Cezairliyan, and V.E. Peletsky, Ed., Survey of Measurement Techniques, Vol 1, Compendium of Thermophysical Property Measurement Methods, Plenum Press, 1984 A.F. Crawley, Density of Liquid Metals and Alloys, Int. Metall. Rev., Review 180, Vol 19, 1974, p 32–48 E.E. Shpil’rain, K.A. Yakimovich, and A.G. Mozgovoi, Apparatus for Continuous Measurement of Temperature Dependence of Density of Molten Metals by the Method of a Suspended Pycnometer at High Temperatures and Pressures, Recommended Measurement Techniques and Practices, Vol 2, Compendium of Thermophysical Property Measurement Methods, Plenum Press, 1992, p 601–624 S.D. Mark and S.D. Emanuelson, A Thermal Expansion Apparatus with a Silicon Carbide Dilatometer for Temperatures to 1500  C, Ceram. Bull., Vol 37 (No. 4), 1958, p 193–196 L.D. Lucas, Physicochemical Measurements in Metals Research, Part 2, R.A. Rapp, Ed., Interscience, New York, 1970, p 219 P. Parlouer, Calorimetry and Dilatometry at Very High Temperatures. Recent Developments in Instrumentation and Applications, Rev. Int. Hautes Temp. Re´fract., Vol 28, 1992–1993, p 101–117 T. Iida and R.I.L. Guthrie, The Physical Properties of Liquid Metals, Clarendon Press, 1988, p 188 W.K. Rhim and A.J. Rulison, Measuring Surface Tension and Viscosity of a Levitated Drop, Report of National Aeronautics and Space Administration Contract NAS 7–918, 1996, p 1–6 I. Egry, G. Lohofer, P. Neuhaus, and S. Sauerland, Surface Tension Measurements of Liquid Metals Using Levitation, Microgravity, and Image Processing, Int. J. Thermophys., Vol 13 (No. 1), 1992, p 65–74 Y. Bayazitoglu and G.F. Mitchell, Experiments in Acoustic Levitation: Surface Tension Measurements of Deformed Droplets, J. Thermophys. Heat Transf., Vol 9 (No. 4), 1995, p 694–701 W.J. Parker, R.J. Jenkins, C.P. Butler, and G.L. Abbott, Flash Method of Determining Thermal Diffusivity, Heat Capacity and Thermal Conductivity, J. Appl. Phys., Vol 32 (No. 9), 1961, p 1679–1684 R.D. Cowan, Pulse Method of Measuring Thermal Diffusivity at High Temperatures,

480 / Modeling and Analysis of Casting Processes

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49. 50. 51. 52. 53. 54.

J. Appl. Phys., Vol 34 (No. 4), 1963, p 926–927 J.T. Schriempf, A Laser Flash Technique for Determining Thermal Diffusivity of Liquid Metals at Elevated Temperatures, High Temp.—High Press., Vol 4, 1972, p 411–416 K.D. Maglic and R.E. Taylor, The Apparatus for Thermal Diffusivity Measurement by the Laser Pulse Method, Compendium of Thermophysical Property Measurement Methods, Vol 2, 1992, p 281–314 R.E. Taylor and K.D. Maglic, Pulse Method for Thermal Diffusivity Measurement, Survey of Measurement Techniques, Vol 1, Compendium of Thermophysical Property Measurement Methods, Plenum Press, 1984, p 305–336 Y. Qingzhao and W. Likun, Laser Pulse Method of Determining Thermal Diffusivity, Heat Capacity, and Thermal Conductivity, FATPC, 1986, p 325–330 Y. Maeda, H. Sagara, R.P. Tye, M. Masuda, H. Ohta, and Y. Waseda, A High-Temperature System Based on the Laser Flash Method to Measure the Thermal Diffusivity of Melts, Int. J. Thermophys., Vol 17 (No. 1), 1996, p 253–261 G. Ho¨hne, W. Hemminger, and H.-J. Flammershein, Differential Scanning Calorimetry. An Introduction to Practitioners, Springer-Verlag, Berlin, 1996 J.D. James, J.A. Spittle, S.G.R. Brown, and R.W. Evans, A Review of the Measurement Techniques for the Thermal Expansion Coefficient of Metals and Alloys at Elevated temperatures, Meas. Sci. Technol., Vol 12, 2001, p R1–R15 Y. Su, K.C. Mills, and A. Dinsdale, A Model to Calculate Surface Tension of Commercial Alloys, J. Mater. Sci., Vol 40, 2005, p 2185–2190 K.C. Mills, Measurement and Estimation of Physical Properties of Metals at High Temperatures, Fundamentals of Metallurgy, S. Seetharaman, Ed., Woodhead Publishing Limited, Cambridge, England, 2005 “Standard Test Method for Determining Specific Heat Capacity by Differential Scanning Calorimetry,” E 1269, Annual Book of ASTM Standards, ASTM, 2001 “Practice for Temperature Calibration of DSC and DTA,” E 967, ASTM “Standard Practice for Heat Flow Calibration of DSC,” E 968, ASTM “Standard Method for Enthalpy Measurement Validation of Differential Scanning Calorimeters,” E 2253, ASTM “Standard Test Method for Enthalpies of Fusion and Crystallisation by DSC,” E 793, ASTM “Standard Test Method for Specific Heat of Liquids and Solids,” D 2766, Annual Book of ASTM Standards, ASTM, 1995 “Advanced Technical Ceramics. Monolithic Ceramics. Thermo-Physical Properties.

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

65.

66.

67. 68. 69. 70.

Determination of Specific Heat Capacity,” EN 821–3:2005 “Standard Test Method for Linear Thermal Expansion of Solid Materials with a Vitreous Silica Dilatometer,” E 228, Annual Book of ASTM Standards, ASTM, 1995 “Standard Test Method for Linear Thermal Expansion of Rigid Solids with Interferometry,” E 289, Annual Book of ASTM Standards, ASTM, 1999 “Standard Test Method for Linear Thermal Expansion of Solid Materials by Thermomechanical Analysis,” E 831, Annual Book of ASTM Standards, ASTM, 2000 “Advanced Technical Ceramics. Monolithic Ceramics. Thermo-Physical Properties. Determination of Thermal Expansion,” EN 821–1:1995 “Fine Ceramics (Advanced Ceramics, Advanced Technical Ceramics)—Test Method for Linear Thermal Expansion of Monolithic Ceramics by Push-Rod Technique,” ISO 17562:2001 “Standard Test Method for Thermal Conductivity and Thermal Diffusivity by Modulated Temperature Differential Scanning Calorimetry,” E 1952, Annual Book of ASTM Standards, ASTM, 2001 “Standard Test Method for Thermal Diffusivity of Solids by the Laser Flash Method,” E 1461, Annual Book of ASTM Standards, ASTM, 2001 “Standard Test Method for Steady-State Thermal Transmission Properties by Means of the Heat Flow Meter Apparatus,” C 518, Annual Book of ASTM Standards, ASTM, 1998 “Advanced Technical Ceramics. Monolithic Ceramics. Thermo-Physical Properties. Determination of Thermal Diffusivity by the Laser Flash (or Heat Pulse) Method,” EN 821–2:1997 “Fine Ceramics (Advanced Ceramics, Advanced Technical Ceramics)—Determination of Thermal Diffusivity of Monolithic Ceramics by Laser Flash Method,” ISO 18755:2005 “Standard Test Method for Normal Spectral Emittance at Elevated Temperatures,” E 307, Annual Book of ASTM Standards, ASTM, 1972, 2002 “Standard Test Method for Total Normal Spectral Normal Emittance of Surfaces Using Inspection Meter Techniques,” E 408, Annual Book of ASTM Standards, ASTM, 1971, 2002 R.F. Brooks, A.P. Day, K.C. Mills, and P.N. Quested, Int. J. Thermophys., Vol 18, 1997, p 471–480 S. Sauerland, R.F. Brooks, I. Egry, and K. C. Mills, Proc. TMS Conf. Ann. Conf. on Containerless Processing, 1993, p 65–69 K.C. Mills and R.F. Brooks, Mater. Sci. Eng. A, Vol 178, 1994, p 77–81 D.R. Gaskell, Introduction to Metallurgical Thermodynamics, McGraw-Hill, 1973, p 497–501

71. L. Darken and R. Gurry, Physical Chemistry of Metals, McGraw-Hill, 1953, p 125–126 ¨ ber die Spe72. F.E. Neumann, Untersuchung U cifische Wa¨rme der Mineralien, Pogend. Annal., Vol XXIII (No. 1831), p 1–39 73. H. Kopp, Proc. R. Soc. (London), Vol 13, 1863–1864, p 229–239 74. K.A. Jackson, Liquid Metals and Solidification, American Society for Metals, 1958, p 174 75. J. Blumm and J.B. Henderson, High Temp.—High Press., Vol 32, 2003–2004, p 109–113 76. R. Morrell and P. Quested, Evaluation of Piston Dilatometry for Studying the Melting Behaviour of Metals and Alloys, High Temp.—High Press., Vol 35–36, 2003– 2004, p 417–435 77. G. Lang, Density of Liquid Elements, CRC Handbook of Chemistry and Physics, 75th ed., D.R. Lide and H.P.R. Fredererikse, Ed., CRC Press, 1994–1995, p 4– 126 to 4–134 78. J.R. Wilson, Metall. Rev., Vol 10, 1965, p 381 79. L.J. Wittenberg, Viscosity of Liquid Metals, Physicochemical Measurements in Metal Research, Vol IV, Part 2, R.A. Rapp, Ed., Wiley Interscience, 1970, p 193 80. W. Kurz and D.J. Fisher, Fundamentals of Solidification, Trans Tech Publications, Switzerland, 1984, p 15 81. B.C. Allen, Liquid Metals: Chemistry and Physics, S.Z. Beer, Ed., Marcel Dekker, 1972, p 161–197 82. A.S. Skapski, J. Chem. Phys., Vol 16, 1948, p 389 83. B.J. Keene, Review of Data for the Surface Tension of Pure Metals, Int. Mater. Rev., Vol 38, 1993, p 157–192 84. B.J. Keene, Review of Data for Surface Tension of Iron and Its Binary Alloys, Int. Mater. Rev., Vol 33, 1988, p 1–37 85. K.C. Mills and Y.C. Su, Review of Surface Tension Data for Metallic Elements and Alloys, Part 1: Pure Metals, Int. Mater. Rev., Vol 51, 2006, p 329–351 86. R.F. Brooks and P.N. Quested, The Surface Tension of Steels, J. Mater. Sci., Vol 40, 2005, p 2233–2238 87. S.-L. Chen, W. Oldfield, Y.A. Chang, and M.K. Thomas, Metall. Mater. Trans. A, Vol 25, 1994, p 1525–1533 88. F. Zhang, Y.A. Chang, and J.S. Chou, A Thermodynamic Approach to Estimate Titanium Thermophysical Properties, Proceedings of 1997 International Symposium on Liquid Metal Processing and Casting, A. Mitchel and P. Auburtin, Ed., American Vacuum Society, 1997, p 35–59 89. K.S. Yeum, R. Speiser, and D.R. Poirier, Metall. Trans. b, Vol 20, 1989, p 693–703 90. H.-K. Lee, J.P. Hajra, and Z. Frohberg, Metalkd., Vol 83, 1992, p 638–643 91. J.A.V. Buttler, Proc. R. Soc. (London) A, Vol 135, 1932, p 348

Thermophysical Properties / 481 92. J.-S. Chou et al., “Thermophysical and Solidification Properties of Titanium Alloys.” Report TR 98–87, National Center for Excellence in Metalworking Technology, June 30, 1999 93. F. Zhang, S.-L. Chen, and Y.A. Chang, Modeling and Simulation in Metallurgical Engineering and Materials Science, Proc. Int. Conf., MSMM’96, Z.-S. Yu, Ed. (Beijing, China), 1996, p 191–196 94. J.L. Murray, Phase Diagrams of Binary Titanium Alloys, ASM International, 1987 95. J.L. Murray, Bull. Alloy Phase Diagrams, Vol 2, 1981, p 185–192 96. N. Saunders and V.G. Rivlin, Mater. Sci. Technol., Vol 2 1986, p 521 97. L. Kaufman and H. Nesor, Metall. Trans., Vol 5, 1974, p 1623–1629 98. U.R. Kattner and B.P. Burton, Phase Diagrams of Binary Titanium Alloys, ASM International, 1987 99. K. Frisk and P. Gustafson, CALPHAD, Vol 12, 1988, p 247–254 100. K.-J. Zeng, H. Marko, and L. Kaj, CALPHAD, Vol 17, 1993, p 101–107 101. J.-O. Anderson and N. Lange, Metall. Trans. A, Vol 19, 1988 102. A.D. Pelton, J. Nucl. Mater., Vol 201 (No. 218–224), 1993, p 1385–1394 103. W. Huang, Z. Metallkd., Vol 82, 1991, p 391–401 104. I. Takamichi and R.I.L. Guthrie, The Physical Properties of Liquid Metals, Clarendon Press, 1988, p 148–198 105. R.F. Brooks, A.T. Dinsdale, and P.N. Quested, The Measurement of Viscosity of Alloys—A Review of Methods, Data

106. 107. 108.

109. 110. 111.

112. 113. 114. 115. 116.

117.

and Models, Meas. Sci. Technol., Vol 16, 2005, p 354–362 E.N. da C. Andrade, The Theory of the Viscosity of Liquids: I, Philos. Mag., Vol 17 1934, p 497, 698 M. Hirai, Estimation of Viscosity of Liquid Alloys, ISIJ, Vol 33, 2002, p 281–285 D.U. Sichen, J. Byge´n, and S. Seetharaman, A Model for Estimation of Viscosities of Complex Metallic and Ionic Melts, Metall. Trans. B, Vol 25, 1991, p 519 E.T. Turkdogan, Physical Chemistry of High Temperature Technology, Academic Press, 1980, p 109 T. Chapman, AlChE J., Vol 12, 1966, p 395 T. Iida et al., Accurate Predictions for the Viscosities of Several Liquid Transition Metals, Plus Barium and Strontium, Metall. Mater. Trans. B, Vol 37, 2006, p 403–412 N.F. Mott, Proc. R. Soc. (London) A, Vol 146, 1934, p 465 T.E. Faber, Introduction to the Theory of Liquid Metals, Cambridge University Press, 1972 J.M. Ziman, Adv. Phys., Vol 13, 1964, p 89 R. Evans, D.A. Greenwood, and P. Lloyd, Phys. Lett. A, Vol 35, 1971, p 57 R. Evans, B.L. Gyorffy, N. Szabo, and J.M. Ziman. Proceedings of the Second International Conference on Liquid Metals (Tokyo), S. Takeuchi, Ed., Taylor and Francis, London, 1973 Y. Waseda, The Structure of Non-Crystalline Materials: Liquids and Amorphous Solids, McGraw-Hill, 1980

118. S. Takeuchi and H. Endo, J. Jpn. Inst. Met., Vol 26, 1962, p 498 119. J.L. Tomlinson and B.D. Lichter, Trans. Met. Soc. AIME, Vol 245, 1969, p 2261 120. L. Lorentz, Ann. Phys. Chem., Vol 147, 1982, p 429 121. G.T. Meaden, Electrical Resistance of Metals, Plenum, 1965 122. K.C. Mills, B.J. Monaghan, and B.J. Keene, “Thermal Conductivities of molten Metals: Part 1 Pure Metals.” International Materials Reviews 1996, Vol 41, 1996 p 209–242 123. G.H. Geiger and D.R. Poirier, Transport Phenomena in Metallurgy, Addison-Wesley Publishing Co., 1973, p 575–587 124. C.J. Smithells, Radiating Properties of Metals, Metals Reference Book, 7th ed., E.A. Brandes and G.B. Brook, Ed., Butterworth-Heinemann, 1992, p 17–1 to 17–12 125. J. Szekely and N.J. Themelis, Chap. 9, Rate Phenomena in Process Metallurgy, Wiley-Interscience, John Wiley and Sons, Inc., 1971, p 251–300 126. E.M. Sparrow and R.C. Cess, Radiation Heat Transfer, Brooks/Cole, Belmont, CA, 1966 127. H.C. Hottel and A.F. Sarofim, Radiative Transfer, McGraw-Hill, 1967 128. T.J. Love, Radiative Heat Transfer, Merrill, Columbus, OH, 1968 129. M.J. Assael et al. “Reference Data for the Density and Viscocity of Liquid Aluminum and Liquid Iron” J. Phys. Chem. Ref. Data, Vol 35, 2006, p 285–300

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 485-487 DOI: 10.1361/asmhba0005187

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Shape Casting Processes— An Introduction J.L. Jorstad, J.L.J. Technologies, Inc.

SHAPE CASTING, as a metal-forming process, both technically and economically enables some of the most important products employed in every sector of our economy in transportation, housing, appliances, recreation, medical, and other (see the article “History and Trends of Metal Casting” in this Volume). In fact, hardly a system exists that does not depend on shaped castings in one way or another. Shape casting can be divided into several broad categories, for instance, single-use (onetime-only) processes, such as sand, plaster, ceramic, and graphite molding; essentially

Fig. 1

Chart of shape casting processes

unpressurized multiuse (metal mold) processes, such as permanent mold; and high-pressure metal mold methods, such as die casting, squeeze casting, and semisolid processing. The categories and subcategories of shape casting processes are illustrated in Fig. 1 and are described in more detail in other articles of this Volume. Each shape casting process has its niche, perhaps by virtue of the metal (or metals) to which it applies, or maybe based on its overall productivity; there can be numerous reasons to select one process over another. Tables 1 and 2 compare some of the typical capabilities of

shape casting processes (see also Tables 7, 8, and 9 in the article “History and Trends of Metal Casting” in this Volume). The ratings and data are necessarily typical and approximate, because many material and design factors can influence the economical feasibility of a casting process. The tables provide a brief overview only for general approximate comparison. Green sand is by far the most-oftenemployed casting process, simply because it is the dominant process for casting ferrous metals. Ferrous (steel, gray, and ductile iron) account for nearly 75% of all metals cast:

486 / Principles and Practices of Shape Casting Table 1 Shape casting processes ratings chart Capabilities Mold method

Applicable metals

Expendable mold with permanent patterns Bonded sand processes Green sand CO2 sand Cold box Hot box Shell Slurry processes Plaster Ceramic Rammed graphite Nonbonded sand processes Vacuum molding Magnetic molding Expendable mold with expendable patterns Lost foam Investment Replicast Centrifugal casting Metal mold Sand mold Permanent mold and semipermanent mold Gravity Static top pour

Tilt pour Low-pressure permanent mold (LPPM) Standard LPPM Vacuum riserless casting/pressure riserless casting Counterpressure casting/pressure-counterpressure casting Countergravity casting High-pressure die casting (HPDC) Conventional HPDC Hot chamber Cold chamber Vacuum die casting Squeeze casting Indirect Direct Semisolid processing Thixocasting (billet) Rheocasting (slurry) Thixomolding (granular magnesium)

Productivity(a)

Disposable cores

Casting size

Dimensional control

All All All All All

5 2 2 2 3

Yes Yes Yes Yes Yes

Unlimited Unlimited Unlimited Unlimited Small to medium

... ... ... ... ...

Nonferrous All Reactive

1 1 1

Yes Yes Yes

Unlimited Unlimited Unlimited

Outstanding Outstanding ...

Aluminum Ferrous

1 1

NA NA

Unlimited Unlimited

... ...

Aluminum All All

3 2 1

NA NA NA

Medium to large Small to medium Medium to large

... Outstanding ...

Nonferrous All

3 2

Yes Yes

Small to medium Small to medium

... ...

Nonferrous and metal-matrix composites (MMCs) Nonferrous

3

Yes

Medium to large

...

3

Yes

Medium to large

...

Light metals Aluminum

3 3

Yes Yes

Medium Medium

... ...

Aluminum

3

Yes

Medium

...

All

3

...

Small

...

Zn and Mg Nonferrous Aluminum

5 5 4

No No ...

Small Small to medium ...

Outstanding Outstanding ...

Al and MMCs Nonferrous and MMCs

4 1

No No

Small to medium Small to medium

... ...

Aluminum Nonferrous Magnesium

4 5 5

No No No

Small to medium Small to medium Small

Outstanding Outstanding Outstanding

NA, not applicable. (a) Productivity rankings: 1 = low productivity; 5 = high productivity

Dominant metals

Cast steel Gray iron Aluminum Ductile iron Others (magnesium, zinc, copper, etc.)

Percentage cast

29 26 20 17 7

Virtually all sand processes are suitable for casting both ferrous and nonferrous metals, but green sand excels over other sand processes because it is also the most productive. No other sand process can boast hundreds of molds per hour, with the potential for numerous cavities per mold. Percentage of casting tonnage by molding process is: Molding process

All All All All

sand casting permanent and semipermanent mold die casting other

Casting tonnage, %

75 >5 19 <1

On the nonferrous side, high-pressure die casting is the dominant process, largely because it readily accommodates scrap-based secondary alloys. Among metal mold processes, it enjoys the highest productivity (short cycle times). However, every dominant process also has its downside. Green sand does not provide the dimensional accuracy or surface finish of the other bonded-sand processes; thus, they find their niche. The properties of many nonferrous alloys are sensitive to solidification rate; when strength and ductility of those alloys are important, green sand gives way to metal mold processes. High-speed green sand lines make the employment of chills somewhat difficult, so when localized high cooling rates are required in a casting, CO2 or cold box sand processes may provide better local chill placement opportunity. High-pressure die casting (HPDC) is noted for spraying molten metal into the die at very high velocity, creating turbulence and entrapping

cavity air and gases in the solidified casting. Thus, HPDC is not suitable for castings that must be heat treated, welded, or subjected to other elevated-temperature treatments such as porcelain enameling. Variations on HPDC (squeeze casting semisolid processing and high-vacuum die casting) have emerged to overcome those shortcomings, but at a cost premium that renders them somewhat less competitive than HPDC, so there must be a trade-off. The slurry processes provide exceptional dimensional accuracy and surface finish for precision parts having complex contours that would be difficult and expensive to machine. High-pressure die casting and its variations also provide near-net shape castings but require much more expensive tooling so are generally warranted only for high annual volume requirements. The sand and permanent mold processes readily accommodate disposable internal cores,

Shape Casting Processes—An Introduction / 487 Table 2

Comparison of several casting methods

Approximate and depending on the metal Parameter

Green sand casting

Permanent mold cast

Die casting

CO2-core casting

Investment casting

Medium-high Medium-high Shell: g(oz)–115 kg (250 lb) CO2 0.22 kg (½ lb)–Mg (tons) 2:5ð1=10Þ 0.25 (0.01)

Highest Medium g(oz)–45 kg (100 lb)

Relative cost in quantity Relative cost for small number Permissible weight of casting

Low Lowest Up to approx. 900 kg (1 ton)

Low High 45 kg (100 lb)

Lowest Highest 40 kg (85 lb)

Thinnest section castable, mm (in.) Typical dimensional tolerance, mm (in.) (not including parting lines) Relative surface finish

2:5ð1=10Þ 0.3 (0.012)

3:0ð1=8Þ 0.75 (0.03)

0:8ð1=32Þ 0.25 (0.01)

Fair to good

Good

Best

Relative mechanical properties Relative ease of casting complex design Relative ease of changing design in production

Good Fair to good Best

Good Fair Poor

Very good Good Poorest

usually made from sand. The HPDC processes, on the other hand, have very limited ability to accommodate disposable cores. The high metal flow velocities and the high pressures applied during solidification destroy most cores, and those robust enough to withstand HPDC processing parameters cannot then be easily removed from the solidified part. Some salt cores are routinely used in HPDC, as are some low-melting-temperature metal cores, but at this writing (2008), HPDC must be considered very limited with respect to application of disposable cores such as those applied to the sand and permanent mold processes. By assembling simple-to-mold shapes, the lost foam process has the ability to consolidate several complex cast configurations into one, sometimes

enabling a cast part that would be impossible to machine. Lost foam requires no separate disposable internal cores to create internal shapes that would require cores in other processes. While conventional HPDC dominates the nonferrous casting industry, it has been losing ground in recent years to other metal mold processes and to variations on itself that are better able to meet the demands for lightweight structural castings, such as those in the chassis and suspension systems of automobiles and light trucks. Squeeze casting was the first variation on HPDC to prove capable of structural aluminum castings, such as steering knuckles. Now, the low-pressure casting process and its variations, such as vacuum riserless casting/pressure riserless casting and counterpressure casting/

Shell: good CO2: fair Good Good Fair

1:5ð1=16Þ 0.25 (0.01) Very good Fair Best Fair

pressure-counterpressure casting, are equally able to provide crash-worthy products and can achieve better cavity counts than squeeze casting because metal is not injected or solidified under high pressure. However, those processes are best suited for thicker-walled parts, and semisolid processing and high-vacuum die casting are now emerging as the processes for large, very thin-walled (2 mm, or 0.08 in.) structural parts. The dominant processes will continue to dominate for legitimate reasons, and other processes will continue to emerge or expand as the requirements for castings continue to change. That is the nature of the business. Castings of every metal and the processes to make them have always managed to develop and change to meet the needs of the many user industries.

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 488-496 DOI: 10.1361/asmhba0005308

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Patterns and Patternmaking Robert C. Voigt, Pennsylvania State University

A PATTERN is a form made of wood, metal, plastic, or composite materials around which a molding material (usually prepared sand) is formed to shape the casting cavity of a mold. Most patterns are removed from the completed mold halves and used repeatedly to make many duplicate molds. Expendable patterns of such materials as wax or expanded polystyrene are made in quantity and are used only once to produce an individual mold (see the articles “Investment Casting” and Replicast Molding and the discussion of the lost foam process in the article “Lost Foam Casting” in this Volume). The pattern equipment (tooling) needed to make a casting includes the pattern and may also include one or more core boxes. Core boxes are used to make refractory inserts or cores that are placed within a mold cavity to form internal cavities or passageways within the casting. The selection of the type of pattern equipment used to make a casting depends on many factors, including the number of castings to be produced, the size and shape of the casting, the molding or casting process to be used, and other special requirements such as the dimensional accuracy required. It is common for the pattern to be owned by the casting customer rather than the foundry, although the foundry commonly is contracted to produce the pattern either in-house or at a patternshop. Patternmaking begins with the dimensions of the casting required; however, the design and dimensions of the resultant pattern equipment must incorporate other features and take into account various pattern allowances (Ref 1, 2). Patterns must be oversized to correct for the contraction of the metal casting upon cooling and any extra metal to be removed from the machined surfaces of a casting. Incorporating all cast part features into the pattern or creating some features by subsequent machining is a key decision made by the pattern designer. A mold parting line must be selected, and taper or draft must be included on the vertical faces of the pattern to permit the removal of reusable patterns from the mold after mold making. Patterns also incorporate provisions for gating and risering (rigging), which facilitate the flow of metal into the mold cavity and subsequent feeding. For castings to be made with cores, the pattern and core boxes must include projections called

core prints, which are used to support and locate the cores in the mold cavity. Figure 1 illustrates a typical pattern and mold configuration. The cost of pattern equipment for a given casting can vary greatly, depending on pattern material, type of pattern, and sometimes the dimensional accuracy required. However, because the pattern equipment is only a part of the moldmaking process, the least expensive pattern is not necessarily the most economical. Additional pattern cost and quality often lead to lower end costs by a reduction in molding costs and pattern repair costs and by an improvement in overall casting quality. The patterns and patternmaking techniques described in this article are for patterns used to make expendable molds (primarily sand molds). However, many of the principles described also apply to the dies used for other metal casting processes such as die casting and permanent mold casting (see the articles “Die Casting Tooling” and “Permanent Mold Casting” in this Volume). Patterns are the “positive” replica of the shape to be cast. Dies are the “negative” replica or cavity in which cast shapes are produced.

Types of Patterns The type of pattern used for a specific application depends primarily on the number of castings required, the molding or casting process to be used, the size of the pattern, and the casting tolerances that are required. The stage of development of a casting design is also a factor. If the casting is likely to be redesigned, an inexpensive prototype pattern is often used first. Similarly, gating and risering additions to a pattern are often prototyped first, using inexpensive pattern materials. The choice of pattern material is also closely tied to the molding or casting process and the type of pattern to be used. Figure 2 illustrates the most commonly used basic reusable pattern types for a simple pattern shape—a loose pattern, a match plate pattern, and cope and drag patterns. Also indicated are the typical materials and service lives for each pattern type. It should be noted that many factors affect the

number of molds that can be made from an actual production pattern before it must be reworked or discarded. The pattern lives indicated in Fig. 2 are intended only as general guidelines. Actual pattern life can vary considerably. Loose patterns, also called one-piece patterns or solid patterns, are the simplest, least expensive type of reusable pattern and are suited only to very low-quantity production. This type of pattern (Fig. 2a) is most appropriate for experimental or prototype castings and is only very rarely used for short-run production castings. Molding with a loose pattern requires more manual operations and a much higher degree of molder skill than molding with other pattern types. Many foundries are not equipped to use loose patterns. This increases the cost per mold and results in variations in casting quality from mold to mold compared to moldmaking with other pattern types. Gates and risers for loose patterns can be molded by hand or can be constructed with loose pieces and molded with the pattern. Match plate patterns (Fig. 2b) are split patterns in which the cope and drag (top and bottom) portions are mounted on opposite sides of a plate, called the match plate, conforming to the parting line. The pattern, as well as the associated gating and risering system, is often made separately and then mounted on the match plate, but can be cast or machined integrally with the plate. The size of the match plate corresponds to the size of the flask used to make the final mold. Flask pin guides are

Fig. 1

Schematic of a sand mold. The pattern is used to form the mold cavity, the core print for locating the core, the gate, the runner, the riser, and the sprue. A separate core box is used to make the sand core that is inserted into the parted mold before pouring.

Patterns and Patternmaking / 489

Fig. 2

Three types of patterns used to produce a water pump casting in various quantities. (a) Wood pattern on a follow board, good for 20 to 30 castings. (b) Match plate pattern for up to 50,000 castings depending on material. (c) Cope and drag plastic (up to 20,000 castings) or metal (100,000 or more castings) pattern

used to ensure accurate alignment of the match plate pattern in the flask and to assist in pattern removal from the mold. Multiple patterns of small parts can be mounted on a single match plate. Common gates and risers on the match plate can often be shared by the multiple patterns on the match plate. Interlocking features are added to the cope and drag sides of the match plate to ensure accurate alignment of the cope and drag mold halves with each other after removal of the pattern. Match plate patterns are used for low- to high-volume production of small-and mediumsized castings with good dimensional accuracy. The molding operation is simplified considerably by the use of match plate patterns. The decrease in molding costs and increased mold quality offset higher pattern costs for highervolume castings. Large patterns are not usually made as match plate patterns, because of the limitations on flask sizes and the difficulties of molders or molding machines in handling large match plate patterns. Because significant pattern wear may occur over the life of a high-production match plate pattern, the selection of pattern material with adequate wear resistance is important. Also, the match plate pattern must be rigid enough to withstand high squeeze pressures during automated molding (Ref 3, 4). For this reason, metal match plate patterns are commonly used. If the pattern has a flat parting surface, the cope and drag pattern halves can be made separately and fastened to the plate. However, most

metal match plate patterns, whether they have a flat parting surface or not, are directly machined together with the plate or cast integrally with the plate. Correct registry between the pattern impressions on either side of the match plate is critical to avoid cope and drag shift in the final casting. Cope and drag patterns (Fig. 2c) are similar to match plate patterns except that the cope and drag portions of a split pattern are mounted or integrally cast or machined on separate plates. Cope and drag patterns are preferred over match plate patterns for high-volume production or for the production of large castings. Pattern cost is higher than for match plate patterns, but the total molding cost per casting may be lower because the cope and drag mold sections can be made simultaneously in the foundry. Automated vertical parting line molding machines require cope and drag pattern plates because a cope and drag impression is simultaneously made on opposite faces of the same block of molding sand. Successive blocks of sand are then pushed together side by side. The mold cavity is formed from the cope (or ram) impression from one block of sand and the drag (or swing) impression from the next block of sand. Thus, the equivalent of one mold cavity can be made for each block of sand. Because the cope and drag are mounted on separate plates, accurate alignment of the two patterns is essential. The design and manufacture of cope and drag patterns are identical to those of match plate patterns.

Special Patterns. A wide variety of special pattern types and special patternmaking techniques can have specific uses in the foundry. Traditional master patterns are still sometimes used for casting metal patterns and for making plastic patterns. Multiple patterns are cast from the master to make match plate or cope and drag patterns. A wide variety of techniques and pattern materials can be used to create the master pattern and to form the production patterns from the master pattern. Reverse master patterns (negative patterns) are also used with certain pattern materials to create production patterns. The basic steps for producing production patterns from master patterns are illustrated in Fig. 3. Computer numerical control machining of pattern plates has replaced much of the traditional patternmaking using master patterns. A pattern need not always be a complete replica of the final casting. Molds for very large castings are often made from an assembly of chemically bonded sand mold sections (cores) made in numerous core boxes. This is referred to as an all-core mold. In this case, a single pattern is not required; rather, the casting mold is made from multiple core boxes for each section of the mold. These large mold assemblies are made on the foundry floor or in large pits built into the foundry floor (Ref 5). For example, the mold for a large gear can be made from a sectional pattern or core box that is a wedgeshaped segment of the hub, spokes, rim, and teeth. The complete gear mold is assembled from successively made wedge segments. Core Boxes. Cores are primarily used to form internal passageways or cavities in castings. Cores can also be used to form sections of the mold for a cast shape that has external projections or negative draft, which would prevent the pattern from being withdrawn from the mold. Core boxes are the reverse patterns or reverse molds used to make the sand cores that become part of the final mold assembly. The term pattern equipment refers to the pattern itself and to all of the core boxes needed to produce the cores that will be part of the final mold assembly. Because a given casting may have a complex internal shape or many internal passageways, the cost of producing core boxes may exceed that of the associated mold pattern. An example of an intricate full split aluminum core box used in a core blowing machine is shown in Fig. 4. Because many coremaking processes require the use of a gas catalyst to cure the core and binder, special venting considerations must be included in the core box design. Ejection pins can also be included to aid in removing the core from the core box halves. The type of core box used and the material used to make a core box can vary greatly, just as for patterns. Core box construction depends on the type of coremaking process used and the number of cores required. Sand cores that develop strength during the high-temperature curing of thermal-setting binders (shell hot box and warm box cores) require heat-conducting metal core boxes. (Core sand binder

490 / Principles and Practices of Shape Casting

Fig. 3

Steps in producing a match plate pattern from a master pattern. (a) Wood master pattern is constructed with appropriate shrinkage factors included. A reverse from the master pattern. (b) Reverse mold is used to cast the polyurethane pattern on the aluminum match plate. (c) Reverse mold is removed, and a wood gating system is added. Gating can also be cast on with the pattern or made from aluminum stock. Courtesy of W. Allen, Allen Pattern of Michigan

Fig. 4

Full split aluminum core box designed to eliminate pasting of core halves. From Steel Castings Handbook, 5th ed., Steel Founders’ Society of America, 1980

systems and coremaking machines are discussed in the article “Coremaking” in this Volume.) Core boxes can also be subjected to considerable abrasive wear and pressure when core blowing machines deliver sand into the boxes at high air pressures. As for patterns, the lowest cost core box is not necessarily the most economical. Proper casting design to minimize the use of cores is important not only for minimizing core box cost but also for decreasing molding costs and the overall cost of a casting. Tooling for Expendable Patterns and Des. Two common casting techniques that use expendable patterns are investment casting and lost foam casting. In investment casting, a single-use wax pattern is made in metal dies and coated with multiple layers of a ceramic or plaster slurry that hardens to form a thin rigid mold. The mold is heated to fire the ceramic and to melt out the wax, leaving an empty mold cavity of the desired shape.

Close-tolerance precision castings can be made using the investment casting process (see the article “Investment Casting” in this Volume). In lost foam casting, a single-use expanded polystyrene pattern is blown into metal dies, coated with a thin refractory slurry, and placed in an unbonded sand mold. When molten metal is poured into the mold, the advancing metal vaporizes the polystyrene and fills the resulting cavity (see the article “Lost Foam Casting” in this Volume). The dies, or negative patterns, used to produce the wax or polystyrene are usually machined from cast aluminum. They shore many common characteristics with core boxes used in sand casting and with dies produced for die casting and permanent mold casting. A further discussion of wax and expendable polystyrene pattern materials can be found in the Section “Expendable-Mold Casting Processes with Expendable Patterns” in this Volume. Reverse patterns or dies used for high-pressure die casting and permanent mold casting are directly filled with molten metal to make castings. They have much in common with core boxes and with the reverse patterns or dies used to make wax or lost foam patterns. Many of the pattern characteristics described in this article for sand casting with reusable patterns also apply to the design and manufacture of permanent and semipermanent mold and dies used in the permanent mold casting process and the die casting process. These casting processes are discussed in the article “Permanent Mold Casting” in this Volume.

Additional Features of Patterns The primary purpose of the pattern is to produce a mold cavity of the desired shape. However, to produce successful castings, additional pattern features and mold production

considerations must be incorporated into the pattern design. These include parting line considerations and the addition of gates and risers, core prints, and locating points. Pattern allowances for ensuring a dimensionally correct final pattern are discussed in the next section of this article. Parting Line. It is necessary to separate (part) the completed mold halves and to withdraw the pattern from the mold for most sand casting processes. Selection of the most convenient and economical pattern parting line (mold parting surface) and design of the pattern so that it can be readily removed from the mold are important pattern design considerations. On a flat pattern plate, the parting surface is a simple plane. Many complexly shaped castings, however, require irregular parting surfaces to facilitate mold parting. The pattern match plate or cope and drag plates are used to establish the parting surface. Irregular parting surfaces can be readily incorporated with these types of patterns. Some complex pattern shapes cannot be readily removed from the parted mold halves unless the casting is redesigned or cores are used. By selecting the best parting line for a given casting to facilitate pattern removal from the mold, the use of cores to modify the external shape of the casting can be minimized. Gates and risers (rigging) are important factors in the production of quality castings and should be fully considered in the design and construction of patterns. Gating provides for the efficient flow of the molten metal into the mold cavity. Risers are large reservoirs of molten metal attached to the casting. They provide sufficient liquid metal to the casting as it solidifies to prevent internal shrinkage cavities or shrinkage defects in the casting. Gates and risers are commonly attached directly to the pattern and pattern plate and are integrally molded. On production match plate patterns, the gates and risers can be integrally cast on or machined with the pattern plate. The proper design of pattern gates and

Patterns and Patternmaking / 491 risers depends on the casting shape, the type of metal to be poured, and the molding and casting processes to be used. The location, size, and shape of gates and risers should be determined by an experienced foundry engineer, typically with computer software design assistance. This is often a trial-and-error process, especially for complexly shaped castings. Pattern gating and risering is often modified after initial castings are made and evaluated, not only to produce sound castings but also to increase casting yield efficiency. Core Prints. When the production of a cast shape requires cores, provision must be made on the pattern for core prints. Core prints are the features added to the pattern that will eventually locate and anchor the core in the proper position in the mold. The core print is added to the pattern (and removed from the core box) but is not a feature on the final casting, because it is filled with the core itself. Core prints are illustrated in Fig. 1. Locating Points. The foundry, the pattern shop, and the machine shop all use locating points or locating surfaces on the casting to aid in casting inspection and machining. These features must also be incorporated into the pattern, together with machining allowances on all surfaces to be machined.

Pattern Allowances Although a pattern is used to produce a casting of desired dimensions, it is not dimensionally identical to the casting. A number of allowances must be made on the pattern to ensure that the finished casting is dimensionally correct, to ensure that the pattern can be effectively removed from the mold, and to allow for cores to be firmly anchored. Shrinkage allowance is the correction factor built into the pattern to compensate for the contraction of the metal casting as it solidifies and cools to room temperature. The pattern is intentionally made larger than the final desired casting dimensions to allow for solidification and cooling contraction of the casting. This allowance for contraction is sometimes called patternmaker’s shrinkage. The total contraction is volumetric but is usually expressed linearly. Table 1 lists commonly used pattern shrinkage allowances for various casting alloys. Because different shrinkage allowances must be used for the individual types of metals cast, it is not possible to use the same pattern equipment for different cast metals without expecting dimensional changes. It must be emphasized that these shrinkage allowances are only guidelines and can only be applied with considerable knowledge of the actual casting design and the foundry molding techniques to be used. Computer-aided design (CAD) tools can easily apply a general shrinkage rule so that oversized pattern dimensions can be readily created from the original part dimensions. However, it is important to remember that pattern shrinkage allowances such as those

Table 1

Typical pattern shrinkage allowances for various casting materials

Patternmaker and foundry should be consulted before shrinkage is specified. Shrinkage allowance Alloy being cast

Steel Gray cast iron Ductile cast iron Aluminum Brass Magnesium

Allowance

1 1 1 1 1 1

in in in in in in

Approximate shrinkage, %

mm/m

in./ft

1.6 1.0 0.8 1.3 1.4 1.3

15/7 8/4 7/8 13/1 14/4 13/1

3/16 1/10 3/32 5/32 11/64 5/32

64 100 120 77 70 77

listed in Table 1 cannot be used blindly to predict actual casting shrinkage. Other factors, such as casting shape, section thickness, and mold rigidity, significantly influence the actual change in dimensions as the casting solidifies and cools to room temperature. The resistance of the mold to the normal contraction of the casting and the use of different molding materials and methods may require that different shrinkage allowances be used for different patterns (Ref 6, 7). The same pattern used to make castings in a hand-rammed green sand mold, by high-pressure green sand molding, and in a high-strength chemically bonded sand mold will produce castings of varying dimensions because of the great differences in mold wall dilation during casting and solidification. It is even likely that several different shrinkage allowances will be needed in one pattern, depending on constraint conditions. For example, in Fig. 5, two different contraction situations exist. Along dimension X, the casting has virtually no constraint to contraction, and the pattern should be made correspondingly larger along the X dimension surfaces. Dimension Y, however, is restrained from contraction by the core used to make the center hole and will require little or no shrinkage allowance on the pattern dimensions. Contemporary CAD tools now incorporate some capabilities for choosing individual shrinkage rules for individual casting features instead of simply a common shrinkage rule for all features. Because casting design and foundry methods have a great influence on the required shrinkage allowance and on the shrinkage characteristics of the metal itself, the direct involvement of the foundry producing the castings in the determination of the required shrinkage allowances is essential. For high-production-volume castings where tight dimensional tolerances are required, inexpensive prototype patterns are first created, and prototype castings are produced and evaluated for dimensional conformance. The pattern dimensions are then corrected before the final production pattern is produced. Draft is taper required on the vertical faces of a pattern to permit its removal from the sand or other molding medium without tearing of the mold walls. (On vertically parted molds, draft is required on horizontal pattern surfaces.) The amount of draft required depends on the shape and size of the casting, the molding process

Fig. 5

Casting design that would require different shrinkage allowances than those normally used. Dimension X can contract freely, while dimension Y is restrained by the core used to make the hole.

used, the method of mold production, and the condition of the pattern. A draft angle of approximately 1.5
(16 mm/m, or 0.2 in./ft) is often added to design dimensions. The draft angle may be higher when manual molding techniques are used. Interior surfaces usually require somewhat more draft than exterior surfaces, and deep pockets or cavities may require considerably more draft. In special cases, draft can even be eliminated. Vertical walls can sometimes be drawn if the pattern is smooth and clean and if the drawing equipment is properly aligned. Patterns for V-process molding (vacuum molding; see the discussion in the article “Vacuum HighPressure Die Casting” in this Volume) can be made without draft because the smooth plastic film that covers the sand mold face offers almost no resistance to drawing the pattern from the mold (Ref 8). Expendable patterns, such as lost foam and lost wax patterns, can also be designed with straight walls (no draft) or even with negative draft because they do not have to be withdrawn from the mold. This can offer the casting designer greater shape flexibility without the use of expensive cores. The machining allowance (or machine finish allowance) provides for sufficient excess metal on all cast surfaces that require finish machining. The required machining allowance depends on many factors, including the metal cast, the size and shape of the casting, casting surface roughness and surface defects that can be expected, and the distortion and dimensional tolerances of the casting that are expected. Accurate patterns combined with automated molding can often produce close-tolerance castings with a minimum machining allowance that can reduce final machining costs considerably. Table 2 lists suggested machining allowances

492 / Principles and Practices of Shape Casting for low-production-volume castings. However, the machining allowance for a given casting may vary greatly, depending on customer requirements and casting process capabilities. More demanding dimensional requirements should be discussed with the producing foundry prior to specification. Distortion Allowances. Certain less rigid cast shapes and castings with varied section thicknesses sometimes distort when reproduced from straight or perfect patterns. This distortion is caused by the nonuniform contraction stresses during the solidification of irregularly shaped designs. In some casts, this casting distortion appears after machining. Minor distortions can sometimes be corrected by mechanically pressing or straightening castings, but if distortions are consistent and prominent, the pattern shape can be intentionally changed (faked) to counteract the casting distortions. The properly “distorted” pattern will then produce a distortion-free casting. Prior consultation with the foundry is necessary to review their experience with the warpage distortion of similarly shaped castings.

Table 2 Suggested pattern machine finish allowances Allowances, mm (in.) Pattern size, mm (in.)

Bore

Surface

Cope side

For cast irons Up to 152 (6) 152–305 (6–12) 305–510 (12–20) 510–915 (20–36) 915–1524 (36–60)

3.2 (1/8) 3.2 (1/8) 4.8 (3/16) 6.4 (1/4) 7.9 (5/16)

2.4 3.2 4.0 4.8 4.8

(3/32) (1/8) (5/32) (3/16) (3/16)

For cast steels Up to 152 (6) 152–305 (6–12) 305–510 (12–20) 510–915 (20–36) 915–1524 (36–60)

3.2 (1/8) 6.4 (1/4) 6.4 (1/4) 7.1 (9/32) 7.9 (5/16)

3.2 4.8 6.4 6.4 6.4

(1/8) (3/16) (1/4) (1/4) (1/4)

6.4 (1/4) 6.4 (1/4) 7.9 (5/16) 9.6 (3/8) 12.7 (1/2)

For nonferrous alloys Up to 76 (3) 76–152 (3–6) 152–305 (6–12) 305–510 (12–20) 510–915 (20–36) 915–1524 (36–60)

1.6 (1/16) 2.4 (3/32) 2.4 (3/32) 3.2 (1/8) 3.2 (1/8) 4.0 (5/32)

1.6 1.6 1.6 2.4 3.2 3.2

(1/16) (1/16) (1/16) (3/32) (1/8) (1/8)

1.6 (1/16) 2.4 (3/32) 3.2 (1/8) 3.2 (1/8) 4.0 (5/32) 4.8 (3/16)

4.8 6.4 6.4 6.4 7.9

(3/16) (1/4) (1/4) (1/4) (5/16)

Pattern Materials A wide variety of both traditional materials and advanced composite materials are used in the manufacture of reusable patterns and core boxes (Ref 9 –12). These pattern materials vary greatly in their physical characteristics, the techniques used for manufacturing and repairing the patterns, and the applications for which they are best suited. Pattern material selection is based on many factors, including the number of castings to be made, the final casting dimensional accuracy required, the molding or coremaking process to be used, and the size and shape of the casting. General pattern standards concerning the use and specification of patterns and pattern materials have been published by the National Association of Pattern Manufacturers, but they are not very widely used. Table 3 lists some general characteristics and properties for four commonly used pattern materials: wood, aluminum, cast iron, and polyurethane. Improvements in pattern performance and durability sometimes require that patterns be made of a combination of materials (such as wood with metal or plastic inserts on wear surfaces), coated materials, or composite materials. Wood is used for both master patterns and production patterns because it is inexpensive and easy to shape (Ref 1). However, wood patterns, especially those constructed from softer woods, are susceptible to shrinkage or swelling and warpage due to changes in atmospheric moisture. Figure 6 shows how atmospheric humidity can affect the dimensions of woods used in patternmaking. The ideal moisture content of lumber as-received is 5%. When the initial moisture content of unprotected mahogany or northern pine is increased by an additional 5% (due to an atmospheric humidity of 50%), a volumetric expansion of 1.5% results. At an atmospheric humidity of 75%, the volumetric expansion may be 3% or more. From these data, it is apparent that appreciable changes in

Source: Ref 1

Table 3 Characteristics of pattern materials Pattern material Characteristic

Machinability Wear resistance Strength Repairability Corrosion resistance

Wood Aluminum

E P P E E

G G G F E

E, excellent; G, good; F, fair; P, poor

Cast iron

Polyurethane

F E E G P

G E F E E

Fig. 6

Effects of moisture content of wood and atmospheric humidity on the expansion and shrinkage of three pattern woods

humidity can result in loss of dimensional accuracy, which translates to a corresponding loss in the dimensional accuracy of castings. Proper construction of wood patterns and the use of plywood can inhibit warpage somewhat. Good construction also calls for three or more coats of pattern lacquer or paints, which minimize moisture absorption and the resultant volumetric changes in the pattern. Patterns can also be made from compressed wood laminates (primarily plywood). These hardwood laminates weigh approximately twice as much as untreated wood but have a high strength-to-weight ratio. In addition to their strength, they are tough and have much more resistance to abrasion and moisture than untreated wood. They are readily machined with conventional wood patternmaking machinery, with some adjustment in speeds, feeds, and cutting tools. The life of wood laminate patterns can be considerably longer than that of conventional wood patterns. When subjected to high humidity for long periods of time, these patterns swell so slowly that dimensional changes and warping are negligible for many applications. Protective coatings make wood laminate patterns much more impervious to the humidity conditions typical of foundry operations. Machinable plastics, although more expensive than wood laminates, are commonly used for smaller patterns because of their improved dimensional stability and durability. These plastic pattern materials (typically, polyurethanes) can be formed and constructed with standard woodworking equipment and techniques. Metal pattern equipment is particularly well suited to long production runs. Compared to wood patterns, metal patterns and core boxes are more costly, stronger, more abrasion resistant, and are dimensionally stable under changing humidity. They can be fabricated for use with automated high-pressure molding equipment, can be produced to close dimensional tolerances, and can be used for a large number of castings before pattern repair or rework is necessary. Metal patterns and core boxes are normally made from aluminum alloys, gray iron, ductile iron, steel, or stainless steel. Metal patterns are required for shell molds and cores and molds from sands bonded with heat-setting resins. Patterns used for this purpose are typically made of cast iron because aluminum tends to gall and wear quickly at elevated temperatures. Pattern material selection is based on wear resistance, dimensional stability, machinability, and the ability to provide a smooth surface finish after machining. Metal patterns are typically computer numerical control (CNC) machined on conventional machine tools. However, aluminum patterns may be cast to shape. Aluminum patterns are cast to shape in plaster (ceramic) or sand molds made from wood master patterns. The master pattern must be oversized to account for both the shrinkage of

Patterns and Patternmaking / 493 the metal pattern as it cools and the final shrinkage of the metal casting as it cools in the mold made from the metal pattern (double shrinkage allowance). Precision ceramic molding processes with dimensional accuracies of þ0.001  mm/mm (þ0.001 in./in.) can be effectively  used to produce cast-to-size patterns with good as-cast surface finishes and minimum final machining (Ref 13). After metal patterns are cast, gating and flash must be removed, and considerable machining, filling of surface imperfections, and polishing are required so that the final pattern surfaces are smooth and free of imperfections. These finishing operations may require 50% or more of the time needed for pattern construction (Ref 14). Metal patterns and core boxes are typically machined directly from metal stock without the intermediate steps involved in producing a master pattern and the initial casting of semifinished pattern plates. The expanding use of CNC machining and CAD data bases for parts dictates that direct machining of pattern equipment will continue to grow (see the section “Trends in Pattern Design and Manufacture” in this article). The abrasion resistance of metal patterns can be improved even further by the use of abrasion-resistant coatings. Both electrolytically deposited hard chromium platings and electroless nickel platings can increase the life of metal pattern equipment. Polyurethane plastic inserts adhesively bonded to high-wear locations on metal patterns can also increase metal pattern life. Plastic Pattern Equipment. The versatility, desirable properties, and cost effectiveness of plastic pattern equipment have resulted in the continued growth of plastics for patterns and core boxes. Plastics are primarily used for small- and intermediate-sized patterns. They are more durable but are also usually more expensive than wood patterns. Plastic patterns and core boxes have largely replaced wood patterns in medium-volume applications and can also be a cost-effective substitute for metal high-production pattern equipment in some cases. Plastic patterns have excellent dimensional stability and can be produced with less skill and expense than comparable metal patterns. They have been successfully used in high-pressure molding machines (Ref 10). In addition, the polyurethane and epoxy resins typically used for patterns have the following desirable characteristics:  Better

compressive strength, bending strength, abrasion resistance, and impact resistance than wood  Resistance to environmental and chemical attack  Good bonding to other pattern materials for repairs and inserts  Easy release from molding sands Plastic patterns can be produced by traditional master pattern replication methods, or

patterns and match plates can be directly machined from plastic stock. Replication epoxy patterns are suitable for small patterns and core boxes for limited production. Larger or higher-production patterns often require glass fiber reinforcement of the epoxy resin for increased strength and durability. Epoxy pattern life can also be extended with metal coatings. The polyurethane elastomers are used much more extensively than the epoxies and are becoming one of the most widely used pattern materials. The urethanes are more wear resistant and less brittle than the epoxies. They are used for complete patterns and core boxes, for loose pieces on wood or metal pattern plates, as a thin pattern face skin material, and as wear-resistant inserts on wood or metal patterns. Techniques for producing plastic patterns depend on the type of pattern required, the type of resin system used, and the choice of reinforcement or backing, if any. Plastic patterns and gates and risers are usually mounted on metal pattern plates, but one-piece pattern and pattern plates can be manufactured. Because of the high cost of the resins and the distortion caused by the heat generated during the curing process, less expensive rigid backing materials are used whenever possible for medium and large patterns and core boxes. Metal reinforcing frames can also be used to stiffen and support large pattern plates for high-pressure molding equipment. Traditional casting and layup techniques, starting with liquid resins, are still common but have been replaced to a significant extent by direct machining of plastic board stock. When properly made, plastic patterns shrink very little; therefore, the molds do not require double shrinkage allowance, as in the production of cast aluminum patterns from master patterns. This simplifies construction and results in a considerable cost savings for designs with intricate contours. Solid Replication Plastic Patterns. These traditional patterns are manufactured using either a reverse master pattern or a reverse plaster mold made from a master pattern (usually wood). A sealer and a release agent are applied to the negative pattern or negative mold surface to prevent the resin from bonding to the mold. The resin system (usually two-part or threepart) is poured into the mold and allowed to harden. Curing times and temperatures vary greatly, depending on the resin and specific resin formulation used. Cured patterns can be readily stripped and mounted on pattern boards. Fillers are often added to resins to reduce the amount of resin needed to make a pattern, to provide strengthening reinforcement, and/or to control shrinkage. A wide variety of mineral or metal powders are used as filter materials. Other low-density filler materials can be used when strength is not critical for large patterns. By reducing the amount of casting resin for a given volume with fillers, the temperature rise during the early stages of curing can be

lessened to minimize overall pattern shrinkage and distortion. A variation of the solid casting method (face casting) uses a 1 to 3 mm (0.04 to 0.12 in.) face or skin layer of resin that is allowed to cure partially. The remainder of the mold is then filled with a less expensive backing mix that is heavily reinforced with fillers such as silica sand. Thicker solid molds can also be built up in this way rather than with resins alone. Thin, brush-on surface (gel) coats can also be applied to wood patterns to produce inexpensive patterns with considerably longer pattern life than wood alone. Patterns can also be produced with a core. An undersized core (plug) of wood, aluminum, plastic foam, or some other inert material is supported in the reverse pattern mold. The resin is poured or injected under pressure to fill the clearance space between the mold and the core and then cured to create the pattern surface. The use of the core minimizes the amount of resin needed for a given pattern volume, which minimizes shrinkage during curing. This method is most applicable for small- and medium-sized castings of simple shape (Ref 15, 16). A wide variety of techniques have been developed for building composite pattern equipment using plastic resins. Resin skins can be backed or supported by layers of glass fiber reinforcement. The rest of the mold is then filled with a suitable backing material using methods described previously. Machined Plastic Patterns. Patterns can also be readily fabricated from machinable polyurethane boards using the same patternmaking techniques and patternmaking tools used for wood or metal patterns. The additional cost of using urethane stock rather than wood is justified when patterns must be made to close dimensional tolerances. Patterns of this type are usually machined directly from CAD files. Large patterns of this type can be fabricated by using the machinable plastic board only as a facing material, along with a cheaper foam core or other backing material to support the urethane face to be machined. Pattern Coatings. The performance and service lives of wood, metal, and plastic patterns can be improved with an appropriate wearresistant coating. Standard electrolytic plating methods, electroless nickel plating, and metal spraying (metallizing) can all be used (Ref 17, 18). Table 4 illustrates the coating methods that are applicable for the various pattern substrates. Figure 7 illustrates representative pattern wear rates for coated and uncoated pattern materials. In the study, laboratory sand erosion tests were conducted to evaluate the relative wear resistance of various coated and uncoated pattern materials. The test method was thought to simulate the impact abrasion that occurs on patterns from core blowers or other sand-blowing equipment. These results may not simulate other pattern abrasive wear conditions, because wear rates for different materials depend strongly on the angle of attack of the abrading

494 / Principles and Practices of Shape Casting particle. The relative wear resistance of the various pattern materials is indicated by the percentage of weight loss during abrasion testing. Materials with less weight loss are more abrasion resistant and would likely have longer pattern service lives. The formulations of the epoxy and urethane materials are continually being improved; current-generation plastic pattern materials can be expected to have improved wear resistance. The methods for applying metal coatings to the various pattern substrates are briefly described. More detailed descriptions of these coating techniques are available in Surface Engineering, Volume 5 of ASM Handbook. In electrolytic plating, the coating is electrolytically deposited on patterns negatively charged (anode) in aqueous solutions containing metal salts. Pattern coating thicknesses usually range from 0.05 to 0.13 mm (0.002 to 0.005 in.). Coating thickness may vary somewhat throughout the pattern, depending on the distance from the electrode. In electroless nickel plating, a nickel alloy containing 8 to 10% P is chemically plated from a hot aqueous solution of nickel salts and activators. A typical plating cycle consists of 6 to 8 h in an 80
C (180
F) bath. The coatings produced have a high hardness and low coefficient of friction. Average coating thicknesses range from 0.08 to 0.13 mm (0.003 to Table 4 Pattern coating materials and methods used to enhance wear resistance Applicability to pattern substrate material Coating/method

Wood

Epoxy

Aluminum

Cast iron

Metal spraying Tin-zinc alloys Tin-bismuth alloys Aluminum-zinc alloys

X X ...

X X X

... ... ...

... ... ...

Electrolytic plating Nickel Chromium

... ...

... X

X X

X X

Electroless plating Nickel

...

X

X

X

0.005 in.) or more. The as-deposited hardness of electroless nickel can be substantially increased by subsequent heat treatment at temperatures between 290 and 400
C (550 and 750
F). Epoxy pattern equipment can also be electroless nickel coated. Metal Spraying (Metallizing). Metal spraying is the most rapid and least expensive method of applying a thin metal coating on a wood, epoxy, or metal substrate. Because a sprayed metal coating can be applied uniformly or nonuniformly by directing the spray pattern, metal spraying can also be used to rebuild areas of patterns where excessive wear has taken place. Arc spray or plasma spray guns are used to spray atomized liquid metal onto a pattern. Low-melting alloys can even be sprayed on wood patterns. Coating thicknesses of 1.5 to 2 mm (0.060 to 0.080 in.) are common except when pattern repair necessitates thicker coatings. Sprayed metal coatings are often more nonuniform than other plated coatings and therefore may require more finishing and polishing. Metal spraying can also be used to coat a negative mold (usually epoxy coated with a release agent), which is then backed up with cast epoxy. The metallized epoxy pattern is then removed from the negative mold. Pattern Repair and Rebuilding. Pattern rebuilding must be carried out on a regular basis to repair patterns that have been damaged by use or abuse and to build up worn areas of the pattern that are no longer producing castings within tolerance. The rebuilding procedures used depend on the pattern material and the service life required for the rebuilt pattern. Epoxies and urethanes are widely used to rebuild all types of pattern materials. Metal spraying or weld overlaying can also be used to build up worn areas of production metal patterns. Pattern Handling and Storage. Because significant investments are made in pattern equipment, pattern handling and storage must be carefully considered by the foundry. Improper storage and handling can cause pattern deterioration and damage, particularly for wood patterns because humidity and temperature changes can cause severe distortion. The foundry typically has many patterns in inventory. Obsolete or discontinued patterns are routinely scrapped to make room for new patterns after contacting the owner. See the article “Maintenance, Repair, Alterations, and Storage of Patterns and Tooling” in this Volume;

Selection of Pattern Type The type of pattern or core box used and the pattern or core box material used for a given set of castings depend on the following fundamental factors:

Fig. 7

Average weight losses (percent) of pattern materials tested for 12 h in silica erosion. Source: Ref 19

 The number of castings to be produced  The molding or coremaking process to be

employed

 The casting design  The dimensional tolerances required

Both the life and cost of a pattern can vary dramatically, depending on the pattern material and the type of pattern equipment. In the developmental stage of a pattern design, only a few prototype castings need to be produced before modifications are made to the pattern dimensions or to the gating and risering. If such revisions are likely, an inexpensive wood or plastic pattern is often used first. This will enable engineering changes to be made quickly and inexpensively. After the design and the tolerances of the casting have been approved, a permanent pattern is selected based on production quantity and the molding or coremaking processes to be used. Costs that are influenced by pattern equipment selection depend largely on the pattern material and pattern type and are dictated by production quantity. Expensive pattern equipment can often be justified if production quantities are high. The complexity of the pattern and the quality of the material used to make the pattern generally increase with the number of castings to be produced from one set of patterns. For example, an unmounted or loose softwood pattern could be used only for very limited production before it would require repair or replacement. A similar pattern made from a more durable material and mounted on a pattern board would increase the useful life of the pattern dramatically. Table 5 lists the basic types of patterns, indicating typical pattern materials and expected pattern lives for each during automatic green sand molding. Figure 8 indicates the general relationship between casting cost and production rate for various reusable pattern materials. Foundries also use various molding processes, casting processes, and coremaking processes. Therefore, during molding, patterns and core boxes can be subjected to differing degrees of abrasion, temperature, and stress that affect the pattern type and choice of pattern material. For example, the stresses that the pattern encounters during automated high-pressure green sand molding necessitate the use of high-strength pattern materials and rigid pattern construction. Shell molds or cores can be made only from metal patterns because they are heated to temperatures of approximately 260
C (500
F). In V-process molding, a thin polyethylene sheet prevents the sand mold from contacting the pattern during molding, and this allows inexpensive wood patterns to have an almost indefinite pattern life. Tolerances. The type of pattern used also has a significant effect on the casting tolerances that can be obtained and maintained. In general, final casting tolerances can be held within tighter limits as the rigidity and durability of the pattern equipment increase. When close tolerances are required, it may be desirable to construct the pattern equipment so that it can be

Patterns and Patternmaking / 495 Table 5

Approximate life before repair of various pattern materials

Approximate number of castings produced before pattern repairs Pattern

Core box

Small castings (largest dimension: 610 mm, or 24 in.) 200 300 2000 2000 6000

6000

100,000

100,000

Medium castings (largest dimension: 1.8 m, or 72 in.) 100 100 1000 750 3000

3000

Large castings (largest dimension: > 1.8 m, or 72 in.) 50 50 200 150

500

500

Materials

Hardwood patterns and core boxes Hardwood patterns and core boxes, wearing surfaces faced with metal Aluminum patterns and core boxes; plastic match plate patterns; urethane patterns or urethane-lined magnesiumframed core boxes Cast iron patterns and core boxes Hardwood patterns and core boxes Hardwood patterns and core boxes, wearing surfaces faced with metal Aluminum patterns and core boxes; urethane inserts in wear areas Softwood patterns and core boxes Softwood patterns with exposed projections metal faced; softwood core boxes, metal faced Hardwood patterns reinforced with metal; hardwood, metal-faced core boxes

Source: Ref 7

Fig. 8

Generalized casting costs versus production quantity for four pattern materials

dimensionally adjusted based on the results of prototype castings. These adjustments may involve refitting and remachining.

Trends in Pattern Design and Manufacture The continued development of computeraided design and manufacturing (CAD/CAM) technology has had a continuing effect on both pattern design and patternmaking methods (Ref 20–23). The influence of CAD/CAM technology and CNC technology will continue to grow as they become more affordable for the small

foundry and the pattern shop. Maximum benefit can be achieved with CAD/CAM technologies if pattern design, inspection, and manufacture are fully integrated. This integrated patternmaking approach begins with a geometric data base of the part to be cast. The data base may already exist or may have to be constructed from drawings. The pattern designer adds a parting line and draft and machining allowances at the CAD workstation. The part dimensions can be automatically scaled up to the designed pattern dimensions that included the shrinkage allowance by inputting the appropriate scale factor. At the current stage of development of many CAD software systems, necessary pattern features such as the parting line and draft must be added interactively by the pattern designer. To complete the pattern design, appropriate gate and riser sizes will be determined and added, using one of a number of stand-alone software packages. Solidification modeling software is often used to model the solidification of the final resultant casting from geometry data alone before even prototype patterns are made (see the articles in the Section “Modeling and Analysis of Casting Processes” in this Volume). Pattern dimensions, gating, and riser design can all be evaluated and readily modified by adjusting the pattern geometry data base. The use of solidification modeling programs can dramatically reduce or eliminate pattern prototyping stages. Casting design changes and/or modifications are common even after the pattern is built, but these can also be easily performed by adjusting the pattern data base. Direct CNC manufacture of patterns offers many advantages. Close-tolerance patterns can be readily produced using the pattern geometry data base. The CNC part programmer can

retrieve the pattern geometry model and use that model to construct the appropriate cutter paths for pattern machining. Graphical simulation of the machining operation on the CAD terminal can be used to verify the correctness of the CNC part program before actual pattern machining takes place. Any future modifications to the pattern geometry can be readily incorporated. Changes are made to the pattern data base, and the modified CNC output is used to modify the appropriate pattern dimensions. Multiple-impression tooling can be made with the assurance that each impression is dimensionally accurate. An example of the benefits that can be obtained using CAD/CAM patternmaking techniques compared to conventional patternmaking techniques is shown in Table 6. A 17% reduction in the time necessary to produce a final pattern is shown. Complete CNC machining of patterns will continue to expand as the benefits of CAD/ CAM technology continue to be exploited. The close dimensional tolerances and complex geometries required on many patterns are more suited to CNC pattern machining than to manual construction by patternmakers and machinists. Verification of pattern and casting dimensions is also a tedious process when traditional manual layout techniques are used. An automated coordinate-measuring machine (CMM) is an extremely accurate inspection tool that can dramatically improve the speed and reliability of pattern inspection (Ref 21). Such machines provide absolute measurement capabilities in three dimensions simultaneously; this eliminates the need to make tedious comparisons between the pattern and gage blocks or other length standards. With portable robotic arm CMMs, even large patterns or complex patterns with undercuts can be easily inspected. Data storage and printout capabilities allow easy measurement of many important pattern features. For example, cope and drag pattern plate alignment can be readily determined from comparisons of the locating pin and bushing positions for each pattern plate. Pattern dimensions or pattern inserts can also be inspected at specific intervals during pattern use with little downtime for monitoring pattern wear.

REFERENCES 1. Patternmaker’s Manual, American Foundry Society, 2006 2. E. Hamilton, AFS Patternmaker’s Guide, American Foundry Society, 1976 3. W.A. Blower, Pattern Design for High Pressure Molding, Trans. AFS, Vol 78, 1980, p 313–316 4. Patterns for High Pressure Molding, Foundry, Vol 98 (No. 10), Oct 1971, p 83–84 5. Metalcasting and Molding Processes, American Foundry Society, 1987

496 / Principles and Practices of Shape Casting Table 6 Comparison of traditional and CAD/CAM patternmaking times for an aluminum investment die Technique

Time

Traditional methods (accuracy of pattern: þ0.01 in.)  1) Design 50 h 2) Wooden model 50 h 3) Duplicating 40 h 4) Other machining 85 h 5) Polishing and finishing 40 h Total

265 h

CAD/CAM methods (accuracy of pattern: þ0.002 in.)  1) CAD/CAM design 40 h 2) CNC cutter path development 40 h 3) CNC machining 25 h 4) Other machining 85 h 5) Polishing and finishing 30 h Total

220 h

9. 10.

11. 12.

13. 14.

6. D.R. Dreger, Smooth No-Draft Castings, Mach. Des., May 25, 1978, p 63–65 7. M. Blair and T.L. Stevens, Ed., Steel Castings Handbook, 6th ed., and ASM International, 1995 8. J.L. Gaindhar, C.K. Jain, and K. Subbarathnamatah, Prediction of Pattern Dimensions

15.

16.

for V-Process Precision Castings Through Response Surface Methodology, Trans. AFS, Vol 94, 1986, p 343–349 R. Brown, Plastic Patterns for High Pressure Molding, Mod. Cast., Vol 73, Nov 1983, p 41–43 J.W. Francis, Practical Patternmaking Techniques for the Foundry Industry, Br. Foundryman, Vol 73 (No. 9), 1980, p 258–264 J.W. Francis, Practical Patternmaking Techniques for the Foundry Industry, FWP J., Vol 24 (No. 6), 1984, p 29–44 J.D. Pollard, Materials for Today’s Patternshop—A Personal Choice, Foundry Trade J., Vol 158 (No. 3306), May 23, 1985, p 415–419 M.J. Sneden, Who’s Afraid of Cast-to-Size Tooling, Trans. AFS, Vol 85, 1977, p 9–14 R.L. Allen, Cold Box Design Specializing in Urethane Lined Tooling, Trans. AFS, Vol 85, 1977, p 323–326 J. Sheeham and D. Richardson, The HMP Process—A New Method for Producing Plastic Patterns, Trans. AFS, Vol 92, 1984, p 203–208 G. Anderson and T.J. Crowley, Precision Foundry Tooling Utilizing CAD/CAM and

17. 18. 19. 20.

21. 22. 23.

the HMP Process, Trans. AFS, Vol 93, 1985, p 895–900 M.K. Young, Sprayed Metal Foundry Patterns for Short Run and Production Equipment, Trans. AFS, Vol 88, 1980, p 217–223 J.R. Henry, Metallic Coatings for Patterns and Coreboxes, Mod. Cast., April 1983, p 22–24 R.D. Maier and J.F. Wallace, Pattern Material Wear or Erosion Studies, Trans. AFS, Vol 84, 1977, p 161–166 D.R. Westlund, G.R. Anderson, and A.G. Anderson, Applying CAD/CAM to Foundry Tooling, Mod. Cast., Jan 1984, p 20–24 J.D. Taylor, Coordinate Measuring Machine Application in the Pattern Shop, Trans. AFS, Vol 88, 1980, p 195–198 J.R. Woods, NC-CNC and CAM in the Pattern Shop, Trans. AFS, Vol 91, 1983, p 743–746 D.B. Welbourn, CAD/CAM Plays a Major Role in Foundry Economics, Mod. Cast., Vol 77 (No. 9), Sept 1987, p 40–42

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 497-505 DOI: 10.1361/asmhba0005277

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Casting Practice—Guidelines for Effective Production of Reliable Castings Jonh Campbell, The University of Birmingham, England

AS OUR UNDERSTANDING of the casting process is improved and refined, guidelines for effective production of reliable castings continue to evolve. In recent years, there has been a significant increase and improvement in our understanding of the casting process. This has led to new insights and criteria, which the author has steadily added to the list of requirements as they have become known. From an initial list of four rules, now ten rules have been identified that incorporate the latest technology for the production of reliable castings. These are just the start. It is already known that additional rules exist, but these remain to be further researched and clarified. The ten rules of casting, as originally described in Ref 1 and further developed in Ref 2, are proposed as being necessary but not sufficient for the manufacture of reliable castings. It is proposed that they are used in addition to existing conventional technical specifications, such as alloy type, strength, and traceability in accordance with ISO 9000 and so on, and other well-known and wellunderstood conventional foundry controls, such as casting temperature and other parameters. There are also basic reasons for believing that the ten rules have general validity, even though they have not been tested on all cast materials. Some rules have obvious implications and are applicable to all types of metals and alloys, including those based on aluminum, zinc, magnesium, cast irons, steels, air- and vacuum-cast nickel and cobalt, and titanium. Nevertheless, although all materials will probably benefit from the application of the rules; some may benefit more whereas others will be less affected. The rules as outlined here can be viewed as a draft process specification that a buyer of castings may use to purchase the best possible casting quality. Conversely, adherence to these rules is intended to assist the casting industry by helping to ensure the quality and reliability of the castings in a cost-effective way (as opposed to more expensive quality control on the finished product). The rules can help speed up the process of producing the casting right the first time and should contribute in a major way to the reduction of scrap when the casting goes into production. The ten

rules may also help raise the reliability of castings to be equal to that of forgings that can be offered with confidence.

Rule 1: Good-Quality Melt Immediately prior to casting, the melt shall be prepared and treated, if necessary, using the best current practice. Naturally, it is of no use to incorporate the best designs of filling and feeding systems if the original melt is of such poor quality, or perhaps already damaged, that a good casting cannot be made from it. It is a requirement that either the process for the production and treatment of the melt has been shown to produce good-quality liquid, or the melt should be demonstrated to be of good quality. A goodquality liquid is one that is substantially free from nonmetallic inclusions. It should be noted that such melts are not to be assumed and, without proper treatment, are probably rare. (Additional requirements, not part of this specification, may also be placed on the melt, for example, low gas content or low values of particular solute impurities.) An interesting possibility for future specifications for aluminum alloy castings (where, fortunately, residual gas in supersaturated solution does not appear to be harmful) is that a double requirement may be made for the content of dissolved gas in the melt to be high but the percentage of gas porosity in the casting to be low. The meeting of this double requirement will ensure the customer that double-oxide films are not present. This is because these damaging but undetectable defects, if present, will effectively be labeled and made plainly visible on x-ray radiographs and polished sections by the precipitation of dissolved gas.

Maximum meniscus velocity is in the range of 0.4 to 0.6 m/s (1.3 to 2.0 ft/s) for most liquid metals. This maximum velocity may be raised in constrained running systems or thin-section castings. This requirement also implies that the liquid metal must not be allowed to fall more than the critical height (which corresponds to the height of a sessile drop of the liquid metal). Maximum Velocity Requirement. Recent research has demonstrated that if the liquid velocity exceeds a critical velocity, there is a danger that the surface of the liquid metal may be folded over by surface turbulence (Fig. 1). Therefore, there is a chance that the surface oxide film may be folded into the bulk of the liquid if its speed of advance of the liquid front exceeds this critical velocity. The foldedin films constitute initiation sites for gas precipitation, shrinkage cavities, and hot tears, and after being frozen into the casting become effective cracks, lowering the strength and fatigue resistance. The folded films may also create leak paths, causing leakage failures. Thus, the entrainment of the liquid surface by surface turbulence leads to a catalog of problems that beset the foundry engineer. The folding-in of the oxide is a random process, leading to scatter and unreliability in the properties and performance of the casting on a casting-to-casting, day-to-day, and monthto-month basis during a production run.

Rule 2: Liquid Front Damage This is the requirement that the liquid metal front (the meniscus) should not go too fast.

Fig. 1

Effect of surface turbulence in entraining double (folded-over) oxide films, which form cracks and pores in the liquid

498 / Principles and Practices of Shape Casting The critical velocity is close to 0.4 m/s (1.3 ft/s) for dense alloys such as irons, steels, and bronzes; it is 0.5 m/s (1.6 ft/s) for liquid aluminum alloys, and 0.6 m/s (2.0 ft/s) for liquid magnesium alloys. The maximum velocity condition effectively forbids top-gating of castings. This is because liquid aluminum reaches its critical velocity of approximately 0.5 m/s (1.6 ft/s) after falling only 12.5 mm (0.5 in.) under gravity. The critical velocity of liquid iron or steel is exceeded after a fall of only approximately 8 mm (0.3 in.). Such short fall distances are always exceeded in practice in top-gated castings, leading to much incorporation of the surface oxide films and consequent leakage and crack defects. Castings that are made in which velocities everywhere in the mold never exceed the critical velocity are consistently strong, with high fatigue resistance, and are leaktight (if properly fed so as to be free from shrinkage porosity). Experiments on the casting of aluminum have demonstrated that the strength of castings may be reduced by as much as 90% if the critical velocity is exceeded. The corresponding defects in the castings are not always detected by conventional nondestructive testing such as x-ray radiography or dye penetrant, since, despite their large area, the folded oxide films are too thin to detect. The speed requirement automatically excludes conventional pressure die castings, since the filling speeds are over an order of magnitude in excess of the critical velocity. Some recent special developments of highpressure technology are capable of meeting this requirement, however. These include the vertical injection squeeze casting machine by UBE Machinery Corp. and the shot control technique by Buhler, both of which can, in principle, be operated to fill the cavity through large gates at low speeds and without the entrainment of air into the liquid metal. It is necessary to saw, rather than break, such castings from their filling systems. Other uphill filling techniques, such as the Cosworth process and low-pressure systems, are capable of meeting rule 2. Even so, it is regrettable that many low-pressure die casting machines are so poorly controlled on flow rate that the speed of entry into the die greatly exceeds the critical velocity, thus negating one of the most important potential benefits of the low-pressure system. In addition, the critical velocity is practically always exceeded during the filling of the low-pressure furnace itself because of the severe fall of the metal as it is transferred into the pressure vessel, so that the metal is damaged even prior to casting. Low-pressure machines with interchangeable crucibles can avoid this problem. Metals that behave in the same way, suffering damage from the entrainment of their surface films, are suggested to be the zincaluminum alloys and ductile irons. Carbon steels and some stainless steels are thought to be similar, although in some of these systems,

the entrained oxides agglomerate because they are partially molten and thus float to form surface imperfections in the form of slag macroinclusions on the surface. Duplex stainless steels exhibit a tough surface oxide film on the melt that gives special problems if surface entrainment is permitted. For a few materials, particularly alloys based on the Cu10Al-type bronzes, the critical velocities were thought to be much lower, in the region of 0.075 m/s (0.25 ft/s). However, from recent work at the University of Birmingham, this seems to be a mistake, probably resulting from the confusion caused by bubbles entrained in the early part of the filling system. With well-designed filling systems, the aluminum bronzes fulfill the theoretically predicted 0.4 m/s (1.3 ft/s) value for a critical ingate velocity. No-Fall Requirement. It is quickly shown that if liquid aluminum is allowed to fall more than 12.5 mm (0.5 in.), it exceeds the critical 0.5 m/s (1.6 ft/s). Similar critical velocities and critical fall heights can be defined for other liquid metals. The critical fall heights for all liquid metals are in the range of 3 to 15 mm (0.12 to 0.6 in.). Thus, any form of pouring, such as in transfer between ladles (Fig. 2) or top-gating of castings, almost without exception will lead to a violation of the critical velocity requirement. Many forms of gating that enter the mold cavity at the mold joint will also violate this requirement if any significant part of the cavity is below the joint. In fact, for conventional sand and gravity die (permanent mold) casting, the requirement effectively dictates bottom-gating into the mold cavity and excludes any other form of gating at any other height. Also excluded are any filling methods that cause waterfall effects in the mold cavity. Avoiding waterfalls dictates the siting of a separate ingate at every isolated low point on the casting. The initial fall down the sprue in gravityfilled systems does introduce some oxide damage into the metal, especially if using an oversized sprue where the metal is not constrained (Fig. 3). However, for some alloys, much of the oxide introduced in this way is attached to the walls of the sprue and does not appear to find its way through to the mold cavity. This surprising effect is clearly seen in many topgated castings, where most of the oxide damage (and particularly any random leakage problem) is confined to the area of the casting under the point of pouring, where the metal falls. Although extensive damage does not seem to extend into those regions of the casting where the speed of the metal front decreases and where the front travels uphill, there appears to be a carryover of some concentration of defects. Thus, the provision of a filter immediately after the completion of the fall down the sprue is valuable. However, some damage is still expected to pass through the filter. The requirement for no fall of the liquid within the running system (after the metal

Fig. 2

Effect of increasing height on a falling stream of liquid. (a) Oxide film remains intact. (b) Oxide film detaches and accumulates to form a dross ring. (c) Oxide film and air are entrained in the bulk melt.

leaves the base of the sprue) and mold cavity implies that metal must be designed to go only uphill after it leaves the base of the sprue. Thus, the runner must be in the drag and the gates must be in the cope for a horizontal singlejointed mold. (If the runner is in the cope, there is the danger that the gates will fill prematurely, before the runner itself is filled; thus, air bubbles are likely to enter the gates.) The no-fall requirement may also exclude some of those filling methods in which the metal slides down a face inside the mold cavity, such as Durville casting-type processes. During the slide down a slope, the critical velocity is easily exceeded if the total loss of height exceeds the critical fall height. This effect is discussed in more detail in the following section. It is noteworthy that these precautions to avoid the entrainment of oxide films also apply

Casting Practice—Guidelines for Effective Production of Reliable Castings / 499

Fig. 3

Oversized sprue that has suffered severe erosion damage because of nonfilling during the pour. A correctly sized sprue shows a bright surface, free from damage.

application of pressure to force it into such narrow gaps. This well-known effect is known as capillary repulsion. The liquid surface is now so constrained that it is not easy to fold over the surface, that is, there is no room for splashing or droplet formation. Thus, the critical velocity is higher, and metal speeds can be raised somewhat without danger of exceeding this higher limit. In fact, tight curvature of the meniscus becomes so important in very thinwalled castings, with walls less than 2 mm (0.08 mm) thickness, that filling can sometimes be without regard to gravity (i.e., can be uphill or downhill), since the effect of gravity is swamped by the effect of surface tension. The dominance of surface tension also makes the uphill filling of such thin sections problematic, because the effective surface tension exceeds the effect of gravity, so that instabilities can occur, whereby the moving parts of the meniscus continue to move because of the reduced thickness of the oxide skin at the advancing location. Other parts of the meniscus that drag back are further suppressed in their advance by the thickening oxide, so that a runaway instability condition occurs. This dendritic advance of the liquid front is no longer controlled by gravity in very thin castings, making the filling of extensive horizontal sections a major filling problem. The problem is relieved by the presence of regularly spaced ribs or other geometrical features that assist in organizing the distribution of liquid and thus avoid the starving of areas that the flow happens to avoid because of random meandering. Such chance avoidance, if prolonged, leads to the development of strong oxide films or even freezing of the liquid front. Thus, the final advance of the liquid to fill such regions is hindered or prevented altogether.

Rule 3: Liquid Front Stop Fig. 4

Unstable advance of a film-forming alloy, showing the formation of laps as the interface intermittently stops and restarts by bursting through and flooding over the surface film

to casting in (so-called) inert gas or even in a (so-called) vacuum. This is because the oxides of aluminum and magnesium (as in aluminum alloys, ductile irons, or high-temperature nickel-base alloys, for example) form so readily that they effectively “getter” the residual oxygen in any contaminated gas or conventional industrial vacuum and form strong films on the surface of the liquid. Rule 2 applies to “normal” castings with walls of thickness over 3 or 4 mm (0.12 or 0.16 in.). For very thin-walled castings, of section thickness less than 2 mm (0.08 in.), the effect of surface tension in controlling filling becomes predominant, the meniscus being confined between walls that are so much closer than the equilibrium curvature that the meniscus is effectively compressed and requires the

This is the requirement that the liquid metal front should not go too slowly and, more exactly, should not stop at any point on the front. The advancing liquid metal meniscus must be kept “alive” (i.e., moving) and therefore free from thick oxide film that may be incorporated into the casting. This is achieved by designing the liquid front to progress only uphill in a continuous uninterrupted upward advance (in the case of gravity-poured casting processes, from the base of the sprue onward). This rule implies:
Only bottom-gating is permissible. No stop-

ping due to arrest of pouring. No downhill flow (either falling or sliding)
No horizontal flow (of any significant extent) The liquid metal front must continue to advance at all points on its surface. If it stops, a thick surface film has the opportunity to form and thus resist the restarting of the flow at some later moment. The incorporation of this film in

the casting as metal flows over it from other regions forms a major defect, often spanning the casting from wall to wall (Fig. 4). This problem can occur in several ways, as detailed subsequently and discussed further in the article “Filling and Feeding System Concepts” in this Volume. If the liquid metal is allowed to either fall or spread horizontally, the unstable propagation of the front in the form of jets or streams bounded by an oxide flow tube leads to a situation where the oxide flow tube grows to a considerable thickness and is finally sealed into the casting as the liquid metal envelops it. A double-oxide film defect is created, with a highly asymmetric double layer, in which one oxide is thick and the other thin (Fig. 4). However, since dry side is arranged against dry side as usual, the defect is once again a crack. Such cracks are extensive and serious. They are different from the fragmented and chaotic double films introduced by surface turbulence. They have predictable geometric shapes, such as planes, cylinders, or riverine forms. In common with the irregular and unpredictable forms produced by surface turbulence, they are normally invisible to x-rays and to dye penetrant testing. They are only revealed by gross mechanical deformation that opens the crack. The deleterious flow tube structure that forms when filling downward or horizontally is usually eliminated when filling vertically. However, even in this favorable mode of filling, a related defect can still occur if the advance of the meniscus is stopped at any time. This effect is discussed as follows. Continuous Uphill Advance of the Meniscus. The uphill motion of the liquid front, if it can be arranged in a mold cavity, assists in keeping all parts of the front “alive;” that is, the meniscus does not become pinned in place with oxide and thus become “dead.” As explained earlier, the front becomes pinned if the rate of advance of the metal front is too slow or if it stops. If this happens, the surface oxide has the opportunity to thicken, becoming so strong that there is the danger that any further advance of the front will be prevented. The breaking through of the meniscus at a weak point will then flood over the fixed, thickened oxide, sealing it into the casting as a major oxide lap defect (Fig. 4). One of the reasons to prevent any metal fall in the mold is not only to avoid exceeding the critical velocity and thus impairing the metal that falls but also to avoid the period when the rise of the metal is interrupted, causing an oxide to form on the stationary front (while the metal level is fixed during the period of the overflow) and on the tube of flow surrounding the fall itself, which is also effectively stationary. The no-fall requirement may also exclude some of those filling methods in which the metal slides down a face inside the mold cavity, such as Durville casting-type processes. These processes can be used to good advantage if used

500 / Principles and Practices of Shape Casting correctly. In such cases, the liquid metal front progresses steadily at all points, expanding and breaking its surface oxide and rolling it out against the mold wall as the front progresses. In this way, the oxide never enters the bulk of the metal; it only continues to augment the oxide already present on the surface of the casting. The production of ingots by the controlled tilting of the mold can maintain the “live” nature of the front at all times as the flow expands into the ingot mold. When a Durville-type pour is carried out badly, the metal can flow down into the mold as a stream. Its boundaries (that is, its outer surface is stationary) are formed as a tube of oxide through which the metal flows while it is running down one of the faces of the mold. This oxide flow tube becomes sealed into the finished casting, creating a curious but serious tubular crack defect. The problem is analogous to the oxide tube defect that forms around the falling metal during any waterfall effect during the filling of the casting. These flow tubes are a common defect seen in a wide variety of castings that exhibit either extensive horizontal or downhill sections. The requirement that the meniscus travels only uphill is sacred. Similar defects that have the appearance of laps also occur on the sidewalls of flows as they meander aimlessly across large horizontal parts of the mold. The rattail defect is a familiar reminder of such wanderings of the metal stream. (In this case, the rattail line is usually formed by the failure of the mold. However, a corresponding line of oxide defect in the metal casting may also be present as an additional problem.) The avoidance of extensive horizontal mold sections is therefore essential for reproducible and defect-free castings. Any horizontal sections should be avoided by the designer or by the caster by tilting the mold. (This is easily provided by some casting techniques such as the Cosworth process, where the mold is held in a rotatable fixture during casting.) The flow across such inclined planes is therefore progressive, if slow, but the continuous advance of the front at all points assists in the goal of keeping the meniscus “alive.”

Rule 4: Bubble Damage No bubbles of air entrained by the filling system should pass through the liquid metal in the mold cavity. This may be achieved by:
Use of an offset step pouring basin design of

sufficient size to facilitate detrainment of entrained air and oxides; preferred use of a stopper to aid the filling of the basin; the use of sprue and runner designed to fill in one pass; avoidance of the use of wells or other volume-increasing features of filling systems; possible use of ceramic foam filter close to sprue/runner junction; possible use of bubble traps
No interruptions to pouring

The passage of a single bubble through an oxidizable melt results in the creation of a bubble trail as a double-oxide crack in the liquid. Thus, even though the bubble may be lost by bursting at the liquid surface, the trail remains as permanent damage in the casting. Poor filling system designs can result in the entrainment of much air into the liquid stream during its traverse through the filling basin, its fall down the sprue, its travel along the runner, and its entry into the mold cavity, if the melt velocity is still too high at that point, or if the melt is allowed to fall inside the mold cavity. The repeated passage of many bubbles through the liquid metal leads to an accumulation of a tangle of bubble trails and, if the density of trails is sufficiently great, residual fragments of entrapped bubbles wedged in among the trails. This mixture is collectively christened as bubble damage (Fig. 5). Bubble damage, as a mass of oxides and fragments of porosity, is probably the most common defect in castings but is almost universally unrecognized. The problem is commonly observed just inside and above the first ingate from the runner and is often mistaken for shrinkage porosity as a result of its irregular form. However, in an aluminum alloy casting, it can be recognized on a fracture surface; instead of shiny dendrite tips characteristic of shrinkage porosity, a series of dark, nonreflective oxidized surfaces interleaved like crumpled pages of sepia-colored paper will be found. (Some associated shrinkage porosity with its dendrites will almost always be present but should not cause the major nonreflective regions to be overlooked. It is easy to fall into this trap because under the microscope, in contrast with the eye-catching regular crystallographic order of the dendrites, the oxidized regions do not attract attention.) In some stainless steels, bubble damage is seen under the microscope as a mixture of bubbles and cracks (a remarkable combination that would normally be difficult to explain as a metallurgical phenomenon), because the high cooling strain leads to high stresses in such a strong material and thus opens up the double-oxide bubble trails. Gravity-Filled Running Systems. In gravityfilled running systems, the requirement to reduce bubbles in the liquid stream during the filling of the casting calls for an offset stepped basin or other advanced basin design. The conventional conical pouring basin cannot be permitted. The cone shape usually introduces approximately 50% air, introducing massive amounts of damage. The requirement also demands properly engineered and manufactured sprues. The sprue must have a calculated taper. Preferably, the taper should not be a straight taper but a hyperbolic shape to follow the shape of the falling stream, maintaining the pressurization of the walls of the sprue at all points during the fall. Parallel or reversed-taper sprues are not permitted without some feature to reduce the damage that they cause. Such damage-limiting

features include a filter at the entrance to the runner, or inverted well, a pork-pie-shaped obstruction designed to aid the backfilling of the sprue. It is mandatory that the taper of the sprue contains no perturbations to upset the smooth fall of the liquid metal. Thus, it must fit well with the pouring basin or other mold or die joints; no steps, ledges, or abrupt changes in direction are permissible. Also, no branching or joining of other ducts, runners, gates, or sprues is allowed. All such features (especially common in investment castings) have the potential for the introduction of air into the stream or the uncontrolled escape of liquid into other parts of the mold cavity. At one time, it was mandatory that each sprue have a sprue well at its base. The well was thought to facilitate the turn of the metal through the right-angle bend into the runner with minimum turbulence. Previous work on well development had been based on water models. However, more recent work at the University of Birmingham using liquid metals has demonstrated that, at best, the well is no better than no well at all and, at worst, causes considerable extra turbulence. Thus, the new filling system designs incorporate no well at the base of the sprue. It is mandatory that no interruption to the pour occurs that leads to uncovering the entrance to the sprue, so that air enters the running system. A provision must be made for the foundry to automatically reject any castings that have suffered such an interrupted pour. Pumped and Low-Pressure Filling Systems. Pumped systems such as the Cosworth process or low-pressure casting systems into sand molds or metal dies are highly favored as having the potential to avoid the entrainment

Fig. 5

Emergence of liquid metal and air bubbles through an ingate, showing the creation of bubble trails and residual bubbles

Casting Practice—Guidelines for Effective Production of Reliable Castings / 501 of bubbles if the processes are carried out under proper control. (Good control of a potentially good process should not be assumed; it must be demonstrated. In addition, bubbles can be released erratically from the interior wall of a tube launder system as well as from the underside of a badly designed distribution plate or from a poorly maintained and cleaned launder tube, if one is used.) Although low-pressure filling systems can, in principle, satisfy the requirement for the complete avoidance of bubbles in the metal, a leaking riser tube in a low-pressure casting machine can lead to a serious violation. The stream of bubbles from a leak in the riser tube will rise up the tube and directly enter the casting. Thus, regular checks of such leakage, and the rejection of castings subjected to such consequent bubble damage, will be required. The other major problem with conventional low-pressure delivery systems where the melt is contained within a pressurized vessel is that the quality of the melt is usually damaged by the filling of the pressure vessel. This is an uncontrolled fall from the foundry transfer ladle, then down a chute, and finally into the melt. This unsatisfactory transfer process introduces much oxide into the metal, only part of which is subsequently removed by a filter at the entrance to the mold cavity. As mentioned previously, this damaging filling procedure is avoided by those low-pressure machine designs that incorporate multiple exchangeable crucibles.

Rule 5: Core Blows

occurs because the clay contains water bonded into its crystallographic structure but is impermeable. When heated by the arrival of the melt, the water vapor from the clay is prevented from escaping into the core or mold, thus forcing the escape path of the steam through the melt. The use of core and mold repair pastes is to be avoided (unless they can be demonstrated to avoid the generation of blows in the melt). To demonstrate that a core, or assembly of cores, does not produce blows may require a procedure such as the removal of all or part of the cope or overlying cores and taking a video recording of the filling of the mold. If there are any such problems, the eruption of core gases will be clearly observable, resulting in the creation of a froth of surface dross (which would normally be entrapped inside the upper walls of the casting). A series of video recordings may be necessary, showing the steady development of solutions to a core-blowing problem and recording how individual remedies resulted in progressive elimination of the problem. The video recording should be retained by the foundry for inspection by the customer for the life of the component. Any change to the filling rate of the casting, core design, core repair procedure, or change of binder will necessitate a repeat of this exercise. For castings with a vertical joint, where a cope cannot be conveniently lifted clear to provide such a view, a special sand mold may be required to carry out the demonstration that the core assembly does not cause blows from the cores at any point. This must be constructed as part of the tooling to commission the casting and justified as an investment in quality assurance.

Fig. 6

Detachment of a bubble from the top of a core, leaving a bubble trail as a permanent legacy of its journey. This bubble may be early enough to escape at the free surface of the rising metal.


No bubbles from the outgassing of cores

should pass through the liquid metal in the mold cavity. Cores must be of sufficiently low gas content and/or adequately vented to prevent bubbles from core blows.
No use of clay-based core or mold repair paste (unless demonstrated to be free from the risk of creating bubbles) Serious oxide films, which can cause leaks and mechanical cracking, can be caused by the outgassing of sand cores through the melt, causing bubbles to rise through the metal and leaving an oxidized bubble trail in its wake (Fig. 6). A succession of such bubbles from a core is very damaging to the upper parts of castings (Fig. 7). The bubble from the core contains a variety of gases, including water vapor, that are highly oxidizing to metals such as aluminum and higher-melting-point metals. Bubble trails from core blows are usually particularly noteworthy for their characteristically thick and leathery double-oxide skin, which is probably why core blows result in such efficient leak defects through the upper sections of castings. Real-time x-ray radiography of solidifying castings has revealed considerable volumes of water vapor driven into the melt from claybased core repair and mold repair pastes. This

Rule 6: Shrinkage Damage
No feeding uphill because of unreliable

pressure gradient and complications introduced by convection
Demonstrate good feeding design following all seven feeding rules, by an approved computer solidification model and test castings
Control of the level of flash at mold and core joints, the mold coat thickness, and the temperatures of metal and mold If computer modeling is not carried out, the observance of the author’s seven feeding rules is strongly recommended. Even when computer simulation is available, the seven rules are good guidelines. These rules list the requirements for good feeding under gravity. For example, Fig. 8 shows the penalty of large amounts of porosity if the feeder is too small, and even some penalty if too large. In addition, all five mechanisms for feeding (as opposed to only liquid feeding) should also be used to advantage. For example, solid feeding (a kind of self-feeding by the solidified casting) is especially useful when attempting to achieve soundness in an isolated boss or heavy

Fig. 7

(a) Core blow—a trapped bubble containing core gases. (b) Bubble trail, ending in an exfoliated dross defect as the result of the passage of copious volumes of core gas prior to freezing.

section, where the provision of feed metal by conventional techniques is impossible. Feeding with Gravity. As opposed to filling uphill (which is quite correct), feeding should only be carried out downhill (using the assistance of gravity). Attempts to feed uphill, although possible in principle, can be unreliable in practice and may lead to randomly occurring defects that

502 / Principles and Practices of Shape Casting

Fig. 8

Effect of increasing feeder solidification time on the soundness of a plate casting in Al-12Si alloy

have all the appearance of extensive shrinkage porosity. These occur because of the difficulties caused by two main problems: adverse pressure gradient and adverse density gradient leading to convection. Adverse pressure gradient is dealt with subsequently; the second problem is dealt with in “Rule 7: Convection Damage” in this article. The atmosphere is capable of holding up several meters of the head of a metal. For liquid mercury, the height is approximately 760 mm (30 in.) (the height of the barometer column). There are similar equivalent heights for other liquid metals, allowing for the density difference; thus, the atmosphere will hold up approximately 4 m (13 ft) of liquid aluminum, for example. (While no pore exists, the tensile strength of the liquid will allow the metal, in principle, to greatly exceed this height, since, in the absence of defects, it can withstand tensile stresses of thousands of atmospheres. However, the random initiation of a single minute pore will instantly cause the liquid to “break,” causing such feeding to stop and go into reverse. The casting will then empty down to the level at which atmospheric pressure can support the liquid.) However, if there is a leak path to atmosphere, allowing atmospheric pressure to be applied in the liquid metal inside the mold cavity, the melt will then fall further, attempting to equalize levels in the mold and feeder. Thus, the residual liquid will drain from the casting if the feeder is sited below the casting. This is an efficient way to cast porous castings and sound feeders. Clearly, the initiation of a leak path to atmosphere (via a double-oxide film or via a liquid region in contact with the surface at a hot spot) is rather easy in many castings, making the

whole principle of uphill feeding so risky that it should not be attempted where porosity cannot be tolerated. The recent practice of active feeding, whereby small feeders placed low on the casting are pressurized to cause the casting to feed uphill, is almost certainly allowable in small castings. For larger castings, the technique is subject to the problems of convective flow and is expected to become unreliable. Computer Modeling of Feeding. Computer modeling has demonstrated its usefulness in predicting shrinkage porosity with accuracy. It should be specified as a prior requirement before work is started on making the tooling. This minor delay will have considerable benefits in shortening the overall development time of a new casting and will greatly increase the chance of being “right the first time.” It must be recognized, however, that the computer will not necessarily be capable of any design contribution. Thus, filling and feeding will be required as inputs. Additionally, many computer simulations do not simulate the important effects due to convection, nor do some include conduction. Some neglect, or only crudely allow for, the effect of mold heating by the flow of metal during filling. The belief in the results from such models must therefore be tempered with caution if not skepticism. A further problem surrounds the use of computer predictions of shrinkage porosity when comparing results with real castings. It is common for the castings to exhibit areas of oxide film that closely resemble areas of shrinkage on a radiograph. The misidentification of these regions as shrinkage porosity has led to huge confusion and unjust maligning of computer predictions. It is important when viewing radiographs not to assume that such areas are shrinkage porosity but to report these areas as “appearing to be shrinkage.” Most often, they are not shrinkage porosity. Random Perturbations to Feeding Patterns from Casting to Casting. Flash approximately 1 mm (0.04 in.) thick and only 10 mm (0.4 in.) wide has been demonstrated to have a powerful effect on the cooling of local thin sections up to 10 mm (0.4 in.) thick, speeding up local solidification rates by up to 10 times. Thus, flash must be controlled, or used deliberately, since it has the potential to cut off feeding to more distant sections. The erratic appearance of flash in a production run may introduce uncertainty in the reproducibility of feeding and the consequent reproducibility of the soundness of the casting. Moderate flash on thicker sections is usually less serious, because convection in the solidifying casting conveys the local cooling away, effectively spreading the cooling effect over other parts of the casting, so that an averaging effect over large areas of the casting is created. In general, however, it is desirable that these uncertainties be reduced by good control over mold and core dimensions.

The other known major variable affecting casting soundness in sand and investment castings is the ability of the mold to resist deformation. This effect is well established in the case of cast irons, where high mold hardness leading to good mold rigidity is needed for soundness. However, there is evidence that such a problem exists in castings of copper-base alloys and steels. A standard system, such as statistical process control (SPC) or other technique, must be in place to monitor and facilitate control of such changes. (Permanent molds, such as metal or graphite dies, are relatively free from such problems.) The solidification pattern of castings produced from permanent molds such as gravity dies (permanent molds) and low-pressure dies may be considerably affected by the thickness and type of die coat that is applied. A system to monitor and control such thickness on an SPC system must be in place. Alternatively, for each casting on a case-by-case basis, the thickness of the die coat must be demonstrated to be of no consequence. For some permanent molds, pressure die, and some types of squeeze casting, the feeding pattern is particularly sensitive to mold cooling. Changes to cooling channels in the die or to the cooling spray during die opening must be checked to ensure that corresponding deleterious changes have not been imposed on the casting.

Rule 7: Convection Damage Assess the freezing time in relation to the time for convection to cause damage. Thinand thick-section castings automatically avoid convection problems. For intermediate sections, either reduce the problem by avoiding convective loops in the geometry of the casting and rigging, or eliminate convection by rollover after filling. Convection enhances the difficulty of uphill feeding in medium-section castings, making them extremely resistant to solution by increasing the (uphill) feeding. In fact, increasing the amount of feeding, for example, by increasing the diameter of the feeder neck for uphill feeding, makes the problem worse by increasing the driving force for convection. Many of the current problems of low-pressure casting systems derive from this source. In contrast, feeders with hot metal at the top of the casting and feeding downward under gravity are completely stable and predictable and give reliable results. Convection Damage and Casting-Section Thickness. Damage to the micro- and macrostructure of the casting can occur if the solidification time of the casting is commensurate with the time taken for convection to become established, since extensive remelting can occur. This incubation time appears to be in the region of 2 min for many castings of approximately 10 to 100 kg (22 to 220 lb). In 3 min or more, convection can cause extensive remelting, the

Casting Practice—Guidelines for Effective Production of Reliable Castings / 503 development of flow channels, and the redistribution of heat in castings. Castings that freeze either quickly or slowly are free from convection problems, as indicated subsequently. Thin-section castings can be fed uphill simply because the thin section gives a viscous constraint, which reduces flow, and more rapid freezing, thus allowing convection less time to develop and wreak damage in the casting. Therefore, instability is both suppressed and given insufficient time, so that satisfactory castings can be made. Thick-section castings are also relatively free from convection problems, because the long time available before freezing allows the metal time to convect, reorganizing itself so that the hot metal floats gently into the feeders at the top of the casting, and the cold metal slips to the bottom. Once again, the system reaches a stable condition before any substantial freezing has occurred, and castings are predictable. The convection problem arises in the wide range of intermediate-section castings, such as automotive cylinder heads and wheels, and the larger investment-cast turbine blades in nickelbase alloys and so on. This is an important class of castings. Convection can explain many of the current problems with difficult and apparently intractable feeding problems in such common products. The convective flow takes approximately 1 or 2 min to gather pace and organize itself into rapidly flowing plumes. This occurs at the same time that the casting is attempting to solidify. Thus, the action of the convecting streams will remelt channels through the newly solidified material. These channels will contain a coarse microstructure because of their greatly delayed solidification and may contain shrinkage porosity if unconnected to feed metal. This situation is likely if the feeders solidify before the channels, as undoubtedly happens on occasion. Conventional gravity castings that require a significant amount of feed metal, such as cylinder heads and blocks, and are bottom gated but top fed will dictate large top feeders. This follows because of their inefficiency as a result of being furthest from the ingates and thus containing cold metal, in contrast to the ingate sections at the base of the casting that will be nicely preheated. This unfavorable temperature regime is unstable because of the inverted density gradient in the liquid and thus leads to convective flow and consequent poor predictability of the final temperature distribution and effectiveness of feeding. The upwardly convecting liquid within the flow channels usually has a freezing time close to that of the preheated section beneath that is providing the heat to drive the flow. In many low-pressure systems, artificially heated metal supply systems lead to a constant heat input, so that the convecting streams never solidify. Thus, long after the casting should have completely solidified, the channels empty or partially empty, pouring out liquid metal when the casting is lifted from the casting station.

Alternatively, a technology where the mold is inverted after casting effectively converts the preheated bottom ingate system into a topfeeding system (Fig. 9). Furthermore, the inversion of the system to take the hot metal to the top and the cold at the bottom confers stability on the thermal regime. Convection is eliminated. This is a powerful and reliable system used by such operations as Cosworth and an increasing number of others at the present time (2008). It is expected that techniques involving rollover immediately after casting will become the norm for many castings in the future. The rollover is only reliably carried out while the pressure of filling is maintained. This is because the hydrostatic head of metal above the cores is greatly reduced during this action, creating the danger of core blows. If a bubble does form, it may travel up or down, or both, during the inversion, significantly increasing damage and confusing any identification of its source. Durville-type casting processes (where the rollover is used during casting—actually to effect the filling process) also satisfy the topfeeding requirement. However, in practice, many geometries are accompanied by waterfall effects, if only by the action of the metal sliding in the form of a stable, narrow stream down one side of the mold. Thus, meniscus control is unfortunately often poor. Where the control of the meniscus can be improved to eliminate surface film problems, the Durville-type technique is valuable. An alternative kind of convection, that driven by density differences due to segregation, may lead to other problems, as outlined subsequently.

castings and the concentration of heavy elements, such as tungsten and molybdenum, at the base of large tool steel castings. Strong concentrations of segregated solutes and inclusions are found in channel segregates, which are a feature of large, slowly cooled castings. When extensive and/or intensive, such changes in composition of the casting may cause the alloy of the casting to be locally out of specification. If this is a serious deviation, the coincidence of local brittleness in a highly stressed region of the casting may threaten the serviceability of the product. The possibility of such regions must be assessed prior to casting, if possible, and demonstrated to be within acceptable limits in the cast product.

Rule 8: Segregation Predict segregation to be within the specification limits or agree upon out-of-specification compositional regions with the customer. Avoid channel segregation formation if possible. At regions in which the local cooling rate of the casting changes, such as at a change of section or at a chill or feeder, it is expected that a change in composition of the casting will occur. One of the most well-understood segregations of this type is inverse segregation, which the author prefers to call simply dendritic segregation. In this case, the partitioned solute is segregated preferentially to the face of the mold, especially if this is a chill mold. A similar effect will occur at the junction with a thinner section, which will act as a cooling fin (Fig. 10). However, in a complex thermal field and where the geometry of the casting requires a complex distribution of residual liquid to feed shrinkage, these chemical variations can be complex in distribution and not always easily predicted, except perhaps by a sophisticated computer simulation. Other segregations are driven by gravity and account for the concentration of carbon and other light elements at the top of large ferrous

Fig. 9

Cylinder casting arranged for inversion casting

Fig. 10

Schematic illustration of the effect of chills and unequal section junctions, causing the composition of parts of the casting to be above and below specification, leading to local brittleness and weakness, respectively

504 / Principles and Practices of Shape Casting

Rule 9: Residual Stress No quenching of certain light-alloy castings into water following solution treatment. (Boiling water is also not permitted for these castings, but polymer quenchant or forced-air quench may be acceptable if casting stress is shown to be negligible.) If, when quenching certain castings following high-temperature heat treatment, the time for cooling of the casting outer sections is shorter than the time required for heat to diffuse out from the interior sections (Fig. 11a), then the internal sections will cool and contract after the outer parts of the casting form an effectively rigid frame. Thus, the interior sections go into tension. It is a fact that quenching into water causes high residual stresses in large and complex castings. The most affected products are hollow castings with small openings to the outside world, and with interior walls. Such castings include cylinder heads and some compressor housings. The stresses will be tensile in some regions, mainly in the center of the casting volume, and compressive in others, mainly the outer walls. Small and simple-shaped castings are, in general, not seriously affected. The remainder of this section deals specifically with those castings that are particularly susceptible to the dangers of high quench stress. The use of a boiling water quench has been demonstrated to be of insignificant assistance in reducing the stresses introduced by water quenching. Furthermore, the stresses are not significantly reduced by the subsequent aging treatment. Immediately following the quench, the residual stress in susceptible aluminum alloy castings solution treated and quenched into water are well above the yield point of the alloy. Even after strengthening during the aging treatment, the stress usually remains at approximately 50 þ  20% of the yield stress. Thus, the useful strength of the alloy is reduced from its unstressed state of 100% to 70, 50, or even 30% (Fig. 11b). This massive loss of effective strength, usually affecting the interior sections in the casting, makes it inevitable that residual tensile stresses are a significant cause of casting failure in service. It is a scandal that many national standards for heat treatment specify water quenching, regardless of the size and complexity of the component. This disgraceful situation must be remedied. In the meantime, such dangerous standards should be overridden by agreement with the customer. The reduced mechanical strength when using polymer or forced-air quenching (Fig. 11c) is more than compensated by the benefit of increased reliability from putting unstressed castings into service. Thus, somewhat reduced mechanical strength requirements should be specified by the casting designer and/or customer. The reductions are expected to be in the range of 5 to 10% for strength and hardness.

Fig. 11

(a) Threshold at which aluminum (or steel) castings will suffer residual quench stress as a function of cooling rate between 500 and 200  C (930 and 390  F), and the average distance over which heat will be required to diffuse during the quench. (b) Logic behind the creation of quench stresses revealed as a problem of quench strain of approximately 1% applied to the internal features of the casting. (c) Spectrum of cooling rates commonly available in a 20 mm (0.8 in.) diameter bar of aluminum

Casting Practice—Guidelines for Effective Production of Reliable Castings / 505 features but which may be formed by a difficult-to-place core or a part of the casting that requires extensive dressing by hand. The result is a casting that does not clean up on machining and is thus, perhaps somewhat unjustly, declared to be dimensionally inaccurate. This rule is designed to ensure that all castings are picked up accurately, so that unnecessary scrap is avoided. Different arrangements of location points are required for different geometries of casting. Some of the most important types are listed as follows:
Six points, as illustrated in Fig. 12, are

Fig. 12

Classic six-point location system showing various degrees of sophistication, and the jig or fixture that would be used for dimensional checking or for machining. (a) Basic location system. (b) Halving of length errors. (c) Use of tooling lugs for clamping. (d) Jig or fixture attached to mark-out table or machine tool

Although the strength of the material will be lowered by the slower quench, the strength of the casting acting as a load-bearing component in service will effectively be increased.

Rule 10: Location Points All castings are to be provided with agreedupon location points for dimensional reference and for pickup for machining. Location points have a variety of other names, such as tooling points, pickup points, and so on. It is proposed that the term location points describes their function most accurately. It is essential that every casting has defined locations that will be agreed upon with the machinist and all other parties who must pick up the casting accurately. Otherwise, it is common for an accurate casting to be picked up by the machinist using what appear to be useful

required to define the position of a component with orthogonal datum planes that is designed for essentially rectilinear machining, such as for an automotive cylinder head. (Any fewer points is insufficient, and any more will ensure that some points are potentially in conflict.) Three points define plane A, two define the orthogonal plane B, and one defines the remaining mutually orthogonal plane C.
For cylindrical parts to be picked up in a three-jaw chuck, one axis must be defined as a datum reference, together with four additional tooling points, three of which define the plane orthogonal to the axis, and the last tooling point defines a “clock” position with respect to the axis.
For prismatic shapes, comprising hollow, boxlike parts such as sumps (oil pans), the pickup may be made by averaging locations defined on opposite internal or external walls. This is a lengthier and more expensive system of location often tackled by a sensitive probe on the machine tool, which then calculates the averaged datum planes of the component and orients the cutter paths accordingly. For maximum internal consistency between the tooling points, all should be arranged to be in one-half of the mold, usually the fixed or lower half. However, it is sometimes convenient and correct to have all tooling points in one-half of a core (defined from one-half of the core box) if the machining of the part must be defined in terms of the core. Datums located at the end of a long component, or at the opposite end to a critically located feature, give rise to casting rejections or unnecessary manufacturing difficulties. It is helpful if the part can have its datums for dimensions arranged to go through or near to the center of the part. The slight variations in size of the part, from one part to the next, then remain more easily within tolerance, since any variations in length are halved. Alternatively, the part can be datumed through the feature that is most critical to have in a fixed location, such

as the dipstick boss in the oil pan casting. The problem of variability of other dimensions is then greatly reduced. The tooling points should be defined on the drawing of the part and agreed upon by the manufacturer of the tooling, the caster, and the machinist that all parties will work from only these points when checking dimensions and when picking up the part for machining. The use of the same agreed-upon points by toolmaker, caster, and machinist leads to an integration of manufacturing between these parties. Disputes about dimensions then rarely occur, or if they occur, are easily settled. Casting scrap apparently due to dimensioning faults or faulty pickup for machining usually disappears. ACKNOWLEDGMENT Adapted from J. Campbell, The Ten Casting Rules: Guidelines for the Reliable Production of Reliable Castings—A Draft Process Specification, Proceedings, Materials Solutions Conference ’98 on Aluminum Casting Technology, ASM International, 1998 REFERENCES 1. J. Campbell, The Ten Casting Rules: Guidelines for the Reliable Production of Reliable Castings—A Draft Process Specification, Proceedings, Materials Solutions Conference ’98 on Aluminum Casting Technology, ASM International, 1998 2. J. Campbell, Casting Practice: The 10 Rules of Casting, Butterworth-Heinemann, 2004 SELECTED REFERENCES
J. Campbell, Castings, 2nd ed., Butterworth-

Heinemann, 2003


J. Campbell, Invisible Macrodefects in Cast-

ings, J. Phys. (France) IV, Colloque C7, supplement to J. Phys. (France) III, Vol 3, 1993, p 861–872
N.R. Green and J. Campbell, Statistical Distributions of Fracture Strengths of Cast Al-7Si-Mg Alloy, Mater. Sci. Eng. A, Vol 173, 1993, p 261–266
N.R. Green and J. Campbell, Influence of Oxide Film Filling Defects on the Strength of Al-7Si-Mg Alloy Castings, Trans. AFS, Vol 102, 1994, p 341–347
J. Runyoro, S.M.A. Boutorabi, and J. Campbell, Critical Gate Velocities for Film-Forming Casting Alloys, Trans. AFS, Vol 100, 1992, p 225–234

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 506-512 DOI: 10.1361/asmhba0005221

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Filling and Feeding System Concepts John Campbell, University of Birmingham, England

FILLING the casting cavity with molten metal can be accomplished in various ways. Filling methods differ among the various casting processes, and techniques depend on the geometry of the cast part and the castability of the alloy. In the past, these systems have evolved largely from the experience and capabilities of individual foundry people. However, new research tools and computer simulation have assisted in the design of improved filling systems. This article introduces filling and feeding concepts from the general perspective of what constitutes good casting practice (see the preceding article, “Casting Practice: Guidelines for Effective Production of Reliable Castings,” in this Volume). This short review supplements a more detailed account (Ref 1) and is an attempt to provide a critical overview and update of concepts (Ref 1–4) in filling-system design. The casting industry has never systematically researched its filling-system designs for castings; consequently, current knowledge and practice is poor. This missed opportunity has cost the casting industry and its customers dearly. In particular, conventional unpressurized and pressurized filling systems (designed to fill molds under gravity) perform poorly, creating generous quantities of entrainment defects (bubbles, bifilms, and sand inclusions) and thereby efficiently degrading castings and their properties. Two recent developments that can help to avoid these problems include:
Preprimed filling systems
Naturally pressurized gravity filling systems

Prepriming (or the prefilling) of a filling system, using some variety of two-stage pour, is a powerful technique that deserves wider use. In the absence of prepriming, the new concept of the naturally pressurized gravity filling system is also described in this article in terms of its successes and limitations. It seems to the author that contact pouring in conjunction with either prepriming or a naturally pressurized system may be the only good solution at the present time to the gravity pouring of heavy steel castings using a bottom-teemed ladle. Perhaps the pouring of melts at any stage during the melting or casting process may ultimately be avoided in

new designs of future foundries. Thus, new melt-handling technology would be developed for future foundries whereby countergravity filling of molds would be recommended as capable of routinely producing substantially defect-free castings (see the article “Low-Pressure Countergravity Casting” in this Volume). As noted, this article briefly reviews concepts that may help clarify and quantify objectives for more effective mold-filling designs. The design of improved filling systems is now significantly facilitated by research tools such as the video x-ray radiography technique, in which the flow behavior of real liquid metals can be studied in detail in real molds (Ref 5). In addition, computer simulation has now come of age in this field, and its intelligent use can be invaluable.

Entrainment Defects in GravityPoured Castings Entrainment defects are extremely difficult to eliminate from gravity-poured castings, and the task for the designer of the gravity-driven filling system amounts to a damage-limitation exercise. This can be understood in terms of the so-called critical ingate velocity at which the melt has sufficient energy to entrain its own surface in a phenomenon called surface turbulence. This concept incorporates a further concept called entrainment that is a mechanical folding action, so that entrainment defects such as air and oxides (more specifically, bubbles and bifilms) can be introduced into the melt. The major difference between a bubble and a bifilm is the amount of air that each entraps; the bifilm is a kind of flat bubble that is more like a crack. Whereas the bubble is visible, the bifilm is usually so thin that it is invisible and only detected under a good microscope. Both can be exceedingly damaging to the quality of a casting. The critical fall distance (the fall distance under gravity at which the critical velocity is exceeded for that particular metal) is only approximately 12 mm (0.5 in.) for the light liquid metals, aluminum and magnesium, and approximately 8 mm (0.3 in.) for the dense metals such as copper, nickel, and iron. Thus, the vertical height through which the melt falls

in the running systems of all gravity-poured castings exceeds this velocity by a large margin. Therefore, entrainment defects are extremely difficult to eliminate from gravity pouring.

Traditional Filling Designs The traditional methoding systems have included the so-called unpressurized filling design, in which an attempt is made to reduce the velocity of the metal prior to entering the mold cavity. This worthy aim is attempted by expanding the flow channels. For instance, each doubling of the area is expected to result in halving the speed of flow. In terms of areas of the sprue exit/runner/gate, the relative areas chosen are often 1 to 2 to 2 or 1 to 2 to 4 and so on. The assumption has been that the volume flow rate, Q (m3/s), and velocity, V (m/s), relate to the area, A (m2), of the stream by the simple equation:



Q¼A V

(Eq 1)

The problem with this approach is that Eq 1 is perfectly true for the melt stream itself but not necessarily for the mold channel. Thus, an expansion of the channel to reduce the velocity usually has no effect on velocity; the melt flow progresses at its original speed and cross-sectional area, so the channel simply runs only part full. This is a potential disaster for melt quality, since the velocity is well above the critical velocity, and air and oxides (i.e., more precisely, bubbles and bifilms) can now be entrained. Thus, unpressurized systems are typically poor at reducing velocity and generally deliver poorquality metal into the mold (Ref 1). The filling-system design widely employed in the cast iron industry is the pressurized system. Here, the so-called gating ratio is often approximately 1 to 0.9 to 0.9, allowing the channels to run full, provided air is not introduced in the basin. However, poor conical basin designs and improperly tapered down-sprues, resulting from the poor design of automated green sand molding plants, usually eliminate these advantages, with the result that air is introduced at the entrance to the filling system. Thus, although the potential of the pressurized

Filling and Feeding System Concepts / 507 system was promising, in practice the benefits were eliminated by poor front-end design. Finally, the pressurization effect resulted in very fast delivery of melt into the mold cavity, where the high velocity (once again, much higher than the critical velocity) had the potential to create significant additional damage. Also, damage generated in the mold is the worst kind of damage, since such damage necessarily is incorporated into the casting. (In contrast, much of the damage created in the filling system appears to “hang up” in the filling system, never finding its way into the casting.) This problem of good general principles but poor front ends to the system, or other features that introduce defects, illustrates a significant aspect of filling-system design: The whole of the system must work well. Any small detail that is incorrect can introduce destructive amounts of irreversible damage, a significant proportion of which is likely to arrive in the casting. Unfortunately, gravity filling systems are hypersensitive to small errors as a result of the high velocities involved. This fact has been the source of major problems to the casting industry. This situation contrasts with countergravity filling systems where the melt velocity can be controlled at all stages below the critical velocity, so that it is possible to introduce zero damage during filling, despite the presence of errors such as the ledges arising from the mismatch of channels and so on. It follows that countergravity systems, when properly operated and despite modest errors in the filling system, represent robust technology that can routinely deliver good products and have the potential to revolutionize the performance and profitability of the industry. The poor front-end designs used for many systems have effectively undermined many otherwise reasonable filling-system designs. Similarly, good front-end features, such as the offset stepped pouring basin (the only basin to have been researched and shown to be capable of avoiding air entrainment, Ref 6), have often been mistakenly criticized for poor performance because of the deleterious actions of other features in the system, thus adding to the overall confusion relating to filling-system design. A further deleterious feature that has a long pedigree is the well at the base of the sprue, often said to be placed there to cushion the fall of the metal. However, recent work has illustrated that the high velocity at this location, combined with the generous volume provided by the well, maximizes the damaging churning action of the melt. This turbulence entrains clouds of bubbles and, without doubt, large quantities of oxide. The sprue/runner junction for the naturally pressurized filling design described subsequently is a simple and basic turn, providing no extra volume in the system that can allow surface turbulence. When properly designed and made, this sprue/ runner junction does not contribute a single bubble to the downstream flow. Potential solutions to the central problem of the avoidance of entrainment defects are described in order in the following sections.

Prepriming Techniques The prefilling of the filling system prior to the filling of the mold is a powerful technique for eliminating entrainment problems. It deserves to be used far more widely. There are various partial solutions to the prepriming approach:
If an offset pouring basin is filled with metal

that is allowed to immediately fall into the sprue, the filling of the sprue is ragged and messy, incorporating considerable quantities of air. The resulting air/melt mixture is usually in the region of 50% air and 50% metal for the several seconds that it takes for the basin and sprue to prime. This is because the sprue is designed to work from a particular height of metal in the basin. Initially, the depth of metal in the basin is much less, so that for the first several seconds, the sprue entrance is insufficiently pressurized and thus runs only part full. The use of a stopper in the entrance to the sprue allows the basin to be filled to some minimum design level, prior to lifting the stopper and allowing the melt to enter the sprue (Fig. 1a). When observed by video x-ray radiography, the prefilling of the basin by the use of a stopper is seen to significantly improve those critical early seconds of sprue priming.
The use of a valve at the base of the sprue is valuable to allow the prefilling of the basin and the sprue (Fig. 1b). This technique is even better than the mere prefilling of the basin. There is an additional benefit since, on opening the valve, the melt in the sprue starts to flow from a standing start and must accelerate to its full flow speed. This speeding up from zero occupies only a second or so but is enough to reduce the initial jetting of the melt along the runner. The filling of a disc brake mold with an Al/SiC metal-matrix composite (MMC) was achieved only by such a system, as described by Cox and coworkers (Ref 7). The valve in this case consisted of a thin sheet of ceramic paper that was supported on a ceramic foam filter. After the sprue was filled, the ceramic paper was lifted clear by raising a wire hook attached to the edge of the paper. The use of a sheet steel slide gate in the runner has been reported to raise the quality of aluminum-base MMC castings. The excellent results were convincingly quantified by Weibull statistical analysis (Ref 8).

for a few seconds prior to melting and automatically releasing the flow are known and used in the industry, although little has been published. In laboratory tests, aluminum foil backed by a filter was not successful because the aluminum oxide layer on the aluminum foil was too strong and resisted breakthrough, despite the melting of the aluminum metal in the center of the foil. Thus, bursting discs with strong oxide films should be avoided. Conversely, metal foils made from zinc, copper, nickel, or iron would be expected to be useful. The thickness of the discs would have to be determined by experiment or perhaps by computer simulation. Any of these techniques can be combined with, and will improve the performance of, a naturally pressurized system design, as described in the next section. However, prefilling enhances the performance of even very poorly designed filling systems, which is why the technique is so valuable. The prefilling action is temporarily turbulent until the prefilling action is complete. At that stage, the melt in the running system is expected to contain both bifilms and bubbles. If a short dwell time of 1 or 2 s is provided, little heat is lost, but the melt has time to eliminate its bubbles by forcing them out through the permeable sand mold. (The second and third prepriming systems in the bulleted list can only be used safely with permeable, or sand, molds because the air needs to escape during the prepriming event. In permanent molds, the air cannot escape, and as it quickly heats and expands, it can reverse the melt, ejecting it from the sprue.) Most entrained oxides will already be attached at some point to the mold wall, so that gentle buoyancy or residual advection will assist in hinging the films against the surface of the runner, where, in general, they are expected to remain. Any attachment will not be particularly strong or secure, but the films will lie mainly in the region of low or zero flow rates in the region of the boundary layer against


The use of a valve at the gate into the mold

allows the whole filling system to prefill prior to the melt entering the mold (Fig. 1c). Such a system would be expected to deliver excellent castings even from what may be viewed as a lamentable running system. All of the techniques can be implemented with operator-controlled devices. However, sheet metal bursting discs that arrest the flow

Fig. 1

Prepriming techniques for (a) the basin, using a stopper, (b) the basin and sprue, and (c) the whole filling system. All of the prepriming methods can use devices such as mechanical valves or fusible bursting discs.

508 / Principles and Practices of Shape Casting the mold walls. Thus, much of the oxide entrained during the prepriming stage is likely to remain gently glued to the inside walls of the running system and is not in danger of entering the casting. The bifilms would not be expected to remain in place if air were to be entrained and flow through the system, since slugs of air create buffeting and impact forces by their negative momentum; the system only works well if it remains completely filled with liquid metal. As mentioned previously, the prefilling of any or all of the filling system is extremely valuable and strongly recommended, even if only achieved in part. However, if this cannot be accomplished, then the filling system must be designed to fill without such assistance. This is a significantly more difficult task, since it is a challenge to exchange the 100% air in the system for 100% metal prior to the melt entering the mold cavity. The challenge is even more difficult to meet if this 100% metal is to arrive in the mold cavity in a reasonably undamaged condition (i.e., reasonably free from entrainment defects such as bifilms and bubbles). This is the challenge that the naturally pressurized system attempts to meet.

Naturally Pressurized Filling System The naturally pressurized system is characterized by the narrowness of its flow channels. Its central concept, fundamental to the whole approach of the naturally pressurized system, is that of maximizing constraint on the flowing liquid from nearby walls of the channels. Thus, despite the melt having a velocity well above the critical velocity, it tends to not suffer damage because there is effectively no room for the advancing meniscus to fold over on itself to entrain a defect. The concept of a choke in the system becomes redundant, because the whole system, along its complete length, acts as a choke and is effectively a continuously choked design. The six individual parts of the filling system are:







Offset stepped pouring basin Sprue Sprue/runner junction Runner Gates Feeding via feeders (or running system, if possible)

Offset Stepped Pouring Basin. It is assumed that the offset stepped basin design is used in the naturally pressurized filling system (Fig. 2). While not being exclusive to the naturally pressurized system, it appears to be the best basin design thus far and is capable, in some cases, of delivering 100% metal (i.e., zero content of entrained air bubbles corresponding to a discharge coefficient of 1.00) (Ref 9). It seems to work adequately well for

aluminum-silicon alloys, particularly when a basin response time (i.e., its draining time if simply allowed to empty) of approximately 1 or 2 s is selected. This usually gives adequately slow response so that the pourer can keep the basin filled (Ref 1) and appears to give adequate detrainment time for most of the air entrained by the pour into the basin. A widespread error is to calculate the driving pressure for filling from the height of the basin itself. However, if the basin is only slightly underfilled at any time, the reduced pressurization of the sprue entrance will lead to air being taken down the sprue (although this problem is not apparent to the operator, who is under the impression that the system is working satisfactorily). This is the common reason why most designs of offset stepped basins fail. To avoid this, for the purposes of the design of the whole filling system, it is essential to assume a level of fill in the basin that can be reliably maintained. If the actual fill level is always maintained above this minimum, the sprue can remain properly pressurized and therefore completely full at all times. A doubling of the height of the basin above the minimum level is not too much to leave a good margin of safety. The scheme is illustrated in Fig. 2. Even so, the problem of basin design is not completely solved and, in fact, may not be solvable for some metals. Metals that generate strong oxide films, such as aluminum bronzes and some stainless steels, may overload the system with oxides that cannot be detrained. A mass of strong oxide films waving about in the flow like waterweed in a river will prevent detrainment in the basin and act to funnel bubbles down the sprue. Similarly, the pouring of steel from bottomteemed ladles also may represent an unsolvable situation. In this case, the great depth in the ladle naturally results in an extremely high exit velocity from the bottom nozzle. The high-velocity melt entering the basin creates a churning action that overwhelms the system with oxides and bubbles that have insufficient time to detrain. The use of an argon shroud around the pouring stream is an attempt to address this problem and has some limited beneficial effect. However, for a more rigorous solution, a sophisticated version of contact pour may be required in which the pouring basin is effectively exchanged for a bottom-poured ladle in which the outlet nozzle from the ladle makes direct contact with, and seals against, the entrance to the sprue (Fig. 3). The technique was demonstrated successfully for the first time for a 50,000 kg (55 tons) steel roll casting (Ref 10) and is a promising general method for steel castings. A production system using small, intermediate ladles sitting on top of the molds and able to contain the complete volume of melt required for the pour can be envisioned. The intermediate ladle is filled from the large ladle transferring metal from the melting furnace. A few seconds after the intermediate ladle is filled, its stopper is raised and the mold is filled.

A series of intermediate ladles on a series of molds may be workable, with the early ladles being transferred forward to be used a second time for later castings in the pour series. As an alternative to a stoppered ladle, a basin or ladle with a fusible bursting disc may be simple, economical, and practical.

Fig. 2

Offset stepped basin illustrating the target fill level, the minimum fill level, and other key

features

Fig. 3

Teeming of steel into an intermediate ladle, kept filled to at least a minimum depth, h

Filling and Feeding System Concepts / 509 Sprue. The basis of the sprue design is the prediction of the form of a freely falling liquid stream. This frictionless flow is the starting point for the calculation of the areas for the sprue and the remainder of the running system. Assuming Eq 1, for a volume per second flow rate (Q), the area of the cross-sectional area of the stream (A) after a fall of distance h under the acceleration due to gravity (g) from the upper surface of the melt is given by: A2 ¼ Q2 =ð2ghÞ

(Eq 2)

This area/height relation is the famous result describing a rectangular hyperbola (not a parabola, as has often been assumed in error). It follows that the really important relation giving the profile (i.e., the diameter, D) of a rotationally symmetrical stream at height h is: D4 ¼ 8Q2 =ð2 ghÞ

(Eq 3)

This relation of D versus h1/4 is a significantly more complex curve, shown schematically in Fig. 4 (with height in dimensionless units of 8Q2/(p2gh)), emphasizing the nonstraightness of its taper, particularly near the top of the sprue, and the approach to asymptotes at each limit. Whereas in the past it was common to calculate the entrance and exit sizes of the sprue and connect these with a straight taper, the author now uses the hypobolically tapered sprue for all castings above approximately 300 mm (12 in.) high. This nonlinear taper follows the form of the falling stream, pressurizing at all points of the fall, and effectively excludes air. The simple naturally pressurized concept arises as follows. The theoretical shape (i.e.,

the cross-sectional area of the stream at every point) is based on the theoretical prediction assuming a frictionless fall of the melt (Eq 3). When applying this theoretical shape in practice, a mold is effectively created around it, so as just to touch and contain the flowing stream. The action of a solid surface designed to just touch the surface of the falling melt changes everything. Friction is now introduced for the first time. The natural frictional back-pressure now builds up along the length of the sprue (and all the subsequent channels if contact with the walls is maintained by appropriately narrow sections, as illustrated in Fig. 5(b), causing the liquid to gently pressurize the walls of the channel. This natural pressurization is of enormous value in assisting to keep air out of the system and is a fundamentally important feature of the naturally pressurized design. (In contrast, the traditional oversized running systems illustrated in Fig. 5a do not benefit from the friction to reduce the velocity of flow and usually cannot be pressurized. As a consequence, the systems introduce copious quantities of air and oxides.) Sprue/Runner Junction. At bends in the system, particularly the sprue/runner junction, it is essential to provide a radius to the inside of the turn. The radius should be close to the diameter or width of the sprue exit (Ref 11). If no radius is provided, the junction entrains air to create bubbles and oxides. It has been widely assumed that significant friction is involved at any bends in the system, causing up to a 20% loss of velocity. If true, the runner cross-sectional area should be expandable by 20% to take advantage of the 20% reduction in velocity, giving a gating ratio of R = 1/1.2/1.4. A succession of right-angle bends, when used in this way, can be successful in making significant reductions in final velocity at the end of the system. There is some experimental work (Ref 12) on a series of right-angle bends in a runner that appears to give confirmation of this theoretical prediction from an observed series of speed-reduction ratios close to 1.0, 1.2, 1.4, 1.7, 2.1, 2.5, and

so on. Traditionally, the right angle was always avoided because of the turbulence that it could create. This was true for oversized sections. However, for the slim sections in the naturally pressurized system, the melt is constrained so that little room is available to allow the metal to damage itself by folding over its advancing front to entrain bifilms or bubbles. Therefore, instead of being a threat, the right angle may constitute a valuable speed-reducing feature in the new system. This L-junction has been studied in detail by Hsu and others (Ref 11). Runner. Alternatively, the runner need not be expanded by 20% and instead may have the same area as the sprue, giving R = 1/1/1. This will significantly assist to pressurize the melt in the runner and gates so that there will be a greatly reduced danger of the entrainment of air. If this option is taken, the choking action of the melt entering the runner will lead to a 20% increase in filling time and a further 20% at the gates, giving a total extension of approximately 40% to the filling time due to the losses at the L-shaped turns alone. These losses in the running system are often overlooked, perhaps at least partly explaining why many calculated filling systems appear to run more slowly than predicted. Alternatively, the full 20% increase in area for the runner need not be taken. If only 10% were selected, giving R = 1/1.1/1.2, the runner would benefit from an extra 10% pressurization at each turn that would suppress entrainment defects to some degree, but filling time would be extended by approximately 10 + 10 = 20%. After the initial increase of the area of the runner to allow for losses at the sprue/runner junction, its area can remain constant up to the first division or gate. This constant area will contribute to the gentle pressurization along the length because of the frictional loss accumulating along the length. The discussion about a 20% loss of velocity at a right-angle bend has been called into question by the more recent work of Hsu and coworkers (Ref 11). In the early results of their

Fig. 4

Shape of the falling stream. The straight taper, taken as an example from 50 to 500 mm (2 to 20 in.), is seen to be a poor approximation.

Fig. 5

(a) A poor traditional filling design contrasted with (b) an improved system

510 / Principles and Practices of Shape Casting detailed study of the L-junction, the rule of 20% loss per right-angle junction compared to assumed negligible losses along the straight sections of the system is seen to be oversimplistic. It appears that the frictional back-pressure associated with the bend is real, but the accumulated effect along the sprue and along the runner appears to exceed the frictional loss at the bend. This is probably because the bend acts over a rather short length of flow compared to the longer lengths of the sprue and runner. Hsu’s results suggest that the accumulating frictional losses along the lengths of channels would permit their gentle expansion, allowing a progressive reduction in speed without the penalty of reduced volume flow rate or the loss of pressurization in places. This situation clearly requires more research. The outcome of such targeted research may herald the next development of filling-system design. In the meantime, it seems valuable to be cautious, not expanding the area of the flow channel at any point and thereby ensuring that the system remains pressurized everywhere to ensure the exclusion of air. Gates. As is well known, the placing of a series of equal gates along the length of a uniform runner will favor delivery through the far gates. The delivery through the gates should be balanced as far as possible by systematically reducing the area of the runner. The traditional practice of providing steps, reducing the area of the runner as each gate is passed, is definitely not recommended because of the deflection of the flow and the turbulence caused at sharp steps. Far better is a linear taper, gradually reducing the area of the runner along its length. For many aluminum-silicon alloy castings weighing 10 to 50 kg (22 to 110 lb) and up to 0.5 m (1.5 ft) or so long, approximately 50% reduction in area is often closely correct. (Tapering to zero area is also definitely not recommended. This is an overcorrection that causes the early gates to be favored and exposes foundry personnel to the danger of accidents with dagger-shaped runners!) Because many castings have only a single mold joint and must be gated on the joint, the gates have to be horizontal. If the gates now enter the mold cavity at more than the critical distance above the base of the casting, allowing the melt to fall inside the mold cavity, significant damage is expected to the properties of the casting. Gating at the mold joint can therefore be a major risk to casting integrity. Bottom gating, where the gates enter at the lowest possible level of the casting, is necessary to achieve good and reliable results. However, horizontal gates often cause the metal to jet over the base of the mold. Such uncontrolled directional propagation can lead to metal damage in the form of oxide flow-tube defects or, in the case of silica green sand molds, mold expansion defects such as rattails that damage the mold surface at the edges of the jetting stream. By contrast, vertical bottom gates, in which gravity assists both the filling of the gates and the slowing of the melt as the gates prime, work

well. They work even better if expanded along their length (i.e., expanded in area with increasing height) (Ref 1). It is not so much a luxury but self-interest to have an additional joint beneath the casting, formed by a third mold part or craftily designed core (Fig. 6). Such features usually quickly recover their cost. Whenever possible, gate directly onto a core, particularly if the local section thickness is small, for instance, preferably 4 mm (0.16 in.) thickness or less. This rule will alarm traditional foundry people, because a well-known rule was “never gate onto a core.” Never gating onto a core was correct when poor running systems carried up to 50% air with the metal, because the air would oxidize away the core binder, eroding the core and filling the casting with sand inclusions. However, with a properly designed naturally pressurized system, no air is present. Although the core will clearly become hot, it remains unharmed. The benefit to the flow of metal is the redistribution of melt over a larger area prior to emerging into the free volume of the mold cavity. Thus, the thin wall defined by the core acts as a final extension to the runner system, making a valuable reduction in the final velocity of the melt. Gates require careful design. The requirements are:
The total gate area should, if possible, be

sufficient to ensure that the ingate velocity at the point of entry to the mold is less than 0.5 m/s (1.5 ft/s), although in practice, up to 1.0 m/s (3 ft/s) is often tolerable.
The total number of gates must be sufficient to ensure that the sideways (transverse) velocity inside the mold does not exceed the critical velocity (Ref 1).
It is essential to ensure that the gates do not form a hot-spot at the junction with the casting. The rules for ensuring that a hot spot is not created are well established (Ref 1). For the most susceptible geometry of casting/ ingate junction, the relative geometric modulus should be less than 0.5 or greater than 2.0. Sometimes, however, it is useful to use particularly narrow slot gates, with modulus much less than 0.5 of that of the local piece of the casting. These become hot during filling but rapidly lose their heat to act as

Fig. 6

cooling fins after the filling is complete. This variety of gate can be used to create a valuable temperature gradient for directional solidification and enhanced feeding efficiency (Ref 1). Feeding. The author is not in favor of feeding castings, unless absolutely necessary (Ref 1). Most castings seen in foundries are hugely overfed. This almost certainly is the result of a number of factors. It is understandable and commendable that the founder should exercise caution, but in practice, this caution is carried out to excess for the following reasons:
The damage to the melt by poor running sys-

tems is usually so bad that most of the metal used to prime the system is no longer fit for contributing to the casting. Thus, it is dumped into the feeder, making the feeder oversized for its purpose of feeding. In this case, the primary role of the feeder is merely to act as an overflow. The founder justifies the size of the feeder on the apparently reasonable grounds that if the feeder size is reduced, the casting exhibits unacceptable shrinkage porosity. This is true. However, the real reason is that the melt in the mold cavity contains a major population of large bifilms created by the poorly designed filling system, so that porosity resembling shrinkage porosity is rife. Only by increasing the feeder size (i.e., getting rid of a high proportion of the damaged metal) is the problem apparently alleviated. However, the lesson is clear: Instead of increasing the feeder size, it would be more economic and significantly more effective to improve the filling system to avoid damaging the melt in the first place.
The computer simulation packages that check for solidification shrinkage in the casting are usually programmed with significantly inflated shrinkage factors, resulting in oversized feeders. Because of the first feature in this list, the oversized feeders are currently required and thought to be necessary for feeding because of a lack of awareness of the overflow dump role of the feeder. The problem of the overestimation of feeding percentages has remained unsuspected and become the norm.

Bottom-gated systems achieved by (a) a three-part mold with accurately molded running system, (b) a two-part mold making use of a core, and (c) a two-part mold using standard preformed sections

Filling and Feeding System Concepts / 511 When good filling systems are exchanged for poor originals, the oversized feeders can usually be greatly reduced, substantially increasing the casting yield. It follows that for good metal and a good filling system, simulation packages must reduce their shrinkage factors to more realistic values. The reduction in feeder size is the result of two benefits: the elimination of the role of the feeder as an overflow dump, and the reduction in the sensitivity of the melt to feeding problems because of the reduction in bifilm content. With excellent-quality melt and an excellent filling system, the casting may have such a low bifilm content that porosity cannot be initiated, and the resulting surface sinks (the corresponding collapse of the casting to feed any residual internal shrinkage deficit) may be negligible (i.e., not below the dimensional or machining allowance). In this case, the casting effectively self-feeds without the necessity for additional melt from a feeder. This ideal situation is worth targeting but may not be completely achieved, so some residual feeding may be necessary. When reducing the feeder volume, any reduction in its height should be made with caution, if at all. The metallostatic pressure due to height is valuable to pressurize the melt in the freezing casting and thus suppress the opening of any residual bifilms. Bifilms are usually so thin as to be invisible but are so numerous that they should always be assumed to be present. If bifilms are allowed to open, properties are reduced. It is essential to realize that in the early phases of this effect, the bifilms may not have opened sufficiently to create visible porosity. Thus, the casting may still appear perfectly sound, so that the loss of properties may appear inexplicable. Only if the bifilms open by some fraction of a millimeter will they start to become visible as microporosity. In this case, properties will have fallen even further (Ref 2). If bifilms must be tolerated in melts, tolerable properties will only be maintained if the bifilms are maintained firmly closed. If closed, they can support a shear stress but not a tensile stress at right angles to their plane. Although always damaging to properties, keeping the bifilms closed limits the ensuing damage.

Filters At least one study has definitively shown that the beneficial action of a ceramic foam filter derives approximately 5% from filtration (depending on the quality of the metal) and 95% from improved flow (Ref 13). Therefore, it seems likely that the benefits of using filters in filling systems arise mainly from the reduction in speed and turbulence of the flow. In this way, the melt downstream of a filter remains cleaner not because of filtration but simply by being less turbulent and therefore creating fewer defects. To retain the benefits of the reduction in speed, the filling channels after the filter must

be adjusted in area to suit the new velocity. Thus, if the filter can reduce speed by a factor of 5, an expansion of downstream area by a factor of 4 will retain some pressurization but gain the benefit of speed reduction to one-quarter of the speed entering the filter. The correct sizing of the postfilter part of the runner is usually overlooked (but, as a kind of perverse benefit, is often not necessary because all the ductwork, both incoming and outgoing from the filter, is already grossly oversized). The filter is most valuable in those systems that entrain air in which the front-end design is poor but cannot easily be rectified. This often happens on automatic green sand lines because the sprue and basin are poorly formed. The filter assists the backfilling of the front-end of the system. It is effectively acting as a device to aid prepriming of the system. Naturally, the provision of a bubble trap to divert buoyant phases such as air (and, in the case of ferrous castings, slag) away from the main flow can also be valuable in this case (Fig. 7) (Ref 14). It is essential to control any spraying or jetting action from the exit side of the filter, especially where the incoming flow is under high pressure from a tall sprue. Such jetting creates consequent oxidation damage downstream of the filter. Figure 7(b) illustrates an arrangement in which the melt emerges against gravity, quickly covering the exit surface and therefore quickly suppressing jetting, so that the subsequent flow remains undamaged. The arrangements in Fig. 7(a) and (c) illustrate the provision of shallow (less than the critical fall height) sumps beneath the filter. These quickly fill and thus aid the effective onward transmission of clean metal. Because the most important inclusions in aluminum alloys, and many other alloys, are films (not particles), it is interesting to note that inexpensive glass cloth is often found to be as effective as the various varieties of ceramic block filters. However, the cloth is often difficult to apply in practice, explaining the prevalence of the more costly alternative.

Performance of Alternative Filling Systems With combined reductions in the volume of both filling and feeding systems, metallic yields

Fig. 7

can easily rise from 45 to 70% or more. Provided the metal that is poured is of adequate quality (which is not to be assumed), defects such as porosity and hot tears can be reduced, if not eliminated (Ref 4). In addition, the mechanical and corrosion properties of the casting can be significantly enhanced (Ref 15–17). Although the naturally pressurized filling system gives a significantly superior performance to current unpressurized and pressurized systems, the approach remains far from optimal. More research is needed to further refine the concept to take advantage of the progressive accumulation of frictional back-pressure to progressively expand sections and thus reduce speeds without loss of volume flow rate or loss of pressurization. For this reason, computer simulations of pressure distribution against the walls of the filling system, to ensure the absence of unpressurized areas, will be more important than current simulations of velocity of flow. Alternative filling systems not discussed here include the horizontal transfer of metal into a mold via the level pour system (Ref 1). This is a clever mechanized system that deserves greater use for appropriate boxlike castings and possibly other shapes. Tilt casting can also be valuable if carried out correctly. In particular, the tilt process is usually started from the horizontal level. This is a mistake. It results in a poorly controlled turbulent fill. A start angle of approximately 20 above the horizontal usually avoids this problem (Ref 1). Further problems occur if the melt suffers a waterfall effect at some stage during the tilt because of the geometry of the casting. Thus, there are a number of reasons why many tilt pouring operations are currently not free from significant levels of scrap. Direct pour, in which the melt is poured via a fiber sleeve and filter sited on top of the casting, is risky but may be worth the risk. It has some key advantages: It is extremely simple, takes up minimal space in the mold, and provides a combined filling and feeding action. It provides significant reproducibility to the pouring action and, for this reason, can deliver 100% good or 100% scrap results. Thus, some trial and error may be involved in finding a suitable location where it gives good results for a particular casting. It appears to work best for small fall distances below the filter, its effectiveness decreasing with fall

Filter arrangements combined with bubble traps. Source: Ref 15

512 / Principles and Practices of Shape Casting height up to 200 mm (8 in.), after which its benefit falls to zero (Ref 13). Also not included in this review is the problem of filling very thin sections in which flow is controlled by surface tension and which suffer the defect-generating problem of microjetting (Ref 2), a kind of naturally occurring microturbulence. Despite its potential importance, this phenomenon is not well understood, requiring further research.

Recommended Foundry Practice Every new foundry should, wherever possible and appropriate, eliminate all pouring of metal and employ a good countergravity filling system to achieve mold filling at velocities below the critical velocity, thereby achieving sound and high-performance castings on a routine basis. Processes that include unconstrained pouring by gravity (such as most low-pressure permanent mold techniques in which the pressure vessel is filled by pouring) have problems delivering products of reliable quality. If countergravity cannot be implemented and gravity must be used, then the next-best option is the prepriming of parts of the filling system, or better, if possible, the whole of the filling system if a permeable mold is used. If prepriming is not possible, then adoption of tilt casting (with suitable precautions as listed in Ref 1) or the naturally pressurized filling system is the next-best option. Key features of the naturally pressurized gravity system are
Control air entrainment at entrance to sys-

tem by offset stepped basin or, even better, by contact pour.


During the early stages of filling, target to

fill and pressurize the whole system with melt in one pass (simulate pressure distribution if possible to ensure pressurization at all locations along the length of the running system).
Control entrainment events at the entry to the mold cavity, limiting jetting by reducing velocity to less than 1 m/s (3 ft/s) by such devices as accumulated frictional loss (possibly by appropriate use of a filter), surge control (Ref 1), or some prepriming filling (Ref 1). Pressurize the casting during freezing with high top feeders to suppress the opening of bifilms. The benefits are suppression of porosity (and, to some extent, tears) and maintenance of tolerable properties (excellent properties will only result from the elimination of bifilms).

6. 7. 8.

9. 10.

ACKNOWLEDGMENT

11.

This article is based on a paper first submitted to TMS Shape Casting Symposium 2007 but is published here expanded and updated.

12.

REFERENCES 1. J. Campbell, Casting Practice, Butterworth-Heinemann, 2004 2. J. Campbell, Castings, 2nd ed., Butterworth-Heinemann, 2003 3. J. Campbell; Shape Casting: The John Campbell Symposium, M. Tiryakioglu and P.N. Crepeau, Ed., TMS, 2005, p 3–12 4. J. Campbell, Mater. Sci. Technol., Vol 22, 2006, p 127–143 and correspondence p 999–1008 5. B. Sirrell, M. Holliday, and J. Campbell, The Seventh Conf. on the Modeling of

13. 14. 15. 16. 17.

Casting, Welding and Advanced Solidification Processes (London), M. Cross and J. Campbell, Ed., 1995 X. Yang and J. Campbell, Int. J. Cast Met. Res., Vol 10, 1998, p 239–253 B.M. Cox, D. Doutre, P. Enright, and R. Provencher, Trans. AFS, Vol 102, 1994, p 687–692 M. Emamy, R. Taghiabadi, M. Mahmudi, and J. Campbell, Second Int. Al Casting Technology Symposium, Advances in Aluminum Casting Technology II, Oct 7–10, 2002 (Columbus, OH), ASM International, p 85–90 T. Isawa and J. Campbell, Trans. Jpn. Foundrymen’s Soc., Vol 13, Nov 1999, p 38–49 X. Kang, D. Li, L. Xia, J. Campbell, and Y. Li, Shape Casting: The John Campbell Symposium, M. Tiryakioglu and P.N. Crepeau, Ed., TMS, 2005, p 377–384 F.-Y. Hsu, M. Jolly, and J. Campbell, Second International Symposium, Shape Casting, P.N. Crepeau, M. Tiryakioglu, and J. Campbell Ed., TMS, 2007, p 101–108. H. Nieswaag and H.J.J. Deen, 57th World Foundry Congress (Osaka), 1990, part 10, p 2–9 T. Din, R. Kendrick, and J. Campbell, Trans. AFS, Vol 111, 2003, paper 03-017 A. Habibollah Zadeh and J. Campbell, Trans. AFS, Vol 110, 2002, p 19–35 N.R. Green and J. Campbell, Trans. AFS, Vol 102, 1994, p 341–347 C. Nyahumwa, N.R. Green, and J. Campbell, Trans. AFS, Vol 106, 1998, p 215–223 H. Sina, M. Emamy, M. Saremi, A. Keyvani, M. Mahta, and J. Campbell, The Influence of Ti and Zr on Electrochemical Properties of Aluminum Sacrificial Anodes, Mater. Sci. Eng. A, Vol 431 (No. 1–2), 2006, p 263–276

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 513-521 DOI: 10.1361/asmhba0005355

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Processing and Finishing of Castings After solidification and cooling, further processing and finishing of the castings is required. The required operations and levels of finishing depend on the process (green sand, no-bake sand, lost foam, investment casting, etc.) and on the type of metal being cast. In general, however, the castings must first be removed from the tree assembly, consisting of the casting with additional areas of metal (sprues and risers) used to direct metal flow during pouring. Cut-off saws are usually used for this task. Risers must be removed by either cutting them off or knocking them off by force. Once the castings are separated from the trees or risers, additional finishing operations are done to remove all flash and parting line metal and to achieve the required finish requirements. These finishing operations include:

Shakeout and Core Knockout After solidification and cooling of castings in sand molds, the casting must be removed from the sand mold. In the case of green sand or resin-bonded sand molds, a considerable amount of energy is required to remove castings and break up lumps of sand still strongly bonded together. In contrast, unbonded sand systems (such as lost foam) do not require as severe mechanical agitation for shakeout. Shakeout from bonded sand molds is done by shaker pan conveyors, vibrating deck or grate, or any other type of vibratory, oscillating, or rotational mechanical action. Some castings will crack if shaken out too soon. Others require a controlled cooling rate to attain a

specific microstructure. In establishing the time for shakeout, consideration must be given to the metal, pouring temperature, casting size and design, coring, and sand mixture used. Shakeout equipment is described in the article “Green Sand Molding” in this Volume. With the exception of those castings poured from low-melting-point metals (aluminum or magnesium), the heat of the metal burns out the resin bond in the sand sufficiently for the mechanical action of the shakeout device to cause almost all of the sand to fall away from the casting. Likewise, core knockout, or decoring, is the process of removing internal resinbonded sand cores from castings. Ferrous metal castings are poured at much higher temperatures (1650  C, or 3000  F) than nonferrous

 Shotblasting  Grinding (from coarse to fine, depending on

the casting finish quality)

 Trimming  Machining (milling, drilling)  Quality testing and inspection

Depending on the alloy and requirements, the processing of castings may also include repair welding, heat treatment, and/or hot isostatic pressing. These additional operations and the machining of castings are not discussed in this article. This article briefly describes the general operations of shakeout, grinding, cleaning, and inspection of castings. Particular emphasis is also placed on automation technology, which can be an important productivity factor for foundries. Cleaning (fettling) operations are traditionally a very labor-intensive component of the cost of a casting, often exceeding 60% of the total casting cost. A very high percentage of cleaning operations may still be performed manually. However, robotic technology (Fig. 1) is becoming more prevalent. Until recently, casting variation and the inability to program robots easily in an offline environment prevented many foundries from automating. Both of these problems have been overcome, and user-friendly solutions are now more widely available. The most common application of robots in any manufacturing environment consists of pick-and-place tasks. This is also true in the foundry industry.

Fig. 1

Robotic handling in grinding. Courtesy of Vulcan Engineering Company

514 / Principles and Practices of Shape Casting castings. Consequently, more energy is required to remove core sand from aluminum castings than from gray iron castings. In gray iron, relatively low-energy shaking, vibrating, tumbling, and/or shotblasting is used to force loose core sand out through the casting openings. Core sand in ferrous castings has much less strength because the ambient heat of the metal, as it solidifies in the mold, burns the resin-bonding agents out of the sand. When the resin is burned out, the grains of sand cannot attach to one another, making high-production cleaning of ferrous castings less complex. The major focus of this section is the cleaning of aluminum castings, which are molded in a temperature range of 705 to 815  C (1300 to 1500  F), approximately one-half the temperature required to cast ferrous metals. This comparatively low molding temperature does not allow sufficient heat to transfer from the casting walls to the resin-bonded core sand. Only the cored passageway closest to the metal will show heat penetration and breakdown. Sand remaining on the surfaces of the castings can be removed by abrasive blast cleaning, as described in the section “Blast Cleaning of Castings” in this article. Core Knockout. Casting complexity, type of resin, type of sand, wall thickness, type of core (solid or hollow), metal-to-sand ratio, and gating/ risering affect the amount of energy required to remove the sand core from the casting. The most common method used to clean core sand from aluminum castings is mechanical vibration with pneumatic tools. In most instances, the core sand has less inherent strength than does the metal; therefore, it breaks down without metal damage. Many foundries use hand-held, vertical, A-frame, or fixtured air tools that operate in the 1000 to 3000 blows per minute range. The hand-held air tool method of decoring castings involves an operator placing a casting on the floor or on a table and locating the chisel of the air tool on the risering. When the operator pulls the trigger on the tool to allow air to pass through and create vibration, the casting shakes, and core sand breaks down. Each type of casting and each core sand is unique. For optimal cleaning, the proper type of vibration and/or blows per minute must be suited for each application. This may not be easily determined, so cracked castings or excessive air tool component failure may result. The advantages of hand-held air tools include:  Low cost  Flexibility (can be used on many different

castings)  Effectiveness applications

for

low-volume

cleaning

The disadvantages are:  Noise (100 to 120 dB); hearing protection

required for personnel

 Vibration, which affects operator’s hand,

arm, and body

   

Required maintenance of air tool At times, insufficient energy to clean casting Operator required to hold air tool Creation of silica dust, which affects operator  Insufficient speed for high production The vertical core knockout machine (Fig. 2) involves the use of a specially made air tool that strokes down to clamp the casting for holding and simultaneously vibrates the casting to break down the core sand. When the operator sees that the internal passages are cleared, the foot switch is depressed to stop the vibration, and the air tool is retracted from the casting. These tools are commonly referred to as C-frame models because the supporting structure resembles the letter “C.” This cleaning method uses the vibration-under-compression theory, which is generally applicable for all castings, yet is somewhat slow. The advantages of vertical core knockout tools include:  Low cost  Flexibility (can be effectively used on cast-

ings from 2 to 15 kg, or 5 to 35 lb)

 No adverse effect on operator’s hands, arms,

or body

 Several machines can be run by one operator  Applicability to low-volume production  Relatively small amount of floor space

Fig. 2

Vertical core knockout machine

Fig. 3

A-frame core knockout machine

required  Portability  Easily adjustable frequency and amplitude The disadvantages are:      

Noise (100 to 120 dB) Required maintenance of air tool parts Lower effectiveness on larger castings Insufficient speed for high production At times, inappropriate for a cleaning method Creation of silica dust

The A-frame core knockout machine involves the use of a cast iron clamp/vibrating head and cast iron clamping head mounted on a common shaft and A-frame leg stands and controls (Fig. 3). The cast iron heads are supported by the common mounting shaft and face one another. The left side contains a clamping and vibrating tool, and the right side contains a bumper or an air cushion device to soften the vibration and energy transmission. To operate the A-frame core knockout machine, an operator positions the casting between the clamp heads and depresses a foot switch to simultaneously hold and vibrate the casting to break down the core sand. This method allows the operator to rotate and probe the casting to reduce cleaning cycle time. When the core sand has been removed, the operator depresses the foot switch to stop vibration and unclamp the casting. The A-frame machines use vibration under compression, which is generally suitable for all types of castings. On this machine, both

the clamping pressure and the vibrating pressure are independently adjustable, so the vibration can be adapted to suit the casting. The advantages of A-frame machines include:  Moderate price  Flexibility (can be used on both large and     

small castings, that is, from 2 to 45 kg, or 5 to 100 lb) No harmful effect on operator’s body Several machines can be run by one operator Applicability to low- and high-volume production Rugged construction to withstand abusive foundry practices Approximately 2.5 m2 (25 ft2) of floor space required

Processing and Finishing of Castings / 515 The disadvantages are:  Noise (100 to 120 dB)  Multiple-unit facilities generate sound waves      

that can have harmful effects on internal organs. In medium- to high-production plants, excessive maintenance is required. High replacement part inventory is necessary. Castings must be transported to and from special acoustical rooms. Insufficient speed for high production Ineffectiveness on thin-walled heavily cored castings Creation of silica dust

The multiple-station air tool fixture involves the use of air tools mounted on a supporting structure with holding and locating blocks to position castings. These fixtures are dedicated to the core knockout of a specific casting. These devices either operate by means of vibration under compression, whereby the casting is clamped and vibrated, or they have a 6 to 10 mm (¼ to 3/8 in.) gap surrounding the casting, which undergoes vibration. As the casting is impacted, it moves violently within this tight space. This allows more energy to penetrate the casting from many different directions. The major problem with this method is that loading and unloading of the casting can be difficult because of the varying sizes of risers. The retractable tools employing vibration under compression allow easy loading/unloading and maintenance, yet can reduce cleaning effectiveness. With dedicated fixtures, multiple air tools can be used on a single casting, thus inducing greater amounts of energy to those areas that require it most. Multiple-station fixtures are normally built to clean two, three, or four castings simultaneously. These core knockout fixtures must be isolated from their support structures by vibration-proof mounts. Dedicated fixtures are most commonly used on high-volume-production applications, although certain low-to-medium production run castings may require dedicated tooling because of the complexity of the casting. These units generate extremely high noise levels; therefore, they are often housed in acoustical-control cabinets that are designed to be portable with interchangeable fixtures (Fig. 4). The advantages of acoustically enclosed, dedicated fixture units include:  Low noise levels  High production       

and lower per-casting cleaning cost Easier cleaning Timed cycle Interchangeable fixtures Sand containment Improved safety Portability Location possible near production line

The disadvantages are as follows:  High initial capital expenditure

Fig. 4

Acoustical-control cabinets. (a) With three-station dedicated fixture. (b) With A-frame tooling

 Acoustical enclosures are a nonincome-

producing expense  Maintenance required for air tool components  Few available suppliers  Fixturing dedicated to specific castings (not universal) A high-frequency drive machine involves the use of two motors spinning weights in opposite directions to develop a rapid push-pull movement of a load table. The casting is hydraulically retained on the mounting table. The rapid back-and-forth movement causes a floating piston to move and contact a strike plate that transmits energy to the casting. This high-frequency drive device is said to develop between 5 and 32 g (50 and 315 m s 2, or 160 and 1030 ft s 2) of force. The motors operate at between 2900 and 3600 rpm. The optimal amplitude or stroke is between 5 and 10 mm (3/16 and 3/8 in.). These units have proved to be effective in cleaning some of the most difficult castings. However, they are very loud, and the hydraulic clamping devices have failed in production settings. The advantages of high-frequency drive machines include:




   

Speed Adjustability Interchangeable fixtures Timed cycle




The disadvantages are:    

High cost Self-destruction Dedicated construction Noise (110 to 130 dB)

The high-pressure water blast device incorporates the use of a gimbal-mounted spray gun positioned in a stainless steel enclosure with a high-pressure pumping system. The successful operating pressures lie within a 17 to 105 MPa (2.5 to 15 ksi) range. A casting is located on a rotating, tiltable table, and the water is directed at the core material through the spray gun. The operator moves the casting to blast the water at the cored passageways. This system is effective in cleaning ceramic material from investment castings and complex sand-cored castings, and is also effective for removing burnt-in core sand from internal passages. The advantages of a high-pressure water blast system include:     

Adjustable water pressure Water can be focused directly at core Safety Quiet operation Dedicated fixturing can be used The disadvantages are:

 High cost  Contaminated water disposal

516 / Principles and Practices of Shape Casting     

Low operating speed Maintenance required for nozzle and pump Unreliable water recirculation Expensive spare parts Possible inability to direct water at blind or oblique passageways

Shotblast Cleaning. In the shotblast cleaning method, castings are hung on a tree and placed in an enclosure. Shot is spun off a wheel at the casting. This method is not very effective for cleaning aluminum castings because the energy is not sufficient to break down the cores, and the cycle time is longer than with other methods. Also, mixing sand with spent shot can cause excessive wear to the reclamation system. Nevertheless, some foundrymen do use this process because of the availability of a shotblast machine and/or because the casting requires shotblasting. Core Bake-Out Method. With the core bake-out method, castings are loaded into baskets and placed into the ovens. The castings are heated to the temperature that will completely burn out the resin binders without affecting the metallurgical structure of the casting. Eventually, the sand becomes loose enough to flow out of the castings when moved. This method is normally used when all other processes fail, but it is expensive, time consuming, and requires heat energy. The castings still hold the loose sand. Either the loose sand that does fall out has to be cleaned from the ovens or a sand reclamation unit must be added to the system.

machine or it can hold the grinding tools and apply them to the casting. The advantages of using a robot to manipulate a casting include:  The robot can pick up the casting from a

 





fixed point that can be magazine fed or from several predetermined positions if the castings are palletized. The robot can pick up castings within the limits of its capacity that would, in many cases, be too heavy for a manual operation. The robot can move the casting to a succession of different tools to perform the cleaning operation to the required standard in the minimum amount of time by using the most efficient tool available at each operation. These sequential operations can be carried out without having to stop the cycle to change tools. The robot can put the casting down on a fixture and regrip it to allow access to other surfaces of the casting. For example, the robot can turn the casting around or completely over. Upon completion of cleaning, the robot can place the casting in a fixture, drop it in a

bin, or place it on any predetermined position on a pallet. When the casting weight is high or when a variety of grinding wheels are necessary, it is more feasible to allow the robot to handle the tool rather than the casting. Industrial manipulators, although not programmable, are useful for this application. Tool change downtime is obviously very important with such a system. A number of problems are inherent in any system, regardless of the type used. These problems include the following:  Grinding develops varying load conditions,

which cause inaccuracies in robot positioning. This can be overcome by either an adaptive control system or by a template-driven system, such as that shown in Fig. 6.  Imprecision in casting geometries creates inaccuracies in location.  Corrections must be made for wheel wear; simple light beam sensors are often used to sense the location of the edge of the wheel.

Cleaning Operations Cleaning operations can be divided into two categories: fixed and variable. Fixed grinding is the removal of the material that is present on every casting in a fixed position (that is, feeder pads, flash, gates, and so on). For highvolume production, automatic fixed-stop machines have been specifically designed to grind castings. In contrast, variable cleaning occurs anywhere on the casting and usually results from defects in the mold or cores. Robotic technology is not sufficiently developed for cost-effective automation of variable cleaning operations. Programmable cleaning operations are normally conducted by robots. Figure 5 shows a typical robotic installation. This cell consists of a robot and four specially designed grinding wheels. The wheels are hydraulically driven from a central unit. Wheel wear compensation is automatic such that the working point of the wheel is at the exact same location from cycle to cycle. Before entering the cell, castings are introduced to a dedicated press, where the main riser is removed and specific contour locations are stamped that will serve as reference areas for positioning. The casting is picked up by the robot and moved to each grinding machine. Depending on casting size, the robot can either hold the casting and move it to the grinding

Fig. 5

Schematic of a first-generation robotic system designed to perform fettling operations

Fig. 6

Guidance template used to control the movement of the robotic arm

Processing and Finishing of Castings / 517

Automating Gate Removal and Grinding ((Ref 1) As noted, the improved ability to easily program robots in an offline environment improves automation capabilities for highly labor-intensive operations. Automating gate removal and grinding (Fig. 7) is now very feasible for short-run casting producers. Cells can be easily reconfigured for quick and easy changeover between jobs. General Factors in Automated or Manual Gate Removal. To optimize the gate removal process, there are several things that should be considered. Many of the following items apply to manual and automated gate removal. Consistent gating includes the following considerations:  The more consistent the runner and gating

system, the easier it is to remove the castings from the runner system. Providing gating that has minimal warpage and is flat across the gating attachment area will make manual removal easier but will make automation even more cost-effective.  Limiting the number of gating schemes will reduce the number of fixtures, grippers, and programs that are needed to process parts. In many cases, the program, gripping points, and infeed features can be the same for a variety of parts.  Design gating so that it is easy to automate. With a little forethought, gating can be designed so that it will allow the robot

Fig. 7

gripper to consistently locate the tree at the pickup location. By adding features to the gating that allow the robot to use the same pick points, the same robot program can be used for a large cross section of parts. This eliminates having to teach many programs. Ease of Programming Robotic programming for gate removal is normally a fairly simple task. Some things to consider when automating gate removal are:  Develop a standard set of base programs that

are easily modified for different parts.

 If done correctly, then new programs only

require changing of variables. This greatly reduces the time to generate a program and will keep the cell in production. Flexibility of System to Accommodate Multiple Parts The next area to consider is the ability of the cell to accommodate a large variety of parts. It is also very important that changeover time between parts can be accomplished very quickly. In the ideal situation, the changeover can be done during the cycle time of the robot. There are also several other factors that a properly configured cell will contain to provide flexibility:  The use of vision and/or laser measurement

systems provides an easy method to find and determine the orientation of the runner system. This information is then uploaded to the robot, which reorients the gripper to pick up the part. The use of these systems

Gate removal. Courtesy of Vulcan Engineering Company

eliminates or greatly simplifies fixturing and grippers.  Human/machine interface needs to be userfriendly to allow the operator to easily change over between parts without ever having to use the teach pendant. It should contain all the information that is required to allow the monitoring of alarms, wheel wear, and job information. All programs that are run on the cell should be stored on this computer, with quick upload of the correct programs to the cell.  If locating fixtures or different grippers are required, they should be quick-change. Depending of the length of the job runs, the decision to use manual quick-change or fully automatic tool change must be considered.  For parts that use multiple fixtures or grippers, work objects must be the same so that only one program must be taught. Proper Peripheral Equipment The cell is only as good as the peripheral equipment around the robot. When looking at the overall scope of work to be performed, the first decision that must be made is to determine if the spindle is mounted on the robot or if the part/runner system is picked up by the robot and taken to the saw. Once this is determined, the following items are very important for either configuration:  The cut-off saw must have programmable

force control. This is the most important factor to properly cut the gating system. The saw must have the force control on it because this overcomes the inability of the robot to change velocity dynamically at the rate required by the cutoff operation. Robots are very good at following a path and moving very accurately to a point, but due to scan times from an input device, they cannot react quickly enough to maintain the proper cutting pressure.  Because the abrasive wheels naturally wear down, a measurement system must be used to offset robot programs to compensate for this wear.  The system must be able to track wear to determine the life of the wheel and notify the operator when it is time to change a wheel. Force control is important to properly cut the castings from the runner system. Figure 8(a) shows what happens when the cutoff saw does not have force control. As the arc of contact of the wheel moves into the riser and the velocity or feed rate is too high, the amount of force exerted on the wheel causes an increase in temperature. This increase in temperature does two things: It will cause the wheel to break down at a rapid rate that results in poor wheel usage, and the heat generated is transferred to the casting. If the casting is a heat-sensitive alloy, this can cause heat cracking. To solve the problems created in Fig. 8(a), the programmer attempts to slow down the feed rate.

518 / Principles and Practices of Shape Casting This may seem like a good idea, but the same thing happens as mentioned previously. Because the feed rate is too slow, the wheel does not break down properly to keep the abrasive sharp. The wheel becomes loaded with material, which causes an increase in friction. As this happens, the temperature of the wheel increases, which heats up the resin that holds the abrasive together and allows it to break down prematurely. The friction also causes the part to become hot (Fig. 8b) and again can cause heat cracking in certain alloys. Figure 8(c) shows what happens in the cut when the machine is equipped with force control. The correct cutting force is constantly applied against the part, and the spindle actually moves back and forth as the arc of contact changes along the circumference of the wheel. Because the system is able to cut with the correct force, the results are a cool cut that optimizes wheel life and provides for quicker gate removal. Gate Grinding. After parts have been removed from the runner system, there is remaining gating material that must be removed to complete the process. There are several factors that must be considered to properly grind gates:  Determine the best method to grind: Does the

robot take the casting to the grinding equipment, or is a spindle mounted on the robot, and the grinding medium is taken to the casting?  Determine grinding tolerances that can be achieved. This can depend on a number of

things: part size, gripper or fixture location in relation to the area to be ground, and variation in the casting versus the final tolerance to be achieved.  If the tolerance is very tight or if the casting variation is greater than the final tolerance, what type of measurement devices are required to complete the task?

Choosing the correct method of introducing the part into the cell is very important. Several options are:  Directly mounting the part on the robot. This



Different methods of programming the cell may be considered, depending on the application. The different types of programming methods are:  The teach pendant is used to teach simple

parts that consist of curves and straight grinds.

 Offline programming using solid models

allows for quick generation of multiple paths that have many points along a profile. This method is less disruptive to production because the program is written in the office and then downloaded to the robot. This method is a requirement for parts with complex geometries.  The use of a digitizer is another way to program offline. It is typically used where no solid model of the part exists. A replica of the same diameter of the grinding wheel is mounted on the end of the digitizer. As the programmer moves the digitizer around the part, he uses a foot switch to teach the points along the grinding path. Again, once this program is taught, it is downloaded directly into the robot.







method requires the operator to unload and load the robot after every cycle, which can impact throughput. Fixture tables with part fixture plates are effective for larger-sized parts. Two to four stations can normally be arranged in front of the robot. The robot picks up the fixture plate and takes it to the grinding equipment. While the robot is working, the operator can unload and load the next part on the remaining stations. This method helps maximize the robot grinding time in the cell. Because multiple input stations are used, this frees the operator to do other tasks while the robot works on processing the casting on the fixture tables. Sliding draws are typically used for smaller parts, where several are placed on a fixture and processed one after the other. Usually, there are two draws located next to each other so that the operator can unload or load while the robot is working. A turntable with a work station on each side allows the operator to load and unload a part while the robot is working. These devices are usually configured with quick-change fixturing that allows the operator to quickly change between jobs. Vision systems allow casting to be sent into the cell on belt conveyors; the vision system then tells the robot the orientation of the part. The robot picks up the part and takes it to the grinding machines. This method eliminates the need for costly fixturing. Vision systems, along with quick-change grippers, provide a quick changeover between jobs, which makes a cell very productive.

A large area of cost savings can come from the combination of several functions in a cell. Figure 9 shows a combination cell. This cell combines both the cut-off and the grinding operations in one cell. One advantage of this type of cell is that it reduces the amount of work that would be required for a cut-off cell and a grinding cell in a process. It also makes quick work of a casting that must be cut off from the gating

Fig. 8

Force control in cutting castings from the runner system. (a) Effect when the cutoff saw does not have force control. (b) Overheating when feed rate is too slow. (c) Proper force control of cutting. Source:Ref 1

Fig. 9

Combination cell. Courtesy Engineering Company

of

Vulcan

Processing and Finishing of Castings / 519 system and then requires additional gate removal and grinding to final tolerance.

Flame Cutting For heavy-section steel castings, where risers cannot be removed economically by grinding, impact, and so on, robotic flame cutting is used for riser and gate removal. Unlike cleaning, however, the torch must be held by the robot. As in cleaning, such systems range from semiautomatic to fully programmable robotized systems. Installations can use both conventional and gantry-style robots. Steel thicknesses in excess of 610 mm (24 in.) have been successfully cut. To perform the flame cutting operation, the operator, using the control panel, moves the torch into the general position of the riser. The unit is preprogrammed for six riser contact sizes. The operator then selects the proper size, and the cut is executed. More automated systems use a six-axis robot for complete automation of the cutting operation. Specially designed locating blocks or pads are cast into each casting. Contact location principles are used to provide alignment between the robot and the casting. The robot is programmed using a teach pendant device. In addition to position and path, the operator programming the robot must also define the travel speed between each point, the preheat delay, and the cutting oxygen pressure for each cut. Results indicate a savings of 120 to 180%, depending on the casting to be cut. The following factors must be taken into account when considering an installation or application:  Fixturing: Distortion of gate position due to

rough handling, mold inaccuracy, and so on, leads to errors in position. Some means of sensing the positions of the gates and risers is necessary.  Feedback system: Because of variations in thickness and other defects, such as flash and drops on the casting, an adaptive control system that accounts for variations would be very beneficial.  Automatic torch ignition: This provides an added measure of safety and convenience.  Preheat sensing: One system is equipped with a fiber optic lens that sights through the oxygen orifice. An infrared sensing measurement is made to the computer controlling the robot. If the signal indicates kindling temperature, oxygen is employed, and the cut is continued.

Blast Cleaning of Castings Blast cleaning of castings is a process in which abrasive particles, usually steel shot or grit, are propelled at high velocity to impact the casting surface and thereby forcefully remove surface contaminants. The contaminants are usually adhering mold sand, burned-in sand, heat treat scale, and so on. For aluminum

castings, the process is often used to provide a uniform cosmetic finish in addition to merely cleaning the workpiece. This is especially true of engine components, such as heads and manifolds, that are highly visible. In the case of cast aluminum wheels, die cast transmission cases, and so on, the process prevents leakage by healing surface porosity. Additionally, in some cases, deburring and a pleasing cosmetic surface are obtained. Methods for the handling of castings during blast cleaning depend on production volume, size, shape, and mix of the parts and the ability of the parts to be tumbled. Robots and similar manipulative devices are also used in blastcleaning operations. The usual methods of imparting high velocity to abrasive particles are by the use of either centrifugal wheels or compressed air nozzles. Centrifugal wheels are the most widely used method because of their ability to efficiently propel large volumes of abrasive. For example, a 56 kW (75 hp) centrifugal wheel can accelerate steel shot to 73 m/s (240 ft/s) at 55.8 Mg/ h (123,000 lb/h) flow. To do the same with 13 mm (½ in.) direct-pressure venturi nozzles at 45 kg/min (100 lb/min) per nozzle would require approximately 20 nozzles and a total air flow of 2.45 m3/s (5200 ft3/min) at 550 kPa (80 psi). However, even though the nozzle blast is not as efficient overall as the wheel blast, in some applications it may be more efficient because the blast stream can be more efficiently applied when blasting into small holes to clean the interior areas of a casting. Other reasons for using nozzle blast include portability and suitability for very hard abrasives, such as aluminum oxide. The machine type most often used is a batch tumbling barrel when the production requirement is low to moderate for parts that can be tumbled. Continuous tumbling is used for high-volume production. When castings cannot be tumbled, the machine type most often used is the rotating-hanger-type machine (Fig. 10). The hanger and hook are moved horizontally to a position opposite the centrifugal wheels, where the castings are blasted while rotating about a vertical axis. These machines come in a number of configurations. When higher production rates are required, a turnstile-type hanger machine is sometimes used. Hangers may also be powered or controlled in various ways. When parts are too large to either tumble or hang on a hook, or when production requirements are low, the machine types most often used are either a table-type machine or a room-type machine. Small table machines are used to handle smaller castings that cannot be tumbled. They are also used when size, shape, or low production requirements rule out a hanger machine. The part or parts are placed on the table by a power lift device, and, after the doors are closed, the parts are blasted by one to three centrifugal wheels. A disadvantage of this type of machine is that usually the parts

must be turned over or repositioned, and the blast cycle must be repeated to effect complete cleaning. Room-type machines are used when the requirement is low production on a variety of medium-sized to very large parts. These machines usually consist of a room with power-operated entrance doors. A self-powered car with a rotating table carries the work to be blasted into the room to the blast position. Mounted on the room are two or more centrifugal wheels. While the rotating part is being blasted, either the car moves to three blast positions, or, in some cases, the wheels are oscillated for better blast coverage. As with a table machine, the part or parts may need to be turned over or repositioned for complete coverage. Cleaning Internal Surfaces. Many times when cleaning castings it is important to clean internal surfaces such as the water passages in engine heads and blocks or other surfaces that cannot be impacted directly because of the casting configuration. In these cases, ricocheting abrasive must do the cleaning. To clean internal surfaces, it is usually important to drain the internal passages while blasting so that pocketed abrasive does not cushion the impact. It is also important to have sufficient ricochet velocity to remove the surface contaminants effectively. In the majority of cases, the best way to drain internal passages is to rotate the part on a horizontal axis, as is done on axial-flow machines or other machines that incorporate horizontal rotating devices. Recent laboratory tests and field experience have shown that increasing the wheel speeds improves internal cleaning significantly. At times, the blast from a centrifugal wheel does not effectively reach into all of the internal cavities of a casting. Often, the solution to this situation is to resort to a “spot blast,” that is, use of a compressed-air nozzle to direct the blast into selected casting openings. Here again, high abrasive velocity can be effective. Changing from straight nozzles to long venturi nozzles can result in abrasive velocity increases

Fig. 10

Front view of a typical rotating assembly. The workpiece is rotated it is held in position on the monorail track to full coverage by the blast pattern produced centrifugal wheels.

hanger 360 as provide by the

520 / Principles and Practices of Shape Casting of 30% or more, or, if abrasive velocity is more important than coverage, the nozzle flow can be decreased and the velocity increased. Blast equipment manufacturers can advise nozzle users on velocity increases that are possible with different types of nozzle configurations and flow rate changes.

Table 1

Casting defects, descriptions, and prevention

Casting imperfections or defects

Cold shuts

Prevention

Pour as quickly as possible Design gating system to fill entire mold quickly without an interruption

Possible causes:

Preheat the mold Modify part design Avoid excessively long, thin sections

 Interruption in the pouring operation  Too slow a pouring rate  Improperly designed gating

Inspection After finishing, castings are inspected for surface quality. Inspection can be performed manually by visual checking, manually by template comparison, or by an automated inspection station. Visual inspection is simple yet informative. A visual inspection would include significant dimensional measurements as well as general appearance. Surface discontinuities often indicate the presence of internal discontinuities. Computer-assisted coordinate-measuring machines measure both pattern and casting dimensions. These machines can perform a variety of dimensional checks ranging from basic geometric measurement to parallel and plane projection. The operator simply identifies critical part locations so that the machine can establish a working plane. The coordinate-measuring machine can perform in a few minutes the tedious checks that take 2 to 4 h to be done manually. For small castings produced in reasonable volume, a destructive metallographic inspection on randomly selected samples is practical and economical. This is especially true on a new casting for which foundry practice has not been optimized and a satisfactory repeatability level has not been achieved. For castings of some of the harder and stronger alloys, a hardness test is a good means of estimating the level of mechanical properties. Hardness tests are of less value for the softer tin bronze alloys because hardness tests do not reflect casting soundness and integrity. Inspection also includes various methods of nondestructive testing (NDT) to screen castings for imperfections that may be considered to be defects. The techniques for NDT of casting are briefly summarized as follows, with further information given in the article “Nondestructive Testing of Components” in this Volume. The general types of imperfections or defects in castings are listed in Table 1. As a general rule, the method of inspection applied to some of the first castings made from a new pattern should include all those methods that provide a basis for judgment of the acceptability of the casting for the intended application. Any deficiencies or defects should be reviewed and the degree of perfection defined. This procedure can be repeated on successive production runs until repeatability has been ensured. Liquid penetrant inspection is extensively used as a visual aid for detecting surface flaws. One of the most useful applications, however, is the inspection of alloys susceptible to hot

Description

Appear as folds in the metal — occurs when two streams of cold molten metal meet and do not completely weld

Hot tears and cracks

Hot tears are cracklike defects that occur during solidification due to overstressing of the solidifying metal as thermal gradients develop. Cracks occur during the cooldown of the casting after solidification is complete due to uneven contraction.

Fill mold as quickly as possible Change gating system; e.g., use several smaller gates in place of one large gate Apply thermal management techniques within the mold (e.g., chills or exothermic material) to control solidification direction and rate Insulate the mold to reduce its cooling rate Modify casting design:

 Avoid sharp transitions between thin and thick sections

 Taper thin sections to facilitate establishment of appropriate solidification gradients

 Strengthen the weak section with additional material, ribs, etc. Inclusions

Presence of foreign material in the microstructure of the casting

Modify gating system to include a strainer core to filter out slag

Typical sources:

Avoid metal flow turbulence in the gating system that could cause erosion of the mold Improve hardness of the mold and core

 Furnace slag  Mold and core material Misruns

Incomplete filling of the mold cavity Causes:

   

Too low a pouring temperature Too slow a pouring rate Too low a mold temperature High backpressure from gases combined with low mold permeability  Inadequate gating Porosity

Holes in the cast material Causes:

Control mold and metal temperature Increase the pouring rate Increase the pouring pressure Modify gating system to direct metal to thinner and difficult-to-feed sections more quickly

Pour metal at lowest possible temperature Design gating system for rapid but uniform filling of the mold, providing an escape path for any gas that is generated Select a mold material with higher gas permeability

 Dissolved or entrained gases in the liquid metal

 Gas generation resulting from a reaction between molten metal and the mold material Microshrinkage Liquid metal does not fill all the dendrific interstices, causing the appearance of solidification microshrinkage

Control direction of solidification:

 Design gating system to fill mold cavity so that solidification begins at the extremities and progresses toward the feed gate  Lower the mold temperature and increase the pouring temperature  Add risers, use exothermic toppings to maintain temperature longer  Control cooling rate using chills, insulators, etc. in selected portions of the mold

Source:Ref 2

cracking. Such cracks are virtually undetectable by unaided visual inspection but are readily detectable by liquid penetrant inspection. Pressure testing is used for castings that must be leaktight. Cored-out passages and internal cavities are first sealed off with special fixtures having air inlets. These inlets are used to build up the air pressure on the inside of

the casting. The entire casting is then immersed in a tank of water, or it is covered by a soap solution. Bubbles will mark any point of air leakage. Radiographic inspection is a very effective means of detecting such conditions as cold shuts, internal shrinkage, porosity, core shifts, and inclusions in aluminum alloy castings.

Processing and Finishing of Castings / 521 Radiography can also be used to measure the thickness of specific sections. Aluminum alloy castings are ideally suited to examination by radiography because of their relatively low density; a given thickness of aluminum alloy can be penetrated with approximately one-third the power required for penetrating the same thickness of steel. The typical complexity of shape and varying section thicknesses of the castings may require digital radiography or computed tomography. Magnetic particle inspection can be applied to ferrous metals with excellent sensitivity, although a crack in a ferrous casting can often be seen by visual inspection. Magnetic particle inspection provides good crack delineation, but the method should not be used to detect other defects. Nonrelevant magnetic particle indications occasionally occur on ferrous castings, especially with a strong magnetic field. For example, a properly fused-in steel

chaplet can be indicated as a defect because of the difference in magnetic response between low-carbon steel and cast iron. Even the graphite in cast iron, which is nonmagnetic, can cause a nonrelevant indication. Ultrasonic inspection for both thickness and defects is practical with most ferrous castings except for the high-carbon gray iron castings, which have a high damping capacity and absorb much of the input energy. The measurement of resonant frequency is a good method of inspecting some ductile iron castings for soundness and graphite shape. Electromagnetic testing can be used to distinguish metallurgical differences between castings. Aluminum alloy castings are sometimes inspected by ultrasonic methods to evaluate internal soundness or wall thickness. The principal uses of ultrasonic inspection for aluminum alloy castings include the detection of porosity in castings and internal cracks in ingots.

ACKNOWLEDGMENTS Adapted from Ref 1 and the following:  G.J. Maurer, Jr., Shakeout and Core Knock-

out, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 502–506  J.H. Carpenter, Blast Cleaning of Castings, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 506–520  R.L. Lewis and Y.L. Chu, Foundry Automation, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 566–573

REFERENCES 1. C. Cooper, “Automating Gate Removal and Grinding,” Vulcan Engineering Company 2. Mechanical Engineering Handbook, CRC, 1999

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 525-527 DOI: 10.1361/asmhba0005241

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Introduction—Expendable Mold Processes with Permanent Patterns Brian V. Smith, General Motors Corporation

CASTING can be done with either expendable molds for one-time use or permanent molds for reuse many times. Both expendable and permanent molds must be separable into two or more parts in order to permit withdrawal of either the permanent pattern (from an expendable mold) or the raw casting (in the case of a permanent mold or die). In the case of sand (or other loose granular material) molding, permanent patterns are removed after the sand is bonded in a flask in cope-and-drag sections (Fig. 1). Cores are separate shapes that are placed in the mold to provide castings with contours, cavities, and passages that are not practical or obtainable with molds. Patterns may be permanent (as is typical in sand casting) or expendable (as in lost foam and investment casting). With expendable patterns, the limitation of two or more separable parts of the mold is not necessary. The molding medium must only surround the expendable pattern and maintain its shape during molten-metal pouring and solidification; after the shape has solidified, the sand or other molding medium is shaken off and out of the part. Casting of parts using expendable mold processes with either permanent or expendable patterns is a very versatile molding method that provides tremendous freedom of design in terms of size, shape, and product quality.

article, “Aggregates and Binders for Expendable Molds,” in this Volume). Manufactured ceramic molding media with specific properties are also becoming more popular in various casting processes. The choice of molding method depends on several factors, such as part size and shape, quantity, tooling, and the molten metal being poured into the mold. Table 1 lists a general comparison of casting methods. The characteristics of foundry molds must address four basic requirements:  Be formable into the desired shape around

some type of pattern

 Be able to hold that shape while the molten

metal is introduced into it

 Be able to maintain that shape while the

molten metal solidifies

 Be able to break down and become strippa-

ble after the metal solidifies

The first requirement is that the mold material must flow or be pliable enough to encapsulate the surface around the shape of the pattern. Patterns (Fig. 2) can be made from a permanent-type material (i.e., wood, steel, hard plastic), so that numerous molds can be duplicated before significant pattern wear would alter the shape of the part being cast. Once the mold material is tightly bonded around the pattern, the mold must be stripped and removed from that pattern, while not damaging either the permanent pattern material or the established mold shape. The use of tapered pattern ends permits it to be removed from the sand mold without restriction (Fig. 3). The mold shape, which is typically the reverse or mirror image of the part being cast, must be strong enough to be handled and manipulated, so that it can be combined with other mold shapes (copes, drags, cores, gatings, etc.) that

Expendable mold processes are widely used in casting, and various methods are used to fabricate expendable molds from permanent patterns. The methods include:  Molding of sand with clay, inorganic binders,

or organic resins

 Shell molding of sand with a thin resin    

bonded shell that is baked No-bond vacuum molding of sand Plaster-mold casting Ceramic-mold casting Rammed graphite molding Magnetic (no-bond) molding of ferrous shot

Sand is the most prevalent molding medium, and various types of binders are used to bond the sand into useable molds (see the next

Fig. 1

Cope-and-drag mold halves with cores in place, ready for closing. Source: Ref 1

526 / Expendable Mold Casting Processes with Permanent Patterns Table 1 General comparison of casting processes Weight, kg (lb) Process

Typical materials cast

Sand Shell Expendable pattern Plaster mold Investment Permanent mold Die Centrifugal

All All All Nonferrous (Al, Mg, Zn, Cu) All (high melting point) All Nonferrous (Al, Mg, Zn, Cu) All

Minimum

0.05 0.05 0.05 0.05

(0.11) (0.11) (0.11) (0.11)

0.005 (0.011) 0.5 (1.1) <0.05 (0.11) ...

Maximum

Typical surface finish, mm, Ra (min.)

Porosity(a)

Shape complexity(a)

Dimensional accuracy(a)

Section thickness, mm (in.) Minimum

Maximum

No limit 100 + (220) No limit 50 + (110)

5–25 (200–1000) 1–3 (40–120) 5–20 (200–800) 1–2 (40–80)

4 4 4 3

1–2 2–3 1 1–2

3 2 2 2

3 2 2 1

(0.12) (0.08) (0.08) (0.04)

No limit ... No limit ...

100 + (220) 300 (660) 50 (110)

1–3 (40–120) 2–3 (80–120) 1–2 (40–80)

3 2–3 1–2

1 3–4 3–4

1 1 1

1 (0.04) 2 (0.08) 0.5 (0.02)

75 (3.0) 50 (2.0) 12 (0.5)

5000 + (11,000)

2–10 (80–400)

1–2

3–4

3

2 (0.08)

100 (4.0)

(a) Relative rating: 1, best; 5, worst. These ratings are only general; significant variations can occur, depending on the methods used.

Fig. 2

Wood pattern to produce a rammed graphite mold. Source. Ref 2

Fig. 4

Cold shut (at arrow) in the flange of a machined cast drum. Magnetic particle inspection revealed faint indications of the cold shut in the rough casting.

Fig. 3

Tapered pattern to ease removal

Fig. 5

Chill plates to eliminate porosity

produce the empty cavity of the negative shape of the part being cast. Typically, the plumbing system—sprues, runners, gates—that allows the molten metal to be introduced into the cast part shape is also made from these same molding materials and becomes part of the mold assembly. After complete assembly of this mold package, which could include multiple identical or different parts, the molten metal is introduced in some manner into this mold assembly. The molding material also must be able to withstand the erosion action of the molten metal as it flows through the gating system

and the part geometry without deteriorating or losing its shape. Any portion of the molding medium that may break away could be washed into the molten metal and end up as a defect within the solidified part being cast. The molten metal could also infiltrate the molding medium at the mold/metal interface, producing a nonhomogeneous scab of solidified metal and molding medium on the part being cast (see the article “Common Defects in Various Casting Processes” in this Volume). The third requirement for the molding medium is to maintain shape while the molten

metal solidifies. This is a critical requirement to ensure that the cast part in the mold accurately reproduces the outline of the pattern used to make the mold. During solidification, the thermal conductivity of the molding medium is also important in meeting the material properties of the metal being cast. Foundries must constantly balance the correct thermal conductivity of the molding medium. If heat conducts too quickly from the mold cavity, it could result in freezing of the metal before filling the entire mold cavity (Fig. 4). Conversely, slow heat conduction can result in slow cycle times or improper microstructure of the solidifying metal, which can lead to incorrect material properties for the part being cast. To assist solidification, the mold may include regions with metal chill plates, which are added to assist in quicker solidification in certain sections of cast parts (Fig. 5). Most metals can be heat treated after casting to assist in meeting the material properties of any particular part. Of course, this adds cost to the product, so any compromises that can be implemented in the heat conductance of the molding medium would save this cost. Finally, the fourth basic attribute of the molding medium is being able to be removed from the cast part. The expendable mold must be broken away and stripped from the solidified part (Fig. 6). More cast parts, especially those with intricate internal passages, are rendered unusable during this shakeout step of the casting process than any other place in the foundry. This is why some of the expendable pattern practices (i.e., lost foam and investment casting) may have an advantage over the permanent pattern methods of the expendable molding processes. Sand molding is a prevalent foundry process. Sand is an abundant raw material, although for the foundryman, sand means something much more than a layman’s concept of sand. For the foundryman, sand has numerous properties that he has learned how best to combine or mix with other materials so that it meets the four requirements specified previously for the molding material. Various terms are also used for the different sand molding processes and the methods that bond the millions of sand particles together around the permanent and/or expendable

Introduction—Expendable Mold Processes with Permanent Patterns / 527 are usually never green in color, but black) use natural or synthetic clays with other additives to bond the sand grains together. Green sand molding may be done either with flasks or without flasks (flaskless). Resin-bonded sand mold processes (called cold box, hot box, warm box, and shell molding) were originally based on organic resin binders, although lately even inorganic binders are mixed with the sand and then hardened (or cured) by chemical or thermal reactions to fixate the shapes. Often, these expendable resin-bonded processes are used to produce cores that are placed in permanent molds. This combination of expendable cores with permanent molds is referred to as semipermanent molding. This is used extensively in the nonferrous metal casting area. Besides sand, other materials can also meet the aforementioned requirements for molding media. Slurry molding uses a plaster and/or ceramic material to form the shape around either a permanent or expendable pattern. Graphite molding materials rammed around a pattern have also been used. Holding molding media together with a vacuum source after the media has taken the shape of the pattern is another innovative means of making molds. These methods of expendable molds fabricated from permanent patterns are described in the subsequent articles in this Section. REFERENCES

Fig. 6

Vibratory shakeout of castings

patterns (see the article “Aggregates and Binders for Expendable Molds” in this Volume).

Green sand molding is perhaps the most popular molding process. Green sand molds (which

1. Steel Casting Handbook, 6th ed., Steel Founders’ Society of America and ASM International, 1995 2. S. O’Connor, AMMTIAC Q., Vol 2 (No. 1)

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 528-548 DOI: 10.1361/asmhba0005242

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Aggregates and Binders for Expendable Molds Updated and reviewed by John Jorstad, J.L.J. Technologies, Inc.; Mary Beth Krusiak (deceased), Sand Technology Co.; John O. Serra, Carpenter Brothers Inc.; and Vic La Fay, Hill and Griffith

EXPENDABLE CASTING MOLDS can be fabricated with either permanent or expendable patterns that reproduce the desired configuration of the part to be cast. In sand molding, a pattern is placed in a flask (or molding machine chamber), which is then filled with a mixture of a refractory aggregate (sand or manufactured ceramic) and binder. The aggregate-binder mix surrounds the pattern and is then compacted and bonded or mechanically held together, so that a mold cavity remains intact when the pattern is removed from the flask (or, in the case of a flaskless mold, produced in a chamber that is integral to the molding machine). If the pattern is permanent (i.e., for reuse in making multiple molds), then each mold requires two flasks: one for the cope section and another for the drag section (Fig. 1). The cope and drag sections are then joined together to form the complete mold for casting. Alternatively, the pattern may be expendable, as in the case of lost foam patterns that evaporate during casting. In lost foam casting, the expendable foam pattern can be placed in one flask or container to form the mold. This eliminates the need for cope and drag sections that are required in removing permanent patterns. Expendable pattern methods, such as lost foam casting and investment (lost wax) casting, are also capable of producing more intricate patterns and castings (see the article “Introduction: Expendable Mold Processes with Expendable Patterns” in this Volume). After the part is cast and solidified in an expendable mold, the next step is a shakeout operation, which breaks up the mold and separates it from the cast part. The sand or mold aggregates can be reclaimed, screened, and processed for reuse. A general flow chart of sand processing in a foundry is shown in Fig. 2. The details of shakeout and reclamation depend on the type of casting process and the methods of bonding or holding the mold aggregates together. In lost foam casting, for example, the sand (or other molding material) is unbonded and just flows out of the part, so that

no violent shaking and/or heating of the mold is needed to break up the mold. In contrast, green (clay-bonded) sand or resin-bonded sand molds and cores require more aggressive shakeout and additional processing for reclamation. This article reviews the basic types of mold aggregates and bonding methods for expendable molds and coremaking. Sand processing, mold making, and sand reclamation are discussed in more detail in the subsequent articles “Green Sand Molding,” “No-Bake Sand Molding,” and “Coremaking” in this Volume. The emphasis is on bonded-sand molding methods that include:
Conventional














clay-bonded (green) sand molding with horizontally parted (cope and drag) molds produced by hand ramming, jolt-squeeze operations, or mechanized systems High-pressure (also known as high-density) clay-bonded sands from impact molding, mechanical horizontal parting, or mechanized vertical parting Inorganic no-bake molding with binders based on cement, silicates, plaster, or aluminum phosphate (which is virtually obsolete) Organic no-bake molding with binders based on acid-cured resins of furan or resoles (phenolic), alkaline phenolic (phenolic ester), acrylic epoxies, and phenolic urethane, which is probably the most significant in North America Thermally cured organic binders such as alkyd oil (also known as alkyd urethane or oil sand) and the phenolic (novolac) of the shell process (see the article “Shell Molding and Shell Coremaking” in this Volume). Rammed graphite for casting highly reactive metals such as titanium or zirconium

Of these, the three most important sand molding methods are shell molding, no-bake molding (phenolic urethane, furan, phenolic ester), and high-pressure green sand molding. These three methods account for nearly 95% of the molds made in the world (Ref 1). Core-oil use is minimal at best. Aluminum-phosphate-bonded

molding is virtually obsolete but is briefly described.

Mold Media Expendable molds for metal casting include:
Bonded

sand or manufactured ceramic aggregates
Unbonded aggregates in lost foam casting or vacuum sand molding (see the article “No-Bond Sand Molding” in this Volume)
Ceramic shell molds used in investment castings (see the section “Other Expendable Mold Media” in this article) The aggregate mold-base material and bonding system depends on the type of metal being poured, the type of casting being made, the availability of molding aggregates, the moldand coremaking equipment owned by the foundry, and the quality requirements of the customer. A thorough understanding of all of these factors is necessary to optimize the molding system used in the foundry. In terms of sand molding, the basic requirements of the aggregate material include:
Dimensional and thermal stability at ele-

vated temperatures Suitable particle size and shape Chemically unreactive with molten metals Not readily wetted by molten metals Freedom from volatiles that produce gas upon heating
Economical availability
Consistent cleanliness, composition, and pH
Compatibility with binder systems





Many minerals possess some of these features, but few have them all. The most prevalent mold base is silica sand, due to its wide availability and lower cost relative to other types of sands, such as zircon, olivine, and chromite. Manufactured ceramics (such as mullite pellets) are also used as mold-base aggregates.

Aggregates and Binders for Expendable Molds / 529 grains. The clay-bonded sand is not dried or baked—hence the term green sand. Green sand molding is the least expensive, fastest, and most common of all the currently available molding methods. The mixture of sand and binder can be used immediately after the mixing process that coats the sand grains. Typically, 90% or more of the sand can be reclaimed in green sand operations. Thus, the costs of sand and reclamation rates are an important consideration in comparing green sand with other binder systems. The choice of a specific binding system (or the use of no-bond sand molding) depends on a number of variables, such as cost of new sand, sand reclamation rates, improvements in casting surface quality (with reduced cleaning costs), and the possibility of reduced binder requirement for reclaimed sand. Variables specific to a particular foundry (such as proximity to sand quarry or disposal fees) also influence the choice of a bonding molding. Nonetheless, there are several reasons why other sand binders (Table 1) are a useful alternative to green sand in terms of:





Fig. 1

Major components of a sand mold. (a) Pattern assembly for cope and drag sections of a mold. (b) Cross section of sand mold assembly with core

For bonded-sand molds, selection of the process and type of binder depends on the size and number of cores or molds required, production rates, and equipment. Green (clay-bonded) sand molding and chemically bonded sand molding are considered to be the most basic and widely used moldmaking processes. The molding media for the two methods are prepared quite differently. Chemically bonded media are prepared by coating grains of sand with a binder that is later cured by some type of chemical reaction. In the case of green sand molding, the sand is coated with a clay-water mixture that forms an adhesive bond between the sand

Flexibility in production rate capability High strength for intricate molds and cores Tighter dimensional tolerances Reduced labor and improved casting quality

In general, binders can be classified as either inorganic or organic. Sand molds based on inorganic bonds include green (clay-bonded) sand molding, dry sand molding, skin-dried molds, and loam molding (see the article “Green Sand Molding” in this Volume). Inorganic bonding can also be obtained with thermosetting silicates (warm box, Table 1) and self-setting (no-bake) inorganic compounds such as sodium-silicates, aluminates (cement), and phosphates (Table 1). Organic-based binders for sand molding include no-bake (self-setting) binders, heatcured (thermosetting) resin binders, and cold box (vapor-cured) binders (Table 1). In the no-bake and cold box processes, the binder is cured at room temperature. Heat curing is done with the thermosetting resins for shell molding, hot box, warm box, and oven-bake (core-oil) processes. The term hot box is used because the method has application in coremaking by curing relatively thin sand sections with heated tooling (core boxes)—see the articles “Coremaking” and “Shell Molding and Shell Coremaking” in this Volume.

Foundry Sands As previously noted, silica sands are widely used for foundry molds, due to their wide availability and lower cost relative to other types of sands. However, more expensive sands (such as zircon, olivine, and chromite) have advantages in terms of lower thermal expansion coefficients than silica (Fig. 3). Zircon and chromite also have exceptional heat transfer/chilling

properties for heat extraction. Zircon and olivine extract heat faster than silica but not as fast as chromite. Sands are also mixed with additives to produce satisfactory castings. Sand quality control is of fundamental importance to a foundry’s profit and casting quality. The leading cause of casting defects, scrap, and rework is inadequate control of sand. Given the rates of molding and casting production with modern automated systems, foundries also cannot afford to fail to set up an in-house sand control program, because it will cost more in the long run if sand problems are encountered. A system of foundry sand is primarily made up of recycled, reclaimed, and reused sand. New or reclaimed sand additions are of critical importance. Their most important function is to reduce the concentration of contaminants in the system sand, such as dead clay, excessive fines, ash, oolitic material, metallics, or contamination from a core process. A second function is to maintain the total system volume. The actual production flow varies with the source of the sand, replacement cost, and the binder system. The rejuvenation of this sand is the principal function of the sand preparation system. Further details on sand processing are covered in the article “Green Sand Molding” in this Volume. This section addresses the basic types of sands and properties.

Types of Sand The production of sand for the foundry industry requires a series of mining and refining steps to yield pure, consistent sands. Natural sands contain enough clay that can be mixed with water and used for green sand molding. However, because of the demands of modern high-pressure molding machines and the necessity to exercise close control over every aspect of casting production, the mining of sands for foundry applications requires repeated washing and desliming operations to remove naturally occurring clays. Clay is then added separately if green sand molding is being done. Because the clay and sand are added separately, these sands were once termed synthetic sands— meaning that the clay and sand were formulated separately, rather than being natural sand with clay. The term synthetic sand is no longer commonly used. Silica Sands. The most commonly used aggregate is silica (SiO2) sand. The purest forms of silica are usually white or off-whitecolored sands. Lake sands, bank sands, and dune sands, which are usually tan or gold in color, are also silica sand, but they are less pure, containing some clay and other minerals. For green sand molding, washed silica sands are usually mixed with bentonite clays (and sometimes fireclay, which is kaolin-based clay) and carbons to produce molding sand. Foundry sands of silica are composed almost entirely in the form of quartz. Some impurities may be present, such as ilmenite (FeO-TiO2), magnetite (Fe3O4), or olivine, which is

530 / Expendable Mold Casting Processes with Permanent Patterns

Fig. 2

General flow chart for sand processing

composed of magnesium and ferrous orthosilicate [(Mg,Fe) SiO4]. Density, hardness, and thermal properties of silica are compared with other sands and refractory materials in Tables 2 and 3, Respectively. Thermal Expansion of Silica. As noted, silica undergoes more thermal expansion than other mold sands (Fig. 3). Quartz also undergoes a series of crystallographic transitions as it is heated. The first, at 573  C (1064  F), is accompanied by expansion (Fig. 4), which can cause mold spalling. Above 870  C (1598  F), quartz transforms to tridymite, and the sand may actually contract upon heating. At still higher temperatures (>1470  C, or 2678  F), tridymite transforms to cristobalite. Linear thermal expansion of quartz increases up to 17  106/K (perpendicular to c-axis) and 10  106/K (parallel to c-axis) at 840 K with increasing temperature. Above that temperature, it has a constant coefficient of thermal expansion up to 1000  C (1830  F). Zircon is a naturally occurring zirconium silicate (ZrO2 SiO2) mineral. It has good volume stability for extended periods of exposure to high temperatures. Its primary advantages are a very low thermal expansion, high thermal conductivity and bulk density (which gives it a chilling rate approximately four times that of quartz), and very low reactivity with molten metal. Zircon requires less binder than other sands because its grains are rounded. It is also compatible with all known binders—both organic and inorganic. The very high dimensional and thermal stabilities exhibited by zircon are the reasons it is widely used in steel



foundries and investment foundries making high-temperature alloy components. The high chilling rate is useful for controlling directional solidification by using it locally for selective chilling. Olivine is a naturally occurring magnesium silicate. Olivine minerals (so called because of their characteristic green color) are a solid solution of forsterite (Mg2SiO4) and fayalite (Fe2SiO4). Forsterite and fayalite in approximately equal proportions are the primary constituents of rock called dunite found in Washington state and North Carolina (Ref 4). Other locations of olivine sand include Scandanavia, Africa, and New Zealand (Fig. 5). The physical properties of olivine vary with their solid-solution composition of forsterite (Mg2SiO4) and fayalite (Fig. 6). Crystal structure of olivine is much more complex than other specialty sands. Inclusions of magnetite (Fe3O4), spinel (magnesium, iron, zinc, manganese, aluminates), apatite (Ca5(PO4 CO3)3), liquids, or gases are common. Movable bubbles are found at times. Olivine alters readily; in fact, it is much more commonly found altered than fresh. The most common alteration products are hydrous magnesium silicates (serpentine). The alteration develops in scales and fibers, and finally the entire olivine crystal can be transformed. Therefore, the composition must be specified to control the reproducibility of the sand mixture. Care must be taken to calcine the olivine sand before use to decompose the serpentine content, which contains water (Ref 5). Likewise, the refractory characteristics of olivine



are damaged if the system is allowed to be contaminated by the accumulation of silica from core, for example. The specific heat of olivine is similar to that of silica, but its thermal expansion is far less. Therefore, olivine is used for steel casting to control mold dimensions. Olivine is somewhat less durable than silica, and it is an angular sand. Unlike silica, olivine is chemically basic. It can be bonded with clays and most organic and inorganic binders, but it cannot be used with acid-curing binders. Chromite (FeCr2O4) is a black, angular sand. It is highly refractory and chemically unreactive, and it has good thermal stability and excellent chilling properties. However, it has twice the thermal expansion of zircon sand, and it often contains hydrous impurities that cause pinholing and gas defects in castings. It is necessary to specify the calcium oxide (CaO) and silicon dioxide (SiO2) limits in chromite sand to avoid sintering reactions and reactions with molten metal that cause burn-in (Ref 6). Aluminum Silicates include kyanite, sillimanite, and andalusite, and all have the same chemical composition (Al2SiO5). Kyanite and sillimanite have long been mined and used either raw or calcined to manufacture refractory mortars, cements, castibles, and plastic ramming mixes. Crushed kyanite and sillimanite have been tested and found unacceptable for foundry sand use because of their cleavage and irregular shapes produced by crushing. Minerals of the aluminum silicate group are similar in mineralogy, as shown in Table 4.

L

L L

L

M L

L

L

L

L

H

M

L

M

L

H H

M M

H

H

H

M M

M

H

M

M

H

M

M

H

M

M

M

H H

H H

H

M M M

H H H

Rate of gas evolution (a)

L

L

L

L

L

L

H

L

L ...

L

L

L

L

L

L

L

M

L L

L L L

Degree of thermal plasticity(a)

L

M

L

L

L

L

L

M

M ...

H

H

M

M

M

M

M

H

H M

M M M

Collapsibility speed(a)

P

G

P

P

P

P

P

P

G ...

G

G

F

G

G

G

G

G

G G

F F F

Ease of shakeout (b)

P

P

F

F

P

P

P

F

G G

F

F

G

G

G

G

G

G

F G

G G G

Moisture resistance (b)

30

45 30–60

30–60

... ... ...

...

...

30

...

H

5–60

...

...

H

H H

...

P

...

...

...

...

...

...

F

G G

2–20

...

...

... ...

2–45

...

...

G

2–45

...

...

...

...

1–45

...

H

1–45

...

P

...

...

F F

... ...

1–45

G G G

Resistance to overcure (b)

... ... ...

Strip time(c), min

...

L

H H

H H H

Curing speed (a) C

24

32

24

24

24

24

24

120

24 ...

24

27

32

27

27

27

27

205

230 230

260 260 260



F

75

90

75

75

75

75

75

250

75 ...

75

80

90

80

80

80

80

400

450 450

500 500 500



Optimum temperature

F

F

P

F

F

F

F

P

P ...

P

P

F

P

P

P

P

F

P P

F F F

Clay and fines resistance(b)

F

F

G

F

G

G

G

G

G G

G

G

F

G

G

G

G

F

G G

E E E

Flow ability (b)

H

H

L

L

H

H

M

M

M H

H

H

H

H

H

H

H

H

M M

N N N

Air drying rate(a)

N

N

L

N

N

N

N

M

M ...

M

M

H

M

M

M

M

M

M M

M M M

Pouring smoke (a)

NOTES: (a) H, high; M, medium; L, low; N, none. (b) E, excellent; G, good; F, fair; P, poor. (c) Rapid strip times require special mixing equipment. (d) Use minimum N2 levels for steel. (e) Iron oxide required for steel

Vapor-cured Inorganic Sodium silicate — CO2

Phosphate — oxide cured

Self-setting inorganics Silicates: Sodium silicate — ester cured Sodium silicate — FeSi cured Sodium silicate — 2CaO.SiO2 cured Aluminates: Cement — hydraulic cured Cement (fluid sand) — hydraulic cured Phosphates:

Thermosetting inorganic Silicate (warm box)

Vapor-cured resins (cold box) Phenolic urethane — amine Phenolic/peroxide – SO2 Solvent evaporation — air

Alkyd — organometallic Phenolic — pyridine

Phenolic — acid Urethanes:

High-nitrogen furan — acid Medium-nitrogen furan — acid Low-nitrogen furan — acid Phenolic no-bakes:

Self-setting resins Furan no-bakes:

Organic thermosetting resins Shell processes: Dry blend Warm coat (solvent) Hot coat Hot box processes: Furan Phenolic Oils: Core oil

Relative tensile strength(a)

Table 1 Comparison of mold and binder system characteristics

...

...

...

...

...

...

...

(d)

... ...

(e)

(e)

...

...

...

...

Steel

...

(d) Steel

... ... ...

Metals not recommended

P (sodium silicate)

E-G (phosphates)

G-F (cements)

P (silicates)

E-F (urethanes)

G-F (phenolic nobake)

G-F (furan no-bake)

F (shell)

Ease of Reclaimability(b)

Aggregates and Binders for Expendable Molds / 531

532 / Expendable Mold Casting Processes with Permanent Patterns

Fig. 3

Thermal expansion of silica compared with other mold aggregates

Table 2 Summary of mineralogy and physical properties of foundry sands Property

Quartz

Color

White to yellowishbrown Hexagonal None 7 2.65 1.54–1.56 None

Crystal system Cleavage Mohs hardness Specific gravity Index of refraction Alteration products Melting temperature,  C ( F) Thermal conductivity (K), Btu/hr ft  F Specific heat (C) Btu/lb  F Density (r), lb/ft3 Thermal diffusivity (K/rC) Heat diffusivity (K/rC)1/2





1713 (3115) 0.52 0.28 93 0.02 3.68

Olivine

Chromite

Zircon

Green

Black

White to pale yellowish-brown

Orthorhombic (010) (100) 6.5–7.0 3.2–4.8 1.63–1.69 Hydrous magnesium silicates Variable 1205–1900 (2200–3450) ...

Cubic None 5.5 4.4–5.2 2.00–2.12 Hydrous iron oxide

Tetragonal None 7.5 4.68 1.92–2.02 Metamict zircon

Variable 1760–1980 (3200–3600) 0.63

2200 (3990) 0.57

0.23 150 0.018 4.67

0.198 176 0.0165 4.46

... ... ... ...

All three minerals are thermally stable up to temperature, where they break down to yield 88% mullite and 12% free silica (cristobalite). The mullite is stable to 1810  C (3290  F). The mullite imparts to the products their

desirable properties of high refractoriness, low thermal expansion and resultant resistance to thermal shock, intermediate thermal conductivity, high load-bearing ability even at high temperatures, and resistance to chemical erosion.

Aluminum silicates are blended with zircon sands. Aluminum silicate-zircon blended sands have lower expansion than any of the specialty foundry sands except zircon. The mixture is characterized by relatively coarse, uniform, rounded sand grains with no fines. The grains readily accept binders and can be used with standard coremaking procedures. The mixture has good resistance to shock, high permeability, good chilling action, and refractoriness. Mullite Manufactured from Aluminum Silicates. Another type of mold aggregate is the manufacturing of mullite ceramics from aluminum silicate (Al2SiO5) minerals. Mullite is the only chemically stable intermediate phase at atmospheric pressure in the SiO2-Al2O3 system. It is rare in nature, the most important place of occurrence being the island of Mull, U.K. As a manufactured ceramic, mullite is produced from the high-temperature breakdown of aluminum silicate (Al2SiO5), which occurs in three common forms: kyanite, sillimanite, and andalusite. These forms of aluminum silicate break down at high temperatures to form mullite and silica. Fine mullite particles are produced by calcining aluminum silicates. The grains are highly angular, but additional processing of the fine grains can produce rounded spherical pellets for the base aggregate of foundry molds. Depending on the sintering cycle, the silica may be present as cristobalite or as amorphous silica. The composition may also be varied as either an intermediate-density or low-density ceramic medium (Table 5). These materials have high refractoriness, low thermal expansion, and high resistance to thermal shock. They are widely used in precision investment foundries, often in combination with zircon. Additives. Foundry sands are mixed with sands to improve properties. If done properly, these additives can alleviate the deficiencies of silica, such as its thermal expansion. In addition, silica reacts with molten iron to form a slag-type compound, which can cause burn-in, or the formation of a rough layer of sand and metal that adheres to the casting surface. Additives include carbon additions (such as seacoal), cellulose, and cereal or starches. The additives help control expansion, enhance casting peel and surface finish, reduce moisture sensitivity, and, in some cases, create a reducing atmosphere. Iron molding sands typically use seacoal and seacoal supplements. Steel sands often use cereals and starches rather than seacoal (to prevent carbon pickup). Nonferrous sands usually use cellulose and starches. If low-expansion sands such as chromite, zircon, olivine, or ceramic aggregates are used, carbons are sometimes not necessary. Some nonferrous silica sand systems operate with no carbons and without expansion defects. These organic components of the system sand are normally measured by the percent combustible (loss on ignition) test, which indicates the total amount of carbons. This test is often

Aggregates and Binders for Expendable Molds / 533 Table 3

Density, hardness, and thermal properties of quartzite (silica) compared with other ceramic materials for refractories Maximum service temperature

Melting point Density, g/cm3

Material

Alumina (Al2O3) Graphite (C) Magnesia (MgO) Mullite (3Al2O3 2SiO2) Quartzite (SiO2) Silicon carbide (SiC) Zircon (ZrO2 SiO2) Zirconia (ZrO2)







C





F

C



F

At 100  C (212  F)

At 1000  C (1830  F)

At 1200  C (2200  F)

Coefficient of linear thermal expansion, from 25 to 800  C (77 to 1470  F), 106/ C

Thermal conductivity, W/m

K

Specific heat (mean) at 25 to 1000  C (77 to 1830  F), J/kg  C

Hardness, Mohs scale



3.96 2.2 3.60 2.8

2050 3600(a) 2850 1850

3720 6510(a) 5160 3360

1950 ... 2400 1800

3540 ... 4350 3270

30 180 38 6

6 63 7 4

... ... 2.5 4

8.0 2.2 13.5 5.0

1050 1600 1170 840

9 0.5–1.0 6 7.5–9

2.65 3.2

1400 2200– 2700(b) 2500 2700

2550 3990– 4890(b) 4530 4890

1090 1400– 1700(c) 1870 2400

1990 2550– 3090(c) ... 3400

... ...

... ...

2.1 17

8.6 4.5

1170 840

7 9

... 2.0

... 2.3

4 0.9

4.5 7.5

630 590

4.5–4.7 5.5–5.8

7.5 6.5

(a) Sublimes. (b) Decomposes. (c) Oxidizes

Fig. 4

carbonaceous materials can improve surface finish to castings. Best results are achieved with such materials as seacoal and pitch, which volatilize and deposit a pyrolytic (lustrous) carbon layer on sand at the casting surface (Ref 7). Cellulose is added to control sand expansion and to broaden the allowable water content range. It is usually added in the form of wood flour, ground cereal husks, or nut shells. Cellulose reduces hot compressive strength and provides good collapsibility, thus improving shakeout. At high temperatures, it forms soot (an amorphous form of carbon), which deposits at the mold/metal interface and resists wetting by metal or slags. It also improves the flowability of the sand during molding. Excessive amounts generate smoke and fumes and can cause gas defects. In addition, if present when the clay content drops too low, defects such as cuts, washes, and mold inclusions will occur in the castings. Cereals, Which include corn flour, dextrine, and other starches, are adhesive when wetted and therefore act as a binder. They stiffen the sand and improve its ability to draw deep pockets. However, use of cereals makes shakeout more difficult, and excessive quantities make the sand tough and can cause the sand to form balls in the muller. Because cereals are volatile, they can cause gas defects in castings if used improperly.

Thermal expansion of silica minerals. Source: Ref 3

Sand Properties

Fig. 5

Locations of major deposits for chromite, olivine, and zircon

supplemented with the volatiles test, since this indicates the amount of carbons that come off quickly at low temperatures. Carbonaceous Additions. Carbon can be added in the form of seacoal (finely ground bituminous coal), asphalt, gilsonite (a naturally occurring asphaltite), or proprietary petroleum products. Seacoal changes to coke at high temperatures, expanding three times as it does so;

Fig. 6

Phase diagram of olivine solid-solution series

this action fills voids at the mold/metal interface. Too much carbon in the mold gives smoke, fumes, and gas defects, and the use of asphalt products must be controlled closely because their overuse waterproofs the sand. Carbon is added to the mold to provide a reducing atmosphere and a gas film during pouring that protects against oxidation of the metal and reduces burn-in. The addition of

The composition, size, size distribution, purity, and shape of the sand are important to the success of the moldmaking operation. The selection of sand grain structure dictates the ultimate mold permeability and density, and both of these parameters are critical to the production of quality castings. In general, it is good practice to select sand with a fineness number no greater than necessary to produce good surface finish and prevent metal penetration. Coarser grains reduce the accumulation of fines during recycling. Modern high-density (high-pressure) molding machines also can use coarser sands without compromising surface finish.

534 / Expendable Mold Casting Processes with Permanent Patterns Table 4 Mineral and thermal properties of the aluminum silicate minerals Property

Color Crystal system Cleavage Index of refraction Mohs hardness Specific gravity Mullite inversion,  C ( F)

Table 6

Kranite

Sillimanite

Andalusite

Blue to gray to white Triclinic (100) (010) (001) 1.71–1.73 5–7 3.6–3.7 1200 (2190)

White to crayish-brown Orthorhombic (010) 1.65–1.68 6–7 3.2 1600 (2910)

White to reddish Orthorhombic (110)(100) 1.63–1.69 7.5 3.2 1400 (2550)

Screen scale sieves equivalent

USA series No.

Tyler screen scale sieves, openings per lineal inch

6 8(a) 12 16(a) 20 30 40 50 70 100 140 200 270

6 80(a) 10 14(a) 20 28 35 48 65 100 150 200 270

Sieve opening mm

mm

Sieve p opening, in., ffiffiffi ratio 2, or 1.414

Permissible variation in average opening, þmm 

Wire diameter, mm

3.35 2.36 1.70 1.18 0.850 0.600 0.425 0.300 0.212 0.150 0.106 0.075 0.053

3350 2360 1700 1180 850 600 425 300 212 150 106 75 53

... 0.0937 0.0661 0.0469 0.0331 0.0234 0.0165 0.0117 0.0083 0.0059 0.0041 0.0029 0.0021

0.11 0.08 0.06 0.045 0.035 0.025 0.019 0.014 0.010 0.008 0.006 0.005 0.004

1.23 1.00 0.810 0.650 0.510 0.390 0.290 0.215 0.152 0.110 0.076 0.053 0.037

Note: A fixed pffiffiffi ratio exists between the different sizes of the screen scale. This fixed ratio between the different sizes of the screen scale has been taken as 1:414ð 2Þ. For example, using the USA series equivalent No. 200 as the starting sieve, the width of each successive opening is exactly 1.414 times the opening in the previous sieve. The area or surface of each successive opening in the scale is double that of the next finer sieve or onehalf that of the next coarser sieve. That is, the widths of the successive openings have a constant ratio of 1.414, and the areas of the successive pffiffiffi openings have a constant ratio of 1:414ð 2Þ. This fixed ratio is very convenient; by skipping every other screen, a fixed ratio of width of 2 to 1 exists. (a) These sieves are not normally used for testing foundry sands. Source: Ref 8

Table 7 Typical calculation of American Foundry Society (AFS) grain fineness number Size of sample: 50 g; AFS clay content: 5.9 g, or 11.8%; sand grains: 44.1 g, or 88.2% Amount of 50 g sample retained on sieve USA sieve series No.

6 12 20 30 40 50 70 100 140 200 270 Pan Total

g

%

None None None None 0.20 0.65 1.20 2.25 8.55 11.05 10.90 9.30 44.10

0.0 0.0 0.0 0.0 0.4 1.3 2.4 4.5 17.1 22.1 21.8 18.6 88.2

Multiplier Product

3 5 10 20 30 40 50 70 100 140 200 300 ...

0 0 0 0 12 52 120 315 1,710 3,094 4,360 5,580 15,243

Table 8 Similarity in American Foundry Society (AFS) grain fineness number of two sand samples with different grain size distributions Percentage retained USA sieve No.

Sand A

Sand B

6 12 20 30 40 50 70 100 140 200 270 Pan Total AFS grain fineness No.

0.0 0.0 0.0 1.0 24.0 22.0 16.0 17.0 14.0 4.0 1.7 0.3 100.0 60.0

0.0 0.0 0.0 0.0 1.0 24.0 41.0 24.0 7.0 2.0 0.0 1.0 100.0 60.0

Source: Ref 8

Source: Ref 8

Most mold aggregates are mixtures of new sand and reclaimed sand. The rejuvenation of reclaimed sand is the principal function of the sand preparation system. Adding new sand also helps maintain fineness and permeability by diluting out fines that build up from the breakdown of angular or weak sand grains. Reclaimed sand includes not only reclaimed molding sand but also core sands. Sand returning to the system

from cores (core sand dilution) should be considered as part of the new-sand addition. It is also beneficial to choose the same base sand for molding and coremaking to maintain consistent grain size and distribution when mold and core sands mix during reclamation. Sand Fineness. Most foundries in the United States use the American Foundry Society (AFS) grain fineness number as a general indication of sand fineness. The AFS grain fineness number

Table 5 Properties of intermediate- and low-density manufactured mullite aggregate Intermediate density

Property

Chemical properties, % Al2O3 SiO2 TiO2 Fe2O3 Other

Low density

75 11 3 9 2

48 48 2 1 1

Thermal properties Expansion, % linear change Expansion coefficient, 1  106 in. in.  C Conductivity, W/cm  C Heat capacity, W/(s/g  C) Diffusivity, cm2/s

0.65 6.00

0.61 5.59

0.0066 1.142 0.0028

0.0068 1.180 0.0033

Mineralogical properties, % Mullite Corundum Beta cristobalite Quartz silica

52 48 0 0

75 13 12 0

 

of sand is approximately the number of openings per inch of a given sieve that would just pass the sample if its grains were of uniform size, that is, the weighted average of the sizes of grains in the sample. It is approximately proportional to the surface area per unit weight of sand, exclusive of clay. The AFS grain fineness number is determined by taking the percentage of sand retained on each of a series of standard screens, multiplying each by a multiplier, adding the total, and then dividing by the total percentage of sand retained on the sieves. Table 6 lists the series of sieves used to run the AFS standard sieve analysis. A typical calculation of the AFS fineness number, which includes the multiplier factor, is given in Table 7. It is important to understand that various grain distributions and grain shape classifications can result in similar grain fineness numbers. Table 8 provides a sample sieve analysis demonstrating that two sands assigned the same AFS grain fineness number can have very different grain size distributions. Grain shape determines the surface area of sand grains. As the sand surface area increases, the amount of bonding material must increase if the sand is to be properly bonded. Thus, a change in surface area, perhaps due to a change in sand shape or the percentage of core sand being reclaimed, will result in a corresponding change in the amount of bond required. Rounded grains have a low surface-area-tovolume ratio and are therefore preferred for making cores because they require the least amount of binder. However, when they are recycled into the molding sand system, their shape can be a disadvantage if the molding system normally uses a high percentage of clay and water to facilitate rapid, automatic molding. This is because rounded grains require less binder than the rest of the system sand. Angular sands have the greatest surface area (except for sands that fracture easily and produce a large percentage of small grains

Aggregates and Binders for Expendable Molds / 535 and fines) and therefore require more mulling, bond, and moisture. The angularity of a sand increases with use because the sand is broken down by thermal and mechanical shock. The subangular-to-round classification is most commonly used, and it affords a compromise if shape becomes a factor in the sand system. However, control of grain size distribution is more important than control of grain shape. The grain size distribution, which includes the base sand size distribution plus the distribution of broken grains and fines from both molding sand and core sands, controls both the surface area and the packing density or porosity of the mold. Mold Porosity and Permeability. When molten metal is introduced into a sand mold, gases and steam are generated as a result of the thermal decomposition of the binder and other additives, cores, and contaminants that are present. If the permeability of the mold is not sufficient to allow the escape of the generated gases, mold pressures will increase, impeding the flow of molten metal and causing misruns or cold shuts, or even causing the metal to be blown from the mold. Conversely, if the porosity of the mold is too great, metal may penetrate the sand grains and cause a burn-in defect. Therefore, it is necessary to control mold permeability by proper balancing of sand distribution for adequate mold porosity.

Fig. 7

Grain size distribution controls the permeability of the mold. The highest porosity will result from grains that are all approximately the same size. As the size distribution broadens, there are more grains that are small enough to fill the spaces between large grains. As grains break down through repeated recycling, there are more and more of the smaller grains, and the porosity of the mold decreases. The size of the voids is determined by the size, size distribution, shape, and packing pattern of the grains. Figure 7 illustrates two sizes of rounded sand grains. Figure 8 shows that the voids in a mold face are large for a coarse sand and small for a fine sand, although the total void area per cubic unit of volume is almost the same for both sands. However, these distribution criteria also govern the dimensional stability of the base sand. It is also important to point out that mechanical venting of the mold is much more effective in providing adequate venting of mold gases than relying solely on sand permeability. The fact that gas is generated within the mold cavity is not always a disadvantage. Gas at the moldmetal interface can help prevent metal penetration into the sand. This minimizes burned-on sand grains and resulting problems associated with cleaning and machining the casting. Thus, a balance between mold permeability and gas generation must be maintained. For

Two sizes of rounded sand grains. Original magnification: 35

example, if mold permeability is low because of the fineness of the base sand, the sand additives should be those conducive to the production of a low-volume gas. On the other hand, if permeability is high, it is advantageous to select materials that yield higher levels of gas. Sand quality control is of fundamental importance. Inadequate control of sand is the leading cause of casting defects, scrap, and rework. Various types of sand tests are described in the articles “Green Sand Molding,” “No-Bake Sand Molding,” and “Shell Molding and Shell Coremaking” in this Volume. A basic in-house sand control program, in the long run, is the most effective way for the prompt identification and prevention of sand problems. While testing a system sand is very important, of equal importance is knowledgeable and responsible review of the data and corrective actions when necessary. A good sand control program, with knowledgeable personnel, can help to reduce material usage, scrap, and rework costs since most molding difficulties and casting defects are sand related. Considering the sand laboratory an entry-level job and rotating the people out as soon as they become proficient is not in the best interest of the product quality. Routine review of all test results should include representation from all the various disciplines within the foundry. It should be recognized that to become proficient at sand testing and interpretation of the data requires time and training. It is to the foundry’s benefit to recognize this and to provide the training necessary and realize the value of the personnel responsible for sand control. Included in the control program should be routine monitoring of the dust collection system. Particular attention should be given to the screen distribution, methylene clay content test, and the total combustible level test. Deviations from normal operating levels indicate equipment malfunctions that require corrective action. Screen distribution is important. For system sands comprised of only base sand (plus newsand additions and core sand dilution), if the grain shape remains constant, changes in screen distribution will occur slowly, a major contributor being the condition of the dust collector system. On the other hand, in those systems that have an influx of various types of base sand from varying core processes and sand sources, changes in the base sand distribution will be more dramatic and will cause dimensional problems unless preventive measures are implemented. For these systems, screen distribution should be tested more frequently.

Clays for Green Sand Molding

Fig. 8

Sizes of pores in faces of molds made from coarse sand and from fine sand. Original magnification: 35

Green sand additives can be divided into two categories: clays and carbonaceous materials. The major purpose of the clays is to function as a bonding agent to hold together the sand

536 / Expendable Mold Casting Processes with Permanent Patterns grains during the casting process. The carbonaceous materials reduce penetration and burn-in, control expansion, improve dimensional stability of the mold (seacoal only), improve surface finish, and improve cleanability of the finished casting. Clays normally used in green sand molding are of two general types:
Montmorillonite, or bentonite clays. These

are subdivided into two general types: Western, or sodium, bentonite; and Southern, or calcium, bentonite. The two clays differ in their chemical composition as well as in their physical behavior within a system sand.
Kaolinite, or fireclay as it is normally called The most significant clays used in green sand operations are the bentonites. Western and Southern bentonites differ in the types of ions adsorbed to their surface. Western bentonite has primarily sodium ions adsorbed to the ion exchange sites. Southern bentonite has primarily calcium ions adsorbed to the ion exchange sites. Due to differences in exchangeable ions, soluble salts, and sometimes other compositional differences, the physical characteristics vary also. In general, Western bentonite requires more mulling energy to develop its green strength and has higher hot strength than the

Fig. 9

Structure of montmorillonite. Large closed circles are aluminum, magnesium, sodium, or calcium. Small closed circles are silicon. Large open circles are hydroxyls. Small open circles are oxygen.

Fig. 10

same amount of Southern bentonite. Southern bentonite, at the same concentration, develops green strength with less mulling energy and produces lower hot strength. The fireclays contribute little to green property development but contribute dramatically to dry and hot property development. Bentonites. The most common clays used in bonding green sand molds are bentonites, which are forms of montmorillonite or hydrated aluminum silicate. Montmorillonite is built up of alternating tetrahedra of silicon atoms surrounded by oxygen atoms, and aluminum atoms surrounded by oxygen atoms, as shown in Fig. 9. This is a layered structure, and it produces clay particles that are flat plates. Water is adsorbed on the surfaces of these plates, and this causes bentonite to expand in the presence of water and to contract when dried. As noted, there are two forms of bentonite: Western (sodium) and Southern (calcium). Both are used in foundry sands, but they have somewhat different properties. Bentonites are not electrically neutral and can therefore attract water molecules between the clay plates. Water is also adsorbed on the quartz surfaces. Thus, there is a network of water adsorbed on sand and clay particles that is set up throughout the molding sand. If the clay covers each sand grain entirely, then clay-water bridges form between grains. In the case in which the clay coverage is nonuniform, similar bridges are formed. Western Bentonite. In Western bentonite, some of the aluminum atoms are replaced by sodium atoms. This gives the clay a net negative charge, which increases its activity and its ability to adsorb water. Western bentonite imparts high green and dry strengths to molding sand, and it has advantages for use with ferrous alloys, as follows. First, Western bentonite develops a high degree of plasticity, toughness, and deformation, along with providing good lubricity when mulled thoroughly with water. Molding sand bonded with plasticized Western bentonite squeezed uniformly around a pattern produces excellent mold strengths.

Second, because of its ability to swell with water additions to as much as 13 times its original volume, Western bentonite is an excellent agent between the sand grains after compaction in the mold. It therefore plays an important role in reducing sand expansion defects. Finally, Western bentonite has a high degree of durability. This characteristic allows it to be reused many times in a system sand with the least amount of rebonding additions. In using Western bentonite, it is important to control the clay/water ratio. Failure to do so can result in stiff, tough, difficult-to-mold sand with poor shakeout characteristics. Although these conditions can be alleviated by adding other materials to the molding sand, control of the mixture is preferable. Southern Bentonite. In Southern bentonites, some of the aluminum atoms are replaced by calcium atoms. Again, this increases the ion exchange capability of the clay. Southern bentonite is a lower-swelling clay, and it differs from Western bentonite in the following ways:
It develops a higher green compressive

strength with less mulling time.


Its dry compressive strength is approxi-

mately 30 to 40% lower.


Its hot compressive strength is lower, which

improves shakeout characteristics.


A Southern bentonite bonded sand flows

more easily than Western bentonite and can be squeezed to higher densities with less pressure; it is therefore better for use with complex patterns containing crevices and pockets. Use of Southern bentonite also requires good control of the clay-water mixture. Southern bentonite requires less water than Western bentonite and is less durable. Clay Blends. In practice, it is common to blend Western and Southern bentonites together to optimize the sand properties for the type of casting, the molding equipment, and the metal being poured. Examples of the effect of mixing bentonites on various sand properties are shown in Fig. 10. At high temperatures, bentonites lose

Effect of blending sodium and calcium bentonites on molding sand properties. (a) Dry compression strength. (b) Hot compression strength at 900  C (1650  F). (c) Green compression strength

Aggregates and Binders for Expendable Molds / 537 their adsorbed water and therefore their capacity for bonding. The superior high-temperature properties of Western bentonite are due to the fact that it retains water to higher temperatures than Southern bentonite (Ref 9). However, if the sand mix is heated to more than 600  C (1110  F), water is driven out of the clay crystal structure. This loss is irreversible, and the clay must be discarded. Fireclay consists essentially of kaolinite, a hydrous aluminum silicate that is usually combined with bentonites in molding sand. It is highly refractory but has low plasticity. It improves the hot strength of the mold and allows the water content to be varied over greater ranges. Because of its high hot strength potential, it is used for large castings. It is also used to improve sieve analysis by creating fines whenever the system does not have an optimum wide sieve distribution of the base sand. However, because of its low durability, its use is generally limited. In addition, the need for fireclay can usually be eliminated through close control of sand mixes and materials. Clay-Water Bonds. As noted, bentonites are not electrically neutral. Thus, water molecules between the clay plates provide adhesive attraction within the network of water adsorbed on sand and clay particles. The clay-water bond can also be explained in terms of the specific surface area of the clay, the type and strength of the water bond at the clay surface, and the hydration envelopes of the adsorbed cations. Clay particles hold adsorbed cations on their surfaces. The bonding of cations on clay particles is weak, and ion exchange is possible in the presence of appropriate electrolytes. Therefore, clay particles and ions are surrounded by electric force fields that direct the water dipoles (the water is polarized at the clay surface) and bind the water network. The field strength decreases with increased distance from the surface of the clay, so that the dipoles closest to the clay surface are bonded most strongly. Beyond the distance at which the force field is effective, the water behaves as a liquid and has no bonding action. There is an ideal water content at which all of the water is polarized and active in the bonding process (because the water added to activate the clay bond is called temper water, this is known as the temper point). Above this water content, some of the water will exist as liquid water, which is not involved in bonding. Below this value, there is insufficient water to develop the bond fully. At the temper point, the green strength of the sand is at its maximum, and additions of water beyond this point decrease the strength of the sand/clay/water mixture. The effect of this can be seen quite clearly in Fig. 11. Controlling Clay Properties. All clays become plastic and develop adhesive qualities when mixed with the proper amounts of water. All clays also can be dried and then be made plastic again by the addition of water, provided the drying temperature is not too high.

Fig. 11

Variation of mold properties with water content. (a) Southern bentonite. (b) Western bentonite. (c) Kaolinite

However, if the temperature becomes too high, they cannot be replasticized with water, and the clay is termed “dead clay.” It is this third condition that dictates the durability of the clay in a system sand. Southern bentonites begin to convert to dead clay once temperature exposure is in the range of approximately 315 to 480  C (600 to 900  F). Western bentonites begin to convert to dead clay in the range of 480 to 705  C (900 to 1300  F), so they are more thermally durable (Fig. 12e). Sand systems with bentonites with higher thermal durability require lower new-clay additions. All clays, regardless of type, develop both adhesive and cohesive properties when mixed with water. The amount of adhesive or cohesive property depends on the amount of water added. When the water content is low, the cohesive properties are enhanced and the clays tend to cohere, or stick to themselves. When water content is high, the clays tend to adhere, or stick to the sand grains to be bonded.

In addition to having different bonding, hot strength, and durability characteristics, the various clays have other very distinctive behavior patterns as a result of their differing physical characteristics. The flowability of sands with Western bentonite is lower than that of sands with Southern bentonite because of the greater swelling tendency of the Western bentonite and its tendency for stickiness. Therefore, the proper formulation of clay materials for a green sand system must take into consideration the flowability requirements as well as the shakeout requirements of the sand. Western bentonite causes more difficult shakeout due to its high hot strength. The ratio of clay to water is of critical importance in optimizing the properties of clays. The ratio of shear strength to green compressive strength of a clay-water mixture increases as mulling is increased. In a sand and clay mixture, water is absorbed by the clay up to its maximum capacity.

538 / Expendable Mold Casting Processes with Permanent Patterns of clay to carbons is constant, and fewer bins are needed for additives. Also, there is less fire hazard from the carbons when preblended with the clays. Some flexibility, however, is lost as the clay and carbon levels cannot be changed independently when using a preblend.

Inorganic Binders In addition to clay, sand can be bonded with other types of inorganic compounds based on silicates, aluminates (cement), or phosphates. The curing (or setting) of inorganic binders is activated by chemical or thermal means. Inorganic binders are also used in the production of ceramic shell molds for investment casting. This section discusses the methods of sand bonding with inorganic compounds. Binders for ceramic shell molds are discussed in the section “Other Expendable Mold Media” in this article. Sodium-Silicate-Bonded Sand Molds. The sodium silicate process is an inorganic system made up of a silicate polymer. This system consists of liquid sodium silicate and CO2 gas. Silicate binders are odorless, nonflammable, and suitable for all types of work (high production to large molds). They are applicable to all types of aggregates and produce no noxious gases upon mixing/molding/coring. A minimum of volatile emissions occurs during pouring/ cooling/shakeout. Carbon dioxide is used to precipitate sodium from what is essentially a silicic acid containing large quantities of colloidal sodium. The reaction of this process is:

Fig. 12

Effect of several variables on the efficiency of clay used as a bonding agent in sand molds. (a) Relationship of shear strength, as measured by pressure required for extruding a continuous worm of clay through an orifice, to water content, for three clays. (b) Effect of type and quantity of clay on erosion of sand-clay mixtures. (c) Effect of temperature on shrinkage of various types of clays. (d) Effect of mold temperature during casting on fusion of clay binder. (e) Effect of temperature on loss of combined moisture in clays

Na2 O 2SiO2 þ CO2 Ð Na2 CO3 þ 2SiO2

Continued gassing gives: Na2 O 2SiO2 þ 2CO2 þ H2 O Ð 2Na2 HCO3 þ 2SiO2

As water content increases, green strength increases and reaches a maximum. The moisture or compactability level at maximum green strength is termed the temper point. Any additional water beyond the temper point causes green and shear strengths to drop, is carried as free water in the system sand, and does not contribute to bonding. However, because excess water may contribute to gas defects such as pinholes, it should be avoided. While the ratio of clay to water in a sand mixture controls the ultimate strength of the sand mixture, the origin of the clay has a significant contribution on the strength potential. Clays from different geographic regions, even though they may be classified as being the same, have different strength curves. However, many of these differences are minimized by modern techniques used in the mining of the clays. Once the type of clay is determined for the system sand based on desired properties, economic considerations must be evaluated, because the geographic location of the foundry

will, in part, dictate the type, or the combination, of clays used in the operation. Clays are often blended to provide special properties. The proper combinations of clays allow the formulation of a system sand conducive to the production of quality castings. Of utmost importance in controlling a green sand system is the selection and consistency of the raw materials introduced into the system sand. Acquisition of the basic raw materials should be from reputable sources only, that is, those that have ongoing quality-improvement programs, including the understanding and application of statistical process control techniques. This is critical for a successful control program. Inconsistency in the raw materials used in the system results in sand variations that no amount of attention or corrective action can overcome. Preblends are in common use in the U.S. foundry industry. These are preblended mixtures of clay and carbons introduced as a single addition. Because they are added as a single addition, system control is simplified, the ratio

This shows that continued gassing dehydrates the amorphous silica gel and increases the strength of the mold. Sodium silicate molds are widely used for large cores and castings where there is a premium on mold hardness and dimensional control. The bond breaks down easily at high temperatures and therefore facilitates shakeout. The silicate-bonded sand, after pouring and shakeout, can be reclaimed by mechanical means, and up to 60% of the reclaimed sand can be reused. Wet reclamation of silicate sand systems is also possible. The amount of silicate binder used for cores and molds varies from 3 to 6%, depending on the type of sand, grain fineness, additives in the silicate binder, and degree of sand contaminants. The type of metal poured and its temperature and the amount of erosion resistance the core or mold will have to withstand are additional factors. A clean, rounded sand grain of 55 fineness requires approximately 3 to 4% of a binder. As the sand fineness increases, the amount of binder that must be used to coat each

Aggregates and Binders for Expendable Molds / 539 grain increases. Thus, a sand of 120 to 140 fineness requires from 1.5 to 3.0% more binder than a sand of 55 fineness. Either continuous-type or batch-type mixers can be used with sodium silicates. Overmixing must be avoided. Mixtures normally have a good bench life, which becomes shorter when higher-ratio silicates are used. Hoppers of mixed sand should be covered with plastic sheets or damp sacking to prevent premature hardening (crusting). The initial tensile strength of cores gassed for 5 s with CO2 varies from 255 to 310 kPa (37 to 45 psi), depending on the binder used. When the core stands, strength increases to a maximum of approximately 670 to 1380 kPa (100 to 200 psi) in 24 h. This increase is due to some dehydration of unreacted silicates and continued gelling of the silicate. Under normal conditions, cores and molds do not deteriorate and can be stored. An exception may occur during periods of high humidity when sand mixtures are unable to achieve maximum strength by dehydration during storage after gassing. Strength deterioration in high-humidity conditions is even more prevalent when the binder formulation or the sand mixture contains organic additives such as sugars, starches, or carbohydrates. Under these conditions, a short heating cycle is necessary to obtain the desired strength. When a core is hardened by carefully controlled gassing, and further hardened by dehydration and polymerization during the subsequent 24 h, good strengths are maintained over a long storage period. Washes may be necessary on cores and molds made by the silicate/CO2 process to prevent burned-on sand and metal penetration. Alcohol-based and solvent-type washes are normally used; isopropyl alcohol is preferred to methyl alcohol because it is less toxic and has a higher boiling point. Specially prepared graphite and zircon refractory pastes, which may be diluted with alcohol or solvent in the foundry, are commercially available. Washes on cores and molds promote peel, aid collapsibility, improve casting surface finish, and aid in resisting moisture absorption during periods of high humidity. Water-based washes (preferably with high solids content) can be used if care is taken to dry the core thoroughly immediately after coating. This procedure must be carried out with care due to the softening effect of the water on the silicate bond. Additives may be used in sodium-silicatebonded sand mixtures to improve shakeout or collapsibility. Sugars are commonly used for this purpose; they can be compounded with the silicate binder or added separately to a sand mixture. Small cores and molds are successfully made with core blowers or shooters combined with gassing stations operating on predetermined cycles and automatically controlled. For these applications, CO2 gas is injected through a hollow pattern or double-wall core box, or by means of a mandrel in a core, or through a hood covering the box.

Larger cores can be cured by means of lance pipes of approximately 5 mm ð 3=16 in:Þ in diameter that are open at the lower end. Using a rod, holes are made approximately 150 mm (6 in.) apart; the lance is inserted into each hole successively, and the gas is applied for 10 to 15 s at approximately 170 kPa (25 psig). The gas permeates and cures an area of sand having approximately a 75 mm (3 in.) radius around the hole. Large cores and molds may be gassed using specially designed, gasketed covers or hoods that fit over the flask or box. It is important to place vents properly to ensure the flow of gas through all parts of the mass. Core boxes and patterns may be made of wood, metal, or plastic and should be washed regularly to prevent sticking problems caused by a buildup of sodium silicate. An ideal release agent is formulated with wax or traditional silicone-based core room release agent. The silicate-bonded sand, after pouring and shakeout, may be reclaimed by mechanical means; up to 60% of the reclaimed sand can be reused. Wet reclamation of the silicate sand system is possible but requires a significant amount of water to scrub the sand. Methods of attrition reclamation combined with low-temperature thermal methods have shown some promise (see the discussion on sand reclamation in the article “Green Sand Molding” in this Volume). Phosphate-Bonded Sand Molds. This inorganic binder system, which consists of an acidic, water-soluble, liquid phosphate binder and a powdered metal oxide hardener, was designed to comply with air quality-control regulations. Because its components are inorganic, fumes, smoke, and odor are reduced at pouring and shakeout. The phosphate no-bake binder system has shakeout properties superior to those of the silicate/ester-catalyzed no-bake systems. The bonded sand can be reclaimed easily by either shot blast or dry-attrition reclamation units. Phosphoric acid bonds are used in both ferrous and nonferrous precision casting to produce monolithic molds. The reaction of this bond has the general form: ½MO þ H3 PO4 Ð MðHPO4 Þ þ H2 O

where M is an oxide frit or mixture of frits. The pH must be controlled carefully and kept acidic (Ref 6). The powdered metal oxide hardener is dry-blended with the sand, and the liquefied phosphoric acid is then incorporated. The coated sand is compacted into core or pattern boxes and allowed to harden chemically before removal. The hardener component is an odorless, freeflowing powder. It must be kept dry; in contact with water, it slowly undergoes a mildly alkaline hydration reaction that alters its chemical reactivity and physical state. Under normal ambient conditions, the material is not hydroscopic. Flow properties of the powdered hardener are good. Standard powder feeding equipment can be used to disperse the hardener into sand mixers.

Curing characteristics of the phosphate no-bake binder system depend on the ratio of hardener to binder. Varying the level of hardener can typically control strip times from 25 min to more than 1 h. Recommended phosphoric acid-base binder levels are from 2.5 to 3.0% for molds and 3.5 to 4.0% for core production. The hardener level should be kept within 18 to 33% of the binder weight for best results. Sand type also affects cure speed. Strongly alkaline sands such as olivine tend to accelerate the cure rate. Zircon forms an extremely strong and stable bond with phosphate binders, and, as a result, shakeout is more difficult. High-quality defect-free castings can be produced using the phosphate binder system for molds and cores with a variety of metals, including gray and ductile irons and various steels. Erosion resistance of both washed and unwashed molds is excellent. Veining resistance is good on unwashed surfaces and can be controlled with the proper coating selection. Aluminate (Cement)-Bonded Sand Molds. Portland cement produces sand molds with very good dry strength. When properly cured and dried, this type of mold has little gas-forming tendency. Cement molds have the advantage of relatively high chill rates but must be used carefully to avoid hot tearing. Most cement molds require drying at higher temperatures, but the method is similar to dry sand molding (see the article “Green Sand Molding” in this Volume). Sand reclamation is expensive.

Organic Binders Organic bonds are used in resin-bonded sand systems. Resin-bonded sands are widely used in metal casting, particularly for cores of all sizes and production volumes and for low-volume, high-accuracy molding. The sand is coated with two reactants that form a resin upon the application of heat or a chemical catalyst. The mixture is typically from 0.7 to 4.0 parts (usually 1 to 2 parts) of binder added to 100 parts of sand. The mixture is then compressed into the desired shape of the mold or core, and the binders are hardened, that is, cured, by chemical or thermal reactions to fixate the shapes. Resin binder processes can be classified into the following categories:
No-bake binder systems
Heat-cured binder systems
Cold box binder systems

In the no-bake and cold box processes, the binder is cured at room temperature; in the shell molding, hot box, warm box, and oven-bake processes, heat cures are applied. Selection of the process and type of binder depends on the size and number of cores or molds required, production rates, and equipment. No-bake, heat cured, and cold box binder systems are compared, respectively, in Table 9, Table 10, Table 11.

540 / Expendable Mold Casting Processes with Permanent Patterns Resin Types

Table 9 Comparison of properties of no-bake binder systems Process(a) Acid catalyzed Parameter

Furan

Ester cured

Phenolic

Relative tensile H M strength Rate of gas evolution L M Thermal plasticity L M Ease of shakeout G F Humidity resistance F F Strip time(b), min 3–45 2–45 Optimum (sand) 27 (80) 27 (80) temperature,   C ( F) Clay and fines P P resistance Flowability G F Pouring smoke M M Erosion resistance E E Metals not (d) ... recommended

Alkaline/ phenolic

Silicate

Oil urethane

Phenolic urethane

Polyolisocyanate

Alumina phosphate

L

M

H

M

M

M


Those composed of liquid polymeric binders

L M G E 3–60 27 (80)

L H P P 5–60 24 (75)

M L P G 2–180 32 (90)

H L G G 1–40 27 (80)

H L E G 2–20 27 (80)

L L G P 30–60 32 (90)

that cross link and set up in the presence of a catalyst (thus transforming from a liquid to a solid)
Those composed of two reactants that form a solid polymeric structure in the presence of a catalyst
Those that are heat activated

P

F

F

P

P

F

F L E ...

F N G ...

F H F ...

G M G (e)

G M P(c) (c)

F N G ...

(a) H, high; M, medium; L, low; N; none; E, excellent; G, good; F, fair; P, poor. (b) Rapid strip times required special mixing equipment. (c) Use with nonferrous metals. (d) Use minimum N2 levels for steel. (e) Iron oxide required for steel

Table 10

The types of resins for bonding of foundry sand fall into the three categories of process types:

Fluid-to-solid transition plastics are primarily furfuryl alcohol-based binders that are cured with acid catalysts. The polymers coat the sand when in the liquid form and are mixed with the liquid catalyst just before being placed in the core box. Alternatively, the catalyst can be delivered to the mix as a gas once the sand mix is in the core box. Reaction-based mold resins include:





Comparison of properties of the heat-cured binder systems

Phenolics (phenol/aldehyde) Oil/urethanes Phenolic/polymeric isocyanates Polyol/isocyanate systems

Process(a) Hot box Parameter

Relative tensile strength Rate of gas evolution Thermal plasticity Ease of shakeout Humidity resistance Cure speed Resistance to overcure Optimum core pattern temperature,  C ( F) Clay and fines resistance Flowability Bench life of mixed sand Pouring smoke Metals not recommended

Shell process

Furan

Phenolic

Warm box

Oven bake (core oil)

H M M F E H G 260 (500) F E E M N

H H L G F H F 260 (500) P G F M (b)

H H M F G M F 260 (500) P F F M Steel

H M L G G H F 175 (350) P G F M (b)

M M M G G L P 205 (400) F F G M (c)

(a) H, high; M, medium; L, low; N, none; E, excellent; G, good; F, fair; P, poor. (b) Use minimum N2 levels for steel. (c) Iron oxide required for steel

Table 11

Comparison of properties for cold box binder systems Process(a)

Parameter

Relative tensile strength Rate of gas evolution Thermal plasticity Ease of shakeout Moisture resistance Curing speed Resistance to overcure Optimum temperature,  C ( F) Clay and fines resistance Flowability Bench life of mixed sand Pouring smoke Erosion resistance Veining resistance Metals not recommended

Phenolic urethane

SO2 process (furan/SO2)

FRC process acrylic/epoxy(b)

Phenolic ester

Sodium silicate CO2

H H L G M H G 24 (75) P G F H G F (c)

M L N E H H G 32 (90) P G G L E F ...

H H L G M H G 24 (75) P E E M F G (d)

L L L G M M G 24 (75) P F F L E G ...

L M H P L M P 24 (75) F P F N G F ...

(a) H, high; M, medium; L, low; N, none; E, excellent; G, good; F, fair; P, poor. (b) FRC, Free radical cure. (c) Iron oxide required for steel. (d) Binder selection available for type of alloy

Curing catalysts include esters, amines, and acids, which can be delivered to the core mix either as liquids or gases. Heat-activated plastics are primarily thermoplastics or thermosetting resins such as novolacs, furans (furfuryl alcohols), phenols, and linseed oils. They are applied as dry powders to the sand, and the mix is heated, at which time the powders melt, flow over the sand, and then undergo a thermosetting reaction. Alternatively, they may consist of two liquids that react to form a solid in the presence of heat. Most binder systems are proprietary. The major ingredients are often mixed with nonreactive materials to control the reaction rate. The reactants are often dissolved in solvents to facilitate handling. Although various materials and schemes are used to form organic bonds in mold-and coremaking, the technology rests on only a few compounds. The presence of so many different systems allows casting producers to tailor the bonding system to the particular application. However, selection of the bonding system requires care. Care must also be taken in controlling process parameters because the systems are sensitive to variations in temperature and humidity. Consideration must also be given to environmental issues in the selection of the system because some organic systems emit noxious odors and fumes.

No-Bake Processes A no-bake process is based on the ambienttemperature cure of two or more binder components after they are combined on sand. Curing

Aggregates and Binders for Expendable Molds / 541 of the binder system begins immediately after all components are combined. For a period of time after initial mixing, the sand mix is workable and flowable to allow the filling of the core/mold pattern. After an additional time period, the sand mix cures to the point where it can be removed from the box. The time difference between filling and stripping of the box can range from a few minutes to several hours, depending on the binder system used, curing agent and amount, sand type, and sand temperature. In North America, phenolic-urethane no-bake molding is probably the most significant (Ref 1). Furan Acid-Catalyzed No-Bake. Furfuryl alcohol is the basic raw material used in furan no-bake binders. Furan binders can be modified with urea, formaldehyde, phenol, and a variety of other reactive or nonreactive additives. The great variety of furan binders available provides widely differing performance characteristics for use in various foundry applications. Water content may be as high as 15% and nitrogen content as high as 10% in resins modified with urea. In addition, zero-nitrogen and zerowater binders are available. The choice of a specific binder depends on the type of metal to be cast and the sand performance properties required. The amount of furan no-bake binders used ranges between 0.9 and 2.0% based on sand weight. Acid catalyst levels vary between 20 and 50% based on the weight of the binder. The speed of the curing reaction can be adjusted by changing the catalyst type or percentage, given that the sand type and temperature are constant. The furan no-bake curing mechanism is shown in Fig. 13. Furan no-bake binders provide high dimensional accuracy and a high degree of resistance to sand/metal interface casting defects, yet they decompose readily after the metal has solidified, providing excellent shakeout. Furan nobake binders also exhibit high tensile strength, along with the excellent hot strength needed for flaskless no-bake molding. They often run sand-to-metal ratios of as low as 2 to 1. Phenolic Acid-Catalyzed No-Bake. Phenolic resins are condensation reaction products of phenol(s) and aldehyde(s). Phenolic no-bake resins are those formed from phenol/formaldehyde where the phenol/formaldehyde molar ratio is less than 1. Again, as with furan no-bakes, these resins can be modified with reactive or nonreactive additives. These resins are clear to dark brown in appearance, and their viscosities range from

Fig. 13

medium to high. Sand mixes made with these resins have adequate flowability for the filling of mold patterns or core boxes. Resins of this type contain free phenol and free formaldehyde. Phenol and formaldehyde odors can be expected during sand-mixing operations. One disadvantage of acid-cured phenolic nobake resins is their relatively poor storage stability. Phenolic binders are usually not stored for more than 6 months. Phenolic resole resins contain numerous reactive methylol groups, and these are generally involved in autopolymerization reactions at ambient or slightly elevated temperatures. The storage period can be considerably longer during the winter months if the temperature of storage remains at 20  C (70  F) or lower. The viscosity advances as the binder ages. The catalyst needed for the phenolic no-bake resin is a strong sulfonic acid type. Phosphoric acids will not cure phenolic resins at the rate required for most no-bake foundry applications. The phenolic no-bake reaction mechanism is: Acid Cured Phenolic þ catalyst ! polymer þ Water resin

The catalyst initiates further condensation of the resin and advances the cross-linking reaction. The condensation reactions produce water, which results in a dilution effect on the acid catalyst that tends to slow the rate of cure. Because of this effect, it is necessary to use strong acid catalysts to ensure an acceptable rate of cure and good deep-set properties. Ester-Cured Alkaline Phenolic No-Bake. The ester-cured phenolic binder system is a two-part binder system consisting of a watersoluble alkaline phenolic resin and liquid ester co- reactants. The reaction mechanism is: Alkaline phenolic resin þ Ester co-reactant ! Suspected unstable intermediate ! Splits to form :

on the resin are used to coat washed and dried silica sand in most core and molding operations. Both the resin and co-reactant are water soluble, permitting easy cleanup. Physical strength of the cured sand is not as high as that of the acid-catalyzed and urethane no-bakes at comparable resin contents. However, with care in handling and transporting, good casting results can be obtained. The distinct advantages of the ester-cured phenolic no-bake systems are the reduction of veining defects in castings and excellent erosion resistance. Silicate/Ester-Catalyzed No-Bake. This nobake system consists of the sodium silicate binder and a liquid organic ester that functions as the hardening agent. High-ratio binders with SiO2:Na2O contents of 2.5 to 3.2:1 are employed for this process, and mixtures usually contain 3 to 4% binder. The ester hardeners are materials such as glycerol diacetate and triacetate, or ethylene glycol diacetate; they are low-viscosity liquids with either a sweet or acetic acidlike smell. The normal addition rate for the ester hardener is 10 to 15% based on the weight of sodium silicate and should be added to the sand prior to the addition of the silicate binder. The curing rate depends on the SiO2:Na2O ratio of the silicate binder and the composition of the ester hardener. Suppliers produce blends of ester hardeners giving work times that are controllable from several minutes to 1 h or longer. The hardening reaction, involving the formation of silica gel from the sodium silicate, is a cold process, and no heat or gas is produced. When added to a sand mixture containing the alkaline sodium silicate, the organic esters hydrolyze at a controlled rate, reacting with sodium silicate to form a silica gel that bonds the aggregate. A simplified version of this curing mechanism is: Sodium silicate Liquid ester Cured þ hardener ! polymer ðNa2 SiO3 Þ

Polymerized phenolic resin Alkaline salts and alcohol

A secondary reaction is thought to occur when the partially polymerized resin contacts heat during the pouring operation, yielding an extremely rigid structure. The viscosity of the ester-cured phenolic is similar to that of the acid-catalyzed phenolic no-bakes. It has a shelf life of 4 to 6 months at 20  C (70  F). Typically, 1.5 to 2.0% binder based on sand and 20 to 25% co-reactant based

The furan acid-catalyzed no-bake curing mechanism

Mixed sand must be used before hardening begins. Material that has exceeded the useful work time and feels dry or powdery should be discarded. The use of sand past the useful bench life will result in the production of weak, friable molds and cores that can result in penetration defects. Curing takes several hours to complete after stripping. Large molds may need 16 to 24 h. Strengths can be higher than those of CO2cured molds, and shelf life is better. Although shakeout is easier than with CO2-silicate systems, it is not as good as the other no-bake processes outlined in this article. Odor and gaseous emission levels are low during mixing, pouring, cooling, and shakeout but depend on the extent of organic additives in the mix. Casting defects such as veining and expansion are minimal. Burn-on and penetration are generally more severe than for the other no-bake systems and can be controlled by sand additives and a wash practice.

542 / Expendable Mold Casting Processes with Permanent Patterns Oil urethane no-bake resins (also known as oil-urethane, alkyd-urethane, alkyd-oil-urethane, or polyester-urethane) are three-component systems that consist of part A, an alkyd oil-type resin; part B, a liquid amine/metallic catalyst; and part C, a polymeric methyl di-isocyanate (MDI) (the urethane component). The three-part system uses the part B catalyst to achieve a predictable work/strip time. It can be made into a two-part system by preblending parts A and B when the amount of the part B catalyst added to the resin controls the work/ strip time. Part A can also be modified for better coating action, improved performance in temperature extremes, or better stripability. Part A is normally used at 1 to 2% of sand weight. The part B catalyst, whether added as a separate component or preblended with part A, is 2 to 10% by weight of part A. The part C isocyanate is always 18 to 20% by weight of part A. Although the oil urethane no-bake system is easy to use, the curing mechanisms are difficult to understand because there are two separate curing stages and two curing mechanisms. When the three components are mixed together on the sand, the polyisocyanate (part C) quickly begins to cross link with the alkyd oil resin (part A) at a rate controlled by the urethane catalyst component of part B, as shown in Fig. 14. This action produces a urethane coating on the sand with enough bonded sand strength to strip the pattern and handle the core or mold. The other stage of the curing reaction is similar to a paint- drying mechanism in which oxygen combines with the alkyd-oil resin component and nearly polymerizes it fully at room temperature to form a tough urethane bond. The metallic driers present in the part B catalyst accelerate the oxygenation or drying (slowly at room temperature or quickly at 150

to 205  C, or 300 to 400  F), but because the full cure is oxygen dependent, section size and shape, along with temperature, determine how long it takes to attain a complete cure. The alkyd-oil urethane mechanism is a twostage process involving: Alkyd þ NCO ðpolymeric isocyanateÞ þ Urethane ! Alkyd catalyst urethane ðpartial cross linkÞ Rigid Alkyd þ O2 þ Metallic driers ! cross-linked urethane

For maximum cure and ultimate casting properties, the mold or core should be heated to approximately 150  C (300  F) in a forced air oven for 1 h. The oil urethane no-bake system, with its unique two-stage cure, results in unmatched stripping characteristics and provides foundrymen with a good method of producing large cores and molds that require long work and strip times. The phenolic urethane no-bake (PUN) Binder system has three parts. Part I is a phenol formaldehyde resin dissolved in a special blend of solvents. Part II is a polymeric MDI-type isocyanate, again dissolved in solvents. Part III is an amine catalyst that, depending on strength and amount, regulates the speed of the reaction between parts I and II. The chemical reaction between part I and part II forms a urethane bond and no other by-products. For this reason and because air is not required for setting, the PUN system does not present the problems with through-cure or deep-set found in other no-bake systems. A simplified version of the curing mechanism for phenolic urethane no-bake systems is: Liquid Liquid phenolic resin þ polyisocyanate Part I Part II Liquid amine þ catalyst ¼ Solid resin þ heat Part III

Fig. 14

Effect of increasing oil urethane system (part B) (catalyst) on work time and strip time

Phenolic urethane no-bake binders are widely used for the production of both ferrous and nonferrous castings and can be successfully used for high-production operations or jobbing shops because of their chemical reaction time and ease of operation. Although many types of mixers can be used with PUN binders, zero-retention high-speed continuous mixers are the most widely used. Because the mixing takes place rapidly, the fast strip times (as fast as 30 s) of the PUN system can be used in practice. No mixed sand is retained in the mixer to harden after it is shut off. Further, the mixers can be coordinated with pattern movement, sand compaction, stripping operations, and mold or core finishing and storage to create a simple manual or fully automated no-bake loop.

Total binder level for the PUN system is 0.7 to 2% based on the weight of sand. It is common to offset the ratio of part I to part II at 55 to 45 or 60 to 40. The third-part catalyst level is based on the weight of part I. Depending on the catalyst type and strip time required, 0.4 to 8% catalyst (based on part I) is normally added. Compaction of the mixed sand can be accomplished by vibration, ramming, and tucking. The good flowability of PUN sand mixes allows good density with minimum effort. Because the PUN system cures very rapidly, the time required for the compacted pattern to reach rollover and strip must coincide with the setup or cure time of the sand mix. For certain ferrous applications (most commonly steels), the addition of 2 to 3% iron oxide to the sand mix can improve casting surface finish. This addition is also beneficial in reducing lustrous carbon defects by promoting a less reducing mold atmosphere. The PUN resin system contains approximately 3.0 to 3.8% N (which is approximately 0.04% based on sand). To reduce the chance of nitrogenrelated casting defects, the part I to part II ratio can be offset 60 to 40 in favor of the part I because substantially all the nitrogen is in part II. It has also been shown that as little as 0.25% red iron oxide is effective in suppressing the ferrous casting subsurface porosity associated with nitrogen in the melt and/or evolved from the PUN binder. The polyol-isocyanate system Was developed in the late 1970s for aluminum, magnesium, and other light-alloy foundries. Previously, the binder systems used in light-alloy foundries were the same as those used for the ferrous casting industry. The lower pouring temperatures associated with light alloys are not sufficient to decompose most no-bake binders, and removal of cores from castings is difficult. The polyol-isocyanate system was developed to provide improved shakeout. The nonferrous binders are similar to the PUN system described previously. Part I is a special polyol designed for good thermal breakdown dissolved in solvents. Part II is a polymeric MDI-type isocyanate, again dissolved in solvents. Part III is an amine catalyst that can be used to regulate cure speed. The chemical curing reaction of the polyolisocyanate system is as follows: Liquid Solid resin Liquid polyol þ ¼ þ polyisocyanate resin heat

In practice, polyol-isocyanate binders are used in much the same way as the PUN binders they evolved from. One difference is that the system does not require a catalyst. Several phenol formaldehyde (part I) resins are available that provide strip times from 8 min to over 1 h. For maximum control, however, an amine (Part III) catalyst can be used to regulate strip times to as fast as 3 min.

Aggregates and Binders for Expendable Molds / 543 For light-alloy applications, binder levels range from 0.7 to 1.5% based on sand. Part I and part II should be used at a 50 to 50 ratio for best results. Reactivity, strengths, and work-time-to-strip-time ratio are affected by the same variables as the PUN binders. Because of the fast thermal breakdown of the binder (Fig. 15), the polyol-urethane system is not recommended for ferrous castings.

Shell Process In the shell process, also referred to as the Croning process, the sand grains are coated with phenolic novolac resins and hexamethylenetetramine. In warm coating, dissolved or liquid resins are used, but in hot coating, solid novolac resins are used. The coated, dry, freeflowing sand is compressed and cured in a heated mold at 150 to 280  C (300 to 535  F) for 10 to 30 s. Sands prepared by warm coating cure fast and exhibit excellent properties. Hotcoated sands are generally more free flowing with less tendency toward caking or blocking in storage. See also the article “Shell Molding and Shell Coremaking” in this Volume. Novolac Shell-Molding Binders. Novolacs are thermoplastic, brittle, solid phenolic resins that do not cross link without the help of a cross-linking agent. Novolac compositions can, however, be cured to insoluble cross-linked products by using hexamethylenetetramine or a resole phenolic resin as a hardener. A simplified version of the novolac curing mechanism is: Novolac þ

HexamethHeat ylenetetramine !

Cured polymer þ ammonia

Phenol-formaldehyde novolac resins are the primary resins used for precoating shell process sand. These resins are available as powder, flakes, or granules or as solvent- or waterborne solutions. A lubricant, usually calcium stearate (4 to 6% of resin weight), is added during the resin production or the coating process to improve flowability and release properties. Hexamethylenetetramine, 10 to 17% based on resin weight, is used as a cross-linking agent. Producing Cores and Molds. The shellresin curing mechanism involves the transition

Fig. 15

Collapsibility of polyol-urethane compared to that of phenolic urethane

from one type of solid plastic to another—thermoplastic to thermosetting. This physical conversion must be completed during a brief period of the shell cycle before the heat (necessary to cure the resin) begins to decompose the binder. Pattern temperatures are typically 205 to 315  C (400 to 600  F). Operating within the ideal temperature range provides a good shell thickness, optimum resin flow, and minimal surface decomposition. Higher pattern temperatures of 275 to 315  C (525 to 600  F) are often successfully used to make small cores, because the shell cycle is short enough that little surface definition is lost by decomposition of the resin at the pattern interface during the relatively brief cure cycle generally needed. Various additives are used during the coating operation for specific purposes. They include iron oxide to prevent thermal cracking, to provide chill, and to minimize gas-related defects. The shell process has some advantages over other processes. The better blowability and superior flowability of the lubricant- containing shell sand permit intricate cores to be blown and offer excellent surface reproduction in shell molding. Because the bench life of the coated shell sand is indefinite, machines do not require the removal of process sand at the end of a production period. Storage life of cured cores or molds is excellent. A variety of sands are usable with the process, and nearly all metals and alloys have been successfully cast using shell sand for cores and molds. See the article “Shell molding and shell coremaking” in this Volume.

Hot Box and Warm Box Processes In the hot box and warm box processes, the binder-sand mixture is wet. A liquid thermosetting binder and a latent acid catalyst are mixed with dry sand and blown into a heated core box. The curing temperature depends on the process. Upon heating, the catalyst releases acid, which induces rapid cure; therefore, the core can be removed within 10 to 30 s. After the cores are removed from the pattern, the cure is complete as a result of the exothermic chemical reaction and the heat absorbed by the core. Although many hot box cores require postcuring in an oven to complete the cure, warm box cures require no postbake oven curing. Hot Box Binders. Conventional hot box resins are classified simply as furan or phenolic types. The furan types contain furfuryl alcohol, the phenolic types are based on phenol, and the furan-modified phenolic has both. All conventional hot box binders contain urea and formaldehyde. The furan hot box resin has a fast cure compared to that of the phenolic-type system and can therefore be ejected faster from the core box. Furan resin also provides superior shakeout and presents fewer disposal problems because of the lack of phenol. Typical resin content is 1.5 to 2.0%.

A simplified hot box reaction mechanism is: Solid resin Liquid resin þ Catalyst þ Heat ¼ þ water þ heat

Catalyst selection is based on the acid demand value and other chemical properties of the sand. Sand temperature changes of 11  C þ5 units in the acid (20  F) and/or variations of  demand value of the sand require a catalyst adjustment to maintain optimum performance. When a liquid catalyst is used, many operations have winter and summer grades that can be mixed together during seasonal transitions. Both chloride and nitrate catalysts are used. The chloride catalyst is the more reactive. Therefore, the chloride is the winter grade, and the nitrate the summer grade. Hot box resins have a limited shelf life and increase in viscosity with storage. If possible, containers should be stored out of the sun in a cool place and used on a first-in-first-out basis. Hot box catalysts have indefinite storage lives. Pattern temperature should not vary more than 28  C (50  F). Measurements should be made at the highest and lowest points across the pattern. Most production shops run hot box pattern temperatures of 230 to 290  C (450 to 550  F), but the ideal temperature is between 220 and 245  C (425 and 475  F). The most common mistake made with the hot box process is to run too high a pattern temperature, which causes poor core surfaces. This condition results in a friable core finish that is especially detrimental to thin-section cores. The color of the core surface shows how thoroughly the core is cured and is a good curing guideline. The surface should be slightly yellow or very light brown—not dark brown or black. Overall, the phenolic and furan hot box resins are extensively used in the automotive industry for producing intricate cores and molds that require good tensile strengths for low-cost gray iron castings. Warm Box Binders. The warm box resin is a minimum-water (<5%) furfuryl alcohol-type resin (furfuryl alcohol content: 70%) that is formulated for a nitrogen content of less than 2.5%. Because the resin/sand mix exhibits a high degree of rigid thermoset properties when fully cured, little or no poststrip distortion or sagging occurs. High hot and cold tensile properties are characteristic of warm box sands and generally permit a binder level between 0.8 to 1.8%, or 20% less than the conventional hot box resin content. Warm box catalysts are copper salts based primarily on aromatic sulfonic acids in an aqueous methanol solution. The catalyst amount used is 20 to 35%, based on resin weight. These catalysts are unusual in that they impart an excellent latent property (unreactive at room temperature) to the binder system but still form strong acids when heated. They promote a thorough curing action at temperatures at approximately 65  C (150  F) or higher.

544 / Expendable Mold Casting Processes with Permanent Patterns is normally referred to as a core-oil process. Several types of binders fall into this category. Linseed or vegetable oil binders account for most of the volume, but urea formaldehyde and resole phenolic resins are also used. The sand-coating procedure, sand formulation, coremaking techniques, and general foundry procedures are chemically different but are similar for all types of core-oil processes. Urea formaldehyde is noted for its excellent shakeout and is used in aluminum foundries and shops that operate dielectric curing ovens instead of the hot, forced air ovens. Because of its high nitrogen content and low hot strength, urea formaldehyde has found rather limited application outside of nonferrous shops. Additives. Core mixes generally contain 1% or less cereal, based on sand weight. The cereal is kept to a minimum because it generates gas. Normally, when more green strength is required for core stripping and/or handling, the cereal is mulled along with the water for a longer time. Small additions of Southern and/or Western bentonite (up to ½% based on sand weight) to cereal have also proved useful for developing green strength. An additional benefit is that bentonite evolves far less gas than cereal does. Water is added to the mix to activate the cereal and to create green strength. The amount must be controlled to develop optimum properties, as indicated in Table 12. Baked strength, green strength, and baking rate are influenced by moisture content. A 2% water addition gives optimum results. Operational Considerations. A number of points are essential for effective use of coreoil systems:

A simplified warm box curing reaction mechanism is: Furan Latent Heat resin þ acidðHþ Þ ! Cured furan binder

The binder components remain stable when mixed together in the proper ratios in sand until activated by heat, which decomposes the catalyst and releases the acid that causes the resin to polymerize. The pattern temperatures used range from 150 to 230  C (300 to 450  F). The optimum temperature of 190  C (375  F) is approximately 55  C (100  F) below the operating temperature for hot box binders. Low resin and catalyst viscosity combine to produce a flowable sand mix. Castings produced with warm box binders exhibit casting features that are very similar to those of a furan no-bake system. Good dimensional accuracy and excellent erosion resistance are observed with warm box binders.

Oven-Bake Processes/Core-Oil Binders Core-oil binder is used in combination with a water-activated cereal to produce a coated sand mix that has green strength. Green strength permits the wet sand mix to be blown or hand rammed into a simple vented, relatively lowcost core box at room temperature and to retain its shape when removed from the pattern. The uncured plasticlike cores are generally placed on a flat board or a dryer plate (a supporting structure to maintain the shape of the core) for oven drying. This process translates into a fast method of producing cores or molds. Except for the subsequent drying operations, the cores are, in effect, made almost as fast as they are blown. Types of Core Oil. A binder system that uses water and cereal to develop green strength and then cures or dries in a hot, forced air oven Table 12


Basic mix: A standard core-oil mix contains

approximately 1% cereal, 1 to 3.5% water, 1% binder, and 0.1% of a flowability/release agent.
Mixing order: As indicated in Table 13, the sequence of additions to the muller has a

Effects of moisture on core-oil sand mixes

1% cereal, 1% oil, 90 min bake at 200  C (400  F), AFS 62 GPN silica sand Percentage of moisture Property

Green strength, kPa (psi) Tensile strength, kPa (psi) Percentage baked

Table 13

0

0.5

1.0

1.5

2.0

3.0

4.0

5.0

1.4 (0.2) 1205 (175) 100

3.4 (0.5) 1450 (210) 100

6.2 (0.9) 1825 (265) 100

8.9 (1.3) 2275 (330) 100

9.6 (1.4) 2480 (360) 100

7.6 (1.1) 2345 (340) 97

6.9 (1.0) 2135 (310) 88

6.2 (0.9) 1725 (250) 78

Sequence of muller additions to core-oil sands

1% cereal, 1% oil, 1.5% water, 90 min bake at 200  C (400  F), silica sand Percentage of moisture

Property

All at once

1. Cereal 2. Oil 3. Water

Green strength, kPa (psi)

6.9 (1.0) 1930 (280)

2.0 (0.3) 1345 (195)

Tensile strength, kPa (psi)

1. Cereal 2. Water 3. Oil

1. Oil 2. Cereal 3. Water

1. Oil 2. Water 3. Cereal

1. Water 2. Oil 3. Cereal

1. Water 2. Cereal 3. Oil

10.3 (1.5) 2495 (362)

3.4 (0.5) 1725 (250)

4.1 (0.6) 1450 (210)

3.4 (0.5) 1380 (200)

7.6 (1.1) 2275 (330)

significant effect on core properties. The best order of addition to the sand clearly is (1) cereal and dry additives, (2) water, (3) oil, and (4) flowability/release agent.
Oven drying: Forced hot air is the principal means of curing core-oil binders. Heat causes the binder in the core-oil sand mix to cross link and provide strength. The proper combination of oven temperature, drying humidity, and time determines the final strength, dimensional stability, and surface finish of the core. It is important to note that mixing cores produced by other binder processes in the same drying oven with uncured core-oil cores usually weakens the non-core-oil cores because of the steam evolved from the water used in the core-oil process. Cold Box Processes The term cold box process implies the room-temperature cure of a binder-sand mixture accelerated by a vapor or gas catalyst that is passed through the sand. Several different cold box systems are currently used, employing different binders and gas or vapor catalysts—for example, triethylamine, dimethylethylamine and dimethylisopropylamine for phenolic urethane binders, sulfur dioxide for furan and acrylic epoxy binders, methyl formate for ester-cured alkaline phenolic binders, and carbon dioxide for silicate binders. Phenolic Urethane Cold Box (PUCB). The process uses a three-part binder system consisting of a phenolic resin (part I), a polymeric isocyanate (part II), and a tertiary amine vapor catalyst (part III). Sand is coated with the part I and part II components and compacted into a pattern at room temperature. The tertiary amine catalyst is introduced through vents in the pattern to harden the contained sand mix. The catalyst gas cycle is followed by an air purge cycle that forces the amine gas through the sand mass and removes residual amine from the hardened core. It is recommended that the exhaust from the core box be scrubbed chemically to remove the amine. The reactive component of the PUCB part I is phenolic resin. It is dissolved in solvents to yield a low-viscosity resin solution to facilitate coating the sand and blending it with the second component. Part II is a polymeric MDI-type isocyanate that again is blended with solvents to form a low-viscosity resin solution. The hydroxyl groups provided by the phenolic resin in part I react with the isocyanate groups in part II in the presence of the amine catalyst to form the solid urethane resin. It is this urethane polymer that bonds the sand grains together and gives the PUCB process its unique properties. A simplified curing reaction mechanism for the PUCB process is: Phenolic Polyiso- Vapor amine catalyst þ cyanate ! resin Urethane

The urethane reaction does not produce water or any other by- product. The system contains 3

Aggregates and Binders for Expendable Molds / 545 to 4% N, which is introduced from the part II polymeric isocyanate component. The organic resins and solvents in the PUCB system make it high in carbon content, which contributes to the formation of lustrous carbon and a reducing mold atmosphere during casting. The PUCB process can be used with all of the sands commonly used for coremaking in the foundry industry. Some consideration must be given, however, to the effects of sand temperature, chemistry, and moisture content on the resin performance of PUCB. The ideal sand temperature is 20 to 25  C (70 to 80  F). Lower temperature can reduce mixing efficiency and increase cure times. Higher sand temperatures reduce gassing cycles and the amount of catalyst required but shorten the usable life of the coated sand mix. A maximum sand moisture content of 0.2% is acceptable for the PUCB process at room temperature (20  C, or 70  F), but when the sand temperature rises to 30  C (90  F), the moisture content of the sand must be kept at less than 0.1% for the process to function properly. All types of popular sand-mixing equipment can be used with the PUCB process. A sand delivery system that causes the least amount of aeration is best. Typically, 1.5% total binder, consisting of equal parts of part I and part II components, is used on a washed and dried sand for ferrous castings. Many foundries prefer to offset the ratio and use slightly more part I for various technical reasons. For the casting of aluminum, magnesium, and other low-pouring-temperature alloys, binder levels of 1% and less are used to facilitate shakeout. The volatile liquid tertiary amines commonly used to cure the PUCB binders are triethylamine, dimethylethylamine, or dimethylisopropylamine. Various designs of generators vaporize and blend these amines with carrier gas and deliver them to the core machine. The best generators provide a consistently high concentration of amine to facilitate fast, predictable cure cycles. The exhaust from the core box is delivered to a chemical scrubber, and the amine is removed by reacting it with dilute sulfuric acid to form an amine sulfate salt. In larger foundry operations, concentrated liquor from the scrubber has been recycled through chemical processing to convert it back into usable amine, thus providing economic and ecological advantages. Certain sand additives can be used with the PUCB system to eliminate specific casting defects. Veining in ferrous and brass castings can be substantially reduced by the addition of 1 to 2% proprietary clay-sugar blends or 1 to 3% iron oxide. Black and red iron oxide additions of 2 to 3% are recommended for steel castings. Red iron oxide at levels as low as 0.25% can be effective in eliminating binder-induced subsurface pinhole porosity in alloys prone to those defects. SO2 Process (Furan/SO2). The sulfur dioxide (SO2) process can be described as a rapid-

curing, gas-activated, furan no-bake. Various furfuryl alcohol-based resins, blended with an adhesion promoter, are used to coat the sand in the range of 0.9 to 1.5%. Organic hydroperoxides at 30 to 50% by weight of the resin are added to the sand mixture and/or blended with the resin. Methanol-diluted silane (5 to 10%, based on resin weight) is used to increase strength, to improve shelf life, and to increase humidity resistance. Once the sand is in place, SO2 gas is introduced. It reacts with the peroxide and water in the furan resin, causing an in situ formation of a complex group of acids that cures the furan binder. The simplified curing reaction mechanism for the SO2 process is: Furan binder

þ

Hþ

Cross linking þ ðpolycondensationÞ

Heat ! dehydration

The curing reaction begins when the SO2 first contacts the peroxide and continues even after removal from the pattern. As the catalyzed furan/SO2 resin bonded sand ages, the water of condensation resulting from the furan polymerization continues to dissipate from the stillcuring sand until the reaction is complete. Upon curing, the sand changes from a light color to dark green or black. The recommended sand temperature for the SO2 process is 25 to 40  C (80 to 100  F). Lower temperatures reduce cure speed and may produce partially cured cores. Higher temperatures promote evaporation of solvents and reduce mixedsand bench life. Hot purging with air at 95  C (200  F) is required to achieve optimum cure. The SO2 process develops approximately 20 to 50% of its overall tensile strength upon ejection from the core box. It then builds strength rapidly to 85 to 95% of overall tensile strength after approximately 1 h. Bench life of the mixed sand is 12 to 24 h. In typical core blowing operations, relatively low blow pressures of 275 to 415 kPa (40 to 60 psi) are possible because excellent flowability is characteristic of the system. The free radical cure (FRC) process includes all acrylic and acrylic-epoxy functional binders. The binders are cured using an organic hydroperoxide and SO2. A variety of acrylic-epoxy binders have been developed for both ferrous and nonferrous applications, ranging from 100% acrylic binders to approximately 30 to 70 ratios of acrylic-epoxy blends. Sand performance and casting properties are influenced by the ratio of acrylic to epoxy functional components present in the binder system. Acrylic binders are based on acrylic functional components. When combined with small amounts of organic hydroperoxides, acrylic binders can be cured through the application of a wide range of concentrations of sulfur dioxide with inert gas carriers such as nitrogen (1 to 100% SO2).

Acrylic binders are primarily used in lightmetal applications because of their good shakeout properties; however, they have specific applications in ferrous metals where veining defects are troublesome with other binder systems. When acrylic binders are used in ferrous applications, a refractory coating and nonturbulent gating design are recommended to reduce the threat of erosion. Acrylic-epoxy binders are blends of acrylic and epoxy functional components. The acrylic-epoxy binder systems offer alternatives to existing cold box systems. The process uses a two-part liquid binder system consisting of (1) an unsaturated resin and/or monomer and (2) an organic hydroperoxide with an epoxy resin. The mixed sand, once exposed to SO2 gas, yields a cured polymer. A simplified version of this reaction mechanism is: Unsaturated polymer Epoxy resin þ and monomer ! SO2 organic hydroperoxide

Cured polymer

Varying the acrylic-epoxy composition influences such core- and moldmaking properties as rigidity, cure speed, SO2 consumption, humidity resistance, tensile and transverse strengths, and core release. In addition, casting properties such as veining resistance, shakeout, erosion resistance, and surface finish can be influenced by changing the acrylic-epoxy composition. The FRC process employs a variety of binder systems, including specific systems for ferrous and nonferrous casting applications. Binder levels range from 0.6 to 1.8%, depending on the type of sand used and the type of metal poured. Because the FRC process components do not react until SO2 is introduced to the pattern, the prepared sand mix has an extremely long bench life when compared to other cold box and hot box binder systems. This feature minimizes waste sand, provides for consistent flowability, and, most important, decreases the machine downtime because sand magazines, hoppers, and mixers do not have to be cleaned on a daily basis. Tooling and equipment requirements are similar to those of the other cold box processes. Gassing units used for the phenolic-urethane or the SO2 process are replaced or simply converted for use with the FRC system. Substitutions of caustic solutions for acid solutions are made in the wet scrubbers when the FRC process replaces the phenolic urethane system. Disposal of SO2 Gas. Scrubbing of the gas is accomplished with a wet scrubbing unit that uses a shower of water and a sodium hydroxide. A 5 to 10% solution of sodium hydroxide at a pH of 8 to 14 provides efficient neutralization of the SO2 and prevents the by-product (sodium sulfite) from precipitating out of solution. Higher sodium hydroxide concentrations will cause precipitation of the neutralized product. Stoichiometrically, 0.58 kg (1.27 lb) of sodium hydroxide is required to neutralize 0.45 kg (1.0 lb) of SO2.

546 / Expendable Mold Casting Processes with Permanent Patterns Casting dimensional accuracy and resistance to veining of ferrous and nonferrous metals are influenced by the thermal expansion of sand and by the hot distortion characteristics of the binder system. The antiveining feature of specific FRC binders has eliminated the need for specialty sands, sand additives, and special slurry applications in iron and steel castings. Further, the lack of nitrogen and the absence of water minimize the nitrogen and/or hydrogen pinholing porosity formation often associated with cold box systems containing water or nitrogen. The phenolic ester cold box (PECB) process Was introduced to the foundry industry in 1984. A two-part system, it consists of a water-soluble alkaline phenolic resole resin and a volatile ester vapor co-reactant. Sand is coated with the phenolic resin and blown into the core box. The liquid ester co-reactant is vaporized and injected as gas through the sand mix. The theoretical reaction is as follows: Alkaline Ester phenolic resin þ co-reactant Polymerized ! phenolic resin Reactive intermediate

Because the ester is consumed in the curing reaction, purging of excess ester vapor can be accomplished with the minimum volume of purge air. However, purge air helps to distribute the ester vapor throughout the sand mix. Methyl formate is the preferred ester for curing the phenolic resin because it is volatile and vaporized more easily than other esters. Methyl formate is readily available and relatively inexpensive. The alkaline-phenolic binder is a low-viscosity (0.1 to 0.2 Pa 7 s) liquid at typically 50 to 60% solids in aqueous solution. The system contains less than 0.1% N and produces a less reducing mold atmosphere than the cold box binder systems. The PECB system is affected by the physical characteristics of the sand, such as grain fineness, grain shape, and screen distribution. The best strengths are achieved with high-purity, washed and dried, round-grain silica sands. Because of the alkaline nature of the resin, however, it is not very sensitive to sand acid demand value. The fact that it is a water-soluble aqueous resin makes the system less sensitive to moisture, and it can tolerate up to 0.3% water in the sand. The PECB system can be mixed using conventional mullers and continuous mixers. Binder levels vary, depending on sand type, but 1.75 to 2.5% resin is typically used with washed and dried silica sands. For sufficient handling strength, somewhat higher binder levels are required than for other gas-cured organic binder systems. Methyl formate is volatilized in generating equipment designed especially for the PECB process. Because the methyl formate is not a

catalyst but a co-reactant of the system, the generating equipment must be capable of delivering a large volume of highly concentrated vapor to promote cure. Stoichiometrically, approximately 15% methyl formate based on phenolic resin is required to harden the binder/sand mixture. In practice, the methyl formate requirement ranges from 30 to 80% and is largely dependent on venting and negative exhaust on the tooling. Lower gassing pressures and longer curing times promote the most efficient use of the ester co-reactant. Despite the low handling strengths characteristic of the PECB system, castings made from all alloys show good surface finish. The erosion resistance and veining resistance of the PECB system are better than those of the phenolic urethane and phenolic hot box systems. A coating is recommended to control penetration defects. The PECB system does not usually require sand mix additives such as iron oxides or sugars to reduce veining and to control nitrogen defects. Because of the alkaline nature of the PECB sand, care must be taken so that it does not contaminate sand systems that are sensitive to pH change, especially if the reclaimed sand is used for other binder processes.

mixing should result in uniform coating of the graphite grains with pitch and the water-syrup mixture. Pitch or syrup in excess of that required for coating of the graphite particles may cause balling of the mold mixture. Pitch balls are aggregates of excess graphite and graphite fines containing a greater-than-average amount of syrup. Mold mixtures become more susceptible to balling as the percentage of reclaimed graphite fines in the mixture increases. Therefore, control of graphite particle size distribution is crucial. Air drying Influences both the green and baked strengths of the mold (Fig. 17). Molds are

Other Expendable Mold Media Besides bonded or unbonded sand, other media used for expendable molds include ceramic shells (for investment casting) and rammed graphite for casting reactive metals such as titanium or zirconium.

Fig. 16

Graphite mold mixture resembling green sand in appearance

Fig. 17

Effect of air drying time on (a) green strength and (b) baked strength of graphite molds

Rammed Graphite Rammed graphite is used to produce molds for the casting of reactive metals and alloys such as titanium and zirconium. Because such metals and alloys react vigorously with silica, conventional molding sands cannot be used. The mold medium has the appearance of green sand (Fig. 16). The mold mixture is pneumatically tamped (rammed) around a pattern, stripped from the pattern, air dried, baked, and fired. The mold medium is manufactured from a mixture consisting primarily of finely divided graphite having a closely controlled particle size and size distribution. To this mixture, nominally 10% water, 10% pitch, 7% syrup, and 3% starch are added. These ingredients are required for good coating of the graphite grains and for the development of optimal mold properties. Although a correct binder system can be developed for any graphite particle size distribution, mold permeability will vary with the type and amount of binder used. Mold shrinkage can be predicted and controlled if ramming pressure and binder composition are constant. Graphite mixing problems may also be related to the pitch and syrup ingredients. Good

Aggregates and Binders for Expendable Molds / 547 air dried on racks; after 24 h of air drying, the mold reaches approximately 13% of its maximum green strength. Maximum green strength is achieved after 94 h of air drying. Baking. After air drying, molds are baked at 175  C (350  F). Baking time depends on mold section thickness and geometry. All moisture must be removed during baking to prevent steam generation and potential mold cracking during firing. Firing removes all of the remaining volatile carbonaceous materials from the mold and develops final bonds. The strength and hardness of the mold increase with firing temperature (Fig. 18); best results are obtained at minimum firing temperatures of 870  C (1600  F). Firing is performed in a reducing atmosphere to prevent oxidation of the graphite. The fired mold is rigid and is similar to ceramic in texture and hardness. Storage. After firing, molds should be stored in a controlled environment (minimum temperature of 25  C, or 75  F, and relative humidity below 40%) to prevent moisture absorption. Moisture absorption is particularly rapid at high relative humidities and low temperatures.

Ceramic Shell Molds The three types of media used for most applications for ceramic shell molding include siliceous compounds (for example, silica itself), zircon, and various aluminum silicates (composed of mullite and usually free silica). See the article “Investment Casting” in this Volume for more information on expendable ceramic shell molds. Colloidal Silica Bonds. Colloidal silica is based on water. Colloidal silica particles are approximately 4 to 40 nm in diameter and form a sol in water. Their stability is determined by surface charge, pH, particle size, concentration, and electrolyte content. The silica is spherical and amorphous, and it contains a small amount of a radical, such as a hydroxide, to impart a negative charge to each particle so that they repel each other and do not settle out. When water evaporates from

these sols (as happens when the mold layers are dried), the silica particles are forced close enough together for hydroxyl groups to condense, splitting out the water and forming siloxane bonds between the aggregates. Colloidal silica bonds are used in investment casting. More refractory binders, such as colloidal alumina and colloidal zirconia, also have been made available for directional solidification processes, which subject ceramic shell molds to high temperatures for relatively long times. However, both are inferior to colloidal silica in room-temperature bonding properties. Ethyl Silicate. An alternate type of silica bond can be produced from hydrolyzed ethyl silicates. These are precipitation bonds such that: ½n SiðOHÞ4  Ð ½SiOðOHÞ2n þ nH2 O

The precipitated silicate bond is a gel that comes out of suspension by a change in binder ion concentration. Hydrolyzed ethyl silicate is manufactured by the reaction of silicon tetrachloride with ethyl alcohol. Two types of ethyl silicate are commonly available. Ethyl silicate 30, the first type, is a mixture of tetraethyl orthosilicate and polysilicates containing approximately 28% silica. Ethyl silicate 40, the second type, is a mixture of branched silicate polymers containing approximately 40% silica. Slurries formed from these ethyl silicates and mold aggregates, such as fused silica or zircon, are very sensitive to changes in pH. The slurries are normally kept at a pH of approximately 3. To gel them around a pattern, they are exposed to ammonia vapor, and their pH increases to 5, where they gel. The shells are then fired, and the ethyl alcohol evaporates or burns off, causing the silica binders to condense and form the silica bond. Ethyl silicate molds have an advantage over colloidal silica in that they do not require long drying times between dips and can be used for monolithic block molds. However, the mold strength of these molds is much less than that of colloidal silicate-bonded molds because of the fine craze cracking that occurs during firing.

On the other hand, this fine network of cracks is also responsible for the high dimensional reproducibility of castings made in block molds. The bonding reaction is fairly rapid and highly exothermic. Its progress can be monitored with temperature measurements. The reaction forms complex silicic acids that are capable of condensing to form coherent gels having good bonding properties. This process is promoted by drying (concentrating) or by adding an alkali such as ammonia. The result is similar in many ways to that obtained using colloidal silica. Prehydrolyzed grades of ethyl silicate having reasonable shelf life are also available, making it unnecessary for foundries to perform this chemical operation themselves. Ethyl silicate, with its alcohol base, dries much faster than colloidal silica. It is, however, more expensive and poses tire and environmental hazards. It is best used in applications involving rapid dipping cycles. Ethyl silicate slurries are readily gelled by exposure to an ammonia atmosphere; this permits dips to be applied very quickly, yet still provides proper drying because of the high volatility of the alcohol. Hybrid binders have been developed in an effort to combine the advantages of colloidal silica and ethyl silicate. The water required for hydrolysis of the ethyl silicate is supplied by using colloidal silica, providing an additional source of silica for improved strength. Less flammable solvents can be substituted for alcohol. Liquid sodium silicate Solutions are used where a very inexpensive binder is desired. Upon evaporation, these binders form a strong, glassy bond. Sodium silicate binders have poor refractoriness, which greatly limits their sphere of application. In addition, they are not resistant to the steam atmosphere of dewaxing autoclaves. Nevertheless, they have found some use, sometimes in conjunction with colloidal silica or ethyl silicate. Other Binder Materials. The operation of directional solidification processes, which subject the mold to high temperatures for relatively long times, along with the introduction of even more reactive superalloys, has led to interest in more refractory binders. Colloidal alumina and colloidal zirconia binders have been made available for this purpose. Both, however, are inferior to colloidal silica in room-temperature bonding properties. ACKNOWLEDGMENTS This article was revised and updated from:
J.J. Archibald and R.L. Smith, Resin Binder

Fig. 18

Effect of firing temperature on the (a) hardness and (b) strength of graphite molds

Processes, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 214–221
T.S. Piwonka, Aggregate Molding Materials, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 208–211
T.S. Piwonka, Bonds Formed in Molding Aggregates, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 212–213

548 / Expendable Mold Casting Processes with Permanent Patterns REFERENCES 1. W. D. Scott, private communication, Dec 2007 2. Reclamation of No-Bake Organic and Sodium Silicate Sands, Steel Founders’ Society of America, 1978. 3. J.A. Rassenfoss, Mold Materials for Ferrous Castings, 1977 Hoyt Lecture, Trans. AFS, Vol 85,

4. Casting Copper-Base Alloys, 2nd ed., American Foundry Society, 2007, p 31 5. E.L. Kotzin, Trans. AFS, Vol 90, 1982, p 103 6. W.F. Wales, Trans. AFS, Vol 81, 1973, p 249 7. I. Bindernagel, A. Kolorz, and K. Orths, Trans. AFS, Vol 83, 1975, p 557

8. Particle Size Distribution of Foundry Sand Mixtures, Mold and Core Test Handbook, American Foundrymen’s Society 9. F. Hofmann, Trans. AFS, Vol 93, 1985, p 377

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 549-566 DOI: 10.1361/asmhba0005243

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Green Sand Molding GREEN SAND MOLDING is one of many methods available to the foundryman for making a mold into which molten metal can be poured. Green sand molding and chemically bonded sand molding are considered to be the most basic and widely used moldmaking processes. The molding media for the two methods are prepared quite differently. Chemically bonded media are prepared by coating grains of sand with a binder that is later cured by some type of chemical reaction. Green sand media are prepared by coating the grains of sand with a clay-water mixture that binds the sand into a rigid mass by the application of force. Green sand molding is the least expensive, fastest, and most common of all the currently available molding methods. This article describes the sand system formulation, preparation, mulling, mold fabrication, and handling of green sand molds. Dry sand and loam molding are also briefly discussed. In addition, reclamation and recycling of sands from green molding are discussed. Reuse of sand is an important economic consideration when selecting a sand and binder system. Besides green sand molding, the other major sand molding methods are shell molding and no-bake molding. The shell and no-bake methods are discussed in subsequent articles in this Volume. The base aggregates and types of binders are discussed in the article “Aggregates and Binders for Expendable Molds” in this Volume. Some advantages of green sand molding include:  The molding material costs are low.  It lends itself readily to high-production

automatic molding.

 Wood or plastic patterns can be used in hand

molding, sand slinging, manual jolt, or squeeze machine molding.  High-pressure molding produces a well-compacted mold, which can produce better surface finishes and casting dimensional tolerances. The disadvantages include:  If the mold is not properly compacted, the

dimensional accuracy of the casting can be impaired, and the surface finish may be poor.  High-pressure green sand molding requires metal pattern equipment, which adds to the cost.

Sand Molding The basic steps involved in making a sandflask mold from a permanent pattern are shown in Fig. 1. The sequence begins with a mechanical drawing of the desired part. Patterns are then produced and mounted on pattern plates. Both the cope and drag patterns include core prints, which will produce cavities in the mold to accommodate extensions on either end of the core. These extensions fit solidly into the core prints to hold the core in place during pouring. The gate or passageway in the sand mold through which the molten metal will enter the mold cavity is usually mounted on the drag pattern plate. Locating pins on either end of the pattern plates allow for accurately setting the flask over the plate. Cores are produced separately by a variety of methods. Figure 1 shows the core boxes, which are rammed with a mixture of sand and core binder (see the article “Coremaking” in this Volume). If the cores must be assembled from separately made components, they are pasted together after curing. They are then ready to be inserted into the sand mold. The mold is made by placing a flask (an open metal box) over the cope pattern plate. Before molding can begin, risers are added to the pattern at predetermined points to control solidification and supply liquid metal to the casting to compensate for the shrinkage that takes place during cooling and solidification. Thus, any shrinkage voids form in the risers, and the casting will be sound. A hole or holes (called sprues) must also be formed in the cope section of the mold to provide a channel through which the molten metal can enter the gating system and the mold cavity. The cope half of the mold is produced by ramming sand into the flask, which is located on the pattern plate with pins. The flask full of sand is then drawn away from the pattern board, and the riser and sprue pieces removed. A flask is subsequently placed over the drag pattern plate, using the locating pins on the plate. Sand is rammed around the pattern, and a bottom board is placed on top of the flask full of sand. The pattern, flask, and bottom board are then rolled over 180 , and the pattern is withdrawn. The completed core is set into the core prints in the drag half of the mold, and the cope half

of the mold is set on top of the drag. Proper alignment of the mold cavity in the cope and drag portions of the mold is ensured by the use of closing pins, which align the two flasks. The flasks can be clamped together, or weights can be placed on top of the cope, to counteract the buoyant force of the liquid metal, which would otherwise tend to float the cope off the drag during pouring. Metal is then poured into the mold cavity through the sprue and allowed to solidify. The casting is shaken from the sand with the sprue, gating system, and risers attached (Fig. 1). Following shakeout, the flasks, bottom boards, and clamps are cycled back to the molding station, while the casting is moved through the production process. When the gates and risers are removed from the casting, they are returned to the furnace to be remelted. After cleaning, finishing, and heat treating, the castings are completed. The manufacture of sand castings is influenced by a number of variables that will have a great impact on the final dimensions of a cast part. Some of these variables are common to other casting processes, but many are unique to the molding method. Metal contraction, mold influences, and metal variables are described as follows. Metal Contraction. Most alloys undergo a large reduction in volume during the change from the molten state to the form of a solid casting at room temperature. In aluminum alloys, the transition may be as much as 10%, requiring careful planning on the part of the patternmaker and foundry engineer to ensure the production of a dimensionally accurate casting with the required integrity. Average values for the contraction of each metal are known and can be built into the pattern construction. The skill of the patternmaker enters into the equation when it becomes necessary to estimate where nonstandard shrinkage allowances are required because of mold restraint imposed by external gating systems, rigid cores, chills to influence directional solidification, and so on. The usual practice is to refine the gating and riser system until a sound casting meeting all of the quality requirements is produced. At this point, adjustments are made to the pattern equipment and locating surfaces to bring the final shape into the profile required by the casting designer. These adjustments are, by necessity, a

550 / Expendable Mold Casting Processes with Permanent Patterns

Fig. 1

Basic steps of sand molding in a flask with permanent pattern

trial-and-error process that can be greatly complicated by multiple pattern impressions and core box cavities. If the true potential of the process is to be realized, the adjustments are an absolute necessity. Mold Influences. The packed density of the mold aggregate is a major factor in maintaining uniform dimensions in the final casting. If variations in density, as reflected by mold hardness, are allowed to occur, the moltenmetal mass will force the mold surface to yield, and inconsistent dimensions will result. The magnitude of the problem is a function of the specific gravity and temperature of the alloy being poured, with steel castings showing the greatest variation and magnesium showing the least of the common alloys.

A second problem is to ensure that the mold assembly maintains the relative position of each component during closing, handling between molding and pouring stations, and the entry of molten metal. If the match between mold sections is not maintained, it will be reflected as a shift in the final casting. The only effective method of preventing this error is to pay constant attention to careful handling and maintenance practices in the foundry. It is suggested that the designer anticipate the problem, tolerance the part accordingly, and specify a parting line that will minimize the impact on the finished item. Cores may be considered a second mold component in the assembly, and they will introduce an additional element of variation. Again,

depending on the nature and condition of the core equipment, the impact on the total tolerance range of the casting will be a function of the core manufacturing process as well as the way the core is positioned and located in the mold assembly. If multiple cores are used in the mold and are prefabricated into subassemblies, the potential does exist for careful gaging and even dimensional corrections by selective stock removal. Metal Variables. As with all other elements of process control, those dealing with the alloy, its composition, its temperature, and the practices used to introduce the metal into the mold can have a large influence on the dimensions of the final casting. The tolerance potential of each major alloy group can vary significantly

Green Sand Molding / 551 for any one molding method. This is illustrated in Fig. 2, which shows the normal expected linear deviation from blueprint dimensions for steel and malleable castings made in green sand. Although the reported variation is large, it must again be remembered that it represents a data base from a family of molding operations and should not be construed as an average capability for the process or alloy family. The best source of information concerning tolerances is the prospective production foundry under consideration as a final source for the casting. Patterns. Regardless of molding method for sand casting, the accuracy of the shape will not be any better than the pattern equipment used to form the cavity. The better the casting process, the more faithful the dimensional reproduction of the mold cavity created by the pattern equipment and the more accurate the final shape produced. It is at this point in the manufacturing sequence that tolerance decisions made by the designer first come into play. If low production is expected and tooling costs are to be minimized, the assignment of wide tolerances will allow the use of loose (unmounted) wood patterns to fabricate the casting. Green sand floor molding, traditionally the process chosen for the low-volume manufacture of castings, will normally use an unmounted, loose wood pattern with very limited accuracy. Sand is packed around the pattern by hand, air rammer, or sand slingeres, and the pattern is withdrawn to form the cavity for the molten metal. Much of the tolerance potential of the process is lost by molder operations that fall into the category of art rather than science. Examples include variations in density

of the hand-rammed mold (which can yield and swell when molten metal is poured into the cavity) to the need to manually rap the pattern to loosen it in the sand before withdrawing it. All of these factors contribute to a great loss of tolerance control in the final shape. As the anticipated production run lengthens, tooling with a longer life is required, and more restrictive casting tolerances may be specified without greatly influencing casting prices or lead times. The typical types of patterns used may range from:  Unmounted, loose wood patterns through

mounted wood patterns

 Impressions that have been replicated in a

plaster molding operation and cast on an aluminum match plate  Impressions machined from solid and replicated in plastic materials before mounting on a metal plate  Metal impressions that have been duplicated by an accurate machining process and then mounted on metal plates  Multiple impressions machined from a solid blank of material with the operations and dimensions controlled by direct transfer of computer-aided design and manufacture (CAD-CAM) data In general, increased accuracy will follow through each of the aforementioned stages of pattern equipment options, and the costs progressively increase as well. The continuing advances in CAD-CAM capabilities promise to further improve tooling, molding, and casting in general. For example,

rapid tooling technology (such as three-dimensional, or 3-D, sand printing laser sintering) is a relatively new approach in low-volume production tooling. The method of 3-D sand printing produces sand molds from the digital pattern of a CAD system, so that a physical pattern is not needed for mold fabrication.

System Control and Formulation Sand, clay, water, and sometimes carbon additives are the primary components of a green sand mold. A realistic approach in the control of this complex system is to target the variables for which actual control can be implemented and realized. In simpler terms, one must control those system parameters that are directly affected by actions taken on the foundry floor. An effective sand control program should also include a testing program for these four basic ingredients of green sand molds (see the section “Sand System Testing” in this article). Next in importance is the condition of the sand processing equipment and sand system engineering. This includes the sand muller or mixer, sand cooling equipment, magnetic separation and/or screening, and dust collection equipment. The coating action and strength development is controlled by the condition of the mixing and/or the muller equipment. Failure to monitor the equipment and to maintain it adds appreciably to variations in the sand and a loss of casting quality. Third is the identification of the critical primary and secondary control parameters. The primary control parameters for the system sand are:  Determination of the organic components

measured by the total combustible and/or percent volatile tests  Determination of clay content measured by the methylene blue titration method  Percent compactability of the sand at the molding machine

Fig. 2

Normal expected linear deviation from blueprint dimensions of steel and malleable castings made from green molding sand. Source: Ref 1

For the primary control parameters, actual sampling of system sand should be accomplished as close to the point in time of use, preferably at the molding machine, as is practical without compromising worker safety. Provision should be made to obtain a test analysis of each revolution of the system sand. However, it is more important to react properly to the available test results than to be concerned with the quantity of the available test data. Secondary control parameters are as important as the primary factors but can be monitored over a period of time. It is necessary to realize that problems develop gradually in system sand. With this in mind, the use of a trend-line analysis technique as a tool to better understand the effects of subtle changes in a system sand can be of significant benefit. The specific secondary tests that should be run on a system sand include moisture, permeability, green

552 / Expendable Mold Casting Processes with Permanent Patterns compressive strength, American Foundrymen’s Society (AFS) clay, and AFS grain fineness and distribution (see the section “Sand System Testing” in this article). Included in the control program for a green sand system should be routine monitoring of the dust collection system. Particular attention should be given to the screen distribution, methylene clay content test, and the total combustible level test. Deviations from normal operating levels indicate equipment malfunctions that require corrective action. Sands. Sand, of course, is the base aggregate and can include any refractory grain such as silica sands (pure silica, lake sand, bank sand, or dune sand) or specialty sands such as chromite, zircon, olivine, carbon sand, or ceramic aggregates. The most common aggregate is silica due to cost and availability, but zircon, olivine, chromite, carbon, and ceramic aggregates are sometimes used for special applications, although these are more expensive. All of these aggregates have expansions lower than silica (see the article “Aggregates and Binders for Expendable Molds” in this Volume). Zircon and chromite have exceptional heat transfer/chilling properties for heat extraction. Zircon and olivine extract heat faster than silica but not as fast as chromite. Carbon and ceramic aggregates do not extract much more rapidly, but they do have lower expansion than silica. All of these aggregates are generally more expensive than silica, but they reduce expansion, burn-in/burn-on, and dimensional-related casting defects. The sand in a green sand molding system is primarily made up of recycled, reclaimed, and reused sand. Sand for the green sand molding system is recovered from the shakeout and cooled, cleaned, and screened. The rejuvenation of this sand is the principal function of the sand preparation system. New sand also must be added to the sand system. Clays. Clay (usually bentonites/montmorillonite clays and sometimes kaolin fireclay) is used as the bonding agent that holds the sand grains together, giving the sand strength and making the sand moldable. Calcined clay is sometimes used in the production of very large castings in dry sand molds because of its extremely low thermal expansion. With the increased use of high-pressure, high-density molding machines (to meet the demands for casting accuracy and integrity), the montmorillonite, or bentonite, clays are used primarily because of their increased durability when heated, higher bonding strength, and plasticity. Except for relatively few applications (such as thin-section castings), fireclay (kaolin) is unsuitable for use with high-pressure, highdensity molding machines. Western bentonites from Wyoming and South Dakota provide high hot strength, but Western bentonite causes more mold shrinkage and is less flowable. Southern bentonites, from Alabama, develop green and shear strength faster (with less mulling time) than Western

bentonites but do not have as much hot strength and are not as thermally durable. Steel sands typically use all Western bentonite to provide the hot strength needed for the high pouring temperature of steel. Iron sands typically use blends of Western and Southern bentonite. Nonferrous sands typically use Southern bentonites and may include a small amount of Western bentonite. The live (bonding) clay content of the system sand is normally measured by the methylene blue titration method. This method of determining clay content is based on the ability of a test sample of the system sand to absorb the methylene blue dye. In this test, the dye is added to the test sample by a buret. The end point of the titration is read by the technician as a “bursting halo” when a drop of the test material is placed onto a hardened piece of filter paper with a stirring rod. The “halo” is an indication that the maximum dye-absorbing capacity of the clay has been reached. The amount of dye required to reach the end point (measured in milliliters) is compared against a known standard mixture. Water must be added and mulled into the clay in order for the clay to function as a bonding agent. Ensuring regular muller maintenance to provide good mulling is of fundamental importance. Water also functions in the casting process as a coolant. The key test used to control water for iron and nonferrous sands is compactability. For steel sands, which are wetter, the moldability test is sometimes better than the compactability test because it is more sensitive on the wet side of temper. Iron and nonferrous molding sands are typically targeted for approximately 40% compactability for automated molding systems, and approximately 45% for hand-rammed operations. Steel molding sands are run on the wet side (approximately 55% compactability) because a wet sand produces higher hot strength. Although compactability is the key control characteristic for tempering the sand, moisture should also be monitored. A system sand that has a constant moisture at the target compactability has consistent amounts of the waterabsorbing components such as bentonite (clay), carbons, dead clay, ash, fines, and so on. Variable moisture contents at the target compactability indicate poor control of these basic moisture-absorbing sand components. Additives. Most system sands contain additives other than clays; monitoring those additions is critical to the production of quality castings. Carbon additives are either single or multiple component. The relative concentration of the carbon additive must remain constant for a laboratory-derived measurement to be of any value. For example, casting finish produced in a system sand with a total combustible value of 3.50% derived from a carbon additive based totally on seacoal can differ drastically from a similar system sand with an equivalent total combustible value derived from a multiple-

component carbon additive. In cases in which the benefits can be justified, the use of preblended sand additives (clay premixed with carbons) is recommended. Preblends ensure a consistent ratio in the addition of clay and carbons. Fewer additive bins are then required, and the fire hazard of carbons is less when mixed with clay. Flexibility in control is reduced, however, because the proportions are fixed. If a preblend is used, both clay and carbons together can be either increased or reduced, but it is not possible to reduce one and increase the other. This can be done by changing the preblend formulation, but it makes the implementation of changes less immediate. Despite the loss of flexibility, the benefits previously mentioned make preblend usage a common and growing method of additive addition.

Influence of Molding Equipment The type of molding equipment used is also critical in the selection of a green sand system. Green sand molding can be divided into three basic types:  Low-density

and low-pressure molding includes manually operated jolt-squeeze units. The green compressive strengths of the sand for these units is generally in the low-to-mid teens (given in pounds per square inch). Methylene blue clay contents typically range from 5 to 8%.  Medium-density units include automatic or semiautomatic units with rigid flasks that combine jolting action and hydraulic squeeze pressures. These usually require green strengths of approximately 105 to 175 kPa (15 to 25 psi), depending on the molding machine and patterns. Methylene blue clay contents typically range from 8 to 10%.  High-pressure or high-density units produce molds with hardness values in the high 90s. Many, but not all, of these units incorporate flaskless molding technology. Hydraulic and gas or air impact are used for sand compaction. Some fill the flask pneumatically prior to compaction. Green strengths for these systems generally run in the 175 to 240 kPa (25 to 35 psi) range. Methylene blue clay levels typically run from 8 to 12%. The third group, that is, the high-pressure or high-density molding methods, dominate the high-production and highly automated foundry today (2008). This type of green sand molding lends itself to both flask and flaskless system designs. Some of the major criteria that dictate whether a system is designed as a flask or a flaskless unit is the amount of metal to be poured in each mold and the total mold area that is required. It is not within the scope of this section to deal with all the factors that influence the selection of a molding system. However, all of the systems in this category can be used for

Green Sand Molding / 553 high-production repetitive work. Dimensional control of castings with these processes is quite good, and the economics of their operation have yet to be matched by any other. Unfortunately, these systems represent a significant capital investment. As automation and production rates have increased, the capability for the precise placement of the sand on the pattern has diminished. Thus, a critical characteristic for today’s (2008) high-production sands is the capability of uniformly covering the pattern surface before application of the compaction energy. Therefore, the sand flowability and consistency are of utmost importance in the formulation of the system sand.

Maintaining Sand System Quality New or reclaimed sand additions to the system sand are of critical importance. Their most important function is to reduce the concentration of contaminants in the system sand, such as dead clay, excessive fines, ash, oolitic material, metallics, or contamination from a core process. New sand helps to dilute the buildup of dead clay and ash that will lower the refractoriness of the sand if allowed to accumulate. Adding new sand also helps maintain fineness and permeability by diluting out fines that build up from the breakdown of angular or weak sand grains. New sand additions also maintain the total system volume to compensate for sand that has been lost from spillage or carried away in deep pockets of the casting. The introduction of additives to the system sand should be based on a programmed approach, which can greatly reduce system variations. Additives are consumed at identifiable rates. Core sand dilution can be determined as well as any raw virgin sand additions. Losses due to sand carryout, dust collection loss, and materials handling can be calculated or approximated. Clay control can be enhanced by close monitoring of clay additions as a material balance. Simply knowing what goes into the sand can result in a significant reduction in system sand variations. By monitoring sand system variations, a predictable quantity of bonding material can be determined and added to the system. This approach helps minimize system variations and provides a backup to clay level analysis in the laboratory. The material balance and methylene blue clay level should show the same trends. If they do not track together, the reasons should be investigated. All ingredients added to the system sand should be added gravimetrically. Volumetric or timed additions, such as auger feeding of dry additives, are generally not consistent enough to ensure adequate control and only contribute to overall consistency problems. This applies to the addition of new or reclaimed sand as well as to the return system sand for a batch system.

Sand returning to the system from cores (core sand dilution) also should be considered as part of the new sand addition. The size and shape of core sand, and the binders used for core sands, are frequently quite different from those used for molding sands, and this must be considered for maintaining the sand system. Also, over time, the sand in the system breaks down as a result of mechanical attrition and thermal cycling and therefore changes in size, size distribution, and shape. Daily variations in the product mix affect the ratio of core sand to molding sand being recycled. Typically, the sum of the new plus core sand input for steel, iron, and nonferrous sands is 230, 135, and 45 kg/ton (500, 300, and 100 lb/ton) of metal poured, respectively. If the levels are much lower, contaminants such as dead clay, oolitic material, fines, and metallics will build up. If the additions are considerably higher than these levels, the sand may become too clean and brittle, which lowers some strength properties and makes the sand brittle. Sand systems with excessive core sand dilution have large amounts of sand carryout that must be removed from the system since the system is a fixed volume. In this case, the sand remains in the system a shorter time and therefore does not receive the benefits of cumulative mulling upon recirculating. The clay is less developed and is used less efficiently in such a system. As a general guideline, the new sand plus core sand entering a green sand system should be 135 kg/ton (300 lb/ton) of metal poured for iron 45 kg/ton (100 lb/ton) of metal poured for nonferrous, and 225 kg/ton (500 lb/ton) of metal poured for steel. Clay and carbons must be added with the new sand. The clay and carbons added must be enough to replace the dead clay and carbons and raise the new sand and/or core sand to the target clay level. The base aggregate is monitored with the sieve analysis (AFS grain fineness and distribution) and the permeability test, which measures the venting characteristics of the sand that are secondary to mechanical venting to relieve gases from steam, carbons, and burning cores from the mold to prevent gas defects, cold shuts, and misruns. Sand System Testing. A complete sand control program should be designed to control the four basic ingredients of green sand molds. It should include testing for:         

Sand temperature Moisture Compactability Permeability Green compression Methylene blue clay content AFS or 25 mm clay content Combustibles (loss on ignition, or LOI) Sieve analysis

Detailed procedures for these tests can be found in the AFS Mold and Core Test Handbook.

These tests should be run as often as practical on a daily basis. Additional tests can be conducted when applicable according to standard procedures outlined in the AFS Mold and Core Test Handbook. The working bond, available bond, and mulling efficiency can be determined from moisture, compactability, and green strength and should be monitored and charted on an ongoing basis. Compactability, methylene blue clay, and LOI are the primary control parameters that measure temper and the quantity of live clay and carbons. The remaining parameters measured on the system sand (i.e., temperature, moisture, green strength, permeability, and AFS grain fineness and distribution) are secondary parameters. Significant changes in these secondary parameters can indicate equipment problems. For example, a drop in green strength may indicate a need for muller maintenance (although it is important to note that overtempering a green sand will also cause green strength to drop). Typically, sand temperature, moisture, compactability, permeability, and green compression are run hourly. Compactability is best controlled on the shop floor with a compactability controller, but the laboratory test should still be run hourly as a check on the controller. Methylene blue clay, AFS or 25 mm clay, the sieve analysis, and LOI should be run at least once per shift. In making adjustments to the sand, it is important to adjust only when there is a trend of at least three tests moving in the same direction, and to consider the cycle time of the sand. If these are not considered and adjustments are made, this can actually introduce more variation in control. While testing the system sand is very important, of equal importance is knowledgeable and responsible review of the data and corrective actions when necessary. Routine review of all test results should include representation from all the various disciplines within the foundry. It should be recognized that to become proficient at sand testing and interpretation of the data requires time and training. It is to the foundry’s benefit to recognize this and to provide the training necessary and realize the value of the personnel responsible for sand control. Considering the sand laboratory an entry-level job and rotating the people out as soon as they become proficient is not in the best interest of the quality. A good sand control program, with knowledgeable personnel, can help to reduce material usage, scrap, and rework costs since most molding difficulties and casting defects are sand related. Relying solely on outside laboratories to perform sand tests also does not provide timely enough control. Considering the fact that inadequate control of sand is the leading cause of casting defects, scrap, and rework, and considering the molding and casting rates on automated systems today (2008), foundries cannot afford to fail to set up at least a basic in-house sand control program, since it will cost more

554 / Expendable Mold Casting Processes with Permanent Patterns in the long run if sand problems are encountered. Sand control is of fundamental importance to a foundry’s profit and casting quality. Clay Tests. Live clay (good clay available for bonding) is normally measured by the methylene blue titration method. This method is based on the ability of a test sample of the system sand to absorb the methylene blue dye. The AFS clay test and the faster 25 mm clay test are two different test methods to measure the sum of the live clay, dead clay, inert fines, and silt. For sands with high carbons, it is important to note that a significant amount of carbons may also be removed in the claywash for determining AFS clay/25 mm clay. This causes an error in the estimation of dead clay as AFS or 25 mm clay minus the methylene blue (live) clay. Adjusted clay is a calculated parameter that can be used to provide a correction to subtract the carbons out of the AFS or 25 mm clay value. To calculate adjusted clay, an LOI afterwash test must be run on a clay-washed sample. This is usually done on the clay-washed sand after a sieve analysis is performed on the sample. Adjusted clay = AFS or 25 mm clay LOI + LOI after wash. Dead clay (including any dead clay particles and silt) can then be calculated as the adjusted clay minus the methylene blue clay. Compactibility is the only physical characteristic of a system sand that a molding machine realizes, and it is of paramount importance that it be controlled within a tight band on a constant basis. Compactability controllers are commonly implemented for this purpose. The desired compactability range will be specific to each foundry, but in general, 40% is typical for iron and nonferrous applications, where steel must be run higher. The range chosen should allow optimum strength of the clay as well as adequate moisture levels to minimize sand friability while satisfying the requirements of the molding equipment. If too high, however, gas defects, swells, poor flowability, and other problems can result. Compactability is an indirect measure of bulk density. If the sand compactability varies, mold heights or cake size on the molding machine will vary due to the fact that a fixed volume of sand is compacted. Screen distribution is important. For system sands comprised of only base sand (plus new sand additions and core sand dilution), if the grain shape remains constant, changes in screen distribution will occur slowly, a major contributor being the condition of the dust collector system. On the other hand, in those systems that have an influx of various types of base sand from varying core processes and sand sources, changes in the base sand distribution will be more dramatic and will cause dimensional problems unless preventive measures are implemented. For these systems, screen distribution should be tested more frequently. Changes in raw material quality or consistency of the clay can cause variable sand properties. The sieve analysis may indicate fines generation or overefficient fines removal,

segregation problems, changes in fineness due to core dilution, or changes in the fineness of the new sand. A reduction or buildup of dead clay may be the result of changes to the product mix that cause large changes in the sand-tometal ratio and the amount of burnout of the clay.

Molding Methods Green sand molds can be made in a number of ways. The optimum method depends on the type of casting, its size, and the required production. When only a few castings are required, it may be more economical to have a loose pattern made and to have the green sand mold made by hand. Hand ramming is the oldest and slowest method of making a mold. Unfortunately, it is becoming increasingly difficult to locate a foundry with hand molding skills. In most cases, the pattern will at least be mounted on some kind of board to facilitate fabrication of the molds. Modern foundries use more automated equipment to minimize the labor/skill elements in the preparation of a mold. Machines perform the many tasks necessary to prepare the sand mass to receive metal, but the sequence and method of performing individual steps in the molding operation can vary widely. All of these steps will influence the ultimate tolerance capability of the molding process, as will the compatibility of the part design to the actual molding equipment selected. Depending on the chosen method of filling the mold preform before compaction and the means available to achieve compaction, the tolerance capability of the equipment can vary widely for any one type of casting. A flat casting can be manufactured to very close dimensional tolerances on one type of molding machine. This same machine may not be able to achieve equal hardness in the mold mass for a part with deep pockets or changes in elevation and will thus sacrifice much of the tolerance capability gained.

half of a mold); this allows for easy removal of the finished mold. Tight flasks are designed as one-piece units that have no clamps. This type of flask remains with the mold during the pouring operation and, normally, until the shakeout operation. Regardless of the type used, the flask must become more rigid as molding pressure increases. Flexing or movement in the sidewall of the flask will adversely affect the accuracy of mold dimensions, flask-to-pattern alignment, and flask-to-flask alignment during closing of the mold halves. Flaskless Molds. Flaskless molding equipment (Fig. 3) has become increasingly popular, especially when molds of less than 160 kg (350 lb) are being considered. As the name implies, the flaskless molding machine has no flask. Rather, the flask is replaced with a box or molding chamber that is an integral part of the molding machine. Flaskless molding gives the foundryman additional versatility in the molding operation. With flaskless molding, a number of things are simplified. Until the advent of flaskless molding, most molds were filled with sand by gravity. With the tighter and more repeatable tolerances that result from the molding chamber being an integral part of the molding machine, other filling methods become more practical. In addition, the parting line of the mold need not be horizontal but can then be easily placed in the vertical orientation.

First-Generation Machines Jolt-type molding machines (Fig. 4) operate with the pattern mounted on a pattern plate (or plates), which in turn is fastened to the machine table. The table is fastened to the top of an operating air piston. A flask is placed on the pattern and is positively located by pins relative to the pattern. The flask is filled with sand, and the machine starts the jolt operation. This is usually accomplished by alternately applying

Mold Types There are two basic types of green sand molds: flask and flaskless. Flask Molds. A flask can be defined as the container that is positioned on the pattern (or platen in some cases) and into which the prepared sand is placed before the molding operation. Although there are flasks termed slip flasks, which are slid up the mold as the depth of compacted sand becomes deeper, the most common types of flasks are snap flasks and tight flasks. Snap flasks are usually square or rectangular. Diagonal corners are held together with camaction clamps. The clamps are moved to the open position after the mold is made, the cores have been set, and the cope (the upper half of a mold) has been placed on the drag (the lower

Fig. 3

Automatic flaskless high-production molding machine

Green Sand Molding / 555

Fig. 4

Fig. 5

Fig. 8

Primary components of a match plate (squeezer) pattern

Fig. 9

Drag half of a cope and drag pattern

Primary components of a jolt-type molding machine

Jolt-squeeze molding machine

and releasing air pressure to the jolt piston, which causes the flask, sand, and pattern to lift a few inches and then fall to a stop, producing a sharp jolt. This process is repeated a predetermined number of times, depending on sand conditions and pattern configuration. Because the sand is compacted by its own weight, mold density will be substantially less at the top of a tall pattern. The packing that results from the jolting action will normally be augmented by some type of supplemental compaction, usually hand or pneumatic ramming. When ramming is complete, push-off pins, bearing against the bottom edges of the flask, lift the flask and completed mold half off the pattern. Various mechanisms are used to lift the mold from the

Fig. 6

Jolt-squeeze molding machine with solid squeeze heads

Fig. 7

Jolt-squeeze molding machine with compensating heads

pattern and turn it over (in the case of the drag mold) or turn it for finishing operations (in the case of the cope mold). Jolt-squeeze molding machines (Fig. 5) operate in much the same manner as jolt-type molding machines. The main difference is that the supplemental compaction takes place as the result of a squeeze head being forced into the molding flask, thus compacting the loose sand at the top. The required pressure can be applied pneumatically or hydraulically. In many cases, the squeeze head will be one piece (Fig. 6) and may even have built-up areas to provide more compaction in deep areas that are hard to ram. In other cases, the squeeze head may be of the compensating type, which consists of a number of individual cylinders, each exerting a specified force on the rear mold face (Fig. 7). Some machines exert the same force on all areas of the mold, while other machines allow the operator to adjust squeezing pressure in zones. Jolt-squeeze machines are

available in many sizes and are suitable for many different purposes and production levels. They can be operated manually or automatically. The operator has the option of independently adjusting the number of jolts from zero to any number and adjusting the squeeze pressure from zero up to pressure that is considered excessive. Hand or pneumatic ramming is often combined with this process; supplemental ramming normally takes place after jolting but before squeezing. Sand slinger molding machines deliver the sand into the mold at high velocity from a rotating impeller. Molds made by this method can have very high strengths because a very dense mold can be made. Density is a function of sand velocity and the thickness through which the high-velocity sand must compact previously placed sand. Sand slingers may or may not be portable. Some ride on rails to the mold, while others have the molds brought to the slinger. Generally speaking, larger molds have the slinger brought to the mold, while smaller molds are brought to the molding station. Although slingers are useful in producing larger molds, it should be noted that the sand entry location and angle are critical to the production of good molds. Entry location is controlled by the operator, while entry angle and, to some extent, location are controlled by internal adjustment. It is extremely important that these adjustments be maintained in accordance with the appropriate maintenance manual. Error can and does lead to soft spots in the mold or to excessive pattern wear. A considerable amount of operator skill is required to achieve consistent results. A number of variations are possible in the aforementioned methods. Smaller patterns (resulting in smaller molds) can be constructed such that both the cope and drag impressions are mounted on opposite sides of the same plate. These squeezer or matchplate patterns (Fig. 8) are often used to produce molds with any combination of hand ramming, jolting, and squeezing, just as cope and drag patterns are (Fig. 9).

556 / Expendable Mold Casting Processes with Permanent Patterns Second-Generation Machines Foundry technology has progressed rapidly since the mid-1950s, and molding methods have been a large part of this progression. It was not until the early 1960s that high-pressure molding machines were developed. Depending on design, this new generation of molding machine would accept either match plate or cope and drag patterns and can be of the tight flask or flaskless configuration. Along with increased levels of foundry technology comes the demand for more accurate and higher-integrity castings. In any case, modern molding machines, metal technology, sand technology, and support equipment technology assist the foundryman in supplying these demands. Rap-jolt machines were among the first of the newer high-pressure molding machines. These machines are similar in many respects to jolt-squeeze machines. Rap-jolt machines have the option of jolting the mold as described previously and/or rapping the mold. Rapping is accomplished by rapidly striking the bottom of the platen on which the pattern is mounted with a weight. The force imparted to the platen/flask/ mold combination may not exceed 1 g, or separation between the flask and pattern will occur. Therefore, there is very little if any vertical movement of the pattern and flask. This method allows for the possibility of squeezing and rapping simultaneously. Some machines of this type allow the operator to jolt prior to the rapjolt operation. Depending on the individual molding machine, any one or any combination of the operations can be used to make the mold. The equipment described thus far has made use of some type of flask—either the snap or tight flask configuration. Match Plate Pattern Machines. Automatic molding machines that use match plates have been used in both tight flask and flaskless designs. Because the patterns do not have the strength to withstand the pressure exerted during compaction without flexing, both the cope and drag must be squeezed simultaneously. Some match plate machines (Fig. 10) fill both the cope and drag by gravity. This type of machine will close up the molding chambers to the pattern and then rotate the assembly so that the drag surface of the pattern is facing up. Sand is then dropped into the drag chamber, and a sealing plate (usually aluminum) is inserted. The molding chamber/pattern assembly is then rotated so that the cope pattern face is up, and the cope chamber is filled with sand. The mold is then compacted by squeezing, the molds are withdrawn from the pattern, and the pattern is removed. The open mold is then available for any finishing work or core setting. The mold is then closed and removed from the molding chambers. Other match plate machines fill cope and drag molding chambers simultaneously by blowing the sand into the cavity (Fig. 11). After the blowing operation, the mold is compacted

by a squeezing operation. After squeezing, the mold halves are withdrawn from the pattern and are available for any necessary finishing or core setting operations. Depending on the design of the machine, it may or may not be necessary to add to the machine cycle time to complete these operations. Match plate pattern machines are available in tight flask and flaskless designs. These machines normally use gravity fill of both cope and drag molds. The cope is filled in much the same manner as for a flaskless machine. The drag is filled by sealing the bottom of the drag flask prior to the gravity-fill operation. The drag flask, still sealed, is then closed to the pattern, and the mold is compacted by squeezing. The squeeze pressure is applied by individual cylinders, each covering a small area of the mold. Cope and Drag Machines. Automatic molding machines that use cope and drag patterns can also be used in tight flask and flaskless designs. Because the patterns normally do not have the strength to withstand the pressure exerted during compaction without flexing, the pattern plates are usually mounted against a platen or grid. In most cases, the cope and drag mold halves are filled and compacted with the pattern facing up. Except in the case of special finishing operations to the cope half of the mold, it is not necessary to rotate either the patterns or the cope half of the mold. However, it is necessary to turn the drag half of the mold over to allow for setting of cores, close up, and pouring. Most of the first automatic molding machines were automated versions of the rap-jolt machine mentioned earlier. The automation of rollover, transportation, and, in some cases, core setting has greatly increased the rate at which these machines produce molds. A relatively common method of compaction in tight flask machines uses pressure from one or several compensating squeeze heads, as shown in Fig. 7. This pressure is normally adjustable in order to optimize the molding conditions. The mold halves can be filled by gravity or the sand blown in using air pressure. Pressure Wave Method. More recent designs use pressure wave technology as the compaction method. These designs normally fill the flasks with sand by gravity. The top of the mold is sealed by a chamber. The chamber then emits a pressure wave, either by rapid release of air pressure or by an explosion of a combustible gas mixture (Fig. 12). As the pressure wave hits the back side of the mold, the sand grains are accelerated toward the pattern. The pattern immediately stops the downward movement of the sand grains, causing the kinetic energy of the mass to compact the sand. Molds made using this method are most dense at the pattern face and progressively less dense as distance increases from the pattern face. There is no need for additional compaction by the application of squeeze pressure. Horizontal flaskless molding machines are a relatively recent design. The patterns are

Fig. 10

Gravity-fill pressure squeeze molding machine using match plate patterns

Fig. 11

Blow-fill pressure squeeze molding machine using match plate patterns

Fig. 12

Pressure wave molding machine that compacts sand by the rapid release of air pressure or an explosive combustible gas mixture. Part (a) shows the mold filled by gravity prior to being compacted by the pressure wave at (b).

mounted in these machines on a hollow pattern carrier (Fig. 13). A grid supports the underside of the pattern to avoid flexing during compaction. The molding chambers are formed by the pattern, the four sides of the molding chamber, and a plate with a sand injection slot. Vacuum is used to evacuate the chamber formed by the

Green Sand Molding / 557 pattern carrier and the pattern plates. Vents in the pattern carrier and pattern plates allow the vacuum into the molding chambers, which causes sand to flow into the molding chambers. Upon completion of the filling sequence, the mold is compacted by squeeze pressure, and the molds are withdrawn from the pattern. The pattern carrier retracts as the drag half of the mold swings out for blowout and/or core setting while another mold is being produced. Because the molds are produced in the same attitude as they will be used, there is no need to turn either half of the mold over. Vertically parted molding machines have been commercially available since the mid 1960s. Like their horizontal counterparts, vertical machines have undergone a number of design changes as electronic technology has improved. Molds are made in these machines by closing the ends of a four-sided chamber with the patterns, which in turn are mounted on platens. The top chamber wall has a slot through which molding sand is blown. After the molding chamber is filled with sand, it is subsequently compacted by squeeze pressure (Fig. 14). Blow and squeeze pressure are both adjustable to optimize molding conditions. After compaction, one of the platens with its mounted pattern swings out of the way, allowing the other platen and pattern to push out

the newly made mold to join with previously made molds. At this position, the mold is available for core setting. Blow off is accomplished automatically. Some models are capable of porting vacuum to the back side of the pattern to assist in the filling of deep pockets. The vertically parted molding machines that are available are flaskless by nature. However, many deliver the mold to a device that will provide added physical support for the mold sides, thus increasing their flexibility.

Other Related Molding Methods Before the advent of the no-bake systems, a number of methods were particularly adapted to medium or large castings. These include dry molding, loam molding, and dried-skin molding. Skin drying is an intermediate between green sand and dry sand type of molding. The process is similar to dry sand molding in that the same type of sand mixtures and equipment is used. After coating the surface with a refractory wash, the molds are dried to a depth of 6 to 12 mm (0.25 to 0.5 in.). Skin-dried molds have some characteristics of green sand molds and some of dry sand molds, such as ease of shakeout and firm mold face, respectively. Skin-Dried Molding. In this type of molding, the skin of the mold cavity or parting line is heated to a temperature where the water in the molding sand is driven back into the mold. This movement of the water, in effect, leaves a dry sand layer on the mold surface. Special refractory coatings are brushed or sprayed onto the surface that will be in contact with the molten steel. The refractory used in this coating is usually finely crushed zircon, chromite, or mullite. When dried, this coating provides high-temperature resistance and enhances separation of the mold from the casting. Dry Sand Molding. Almost all dry sand molding has been replaced by the no-bake molding. The essential difference between dry sand and green sand molding is that the moisture in the mold sand is removed prior to pouring the metal. Dry sand molding is more applicable to medium and large castings than to small castings. The molds are stronger and more rigid than green sand molds. They can

therefore withstand more handling and resist the static pressure of molten metal, which may cause green sand molds to deform or swell. In addition, they may be exposed to the atmosphere for long periods of time without detrimental effect. Such exposure may be necessary for placing and fitting a large number of cores. Seacoal is the most common carbon material used in green sand; pitch is the most common in dry sand. Other materials in dry sand are gilsonite, cereal (corn flour), molasses, dextrine, glutrin, and resin. These additives thermoset at the baking/drying temperature (150 to 315  C, or 300 to 600  F) to produce high dry strength and rigid mold walls. The base sand is normally coarser than in green sand to facilitate natural venting and mold drying. The sand mixture for dry sand molds can be cured (hardened) by baking in an oven or using forced hot air or stoves placed in the mold. Dry sand molds are coated with refractory washes. Water and/or solvent carriers using graphite, silica, or zircon are the most frequently used. It is common to apply several coats. Dry sand molds were made by a variety of methods. Large sand compaction machines of the jolt, roll-over, and draw-type or jolt-onlytype compact the sand in conjunction with tucking, hand peening, and air ramming. Like slingees for green sand floor molding (Fig. 15), slingers for dry molding throw and compact the sand by means of centrifugal force. A variety of sizes are available. Some supplemental hand ramming may be required. Floor Molding. This molding method uses larger flasks, normally requiring the services of an overhead crane. The molds are made by a combination of mechanical equipment (slinger), hand peening, and pneumatic hand-operated rammers. Sand must be placed in the flasks in layers, and care must be taken by the molder to make certain that each layer knits and adheres to the other and is of uniform hardness. Pit Molding. This method is used for very large castings when flasks are impractical. Pits are normally constructed of concrete walls and sometimes floors to withstand great pressures during pouring. Because the drag part of a pit cannot be rolled over, the sand under the pattern must be rammed or bedded in, or the

Fig. 13

Vacuum-fill pressure squeeze machine that uses cope and drag patterns

Fig. 14

Blow-fill pressure squeeze molding machine making vertically parted molds. (a) Molding chamber filled with sand. (b) Sand compacted by squeeze pressure. (c) Finished sand mold pushed out of molding chamber

558 / Expendable Mold Casting Processes with Permanent Patterns

Fig. 16 Fig. 15

Primary components of a continuous muller

Fig. 17

Essential components of a conventional vertical wheel batch-type muller

Fig. 18

High-speed batch-type muller with horizontal wheels

Sand slinger for filling large floor molds

bottom must be constructed with dry sand cores. A bed of coke, cinders, or other means of venting the pit bottom must be provided. Once the pattern is in place, mold-making procedures are the same as those for floor molding. Loam molding is one of the oldest methods known. It requires the skill of an experienced craftsman and is seldom used today (2008) because of the lack of these craftsmen and the advent of the no-bake systems that have a number of advantages over loam molding, especially greatly reduced production time. Loam molding is particularly adapted to medium or large castings, usually those of circular shape. The equipment required is relatively simple. No pattern is used; however, a flask or pit may be necessary to support the finished mold. Replacing the pattern is one or more sweeps conforming to the shape of the mold/core. The sweep is attached to an upright, rotating spindle. A bed plate, a cover plate, and arbors for cores are also necessary items. Materials for making the molds consist of:  Soft bricks capable of absorbing water  Molding sand that is quite coarse (some-

times referred to as gravel) and contains a high percentage of clay  Sawdust, straw, hay, hair, or cloth clippings may be used. These materials contribute to mold strength and burn during casting, thus enhancing collapsibility (disintegration of mold/core). The molding sand is mixed with water to which molasses may be added until a thick loam slurry resembling concrete is obtained. This mixture is used along with the bricks to build up the general shape, using the same technique that a mason would use. Once the general shape is attained, the bricks are covered with a layer of loam, and the sweep is used to form the outside surface. A layer of loam from 6 to 25 mm (0.25 to 1 in.) thick remains. The mold/ core is then baked until thoroughly dry, after which a refractory coating is applied. During the drying process, numerous small cracks form on the surface of the low-moisture-content clay. These small cracks produce a surface that is very receptive to the refractory coating, which

may be brushed, swabbed, sprayed, or actually troweled on. After application, the coating is brushed with molasses and water. A camel hair brush may be used to impart an exceptionally smooth surface. The coating must be dried prior to assembly. Loam molds impart a superior peel to castings. Before the advent of spun cast pipe in metal molds, large- diameter pipe was made in pits (pit cast pipe). The cores for these molds were made from loam. Hay rope was wrapped around a metal mandrel, and the loam slurry was applied, smoothed, and dried, after which a refractory coating was applied. A typical screen analysis of a loam molding sand is as follows: Screen No.

6 12 20 30 40 50 70 100 140 200 270 Pan ( 270) AFS clay

Percent retained

4.5 12.5 25.3 15.0 10.4 7.7 3.0 2.4 1.4 0.7 0.6 1.2 15.3

Green Sand Media Preparation Prior to molding, sand, clay, water, and carbonaceous materials are combined in a mixing device. The length of time these ingredients are left in the mixing device is best determined by the type of device used and the desired sand properties. These devices are called either mullers or mixers. As with molding equipment, different types of mixing equipment can be used. Most foundries use either a continuous or a batch-type muller. Continuous Muller. In a continuous muller (Fig. 16), the sand is fed to the muller into one bowl, and it exits through a door in the other bowl. In most cases, sand is fed into the muller in a regulated, continuous stream, and discharge is controlled based on the power

draw of the muller motor. As power draw reaches a predetermined level, the discharge door opens for a short period, allowing some of the sand to leave the muller. On average, sand will pass through both bowls twice before it is ejected. It is possible, however, that a small percentage of sand will pass directly from the input to the exit in one pass. This type of muller is designed to produce large quantities of sand continuously. Batch-Type Muller. Although not a new design, the batch- type muller (Fig. 17) can produce high-quality molding sand. It is equipped with plows to move the sand mass under the large, weighted rolling wheels, which are vertically oriented. This kneading action provides the capability of consistent control but not short cycle times. The high-speed batch muller shown in Fig. 18 has been adapted to meet the requirements of both high-production molding lines and jobbing foundries. Sand is plowed from the floor of the muller up to the position where the rubber-tired wheels knead the sand against the rubber-lined sidewall. However, not all of the mulling action takes place as a result of the wheels; much of the mulling action takes place as a result of the amount of sand in the muller. For maximum mulling efficiency, the weight of sand charged into this type of muller should not be too small. The manufacturer should be able to provide the necessary information. This type of muller also offers

Green Sand Molding / 559 the possibility of cooling the aggregate. A blower can be installed that will force air into the bottom of the unit. As the air flows through the aggregate, it picks up moisture and is exhausted through the top of the muller. By adding sufficient amounts of water, the sand can be evaporatively cooled. Intensive Mixer. Another type of mixer that is increasing in use is the intensive mixer (Fig. 19). The intensive mixer uses a rotating bowl into which the sand is charged. Inside the rotating bowl are one or two driven rotors that are available with different designs, depending on requirements. To avoid internal buildup, the unit is equipped with a scraper that directs sand away from the sidewalls and back toward the center of the mixer. This type of mixer is available in either batch or continuous configurations. Mixing Variables. Regardless of the type of mixing/mulling equipment chosen, muller input and output must be closely controlled. Clay, combustible material, return sand, and new sand must all be added consistently and in amounts indicated by sand test results. Some sand systems are run on a volumetric basis, while others are run on a weight basis. The preferred method is to add return sand and additives by weight, thus affording closer control. Water additions are controlled in a number of different ways. Some equipment samples the sand going into the muller and bases water additions on testing those samples for heat and moisture content. Other water addition equipment samples the sand inside the muller and adds additional amounts of water as needed. Still other systems use a combination of these. Regardless of the type of equipment used, a minimum of 80% of the total added water should go into the muller immediately before the sand or at least with the sand to allow the maximum amount of time for cooling (if applicable) and/or clay activation and to provide the maximum control Ref 2. Most mulling equipment does not discharge prepared sand in its most flowable condition. Even if it did, prepared sand will normally be routed through at least one hopper (probably two or more) and will be transferred from one belt conveyor to another. Each time the sand is dropped into a hopper or onto another belt, some prepacking takes place. To provide a more flowable sand at the pattern face, operators of manual molding machines may, at the expense of time, riddle the first sand that enters the flask. In addition, again at the expense of time, an operator may even hand tuck the deeper pockets. Automatic molding equipment affords little or no opportunity for this special treatment. For these reasons, each molding station should be equipped with a good aerator to perform the final conditioning of the prepared sand. The aerator should be located on the last belt feeding the molding equipment in order to avoid any subsequent prepacking of the aggregate. It is suggested that prepared sand always be conveyed to the molding equipment by belt conveyors. Other methods of conveying sand

Fig. 19

Top (a) and side (b) views of an intensive mixer. The top view illustrates the loop pattern created by the mixing pan (1) rotating clockwise and the rotating mixing tools [movable mixing star (2) and stationary wall scraper] rotating either clockwise or counterclockwise while mounted eccentrically inside the mixing pan. The optional high-energy rotor (3) intensifies the mixing action. Other components include the discharge opening (4), rotary discharge table (5), discharge plow (6), and either one or two discharge chutes (7).

have been known to introduce unwanted moisture into the aggregate or to scrub clay from the coated sand grains. Prepared sand systems should always be designed with the receiving hoppers large enough to accept the total amount of sand from the feeding equipment. For example, the prepared sand hopper at the molding machine should always have enough capacity to receive all the sand from the muller (or another hopper). Thus, belts can be kept running, minimizing the drying time of the prepared sand on the belt. In no case should

sand that will be used for molding be allowed to dry on a delivery belt. At best, this drying will be intermittent and will lead to inconsistent results at the molding station. The requirements of dimensional stability and casting integrity do not tolerate poor or inconsistent control of molding sand. The old hand-squeeze method of testing is not capable of controlling sand within the necessary specifications. Manual operation of the muller can and does lead to incorrect assumptions and corrections. For example, a sand that feels as though

560 / Expendable Mold Casting Processes with Permanent Patterns it does not have enough body indicates a system that is beginning to run low on clay. The operator is then likely to add additional water. When this additional water is added, the sand may seem to have the correct body, but in fact, it will have an excessive amount of moisture. Defects resulting from oxidation will increase, green sand pockets will be harder to fill, mold wall movement will increase and cause shrinkage defects, casting surfaces will become rougher, and shakeout may become more difficult. Sand test samples should be taken at the point where the sand enters the molding machine. Samples taken from other points, such as at the muller discharge, will provide a false indication of production conditions. The properties tested often vary from foundry to foundry, as will the necessary frequency of testing. Those properties often include (but are not limited to) moisture content, active clay content, loss on ignition, grain size distribution, compression strength, permeability, and compactability. In some cases, green shear and wet tensile tests are also performed. All these tests provide the operating personnel with useful information. Although the primary function of this section is not sand testing, it is necessary to mention another sand test that is widely used in Europe and is gaining acceptance in the United States. It is recommended that green tensile tests be conducted along with the other tests. Green tensile strength does not follow the same conditions that apply to green compression and shear strength. Although not documented, tensile strength decreases more rapidly and earlier than the other strength properties. This, combined with the fact that many molding defects (stickers, drops, and so on) are the result of poor tensile strength, makes the test well worth running. The test has not found widespread use, probably because it is more difficult to conduct and it requires expensive additions to the sand test equipment. Fortunately, another test has been developed whose results have a numerical relationship to green tensile. This test is spalling (splitting) strength. The manufacturers of sand test equipment can supply the necessary information on this test.

Molding Problems The use of molding machines has been the subject of much discussion and, in some cases, disagreement. There are a number of pitfalls the operator can fall into. The first and most common pitfall involves compaction force, which is adjustable hydraulic pressure on most machines. Unfortunately, it is often mistakenly held that greater pressure means better molds. Although a certain amount of pressure is necessary to compact the mold to arrive at the necessary strength and stability, it can easily be overdone. As pressures on the mold face exceed 1050 kPa (150 psi) and especially as they near

1400 kPa (200 psi), a number of detrimental effects can be noticed. Springback. As mentioned earlier, the muller is used to coat the sand grains evenly with clay that has been made plastic by water absorption. Ideally, this coating of clay will completely cover each sand grain. The clay, expanded and made plastic by the water, is compacted by the molding machine with enough pressure to cause the clay that coats one grain to cohere to the clay that coats an adjacent grain, thus forming clay bridges. As compaction pressure is increased, these bridges gain more area and become stronger, at least up to a point. As with most plasticized materials, the clay does have a certain amount of memory. As compaction pressure is released, the clay will attempt to recover some of its former shape. The mold will grow very slightly as pressure is released, even when pressures are quite low. This springback cannot be avoided. As compaction pressures increase above 1050 kPa (150 psi) and approach 1400 kPa (200 psi), pattern release from the mold can become a serious problem. Not only does the mold swell slightly, causing pockets to become tight in the pattern, but the growth can also reach such a magnitude that the clay bridges are partially broken. When these bridges are fractured, tensile strength suffers drastically as greater compaction pressure is applied. The deeper pockets then become increasingly difficult to draw. Expansion Defects. Tensile strength is not the only problem with high compaction pressures. As compaction pressure increases, more bentonite is squeezed into the voids that exist between the sand grains. When the grainto-grain distance becomes too small, there is insufficient bentonite/water to contract as the sand grains expand during the heating by pouring. The result is expansion defects such as rattails, buckles, and scabs. Similar problems can exist with machines that use some form of the pressure wave principle. The problem of springback may not be quite as great, because the mold becomes progressively less dense as distance away from the pattern face increases. However, if the compaction energy becomes too high, the sand grains may actually be stripped off of the bentonite coating, especially on the pattern face, giving rise not only to expansion defects but also to casting surface finish problems. Parting Sprays. The tendency of the mold to stick to the pattern is often combated by spraying the pattern with some kind of parting spray (usually proprietary). This helps draw the pattern from the completed mold and greatly reduces the likelihood of a buildup of bentonite and fines on the pattern surfaces. Unfortunately, there is a tendency to use too much of the spray. Like compaction pressure, the right amount of parting spray is advantageous, but excessive amounts cause a number of problems, including stickers, rough casting finish,

penetration, blows, and sand inclusions. Generally, the pattern should be sprayed with a very light coating once every second or third mold. The operating foundryman should experiment with the optimum spray frequency and spray no more than is absolutely necessary. Pattern Heating. Although the use of a parting spray is effective, a better first step is to heat the patterns to 6 to 11  C (10 to 20  F) above the temperature of the molding sand. Cold patterns will cause the moisture in the molding sand to condense on the pattern face, which makes the sand stick to the pattern. Many molding machines do not have any provision for heating the pattern during the production run. Whether the molding machine has the heating capability or not, the pattern should be preheated to the proper temperature prior to the start of the production run to assist in a rapid start-up.

Mold Finishing After the mold has been compacted and the pattern removed, the mold is ready for the finishing operations. These operations usually consist of blowing out any loose sand, looking for any molding defects, drilling the sprue cup (if applicable), and setting any necessary cores. Once the finishing operations are complete, the cope can be accurately placed on the drag and the mold sent to the pouring station. Mold blowoff is one of the areas that can cause surface finish problems if not property controlled. Excessive amounts of air blown onto the mold can cause localized drying of the mold surface. On the other hand, if the air contains large amounts of moisture, the mold face can become excessively wet, giving rise to rough finish and burn-in penetration, just as excessive amounts of water in the molding sand or excessive amounts of parting spray will. Core setting is another part of the process that has undergone tremendous change in the last few years. The increased demand for more accurate castings has affected cores and core setting just as it has molding. With mold hardness in the 85+ range, it is no longer possible to press an oversized core into undersized core prints. From the viewpoint of casting accuracy as well as cleaning room costs, it is equally unacceptable to place undersized cores in an oversized core print. Modern core processes allow the possibility of making the same size core every time, just as the same size mold can be made every time. Thus, it becomes apparent that every cavity in the core box and every impression on the pattern must be as close as possible to the same dimensions. This obviously places greater demands on the supplier of patterns and core boxes as well as on the process itself. Mold closing is the next step in the operation. Some molding machines close the mold inside the machine, while others close the mold just outside the machine. Still others use a separate piece of equipment to perform this

Green Sand Molding / 561

Shakeout

Vibrating equipment is either a single vibrating mass or multiple vibrating masses, with either brute-force or natural-frequency drives. In a brute-force-driven system, vibrations are generated continuously and directly by the drive, which is usually electric-motor powered. Outboard rotating weights attached to a concentric shaft produce out-of-balance centrifugal forces required to vibrate the shakeout or conveyor. Natural- or resonant-frequency systems use springs to amplify a relatively small drive force to create vibrations of large, heavy masses. The force-amplifying spring system can distribute vibrations along great conveyor lengths. The electric-motor power sources are relatively small compared to those of brute-force systems. Shakeout devices are available in a number of different configurations. Many of the devices available are of the flat deck, vibratory type. They range from normal intensity, frequency, and travel to high-intensity units that use a very short travel but high frequency. Some shakeout units are rotary in nature and, depending on design, can also provide the added function of cooling the sand. Another type is the vibratory barrel. Vibrating conveyors use some periodic motion of a trough to produce conveying action. The motion of the pan can be linear, circular, elliptical, or helical. The stroke, frequency, and motion pattern are selected to meet specific requirements, from the highintensity near-vertical action of a shakeout conveyor to the shuffle action of a transfer-type feeder conveyor. Direction of flow can be horizontal, up an incline, down a decline, around right-angle turns, or in a complete circle. Castings can be transported without bouncing, and sand can be moved without dusting. Deck-type shakeouts (Fig. 20) are available in a number of different configurations for various applications. The first is the stationary type. Stationary refers to the casting and sprue, not the shakeout itself. This type of shakeout is normally used by bringing the mold to the shakeout device; therefore, its primary application is for larger molds and low-to-medium production lines. The deck-type shakeout is also available as a unit that provides the function of conveying the castings from one end of the unit to the other. As mentioned earlier, either type is available in a variety of strokes, intensities, and frequencies. Selection of the shakeout is a function of casting design. Heavier castings

After the castings have cooled sufficiently, they must be shaken out and separated from the sand mold. The energy needed to separate the casting from the mold may be supplied by a shaker pan conveyor, vibrating deck or grate, or any other type of vibratory, oscillating, or rotational mechanical action. For small castings, the vibratory shakeout action may be combined with a means of conveying the casting to the location at which the next operation is performed.

Fig. 21

function totally external to the molding machine. The halves should not be allowed to remain separated any longer than absolutely necessary. Separation of the mold halves for excessive periods of time will allow the mold faces to dry, and this can lead to cuts, washes, and a general degradation of the casting surface. Transportation of the mold to the closing station is critical. Jarring of the mold can cause green sand pockets to break away from the mold. In some cases, the pocket may not break away until the mold is poured; thus, the mold, the cores, and the metal are wasted. Rough transportation can easily cause heavy cores to shift off location, which can cause errors in casting dimensions, broken cores, and excessive metal around coreprints. Mold closing is just as critical as mold transportation, and the same basic rules apply. The mold must be treated as smoothly and gently as possible to avoid the same type of defects. The mold-guiding mechanism is as important at mold closing as it is during manufacture of the mold. Too often, the accuracy and smoothness with which the mold is closed is overlooked. Again, the results can be drops and core movement as well as mold shift, crush, casting dimension problems, and so on. After the mold is closed, mold transportation again becomes important. The finished mold must be carefully transported to the pouring area, or problems such as those already mentioned will likely occur. After the mold has been poured, the molten metal must be given time to solidify and cool to the proper temperature before it is removed from the mold. During the solidifying process, the mold halves must be held solidly together. Any movement will introduce the possibility of casting inaccuracies or increased demands for feed metal or both. The loss of casting accuracy has obvious consequences. The requirement for additional feed metal has the consequence that shrinkage cavities may form and will probably not be evident from the casting exterior. Cooling time after solidification is critical for many casting/alloy combinations. Insufficient cooling time can lead not only to dimensional problems due to lack of casting rigidity but also to hardness and internal stress problems, even to the point of cracking the casting.

Rotary-type shakeout system

can be quite successfully run using a longerstroke shakeout, while thin-wall castings may require a short-stroke high-frequency unit to prevent breakage or damage to the casting. Rotary-type shakeouts (Fig. 21) are also available in different configurations. The sand may exit at the same end that the sand and castings enter the unit, or it may exit at the opposite end. This type of shakeout also provides the function of conveying the castings from one end of the unit to the other. Rotational speed is adjustable on most units to allow flexibility in shakeout intensity. In general, as rotational speed decreases, intensity decreases and castings are less likely to be damaged. Light thinsection castings may not be suitable for this type of shakeout. Although the castings themselves may not damage each other, the sprue is sometimes heavy enough that it can damage the castings. Rotary plus cooling type shakeouts are also available in a configuration that not only holds the sand and castings together for an extended period but also affords the opportunity to cool the molding aggregate. This type of device is designed such that the castings and sand are held together throughout the length of the drum (Fig. 22). The castings and sprue aid in the breakdown of lumps. Sand temperature samples are normally taken somewhere along the length of the drum to determine the amount of water necessary for cooling the sand. The cool sand in turn cools the castings, often down to a temperature that can be comfortably handled at the exit of the drum. Sand and castings are separated at the exit of the drum. As with rotary shakeouts, sprues can damage certain types of castings, especially as wall sections become thinner. In-mold cooling time can become more critical when the castings and sand are kept together in the cooling device. When castings are too hot, hardness problems can result. In some cases, stresses

Fig. 20

Flat deck vibratory-type shakeout device

562 / Expendable Mold Casting Processes with Permanent Patterns can also be introduced into the castings because of the rapid quenching of the casting in the molding sand. Vibrating drum type shakeouts (Fig. 23) combine the operating principles of rotating drum and vibrating deck units. The vibrating section is round in cross section, but it does not rotate. Instead, a rotating action is imparted to the sand and castings by the vibratory action. As the drum vibrates, material is constantly agitated to produce particle migration in both axial and transverse directions. The drum can be designed to provide a very rapid blending action or a gentle folding action, depending on process requirements. Because air can be forcibly exhausted from the drum and because the surface of the sand within the drum is constantly changing, a limited amount of cooling is possible. Additional information on shakeout is available in the article “Processing and Finishing of Castings” in this Volume.

Sand/Casting Recovery After shakeout, it is necessary to recover and recycle the sand. Historically, the sand is returned from the shakeout to a storage bin, where it is kept until the next time it is mixed

with additional clay, water, and carbonaceous materials. Unfortunately, sand-to-metal ratios of 3:1 to 6:1 are quite common. Sand-to-metal ratios in this range, combined with cooling times that allow the castings to become cool enough to separate from the sand, can easily create return sand temperatures of 120  C (250  F) and above. High sand temperatures cause innumerable problems not only with regard to molding and surface finish but also for the system itself. Bentonite does not absorb water and become plastic to develop the necessary cohesive and adhesive strengths when sand is above 45 to 50  C (115 to 120  F). Therefore, the molding sand must be below these temperatures long enough for the muller to provide the necessary input of energy to coat the sand grains properly. Hot sand, usually above 50  C (120  F), is difficult to temper and bond, and when above 70  C (160  F), hot sands are impossible to rebond. Unfortunately, sand is not easily cooled, especially in the quantity necessary to keep a molding line running. Molding sand is a relatively good insulator and therefore tends to hold heat for long periods of time. Storage quantity is therefore not the answer. Not only does sand stored in a bin hold its heat for long periods of time, but it also cools from the outside toward

Fig. 22

Rotary plus cooling type shakeout system in which the castings and water-cooled mold sand are separated at the drum exit

Fig. 23

Front (a) and side (b) views of a vibratory drum type shakeout system

the center. As it cools in this manner, moisture tends to migrate toward the cooler sand, which causes it to cake on the outside walls. As time goes on, the caking on the outside wall becomes thicker until only a small portion of the sand is actually being circulated through the system. Vibrators and bin poppers have been designed and can be of some help in combating this rat-holing tendency of return sand bins, but the ideal situation would be to cool the sand prior to storage. Evaporative cooling is the only practical method of cooling the amount of sand needed in green sand systems. Hot sands must therefore have water added in amounts that exceed those required for tempering if both cooling and tempering are to take place. In addition, an ample supply of air must be present to carry away the heated water vapor. The cooling of molding sand may be regarded as a two-stage process, although no sharp line separates the stages. At temperatures in excess of approximately 70  C (160  F), added water causes a flash evaporation cooling effect. Temperature will continue to decrease fairly rapidly to approximately 60  C (140  F), but more slowly after that. As sand temperature approaches ambient temperature, further cooling becomes more difficult and time-consuming. Conditions of high ambient temperature, especially when combined with high ambient humidity, can substantially reduce the effectiveness of cooling devices. Therefore, ambient conditions should be considered carefully when sand systems are being designed or modified. Some mullers have the capability of blowing air through the sand mixture and will cool the sand very effectively. However, there are some disadvantages to this method. It must be kept in mind that mulling (coating sand grains with bentonite) does not take place until the mixture is cool enough to be tempered and bonded. Cooling time must therefore be added to the mulling time. Although Western bentonite provides the mold stability needed by most foundries, it does require more time and energy to absorb water and develop the necessary properties. Thus, the job of the muller or mixer becomes even more difficult and time-consuming. The storage bin will still have the tendency to rat hole, thus returning sand more quickly and hotter to the muller and further aggravating the situation. Control of solid additives and water becomes more difficult as the molding sand becomes hotter. However, this is not an impossible situation; this method is used quite effectively in a number of foundries. A few steps can be taken to provide some amount of cooling to the return sand in an existing system and to keep equipment costs as low as possible. For example, water can be fogged on the return sand, preferably as early as possible. Chains can then be dragged through the aggregate and/or plows can be used to turn the mixture over. Additional air can be introduced by fans or other sources to enhance cooling.

Green Sand Molding / 563 Elevators can be vented to enhance air flow, but this provides little help because the sand is being conveyed in solid buckets. The only assistance realized will be at the transfer points. Although these and similar methods do help to reduce return sand temperature, they are generally of only marginal value. An effective job of cooling return sand normally requires the addition of water, along with forced air being blown or pulled through the aggregate by some type of auxiliary cooling device. A number of auxiliary cooling devices are available that use forced air for evaporative cooling. These units should always be placed as close as possible to the casting shakeout. In fact, one type of unit, the shakeout-cooling drum, combines the functions of shakeout and sand cooling. Cooling the sand at or near the shakeout enables tighter control, reduces the tendency toward rat holing in the return sand bin, and reduces the demand for cooling on the muller. Because many muller designs make no provision for cooling, adequate external cooling is not only desirable but necessary. Cooling the sand as early as possible reduces the total cycle time of the muller by reducing or eliminating the time necessary for cooling and provides a method for making mulling time more efficient. Southern bentonite can be mulled in very quickly if the aggregate temperature is low enough. As mentioned earlier, Western bentonite is not mulled in very quickly, because it must swell by such a large amount. For this reason, it is advisable to keep the bentonite swelled and as active as possible. Many of the auxiliary cooling devices can be controlled to the point where the level of return sand moisture will be such that the Western bentonite will remain activated. Normally, a retained moisture level of 1.8 to 2.0% will not only keep bentonites activated but will also reduce the amount of dusting at transfer points, thus reducing the load on dust collection equipment. Cooling Devices. As mentioned earlier, it is possible to realize some cooling by adding water to a return sand belt and then using some method of turning the sand over at various places along the length of the belt. There are mechanized devices (Fig. 24) that perform similar functions and provide air flow through the sand. The effectiveness of these methods is often somewhat limited because of conveyor belt lengths; as belt lengths become shorter, the method becomes less effective. Difficult sand temperature problems will require more serious measures. Drums used as cooling units are among the oldest of the effective devices (Fig. 25). A cooling drum does not keep the sand and castings together; instead, this is a separate piece of equipment through which sand from the shakeout flows. As with other cooling devices, water must be added to the molding sand to allow the air moving through the drum to provide the necessary cooling by evaporation.

The fluid bed cooler is a vibratory type of conveyor through which the sand flows in a more or less continuous but controlled stream. Air is pumped through the sand from underneath, causing the necessary evaporation and cooling. Similar to the continuous muller shown in Fig. 17, the figure-eight cooler is designed so that air can be pumped through it and provide the necessary cooling. This device has been used directly above the muller, but a more desirable location would be as close to the shakeout as possible for the reasons already mentioned. Regardless of the equipment used, it is necessary to control the moisture additions so that sufficient moisture is available for cooling and bentonite activation without getting the return

Fig. 24

Fig. 25

sand so wet that problems will be experienced with plugging up of the sand system. The movement of air through the aggregate will almost certainly remove some of the finer material. The higher the velocity of air movement, the better the cooling, but also the greater the loss of that fine material. The loss of a certain amount of that material (such as dead, burnt clay and ash) can be beneficial. Unfortunately, a number of beneficial materials can also be lost, such as the finer grains of sand, coal dust, and bentonite. Any cooling device should be planned with a solids separator on the exhaust air so that these materials can be collected and fed back into the system at a controlled rate. This will improve surface finish, and trapping and using the bentonites and coal dust will provide economic benefits.

Mechanized sand cooler used in high-production molding lines

Cutaway view of a sand cooling drum system. Sequence of operations proceeds from right to left: 1, hot shakeout and spill sand enter, and helical flights convey sand forward to begin blending process; 2, cascading effect provides sand cooling as well as sand homogenization; 3, blended and cooled sand is discharged onto perforated cylinder, which screens off tramp metal and core butts while passing sand; 4, replaceable screen passes sand to discharge onto conveyor; 5, lumps that do not pass final screen carry across to lifter paddles for discharge into overburden chamber

564 / Expendable Mold Casting Processes with Permanent Patterns Metal Separation and Screening. The shakeout does the primary job of separating the sand from the sprue and castings. Smaller pieces of metal can easily slip through the grating of the shakeout device and be processed along with the sand. This will cause casting defects, and it may damage the equipment. Therefore, it is advisable to remove as much of the tramp metal as possible. When magnetic metals such as most irons and steels are being cast, the job is relatively easily accomplished with magnets. The suggested practice is to install an over-belt magnet somewhere along the length of a conveyor belt and a pulley magnet at the discharge end of the same belt. Placing both magnets on the same belt allows more complete separation of the magnetic particles. Nonmagnetic alloys present a different problem. Devices are available that separate the metallic particles based on density differences, but the most common method is to use screens. Multiple screens are often used, and the mesh size from screen to screen becomes progressively finer. Lumps are found in all sand systems and consist of system sand or core parts that have not been sufficiently heated to break down the binder. For this reason, it is necessary to have a good screen in all systems. The opening size in the screen should be as fine as is practical for the system involved. Two basic types of screens are in use: flat deck and rotary. The flat deck type is usually vibratory in nature and has the added function of providing further lump reduction as well as the screening function. The rotary type of screen is normally a large barrel that continually rotates. The exterior of the barrel has the desired size of holes in it to provide the screening action. Because of the tumbling action within the screen, lump reduction similar to that obtained with the vibrating flat deck can be expected. In both cases, the size of the screen should be as fine as is practical. After the sand has been cooled, the tramp metal removed, and the core butts and lumps removed, the sand is ready to be returned to the storage hopper to be used again.

Sand Reclamation The economics of a foundry operation require sand reclamation to reduce the costs associated with new sand and the costs of landfill use, and to reduce the problems associated with the control of environmentally undesirable contaminants in the discarded sand. In addition, tangible operational advantages result from sand reclamation. These begin with the ability to select the best sand, in terms of grain durability and other factors, for the casting process, knowing that most of it will be reclaimed during operation. A properly designed sand reclamation system begins with green sand and converts it to a product that is usable as a replacement for

new sand or for use in cores. The first step is the removal of tramp and foreign materials, such as core rods, metal spills, slag, and paper, and the disintegration of lumps of sand. Then, organic and inorganic binders are removed by mechanical attrition (scrubbing) and/or thermal methods. Dead clay is removed as fines. The sand is then brought up to specification by the addition of new sand, clay, and other sand additives or is used to produce cores. If used in cores, the base sand must be a lowacid-demand-value sand since carbonates in some base sands will convert to alkaline oxides that will negatively react with organic resin binders, and the reclamation process must be one that cleans the sand to a cleanliness close to that of new sand. Sand reclamation systems must be selected with regard to their cost, the specifications for the system sand, system capacity, compatibility with the sand system, metal being poured, and core mixes being used. It is important for the foundry to have a clear understanding of its needs in sand reclamation before calling in vendors. A variety of reclamation systems are described as follows. Wet Washing/Scrubbing. The cores of large castings can be removed by high-velocity jets of water. In the process, the cores are broken down into grains, and some binder is removed. Excess molding sand can be added and washed simultaneously. If the shakeout system is dry, the sand is charged into an agitator system where the solid content is held between 25 and 35%. Excess molding sand may be blended with the core knockout material. A similar system uses intensive scrubbing with a solids content of 75 to 80% and units in series. This latter method is superior because of closer and more frequent grain-to-grain contact. After washing, the sand is classified and may be used either wet (naturally drained to 4 to 5% moisture) and added to a green sand system sand, dried for use in cores, or used for facing sand. Facing sand mixes derive their name from the fact that they are used in limited quantities against the face of the pattern or core box. They have properties usually different from those of the backup or system sand and contain additives or contaminants that are not in the system (backing) sand. Facing sands are designed to perform special functions, such as providing higher green strength for lifting deep pockets, higher deformation for limited draft patterns, and special carbons that enhance skin drying. They are also used in some grades of steel, such as manganese steel, where contaminants in the system sand may react at the mold-metal interface and cause craze cracking or other defects. The wet system has limitations in that only a portion of the binder, clay, and carbon is removed. The product, however, is excellent for use as a makeup sand in systems. Because of the wastewater stream, and the need for a large amount of energy to dry the sand if used in a core process, wet reclamation is the least

common of the reclamation methods mentioned in this section. Dry Scrubbing/Attrition. This method is widely used, and there is a large variety of equipment available in price ranges and capacities adaptable to most binder systems and foundry capacities. Dry scrubbing may be divided into pneumatic, mechanical, and combined thermal-calcining/thermal-dry scrubbing systems. In pneumatic scrubbing, grains of sand are agitated in streams of air normally confined in vertical steel tubes called cells. The grains of sand are propelled upward and rub and impact each other, thus removing the binder. In some systems, grains are impacted against a steel target. Banks of tubes may be used depending on capacity and degree of cleanliness desired. Retention time can be regulated, and fines are removed through dust collectors. In mechanical scrubbing, the equipment available offers foundrymen a number of techniques for consideration. An impeller may be used to accelerate the sand grains at a controlled velocity in a horizontal or vertical plane against a metal plate. The sand grains impact each other and metal targets, thereby removing the binder. The speed of rotation has some control on impact energy. The binder and fines are removed by exhaust systems, and screen analysis is controlled by air gates and/or air wash separators. Additional equipment options include:  A variety of drum types with internal baffles,

impactors, and disintegrators to reduce lumps to grains and to remove binder  Vibrating screens with a series of decks for reducing lumps to grains, with recirculating features and removal of dust and fines  Shot-blast cleaning equipment that may be incorporated with other specially designed units to form a complete casting cleaning/ sand reclamation unit  Vibro-energy systems that use synchronous and diametric vibration. Separation of the binder from the sand grains is caused by frictional and compressive forces. One special, advantageous feature of these systems is the small number of moving parts. Thermal-Calcining/Thermal-Dry Scrubbing Combinations. As mentioned, the selection of a base sand is important if thermal reclamation is used. Pure silica sands are sometimes better than lake or other silica sands because the latter contain calcium carbonate that converts to calcium oxide that can elevate acid demand value (ADV) and pH to detrimental levels if intended for use with organic resin binders. Generally, thermal reclamation systems offer the best reclamation for the organic and clay-bonded systems. Grain surfaces are not smooth; they have numerous crevices and indentations. The application of heat with sufficient oxygen oxidizes the binders or burns them off. In attrition, only because there is no contact

Green Sand Molding / 565 in the crevices, the binder remains. Heat offers the simplest method of reducing the encrusted grains of molding sand to pure grains. Both horizontal and vertical fluidized bed and rotary kiln systems are available. In the horizontal rotary kiln, material is fed into one end (usually the cold one) and moved progressively through the heat zone by rotation assisted by baffles, flights, or other mechanical means. Some mechanical scrubbing also occurs. Some systems incorporate heat-exchanger technology to considerably reduce the energy required. The latest technology also includes provision for recovery of metal entrained in the sand and collection and detoxification of the process wastes for suitable nonhazardous waste disposal. Several fluidized bed system designs are available. Some use preheating chambers and hot air recuperation. A drying compartment may also be added. Sand is introduced into the top (preheating) chamber of the reactor and is lifted by the hot air stream from below until it assumes some of the characteristics of a fluid. The hot air coming in contact with the sand grains burns the organics and calcines the clay. At the same time, some attrition takes place. A correct pressure differential must be maintained between the compartments if more than one is used in order to ensure downward flow of sand; otherwise, gravity flow must be provided. Fluidizing is a very good method for cooling sand when using cool air. Multiple-Hearth Furnace/Vertical Shaft Furnace. The multiple-hearth furnace consists of circular refractory hearths placed one above the other and enclosed in a refractory-lined steel shell. A vertical rotating shaft through the center of the furnace is equipped with aircooled alloy arms containing rabble blades (plows) that stir the sand and move it in a spiral path across each hearth. Alternate hearths are “in” or “out.” That is, sand is repeatedly moved outward from the center of a given hearth to the periphery, where it drops through holes to the next hearth. This action gives excellent contact between sand grains and the heated gases. Material is fed into the top of the furnace. It makes its way to the bottom in a zigzag fashion, while the hot gases rise countercurrently, burning the organic material and calcining clay, if one or both are present. Discharge can be directly from the bottom hearth into a tube cooler, or other cooling methods may be used. The units are best suited to large tonnages, that is, 4500 kg (5 tons) or more. They are extremely rugged and relatively maintenance free. Combinations of systems may also be used, for example, thermal methods followed by dry attrition scrub to remove calcined clay from molding sand or undesirable chemicals and oxides from core processes. Also, commercial centers for sand reclamation are in operation and may be used by smaller foundries. If commercial centers are used, however, care must be taken by the reclamation facility to return the

same sand to the foundry customer, because sands from different foundries may have different grain distributions, grain shapes, and so on. Reclamation Effects on Base Sands. The most important tests for clay-bonded reclaimed sand are screen distribution, AFS clay, ADV, and pH. Reclamation also results in rounding of angular sands. This is clearly the case for reclaimed olivine sands, which are extremely angular compared to the more rounded silica grains (Fig. 26). Effects of reclamation on chromite are shown in Fig. 27. Figure 28 also shows the appearance of a sand before casting, after molding, and after reclamation.

Fig. 26

Fig. 27

ACKNOWLEDGMENTS Portions of this article were adapted from content contributions from Mary Beth Krusiak, Sand Technology Co., and John Jorstad, J.L.J. Technologies, Inc. Portions were also adapted from:  R.B. Brown, Green Sand Molding Equipment

and Processing, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 341–351  P.O’Meara, L.E. Wile, J.J. Archibald, R.L. Smith, and T.S. Piwonka, Bonded Sand Molds, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 222–230

Comparison of (a) angular olivine sand grains and (b) rounded silica sand grains

Chromite sand (a) before and (b) after reclamation. Note the smaller, more rounded grains due to reclamation. Scanning electron microscope. Original magnification: both 50. Source: Ref 3

566 / Expendable Mold Casting Processes with Permanent Patterns REFERENCES 1. Design of Ferrous Castings, American Foundrymen’s Society, 1963, p 103 2. M.J. O’Brien, Cause and Effect in Sand Systems, Trans. AFS, Vol 82, 1974, p 593–598 3. M.J. Granlund, Base Sand Reclamation, Trans. AFS, Vol 92, 1984, p 177–198 SELECTED REFERENCE  W.D. Scott, W.C. Easterly, P. Lodge, and

Fig. 28

Influence of sand reclamation on the appearance of green sand. (a) After molding (no reclamation). (b) After thermal reclamation. (c) New sand

C.S. Blackburn, Foundryman’s Guide to Sand Compaction, Transaction of the American Foundry Society and One Hundred Eighth Annual Metalcasting Conference, June 2004 (Rosemont), p 609–621

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 567-580 DOI: 10.1361/asmhba0005354

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

No-Bake Sand Molding* NO-BAKE SAND MOLDS are based on curing of inorganic or organic binders with either gaseous catalysts or liquid catalysts. Vapor-cured methods include:
Sodium silicate binders cured with carbon

dioxide


Amine-cured phenolic urethane
SO2 gas-cured acrylic epoxy

The vapor-cured binders are also referred to as cold box methods when used for coremaking. The liquid-cured binders, which are self-setting, include the following:
Sodium silicates cured with catalysts with-

out CO2


Furan binders based on furfuryl alcohol
Acid-cured phenolics
Urethanes based on alkyd oil urethanes or

phenolic urethanes Typically, the self-setting binders are referred to as no-bake methods in coremaking. However, neither vapor-cured nor the self-setting binders require baking or heat activation for curing. Therefore, the term nobake is used in this article with reference to both methods. The next article, “Coremaking,” describes the systems that require baking or heat activation (i.e., the hot box methods). The base binders are described in the article “Aggregates and Binders for Expendable Molds” in this Volume. Of the various no-bake methods, the air-set (liquid catalyst) sand systems are widely used. However, each method has different advantages and disadvantages, as detailed further in this article. Vapor-cured no-bake methods lend themselves to high-production runs and automation, while the air-setting (liquid-cured) nobakes do not lend themselves to high-production runs. This is due to the time required to thoroughly cure the mold or core. Compared to green sand molding, one basic advantage of no-bake binding (by either liquid or vapor curing) is improved dimensional tolerance control. The rigidity of the aggregate mass is hardened in place around the pattern equipment, and the strength conferred by the chemical binders helps prevent distortion of the aggregate mass when metal is poured into the mold.

Nobake also has the potential to offer more uniform castings when compared over the duration of a production run. This has frequently been evidenced by a weight reduction in castings converted from green sand practices to nobake. Depending on the level of control exercised over the green sand, the weight reduction can exceed 5% and can produce cost savings that can more than cover the cost of the resins and the disposal/reclamation of the sand materials. Another major advantage of the no-bake process is that the molds are not friable in the cured state and degrade very slowly in ambient atmospheres. This feature allows an almost unlimited time for mold assembly, careful gaging of core assemblies, venting, surface preparation, treatment of chills, spot repairs, and so on before mold closure. All of the above can contribute to greater dimensional control in the final product and can minimize scrap during production. Sand mixtures in the no-bake system typically have good flowability, although some compaction or vibration may be needed (Fig. 1). Nonetheless, the extent of required compaction is less than that required in green sand molding. This may allow more economical options in patternmaking with wood and, in some cases, plastic (instead of with metal patterns that would be required with the high pressures of high-density green sand molding). Likewise, when very large castings require the use of floor molding, this can be done by nobake as a manual operation with pneumatic tools. In contrast, floor molds of green sand require slingers to achieve adequate compaction (see the article “Green Sand Molding” in this Volume). Each of the major no-bake processes has advantages and disadvantages in moldmaking or coremaking. In terms of shakeout and sand reclamation, for example, resin (organic) nobakes may have an advantage over the inorganic (sodium silicate) nobakes. Most of the organic no-bake systems have excellent shakeout. With the exception of those castings poured from low-melting-point metals (aluminum or magnesium), the heat of the metal burns out the resin bond in the sand sufficiently for the mechanical action of the shakeout device

to cause almost all of the sand to fall away from the casting. Sand reclamation is also higher with organics. However, sodium silicates also have the advantages of good availability (as a non-petroleum-based feedstock) and have several methods for catalyzing the bond reactions with a wide range of work/strip times. This article briefly reviews the major aspects of no-bake sand bonding in terms of coremaking, molding methods, and sand processing. Coremaking is also described in the next article, “Coremaking.” Additional information on the types of organic and inorganic no-bake systems is also provided in the article “Aggregates and Binders for Expendable Molds” in this Volume.

No-Bake Sand Processing Refractory Coatings for Chemically Bonded Sands. Refractory coatings or washes are often necessary or desirable for chemically bonded cores and molds in order to improve surface finish and improve casting quality by minimizing reaction with the molten metal. Such coating materials can be of many kinds, using various filler (refractory) materials and liquid suspension materials. They also may

Fig. 1

No-bake molding

*Portions of this article were adapted with permission of the American Foundry Society (www.afsinc.org) copyright 2007 from the following: Casting Copper Alloys, 2nd ed. (ISBN 10-0-87433-302-4), 2007, p 49–65.

568 / Expendable Mold Casting Processes with Permanent Patterns be applied in a variety of ways, including brushing, swabbing, dipping, flow coating, or spraying. It is important to carefully match the coating material and suspension agent to the specific type of chemically bonded sand it will be used with. If these are incompatible, casting defects can occur. Binder suppliers should be contacted in questionable situations. A general, authoritative reference for the selection of wash coating materials for all foundry applications is the AFS Mold and Core Coatings Manual. A few of the more popular chemically bonded sand systems are described as follows. Silicate Binders. Refractory coatings may be necessary on cores and molds made with silicate binders to prevent sand burn-in/burn-on or penetration. Alcohol-based or solvent-based refractories are normally used. Specially prepared pastes containing graphite or zircon refractory are commercially available; these may be readily diluted in the foundry with alcohol or solvent. The use of refractory coatings on silicate-bonded molds and cores not only promotes better casting peel but also improves collapsibility and aids in suppressing moisture absorption in storage during periods of high humidity. Water-based washes can be used if care is taken to dry the core or mold thoroughly, immediately after coating. This procedure must be conducted carefully because of the softening effect of the water on the silicate bond. Furan Binders. Most commercial core washes, regardless of suspension agent, are compatible with furan no-bake binder systems. The wash may be applied in any manner; however, application should be made only on a surface that has been properly and thoroughly cured. If coatings are applied too soon after stripping, a friable surface results. Regardless of the coating suspension agent, the wash should be thoroughly dried by ignition, evaporation, or heating before the mold is closed and poured. Phenolic Binders. Any suspension agent may be used with phenolic binders, as long as the mold or core is adequately cured before application is made and the wash is thoroughly dried thereafter. Phenolic Urethane No-Bake (PUNB) Binders. While nearly all types of coating materials may be used with the PUNB binders, when using water-based washes it is best to make the application immediately after stripping and to dry it promptly. With either air-dry or lit-off solvent-based coatings, the wash is often applied within 10 to 20 min after stripping, followed by ignition of the solvent. There may be a temporary softening of the sand beneath the coating, without any apparent permanent distortion. As with other binder systems, all wash coatings must be thoroughly dried before pouring. Mixers. There are several types of mixers used with no-bake binder systems. The best mixer to use for a particular operation depends on the binder used and the sand volume required. The batch mixer, used by the foundry

industry for many years, is normally used to mix sand with binders that have long working times or are slow curing. The auger or screw-type continuous mixer is another commonly used piece of equipment. With this type of mixer, catalyst and binder are added to the sand as it is screwed along the trough and mixed together. However, continuous mixers of this type have a tendency to waste a considerable amount of coated and raw sand since, unless it is used within a reasonable length of time, the sand in the mixing trough must be discarded. Another type of mixer is the high-speed continuous and/or batch mixer. This mixer usually has single or multiple mixing chambers of short length with very high throughput of sand and a low level of sand waste. Improved ease of maintenance is one of its main advantages. Zero-retention continuous mixers are ideal for no-bake systems since the mixed sand is available at the push of a button, and minimal amounts of sand remain in the mixer at the end of a mixing cycle. The size of the mixer required will depend on the size of the pattern or core box to be filled and the amount of time available for filling. An important part of the mixing facility is the metering system, which is used to control the additions of various components to the mixer. It is essential that the structural materials of the metering devices be accurate, reliable, and compatible with the components of the no-bake system. Gear pumps with accuracy within  þ2% are recommended for handling liquid resin components. This type of pump provides a constant rate of delivery, whereas the pulsation in delivery rate associated with positive displacement pumps makes them less suitable. Positive displacement pumps should be equipped with a surge chamber if used. To ensure that all systems are functioning properly at the mixer, routine calibrations should be made at the beginning of each shift. The delivery rates of sand (wet sand flows more slowly), sand binders, catalysts, and any special additives should be checked to verify that they are as intended. Sand Processing for Molds and Cores. The production of molds and large cores using the no-bake sand binders is facilitated by the fact that the sand mixtures are highly flowable and thus do not require a high level of molding or coremaking skill for their use. However, uniform mold and core densities are not attained by merely dumping the sand mix into a flask or core box. In handling sand-resin mixtures for no-bake molds or cones, the following points should be carefully noted:
No-bake sand must be used and compacted

immediately upon mixing, before any real hardening commences. Hardening starts as soon as the binder contacts the catalyst in the sand mixture; then, the viscosity of the binder film gradually increases and
















hardening begins. Both the viscosity and the curing action greatly affect the flowability of the sand mixture. The curing reaction is temperature-sensitive; sand that is above 40  C (104  F) will hasten precuring, quickly reducing sand flowability. The sand should be maintained at a uniform temperature, preferably between 25 and 35  C (77 and 95  F). When making larger molds or cores, highoutput mixers are essential in order to prevent undue precuring of the sand mixture before the moldmaking or coremaking operation is completed. With larger molds and cores it is essential that the sand, particularly near the mold or core surfaces, be compacted by tucking or tamping it into position. Generally, only a 50 mm (2 in.) layer can be compacted at a time. Sand near the pattern should not be disturbed as layers of sand are added above it, once the working time is exceeded. Vibration compaction is a good method of consolidating no-bake sands and is most applicable to the manufacture of smaller molds and cores. It is essential that the vibration take place for an optimum amount of time, without overloading the vibrating table. Overvibration is detrimental and can cause ram-off problems. A more satisfactory practice, however, is to hand tuck the sand at critical areas and at the mold-metal interface, and then vibrate during the filling of the core box or flask. Oily-type binders (alkyd resin oil-isocyanate) and low-viscosity binders generally provide more flowable sand mixtures than sands bonded with more viscous binders, such as furan-type binders. Any additions that provide green strength reduce the flowability of the mixture and should be as small as possible. Hardness tests (indentation or scratch) will not determine the degree of compaction of no-bake sands. The only simple method available is to determine the permeability of the surface of the mold or core by means of a portable permeability tester. All compacted sands show a decrease in permeability with increase in bulk density. Many binder systems develop less strength if bulk density is less than a critical value.

Foundry Control Tests of No-Bake Sands. Various tests are used for evaluating and monitoring chemically bonded sands. The consistency and uniformity of production using chemically bonded sand technology is dependent upon narrow control parameters. The effects of upsetting the fine balance necessary when dealing with the interacting chemicals, variations in the acid demand value of the base sand, and minute amounts of moisture in the base sand should serve to emphasize the need for strict in-process controls and a testing program tailored to the specific process and production requirements of each operation.

No-Bake Sand Molding / 569 Mixer Calibration. Although not normally considered a sand test, mixer calibration is the most important control test performed with chemically bonded sands. This test first establishes the exact sand flow from the mixer; it is this flow rate upon which binder and catalyst percentage calculations are established. On large mixers, this rate is usually checked by placing a floor scale with a large container underneath the mixer. Sand is then discharged into the container for a prescribed amount of time and weighed. Normally, this is repeated several times to ensure correct flow rates. Although time-consuming and perhaps awkward, this test is extremely important. Once the flow rate has been established, the binder and catalyst pumps are then calibrated for the correct amounts. If possible, samples are taken from the resin discharge ports in the mixer. If this is not possible, they should be taken from a calibration port that is of the same configuration as the discharge port. This is repeated until the samples procured are of the correct calculated weight. Mixer calibrations should be run at the start of each shift to ensure that the correct sand discharge rate and binder additions are being used. Tensile Strength. Measurements of the tensile strength of the sand mixture should be made periodically throughout the day; the frequency of the testing depends on the process and binder system being used. The purpose of this test is to check for consistency of the sand mix in order to note any trends that may signal a potential production problem. For the test to be effective, it is necessary to develop a correlation between tensile strength and core quality (sample weight, permeability). Work Time/Strip Time. The test for work time and strip time can be very useful in a nobake operation. This test need only be performed infrequently, unless a major change in the system should occur. Tests for determining work time or strip time are usually run in a series in order to develop charts or data that relate these values to catalyst level. In this manner, it is possible to regulate the catalyst pump on the mixer to provide the desired working properties for each job and to avoid the use of sand that has surpassed its work time. Some binder systems must be stripped while still plastic. Exceeding this time causes stripping difficulties, damage, and reduced productivity. Similar tests can be used with vapor-cured sand mixes to document the deterioration of the sand-resin mix with time. In this case, the allowable delay time is referred to as bench life as opposed to work time. Sand Temperature. Chemically bonded sands are extremely sensitive to temperature. A general rule-of-thumb is that for every 10  C (18  F) decrease in sand temperature, the reaction time of the binder is doubled. Thus, if a consistent operation is desired, sand temperatures must be consistent. Operators should know what the temperature is so that corrections can

be made to compensate for any possible temperature changes. Temperatures can be monitored by probes in the surge hoppers before the sand mixers, or by the use of an accurate thermometer placed in sand that has been discharged from the mixer. Pattern temperatures should also be controlled. Grain Fineness. American Foundry Society (AFS) grain fineness number (gfn) tests should be performed periodically to check the screen distribution and especially the fines content of the sand at the mixer. This is an important test, because an increase in fines can cause drastic decreases in the strength of cores and molds. It is best to make this test at the mixer, since sand transportation systems and storage silo configurations can change the fineness value significantly. Moisture in the base sand can cause reduced strengths or retard the cure in chemically bonded sands due to the water reacting with the binders. A lowering of flowability and the degradation of the bond between sand grains occurs due to the moisture film. Moisture levels above 0.25% should be avoided. The moisture test can be performed with the same frequency as the fineness test. This will indicate if a moisture problem is developing in the sand, either from high-moisture contents in the delivered sand or from leaks in the sand silos. This test is also helpful in troubleshooting when the cause of low-strength cores or molds is not readily apparent. Acid Demand Value (ADV). Testing the ADV of the base sand can be helpful in several areas. High-ADV sands retard the cure of acid-cured systems. This can result in poor strengths or no cure. It can also contribute to the acceleration of the cure rate in basic catalyzed systems to a point where the sand is cured before it can be used. The ADV test can also be used to monitor acid buildups in acid-cured reclaimed sands and to check for influxes of basic contaminants into the sand. Loss-on-Ignition (LOI). The LOI test reveals the amount of combustible material in the sand. High levels of combustibles in the base sand can affect both cure rate and strength. The test is most important for monitoring the efficiencies of sand reclamation systems. LOI values above 2% can lead to problems with sand properties and may cause gas defects in castings.

CO2-Cured Sodium Silicate The sodium silicate, carbon dioxide process is the oldest of the gas-cured no-bake systems. Sodium silicate (water glass) is a viscous liquid that can be distributed uniformly over sand grains by conventional mulling techniques. Various additives also may be included in the binder-sand mix to improve characteristics such as sand flow or mold/core collapsibility. The sand mix then requires some some light compaction around the mold pattern or core box.

In the case of cores, the mixture can be hand rammed, blown, or shot into the core box. In the case of molds, the sand mixture can be manually compacted, jolted, or squeezed around the pattern in the flask. Patterns may be wood, plastic, or metal, and they can be cleaned by washing with water. After compaction, curing is accomplished by passing carbon dioxide gas through the compacted sodium silicate-sand mixture. Various methods are used to force CO2 gas through the compacted sand mix. With gassing, a strong intergranular bond can be developed in a few minutes. A similar bond is developed when the sodium silicate-sand compact is dried, or when the compact is allowed to stand, because of the reaction with carbon dioxide in the air. However, hardening in air requires much more time. There are also several less frequently used variations of the sodium silicate processes that are self-hardening without the use of CO2 gas. These methods require the use of either ferrosilicon fines, portland cement, or an ester to harden the mold. In comparison with sand molds and cores produced by other methods, those made by carbon dioxide molding have poor shakeout properties. However, shakeout difficulties can be minimized by close control of the composition of the mold or core mixture. The silicate system has no supply or waste disposal issues. Typical silicate binders are ordorless, nonflammable, nontoxic, and contain no sulfur, nitrogen, or phosphorus. They are suitable for use with all types of aggregates in the production of both molds and cores. They produce no noxious gases and a minimum of emissions during pouring, cooling, or at shakeout. On the negative side, the CO2-silicate system requires control of temperature, CO2 flow rate and gassing time, is sensitive to high relative humidity, cores have a short shelf life, and, as noted, there is poor collapsibility with reclamation difficulty.

Applications With carbon dioxide gas curing, sodium silicate-sand mixtures develop a dry compressive strength that may be in excess of 1.4 MPa (200 psi). By this process, ready-to-use molds or cores can be made in a few minutes; no baking is needed. Often, molds are produced by the carbon dioxide process to eliminate the need for special flask equipment and to obtain closer tolerances in sand castings. The relatively high strength of the molds made by the carbon dioxide process allows them to be used without backup aids, such as jackets, flasks, and bottoms. If the parting line is horizontal, a weight on the cope portion of the mold is usually adequate to prevent runout of metal at the parting line. Clamping is required if the parting line is vertical. Sodium silicate-sand mixtures strengthened by carbon dioxide are used for cores more

570 / Expendable Mold Casting Processes with Permanent Patterns commonly than for molds—the main reason being that the cores are quickly made and are ready to use without baking. Thus, for rush work, cores can be available as quickly as can green sand molds, whereas when using cores that require baking, there is considerable delay even though the mold is already prepared. Because of these characteristics, the carbon dioxide process is especially useful in jobbing foundries. Expendable cores used in permanent mold casting are commonly made by the carbon dioxide process because of the dimensional accuracy obtainable. The process is suitable for all common casting alloys, including aluminum and magnesium. It is most widely used, however, for casting of steel, cast iron, and copper alloys. The process has been successfully adapted to the production of castings ranging in weight from less than 0.5 kg (1 lb) to many tons. In addition to improved tolerances, advantages of carbon dioxide molding and coremaking processes are:
Molding and coremaking practice is similar














to that in green and dry sand molding; thus, the carbon dioxide process is compatible with existing operations. Because of the strength of the sodium silicate-sand mixture, the need for internal sand support in drag and cope elements may be eliminated, particularly if the sand is gassed with the pattern in place. Expensive equipment is not required. The sodium silicate can be mixed with sand, using conventional equipment. Carbon dioxide is readily available, and the gassing equipment is low in cost. No baking equipment is required. Cores are ready to place in the mold when they are removed from the core box. Floor-space requirements are minimal, and indirect-labor costs are small. Greater dimensional accuracy can be attained than with most other coremaking processes. The process when used for making cores can be automated for high speed and long production runs. Both large and small cores and molds can be made using this process. The disadvantages include:


When used for making high-production run

cores, metal core boxes must be used.


Molds and cores are more expensive than

most made by conventional processes.


Bench life of sand mixtures for the carbon

dioxide process is much shorter than for most other mold and core mixes.
Shakeout properties and sand neclamation of carbon dioxide molds and cores are poor compared with those of conventional molds and cores.
Cores and molds cannot be stored for long periods of time because the binder is very

hygroscopic (absorbs water from the air), which weakens the cured binder. Molds and cores deteriorate noticeably after being stored under normal atmospheric conditions for more than approximately 24 h.
Although the poor shakeout and collapsibility can be reduced by adding organic materials that burn out, this in turn increases the gas generated by the core or mold.

Sand-Silicate Mixtures Molds and cores typically use silica sand, although other sands, such as zircon and olivine, can also be used. Ordinarily, silica sand within the fineness range of AFS 55 to 85 gfn is used. Sands for the carbon dioxide process should always be dry, with a maximum moisture content of 0.25%. Sands should also be as clean as possible; otherwise, an excessive amount of binder is required. For best results, the sand should be as free from lime as possible, because lime reacts with sodium silicate. As in other molding and coremaking processes, more binder is required for sands that are angular in grain shape. The required amount of binder (sodium silicate) also increases as the grain size of the sand decreases. For instance, a washed and dried rounded silica sand with a grain size of AFS 55 requires approximately 2.5 to 3% of sodium silicate (by weight). A finer sand (AFS 120 to 140) will require from 2.5 to 3.0% additional binder. In determining the amount of binder for a specific job, it is recommended that test cores be prepared using an average amount of binder. Sodium Silicate. The sodium silicate binders in the carbon dioxide process consist of soda (Na2O), silica (SiO2), and water. Compositions may range from 7 to 28% sodium oxide, 26 to 64% silica, and 17 to 67% water. In general, sodium silicate is selected by weight ratio of SiO2 to Na2O. Sodium silicate does not attack most metals, including steel; therefore, it can be stored in steel drums. Once a drum has been opened, it is advisable to keep an oil film on top to prevent the formation of “skin.” Properties of the silicates vary according to the silica (SiO2)-soda (Na2O) ratio of the sodium silicate base that is used to formulate the binder. Liquid sodium silicates are commercially available with varying proportions of silica and sodium oxide and with wide ranges of viscosities. The most commonly used silicates are those having viscosities between 48 and 52 Baume´ (Be´) and a silica-to-soda ratio (SiO2 to Na2O) between 2.0 and 2.8 to 1 (1.9 to 3.2 may be used in other applications). By controlling this ratio and the water content, it is possible to make various grades of sodium silicate having a wide range of chemical and physical properties. The curing of the silicate process involves two mechanisms. The first reaction is that of sodium silicate with CO2 gas to form sodium

carbonate and silica gel and water. At this stage, the system has reached 20 to 40% of its ultimate strength. The second mechanism is the dehydration of the silica to form a glasslike bond. If the gassed core or mold is allowed to remain at room temperature for 24 h, the strength increases to a maximum of 0.7 to 1.4 MPa (100 to 200 psi) (from, say, an initial tensile strength of 0.26 to 0.31 MPa, or 37 to 45 psi, after gassing for 5 s). This increase is due to some dehydration of unreacted silicates and the continued gelling of the silicate. Under normal conditions, cores will not deteriorate and can be stored for various periods of time. An exception may occur during periods of high humidity when stored cores are unable to achieve maximum strength by dehydration after gassing. Strength deterioration of sodium-silicatebonded cores in high-humidity conditions is most prevalent when the binder formulation of the sand mixture contains organic additives such as sugars, starches, or carbohydrates; a short heating cycle is necessary to obtain the desired strength under these conditions. When a core or mold is hardened by carefully controlled gassing and further hardened by dehydration and polymerization during the subsequent 24 h, good strengths are maintained over a long period of storage. It is always advisable, with sodium-silicate- bonded/CO2-hardened cores, to date the cores so that the oldest cores are used first. Additives. Proper use of sand-system additives is necessary for successful casting. The main reason is to improve shakeout or collapsibility. Other functions of additives include:
Controlling expansion
Preventing burn-on/burn-in







and metal penetration Improving green strength properties Promoting peel of castings and improving casting finish Aiding flowability of the sand mixture Creating proper mold atmosphere Preventing excessive drying out of the sand mixture Preventing sticking of sand mixture to patterns and core boxes

However, additives must be used with caution because the bond strength can be seriously lowered in many instances. Additives also cannot remedy poor formulations or faulty operations. Several different additives have been used to improve properties of the sodium silicate-sand mixture, for example, kaolin clay, aluminum oxide (Al2O3), and invert sugar or sugar in other forms. Sugars are commonly used to improve shakeout. Sugars can be compounded with the silicate binder or added separately to a sand mixture. Improvements in shakeout may also be obtained by reducing the binder level or changing to a higher-ratio silicate. Kaolin clay promotes mold stability and retards fusion of Na2O and SiO2 at approximately

No-Bake Sand Molding / 571 815  C (1500  F). Aluminum oxide improves the hot strength of the bonded sand at temperatures near 815  C (1500  F). Additions of invert sugar reduce the retained strength of the mold or core after pouring. Severe shakeout problems may be encountered without additions of sugar. Molasses has been used instead of invert sugar, but certain nitrogen-containing nonsugars in molasses can react with basic sodium silicate and may cause premature precipitation of silica, with a consequent loss of binding power. This reaction is characterized by the evolution of ammonia gas. The total nitrogen content in molasses ranges from 0.4 to 1.5%. Grades that contain less than 0.5% N can be used successfully as a substitute for invert sugar. Such grades will contain no more than 3.0% crude protein. Because of trade practice, the grade should be specified by crude-protein content (3.0% max) rather than by nitrogen content. In general, any cane molasses except that produced in Florida is acceptable for this purpose. A typical mixture for molds and cores consists of the following:






Sodium silicate, 40 Be´: 3.3% Kaolin clay: 1.7% Al2O3 (325 to 600 mesh): 1.7% Invert sugar, 40 Be´: 2.0% Sand, AFS 70: balance

The aforementioned mixture, which is one of many that are commonly used, has been satisfactory and economical for casting of ferrous metals. Effects of Composition on Strength. An ideal mixture is one that has a reasonably high, although not excessively high, strength at elevated temperature (hot strength) but a low retained strength (strength after heating and cooling). Low retained strength is essential for ease of shakeout. The hot strength of a sodium silicate-sand mixture with no additives decreases rapidly as the mixture is heated and approaches zero strength at just above 815  C (1500  F), as

Fig. 2

shown in Fig. 2(a). Retained strength also decreases markedly at approximately 815  C (1500  F) but increases to a peak at approximately 1040  C (1900  F) (Fig. 2a). These relations between strength and temperature indicate several undesirable conditions for such a mixture:
Hot strength is insufficient, which will prob-

ably result in loss of accuracy in casting dimensions.
Retained strength at approximately 1040  C (1900  F) is excessive. Portions of the mold or core adjacent to the molten metal may easily reach 1040  C (1900  F) when ferrous metals are poured. Thus, high retained strength at this temperature will result in shakeout difficulty.
Retained strength at 205 to 260  C (400 to 500  F) is also high. This temperature is likely to be encountered in the centers of many cores, which will add to shakeout difficulty. To correct the aforementioned undesirable conditions, a combination of organic and inorganic materials is added. Invert sugar, kaolin clay, and aluminum oxide are the materials most commonly added. Separate and combined effects of these additives are shown in Fig. 2(b) to (e). The effect of adding 2% invert sugar to the sodium silicate-sand mixture is shown in Fig. 2(b). The retained strength is greatly reduced for all temperatures, which is desirable, because it increases the ease of shakeout. However, the hot strength is almost zero at temperatures higher than 900  C (1650  F), which is undesirable because when hot strength is too low at 900 to 1095  C (1650 to 2000  F), excessive movement of the mold wall is likely. The individual strengthening effect of adding 2.5% kaolin clay (no invert sugar) is shown in Fig. 2(c). With the clay additive, the retained strength is increased in the lower range of temperature, but not enough to be harmful. Hot strength at 980 or 1040  C (1800 or 1900  F) (Fig. 2c) is so high that early core collapse is prevented, which is likely to result in an

Effect of various additives on hot strength and retained strength of sand bonded with sodium silicate

excessive number of defective castings from hot tears. As shown in Fig. 2(d), the addition of equal amounts of kaolin clay and aluminum oxide (with no invert sugar) decreases the hot strength of the mixture, compared with that shown in Fig. 2(c); retained strength changes only slightly. The combination of hot and retained strengths indicated by the curves in Fig. 2(d) is generally acceptable, because it promotes early collapse of the core, thus minimizing hot tears or similar casting defects and, at the same time, minimizes the likelihood of shakeout difficulty. The combined effects of three additives (invert sugar, kaolin clay, and aluminum oxide) on hot and retained strengths are shown in Fig. 2(e). This combination of additives provides low retained strength, low hot strength in the vicinity of 260  C (500  F), and satisfactorily high hot strength between 815 and 1095  C (1500 and 2000  F). Core Collapsibility. From the data given in Fig. 2 and from the previous discussion, it is evident that the additives used in making silicate-bonded molds and cores can have a marked effect on their hot strength and retained strength at various temperatures. These properties directly affect core collapsibility. As is true for cores of other compositions, it is desirable that silicate-bonded cores have low strength within the temperature range where freezing of the cast metal takes place. This causes the core to collapse as the metal freezes, thereby minimizing the danger of hot tears in the casting. Thus, it follows that the composition of the core mixture (principally the additives) must be varied in accordance with the freezing temperature and hot strength of the metal being cast. For example, the typical mixture given previously has proved successful for casting ferrous metals, but it would not be used for casting of aluminum or magnesium alloys, because these alloys freeze at lower temperatures and have extremely low hot strengths compared with ferrous metals. The difficulties

572 / Expendable Mold Casting Processes with Permanent Patterns often encountered in obtaining optimum collapsibility of silicate-bonded cores partly account for their being used to a lesser extent for lightmetal castings than for ferrous metal castings.

Procedures Continuous mixers or batch mixers can be used with sodium silicates, which can be metered through pumps. Overmixing should be avoided since this may reduce the work time (bench life) of a sand mixture. Mixtures normally have a good bench life, which becomes shorter when higher-ratio silicates are used. Hoppers of mixed sand should be covered with plastic sheets or damp sacking to prevent premature hardening (crusting). Molds or cores can be produced from the prepared mix with mechanized equipment or manual methods. Mixing and Handling. The sand, sodium silicate, and additives for carbon dioxide molds and cores are mixed in conventional mullers. Because the viscosity of sodium silicate increases as temperature decreases, best results are obtained when excessive temperature variations are avoided. Room temperature is usually ideal. Problems are sometimes encountered in cold weather if the sodium silicate has been stored in unheated rooms. Mulling is done at slow speed; total cycle time usually is 3 to 5 min. Common practice is first to mull together the additives and the sand, then to add the sodium silicate and mull for 1 or 2 min. A typical cycle is:
Add invert sugar to sand, mull 1 min
Add kaolin clay and Al2O3, mull 2 min
Add sodium silicate, mull 1 min

The shortest mulling time that will accomplish dispersion is best. Mulling too long results in a dry, crumbly mixture that is not usable. Success with this process depends to a large extent on close scheduling of the mixing and molding or coremaking. If the mixture is allowed to stand after mulling, partial hardening takes place, which causes trouble in the subsequent blowing or ramming operations. If it is necessary for the mulled mixture to be held for any length of time prior to making the cores or molds, covering the mixture with a damp cloth will help to retard the natural hardening. Molding. Procedures for making sandsodium silicate molds and cores are the same as for making other types of sand molds and cores. Likewise, patterns and core boxes are the same as are used for other sand molds and cores, and the same factors influence the choice of pattern equipment. Although the carbon dioxide process is well adapted to low-quantity or even unit production, as in a jobbing foundry, it is often used for mass production. Automated machinery can perform blowing, squeezing, gassing, and ejection of molds and cores. After gassing, molds are usually coated with a refractory wash to improve casting finish.

Gassing Procedures. After the sodium silicate-sand mixture has been compacted in a core box or over a pattern, it should be gassed immediately. When small molds or cores are produced in large quantities, automated machines are integrated to perform compaction and gassing. However, for smaller quantities or large sizes, or both, various other gassing procedures are used. Carbon dioxide gassing is accomplished by spiking the back of the mold and fitting a tube or a rubber cup over the holes. More efficient methods employ a hood over the top of the flask, pattern vents, or use of vacuum chamber and back filling with CO2 once the air has been evacuated. Production rates are better with core blowers and vacuum chambers. Probes, either singly or connected in a manifold arrangement, are used for large cores. Molds having sections no thicker than 30 cm (12 in.) are frequently gassed through the pattern. Small cores or molds can be gassed without probe manifolds but rather by the use of a gastight hood, connected to the source of carbon dioxide gas and placed over the opening in the core box. Gassing is sometimes done by placing the pattern or core box over a plate or receptacle that is connected to a vacuum pump. When the air is evacuated, the carbon dioxide gas permeates the entire section quickly. By this procedure, the mixture is not only gassed in a minimum of time, but less gas is used. Evacuation of the mold/core in a vacuum chamber before introducing the CO2 gas into the chamber also allows use of less binder and avoids overgassing. Although residual sodium normally poisons attempts to rebond the sand, a low binder level (2%) allows mechanically reclaimed sand to be successfully recycled. Gas pressure at the probe (typically 0.14 to 0.28 MPa, or 20 to 40 psi) depends largely on the thickness of the section to be gassed. However, the length of time the sand mixture is in contact with the gas is important; the pressure determines the distance of gas penetration. Considerable care must be used when gassing with CO2, since overgassing and undergassing adversely affect the properties of the cured binder system. Operators have a tendency to overgas some parts of the core or mold to ensure that all parts are cured. Control of the gassing operation becomes critical when the higher-ratio binders are used. Shorter gassing times are needed to harden the higher-ratio binders. These materials are easily overgassed, resulting in molds and cores with a shortened shelf life. Overgassing results in strength loss (Fig. 3). Also, because the mold or core will continue to harden in air after gassing, the gassing time should be no longer than necessary for developing an initial hardness that is sufficient to allow handling. Delay Between Gassing and Pouring. As noted, molds and cores will deteriorate from water pickup if held for more than

Fig. 3

Effect of gassing time on the strength of cores. Data based on gassing cores made from sand mixed with 3.3% sodium silicate, 1.7% aluminum oxide, 1.7% kaolin clay, 2% invert sugar

approximately 12 h before the casting is poured. If gassed cores or molds must be held for a longer time before pouring, water pickup can be minimized by coating them with a core wash. Alternatively, they can be dried by heating.

Air-Setting Sodium Silicates In addition to the gas (CO2) curing of sodium-silicate sand molds and cores, other types of hardening agents have been used with sodium silicates. Air-setting (self-setting) hardening of sodium silicates is done with liquid catalysts such as:
Dicalcium silicates
Ferro silicon (Nishiyama process)
Organic esters

Generally, dicalcium silicate and ferro-silicon systems have the advantage of using lowcost materials that are added to the sand mix in powder form. Sands that are suitable for the CO2 silicate process can be used satisfactorily. Usually, sodium silicate of a 2 to 1 ratio of SiO2 to Na2O is required for use with these powder-type hardeners in addition of 3½ to 5½%. Setting agents required will be between 1 and 3%, depending on the rate of activity and the bench life required. Considerable inconsistencies in setting times are often experienced because of the difficult problems of adding powdered materials successfully to sand mixers, with the result that, generally speaking, self-setting powder hardener systems have been superseded by the organic ester processes. Controlled addition of liquids to sand in continuous mixers is easier than powders. All continuous mixers are supplied with two pumps, but a special powder feeder must be fitted to add the powder hardener. Fine powders are prone to bridging in the supply hopper, and achieving a constant controlled feed is not easy. Powder hardeners, being based on slags or cements, are variable in activity from batch to batch, and their activity decreases with storage time, while liquid hardeners have a long shelf life and have closely controllable reactivity. Dicalcium Silicates (Slag/Cement Hardeners). Many calcium compounds, including

No-Bake Sand Molding / 573 lime, gypsum anhydrite, calcium silicates, and calcium aluminates, react with sodium silicate, causing it to gel. Some of these materials react so fast with sodium silicate that they are difficult to use as foundry binders, since the working time of the mix is too short, but some slags and cements that contain calcium silicates and aluminates give controllable setting reactions. Dicalcium silicate (2 CaO  SiO2) in particular is a useful gelling agent for sodium silicate, since its initial reaction rate with sodium silicate is slow, giving a good work-time-tosetting-time relation. Dicalcium silicate is a major constituent of a number of metallurgical slags as well as certain types of portland cement. Unfortunately, while cements are formulated to have constant setting properties with water, their setting time with sodium silicate may vary widely from one batch to another. The most common source of good-quality dicalcium silicate is ferro-chrome slag. The highly basic slags used in ferro-chrome production are known as falling slags due to their property of collapsing to a fine powder as they cool, owing to a large volume change taking place in the dicalcium silicate crystals as they pass through a phase change. Ferro-chrome slags can be used as made or they can be ground further. A synthetic form of dicalcium silicate is also available for use in foundry binders. This has higher purity and is more active than ferro-chrome slags. The reactivity of the dicalcium silicate to sodium silicate depends on its crystalline form, purity, and particle size, fine particles being more active. Dicalcium silicate slowly loses activity through storage in moist conditions. Sodium silicate additions are usually 3 to 5% (on sand weight), and silica/soda ratios of 2.4 to 2.9 are commonly used. Dicalcium silicate additions are usually 1 to 4% (on sand weight), depending on its reactivity and the setting time required. The setting time varies with:
Reactivity of the powder
Percentage of powder added
Ratio of silicate (higher-ratio silicates set

faster)


Temperature

Setting times are usually 10 min to 2 h. The reaction of dicalcium silicate with sodium silicate is not fully understood, but no heat or gasses are evolved, and the process is very safe. Ferro-Silicon (Nishiyama Process). The ferro silicon should contain between 70 and 80% Si and be finely ground to 300 British standard sieve mesh. The reaction between the silicate and ferro silicon is exothermic. Hydrogen gas is released, and fatal explosions have been reported. Because of this danger, the process is little used. Silicate Organic Esters. This self-setting system is used for the manufacture of small

and heavy castings over a wide range of metal compositions, being especially suited to jobbing and semijobbing work of the larger type. The process is widely used for molding as a facing material, subsequently to be backed up by green sand. Binder System. Hardening is achieved by a two-part chemical reaction between sodium silicate and organic esters. The most significant reaction is a decomposition of the esters by the alkali to form alcohol and an acid, which produces gelling of the sodium silicate. Silicate additions of 2 to 4.5% are commonly used. The higher-ratio silicates of 2.5:1 to 2.8:1 perform better with esters. Ester additions of 10 to 12.5% of the silicate are used. Organic Esters. Many organic esters can be used to gel sodium silicate. The speed of reaction depends on the particular ester used and the ratio of the silicate. For example, commonly used esters are:
Glycerol diacetate (diacetin)—fast
Ethylene glycol diacetate—medium
Glycerol triacetate (triacetin)—slow

These esters, and others, can be blended to give a wide range of setting times. Additions above the recommended levels will not improve setting times but could impair could core strength properties. Temperature of the mixed sand greatly influences setting times, and control of the mixed sand temperature is vital if consistent production speeds are to be achieved. Shelf Life. The storage life of molds and cores made in ester-silicate is considerably better than CO2-silicate. Humid conditions will cause molds and cores to deteriorate, as with any core binder, but storage lives of approximately 3 weeks in normal atmospheric conditions are typical. Suitable Sands. This process is used for bonding all the main types of sand. Olivine, chromite, or zircon sands are widely used in the steel industry for the high-pouring-temperature alloys. In some cases, it has been found possible to replace the most expensive materials with silica sands, because it is possible to obtain glass-hard and -smooth surfaces on the cores and molds. Sand Mixing. Continuous trough mixers are particularly suitable for this process, especially for large work. Batch mixers can be used, but where more than one mix is required for the same core or mold, there may be some danger of partial setting between the mixes. Problems of inconsistent setting times can result from heat generated during milling of the sand, particularly with batch mills, but during cold conditions some advantage may be gained by purposely milling the sand to raise the temperature to a good working level. In recent times, sand temperature controlling units have been introduced into continuous sand mixing plants, and these successfully control mixed sand temperatures to within a few

degrees. Cleanliness of the mill is most important. Contamination by other processes, particularly those that contain acid hardeners, will seriously affect hardening of the sand. Bench Life. The bench life of mixed sand is regarded as half the setting time. Ramming. Freshly mixed sand has good freeflowing properties, enabling good compaction to be achieved readily. For larger work, hand packing and tucking of sand supplied by a continuous mixer is the most popular method used for producing both cores and molds, although most other means of compaction are also used. Vibration techniques in particular give excellent results, producing cores and molds with very good compaction and high surface strength, although problems of ram-off may be experienced. Decoring. Because higher-ratio silicates are used, core removal is generally better than the CO2 process. Break-down additives can be used, provided they do not interfere with the setting of the sand.

Air-Setting Organic Binders The most widely used of the no-bake binder systems uses catalysts in the form of a liquid. The refractory sand is coated with the catalyst, and then the binder is added. The chemical reaction between the binder and catalyst starts immediately on contact. The action can be shortened or lengthened by adjusting the amount of catalyst and the temperature of the sand, binder, or catalyst. The no-bake air-set sand systems are mixed in a continuous mixer and deposited directly from the mixer into the core box or into a flask and over the pattern (Fig. 1). Although these sand systems have good flowability, some form of compaction or vibration may be used to increase mold density. When the cores or molds have sufficiently cured, they can be stripped from the core box or pattern without distortion. The cores and molds are then allowed to thoroughly cure, after which a refractory wash or coating can be applied to help protect the refractory sand from the heat and eroding action of molten steel as it enters the mold cavity. As noted, the no-bake air-set systems do not lend themselves to high-production rates for cores or molds. This is primarily due to the time necessary to thoroughly cure the core or mold. Strip times can range from a few minutes to an hour, depending on the system used. Castings of all sizes and complexity can be made using these processes. In the case of very complex casting shapes, the mold can be assembled with cores to shape the outside surfaces and internal passageways in the casting. Some advantages of no-bake air-set cores and molds include:
Wood and, in some cases, plastic patterns

and core boxes can be used.

574 / Expendable Mold Casting Processes with Permanent Patterns
Due to the rigidity of the mold, good casting






dimensional tolerances can be achieved. Sand mixtures have good flowability. Casting finishes are very good. Less skilled molders and coremakers are required. Most of the systems have excellent shakeout. Cores and molds can be stored indefinitely. The disadvantages are:


Since the binders used in most of these sys-







tems are derived from crude oil, their prices fluctuate and can make these expensive systems to use. The time required to properly cure the molds and cores may slow down production. Careful attention must be paid to the binder and catalyst levels to prevent hot tears due to poor collapsibility of the cores and molds. Disposal of used sands is becoming a problem, causing most foundries to implement reclamation of these sands. Many of these systems give off gases during mixing, molding, curing, and pouring, which necessitates collection and removal.

The three members of the air-setting organic binders are:
Furan no-bake resins
Phenolic no-bake resins
Urethanes (oil and phenolic)

Furan Nobake Furan nobake is the generic name of a family of air-setting foundry binders based on furfuryl alcohol that cure at room temperature. The binder contains two liquids: the furfuryl alcohol and an acid that causes the furfuryl alcohol to resinify. Although the main ingredient of part 1 is mostly furfuryl alcohol, it is modified with urea, formaldehyde, additives such as phenolic resins, or extenders. The urea-modified resins are less costly; however, the urea introduces nitrogen into the system in direct proportion to its concentration. Higher nitrogen levels can cause pinholing problems with many ferrous alloys; however, this is not of great significance for copper-base alloys. Part 2 is an acid catalyst for the furan nobake system. Furan no-bake part 2 ingredients are usually proprietary mixtures of acids, solvents, and sometimes a color-coded dye. The two main classes of catalysts are either phosphoric acid (75%, or 85% phosphoric, and perhaps some sulfuric acid in water) or an organic sulfonic acid (a solution of benzenesulfonic acid, or BSA; toluenesulfonic acid, or TSA; or other sulfonic acids, usually in combination). Solvents for these organic acids are water and methanol. The strength and concentration of the acid affects the speed of the curing reaction.

The catalytic effect of part 2 depends on its residual activity after neutralizing foreign materials, sand ADV, and even acids and basic oxides from various reclamation procedures. For this reason, the acid catalyst should be added to the sand first and thoroughly dispersed. In this manner, when part 2 is added, there will not be excessive concentrations of acid to accelerate the cure rate in local areas. Mixing part 1 and part 2 together will result in an exothermic reaction that can result in an explosion. Sands and Mixing. Washed-and-dried silica sands (and zircon) are readily used with furan no-bake binders; olivine, however, due to its alkaline nature, cannot be recommended. Binder requirements are increased with the fineness or angularity of the sand, as well as by the presence of fines or clays. Mixing is readily accomplished with all the common types of equipment; however, batch mixing is not normally recommended for sand formulations having short working times. Since the hardening process begins as soon as the binder is added to the catalyst-coated sand and is continuous and irreversible, it is important that prepared sand mixtures be used as rapidly as possible. Continuous mixers placed immediately adjacent to the molding station are advisable. The average level of resin binder required is approximately 1% (based on sand weight) and may vary from 0.8 to 1.4%, depending on sand characteristics, mixing equipment, and property requirements. Catalyst levels are usually between 15 and 45% of the weight of resin present in the mix and are varied to control the cure rate. Sand temperature has a significant effect on cure rate and should ideally be held between 21 and 27  C (70 and 80  F). The binder supplier should be consulted if sand temperature control is a problem. Working time and strip time are both measures of the cure rate for self-hardening sand mixtures. Strip time is the duration between mixing and removal of the cured sand from the pattern or core box. Working time is the allowable delay between mixing and utilization of the mixture. Moving or working the sand after this period will break bonds and result in reduced strength. The working time for furan nobakes is usually approximately 50% of the strip time. Strip time can be anywhere from less than 1 min to several hours, depending on the kind and amount of catalyst used and the sand temperature. This system can tolerate the presence of some moisture; curing, in fact, generates small amounts of water within the mold or core. The presence of excessive moisture in the mix will prolong the curing time.

Phenolic Nobake Phenolic no-bake (acid-cured) systems are likewise composed of a liquid resin binder and a liquid catalyst. Unlike the furans, the phenolic resins contain water in amounts ranging from

approximately 8 to 25%. Higher water levels result in lower cure rates and lower tensile strength. The principal type of catalyst used with the phenolic no-bake system is a mixture of sulfonic acids (BSA, TSA, and others). Small amounts of sulfuric acid are sometimes also added to these acid mixtures to accelerate curing. Unlike the furan nobakes, phenolic nobakes should not be catalyzed with phosphoric acid, which results in slow curing and low properties. Typical mixtures with phenolic nobakes consist of approximately 1.5% resin (þ0.5%)  and catalyst levels ranging from 20 to 45% of the weight of resin employed. Recommended procedures for mixing and use, as well as control of cure rates and properties, are essentially the same as discussed for furan nobakes. In general, the strength of sands bonded with phenolic nobakes is somewhat lower than for the furan nobakes.

Alkyd Oil Air-setting urethane no-bake systems are of two different types: alkyd oil urethanes and phenolic urethanes. The former was introduced in 1965 and is the third-highest-selling binder system. Alkyd oil urethane binders are available precatalyzed for a predetermined work/strip time as a two-component system or as a threecomponent system; this provides flexibility through on-site cure speed adjustment by catalyst metering. In either version, the workto-strip-time ratio is approximately 1 to 2; that is, a sand mixture with a 25 min work time can, in most cases, be stripped in 50 to 60 min. While product names and designations vary from one supplier to another, the alkyd (oil) component is often referred to as part A, part 1, or some other primary designation. This first component constitutes the bulk of the resin used and is referred to in the further discussion of this system as part A. The polymeric isocyanate component is, again, variously named and is normally used at levels of 18 to 20% (based on part A). In further references, this isocyanate component is designated as part C. The third component (if purchased separately), the catalyst, is used at levels of from 2 to 10% based on part A and contains metallic driers that act as urethane catalysts. This component is referred to as part B. Curing. When all three component parts are mixed with the sand, two reactions take place. First, the polyisocyanate (part C) crosslinks with the oil-modified alkyd (part A) to produce a urethane bond with sufficient strength to permit stripping and handling. This stage does not require oxygen or removal of reaction products such as water. The second stage of the curing reaction requires oxygen for the further polymerization of the alkyd resin. Remaining solvents also evaporate at this stage. Because this stage is

No-Bake Sand Molding / 575 oxygen-dependent, section size and foundry conditions can affect the curing time. The metallic driers assist the oxygenation reaction as well as the isocyanate reaction, thus controlling the overall cure rate of the alkyd (oil) urethane system. The oil-modified alkyd resin (part A) contains the hydroxyl groups. The percentage of part A used in a typical sand mixture is 1.5% (based on the weight of the sand), although this may vary þ0.5% depending on sand type, sand  purity, and casting requirements. An increase or decrease in part A will correlate directly with the ultimate strengths the system will develop, if parts B and C are kept in the same relative proportions to part A. Part B (the catalyst) contains metallic driers that determine the speed of the reaction as well as the overall curing characteristics of the system. Amines up to 2% are added to the metallic driers when fast strip times are required. An increase in part B will increase the speed of the reaction, thus decreasing the work time. The faster the alkyd (oil) urethane resin system selected, the higher the early tensile strength. The use of a metering pump is recommended for controlling the catalyst addition (part B), especially when low flow rates are required (under 0.8% based on weight of the sand). If a metering pump is used, the catalyst should be pumped into the part A resin line just before entering the mixing chamber. This allows the catalyst to become dispersed in the alkyd (oil) resin and thus does not result in soft spots in the mold or core. In many instances, parts A and B are used premixed, thus eliminating pumping problems associated when small amounts of part B are required. The use of a premix has advantages when sand temperatures can be maintained at a constant temperature. Otherwise, the percentage of part B in the premix may require frequent adjustments to compensate for varying in-plant temperatures. Part C is the polyisocyanate component that co-reacts with the resin, part A, and should not be considered a catalyst because it undergoes a chemical change to form the urethane polymer, and a specific quantity is needed to obtain desirable curing characteristics. If less than this quantity is used, the curing rate as well as the ultimate tensile strength will be impaired. Too large a quantity of part C containing 10% N will increase the possibility of gas defects due to unreacted nitrogen complexes. The recommended amount of part C is between 18 and 20% (based on part A), to account for mixing deficiencies, moisture in the sand, and typical foundry variables. Part C, particularly sensitive to moisture, hardens after prolonged exposure, forming a skin at the air contact surface. Tightly closed containers, moisture absorbers, and nitrogen blankets are commonly used to keep moisture out of part C. Sand. In a typical sand mixture of 1.5% part A (based on weight of sand), 6% part B

(based on weight of part A), and 20% part C (based on weight of part A), the cured core will contain 0.03% N based on sand weight. Although alkyd (oil) urethanes contain nitrogen, the cured system structure does not evolve decomposition products of nitrogen as readily as do urea-containing binder systems. Mixing efficiency is extremely important in the development of the optimum cure strength of the binder system. The viscosity of the various liquid components should be closely monitored. Changes in viscosity can affect pump delivery, mixing efficiency, and ultimate core density and core quality. The variables that have a detrimental effect on this binder system performance are:
High sand temperatures (Heat speeds up the

reaction and reduces the bench life of the mixture.)
High humidity (Moisture reacts with part C, decreasing the flowability of the mixture.)
High ADV of the base sand (Alkaline impurities speed up the reaction.)
Aeration of the sand mixture causes solvents to evaporate, decreasing the flowability of the mixture. Alkyd-oil urethane systems are generally less sensitive to sand types than are other no-bake binder systems because of the nature of the curing reaction. Silica sands and specialty sands such as olivine and zircon are compatible with alkyd-oil urethane binders. For optimum performance, the following recommendations are offered for sand selection:
A rounded grain shape provides the best prop-

erties for a given resin level (tensile strength, scratch hardness, flowability, and density).
Sand fineness between AFS 45 and 75 gfn is preferred. Finer sands increase binder requirements and decrease flowability.
Strongly acidic sands retard cure rate, while alkaline sands accelerate the rate.
The lower the moisture, the faster the cure rate and vice versa. Water contamination of the binder is significantly more detrimental to initial cure than is water contributed by moist sand. Other system variables that will significantly affect the properties and productivity of the alkyd (oil) urethane systems are: Combustibles. Loss-on-ignition (LOI) should be kept below 2% maximum (an issue for mechanical reclamation). Excessive LOI results in higher binder levels, reduced flowability, and may be correlated with casting defects. Temperature. The alkyd-oil urethane reaction is exothermic; thus, any increase or decrease in sand temperature will affect the cure rate. For rapid cure rates at minimal catalyst levels, select a sand temperature between 27 and 32  C (80 and 90  F) and maintain this tempera ture within  þ3  C (þ 5 F) for reproducible cure rates. Approximately 0.5  C (1.0  F) change in

temperature increases or decreases the strip time by 1 min. It is also recommended to maintain pattern temperatures at approximately 30  C (86  F) and resin temperature at 25  C (77  F). Segregation of the grain sizes in the sand during handling and storage affect the tensile strength, flowability, and casting quality at a given binder level. Additives. For most applications, additions of 1.5 to 2% red or black foundry-grade iron oxide will improve the hot strength and minimize surface porosity as well as excessive lustrous carbon deposits. Pouring temperature of the metal, gating practice, binder level, pouring time, and base-sand type all influence the formation of lustrous carbon. Effect of Baking. Oven baking of alkyd-oil urethane-bonded core/molds at 149 to 204  C (300 to 400  F) from 2 to 24 h (depending on casting size) promotes further crosslinking of the binder system, resulting in higher tensile strengths and improved erosion and hot strength properties, while minimizing the volatiles in the sand mixture. Baking at temperatures above 200  C (400  F) is detrimental and should be avoided. Washes. Almost all washes can be applied. Water-based washes should not be applied until 24 h after stripping or after baking, and should be dried immediately after application. Alcohol-based washes should be lit-off as soon as possible; however, volatile organic compound emissions are reducing sales of this class of washes. Washes should penetrate the surface by at least two sand grains. Since this will reduce the measured permeability of the surface to zero, venting becomes important. Vents should be narrow enough (1.6 mm, or 1/16 in.) that metal does not penetrate and cause surface sinks.

Phenolic Urethanes Phenolic urethane no-bake (PUNB) binders were introduced in 1970 and are capable of producing strippable cores or molds within 1 to 15 min of filling the pattern or corebox. Phenolic urethanes are the most significant nobake in North America. Phenolic urethane nobakes are three-component binders. Part 1 is a phenolic resin blended with solvents. Part 2 is a blend of polymeric methylene diisocyanate and solvents. Part 3 contains the catalysts, which are various blends of pyridine derivatives and solvents, to regulate the reaction between parts 1 and 2. The unique feature of PUNBs is the curing reaction. Basically, there is a delay in the curing reaction after the three components are mixed with the sand. This is the work time of the mixture. This delay enables the sand mixture to remain free-flowing until the curing reaction begins. Work time can be as high as 75% of the strip time. Once the reaction starts, it is extremely rapid and thorough. Because this cross-linking

576 / Expendable Mold Casting Processes with Permanent Patterns reaction involves no side products, the curing rate is constant throughout the entire sand mass. The center of the core cures at the same time the edges do, and the same strength is developed in the core and at the edges. There is no prolonged plastic stage during the cure, nor are there soft spots after the cure is completed. Two-thirds of the system ultimate strength is attained after approximately 4 h, and ultimate strengths are achieved at 6 to 12 h after stripping. Since the curing reaction occurs so rapidly, high production rates and rapid equipment turnover are possible. The cure speed of a particular sand/resin mixture is dependent on the percent of catalyst, the catalyst type, and the sand temperature, ADV, and moisture content. The total binder level for PUNBs is based on the weight of the sand and is usually in the range of 0.8 to 1.8 %. If the castings to be poured are sensitive to gas defects, then the normal part 1 to part 2 ratio of 50 to 50 should be shifted to 52.5 to 47.5 or even 60 to 40. This should be discussed with the binder supplier, and some loss of tensile strength can be expected. It is generally recommended that equal amounts of part 1 and part 2 be added to the sand to reach the binder level desired. Catalyst usage is generally 2 to 9% of the weight of part 1, depending on the strip time desired, the type of base sand, the sand temperature, and the type of catalyst being used. If catalyst usage is outside the 2 to 9% range, then a catalyst with a different reactivity should be used. Sand and Mixing. Recommendations for obtaining optimum performance properties with PUNB resin systems are:
Sand selection: Round-grain silica sands

yield optimum performance properties. Specialty sands, including zircon sands and chromite sands, are compatible with PUNBs, as are most subangular lake and bank sands. Acid impurities tend to slow the curing, while basic contamination may speed up the cure.
Sand fineness: An AFS 50 to 80 gfn is recommended with a minimum of fines (material on the U.S.sieve numbers 200,270, and PAN). Poor sand reclamation or sand segregation can result in excessive fines and the need for more binder because of the increased sand surface area.
Sand moisture: Less than 0.25% moisture content is recommended. Moisture levels above this will result in slower cure rates, lower strengths, and variable sand flow rates.
Sand temperature: While the ideal sand temperature for PUNBs is 27 to 32  C (80 to 90  F), sand as cool as 16  C (60  F) or as warm as 38  C (100  F) is usable. The main concern should be the maintenance of a consistent sand temperature not varying more than  þ3  C (þ5  F). When such a condition  exists, cure speeds are consistent and predictable, and the need to adjust catalyst

levels during core/mold production will be minimized. If sand heaters and coolers are needed to maintain this temperature range, then they are strongly recommended.
Sand additives: The use of 1.5 to 2% foundry-grade red or black iron oxide may be required (typically steel) for optimum casting finish from improved hot compression strength and the elimination or reduction of surface porosity and lustrous carbon casting defects. The PUNB resin system can be mixed in most foundry batch mullers, auger-type continuous mixers, and the newer generation of high-speed intensive mixers. Part 1 and the appropriate catalyst are added to the sand first. Because of the small amount of catalyst used, it is recommended that the catalyst be preblended with part 1 prior to addition to the sand. This will ensure that the catalyst is efficiently distributed and that the binder system will perform in an optimum manner. Zero-retention continuous mixers are ideal for the PUNB system, as are conventional batch mixers, provided that only the amount of mixed sand that can be used before the work time expires is prepared at a given time. Mixing times for batch mixers are best determined at the point of use. Overmixing of sand and resin may result in lower tensile strengths due to partial curing of the mixture. Undermixing may result in poor distribution and lower tensile strengths. Properly mixed sand will discharge from a muller as a free-flowing mass that can be distributed and compacted easily, since essentially no green strength exists. When continuous mixing is employed, gear pumps accurate to  þ1% are recommended to pump the two binder components 1 and 2. The preblending of part 1 and catalyst is conveniently accomplished by mating the catalyst pump exit hose to the part 1 hose immediately upstream from its point of discharge into the sand. Part 1 and catalyst should be added to the sand prior to part 2. However the mixing is accomplished, calibration and maintenance of the mixing equipment (mixer blades, sand flow controls, pumps) is essential to a smoothly operating no-bake system. Weighing and pumping devices should be checked for accuracy at least once per shift. Part 1, part 2, catalyst, and sand delivery rates should be calibrated to conform to a predetermined process control plan.

Gas-Cured Organic Binders Vapor-cured organic binders are hybrids since the processes are based on chemistries that are similar to the acid-catalyzed urethane resin system already discussed. When mixed with sand and either rammed or blown into a mold or core box, these binders are cured by introducing into the sand a vaporized liquid or gaseous catalyst at room temperature. No external heat is required.

Currently, there are two gas-cured organic binder systems widely in use: the phenolic urethane-amine system and the furan-SO2 system. Both of these processes generate an offensive odor due to the gases used; therefore, the exhaust gases must be collected and treated to remove the odor. For the most part, these two processes are used to make cores, but there is no reason why they could not be used to make molds. Some advantages of the amine and SO2 process are:
The core sand mixture is very flowable and

thus requires little energy for compaction.


Shakeout and collapsibility are excellent.
Dimensional accuracy is very good.
Casting finish of both the external and cored

surfaces of the casting is excellent.


These processes are used for high-produc-

tion runs since the coremaking cycle is very short.
Cores or molds made with processes have excellent shelf life. The disadvantages include:
Because the majority of the core sand used

in these processes is blown into the core box, the core box must be made of metal.
Special scrubbing or treatment equipment is required to control the offensive odor as the gas catalyst leaves the core box.
Special seals are required at the parting surfaces of the core box to contain the gas catalyst and to prevent it from escaping into the air. The phenolic urethane-amine system has experienced phenomenal growth since its introduction in 1968. Similar to the phenolic urethane no-bake process, the vapor-cured version consists of a three-part resin system:
Phenolic resin
Isocyanate
Amine catalyst

Total resin based on sand varies from 0.8% for aluminum to 1.5% for cores in ferrous castings. Tertiary amines such as triethylamine and diethylamine are used to set the sand. These low-boiling-point amines are vaporized in a carrier gas (6 to 12% in the carrier gas) and introduced to the sand mixture to achieve an almost instantaneous cure. Since liquid amines are flammable, inert carrier gases such as nitrogen and carbon dioxide are normally used to minimize fire or explosion hazards. Typical cure rate values range from 1 kg (2 lb) of sand per second to 4.5 kg (10 lb) per second. The final step in the curing process involves the use of a high-pressure air purge to distribute the amine to all portions of the core box and ultimately flush residual amine vapors from the core cavity. These vapors are collected from the sealed core box and

No-Bake Sand Molding / 577 exhausted into an acid-bath chemical scrubber where they are neutralized. Sand chemistry, temperature, and moisture levels have a significant effect on the system bench life. Basic sands, such as lake sand, bank sand, and olivine, contain alkaline impurities that decrease the bench life of the system. On the other hand, neutral sands normally have a bench life in excess of 3 h. High temperatures result in solvent evaporation, which lowers bench life and may result in friable core surfaces. A usable sand-temperature working range is (10 to 38  C) (50 to 100  F). High moisture levels in the sand also result in decreased bench life and ultimate strengths. In addition to reacting with the isocyanate, moisture decreases the adhesion between the resin film and the sand grain, causing a poor bond. Moisture levels up to 0.2% can be tolerated but should be maintained at minimal values. Acrylic-Epoxy-SO2 System. An earlier furan-SO2 system was basically a gas-cured furan system. Instead of using a liquid catalyst, it relied on the passage of SO2 gas through the uncured sand mix in the presence of an oxidizing agent (in this case, organic peroxide) to create the acid. This produced an instantly cured core or mold exhibiting all the characteristics of a conventional furan no-bake system. This system has been largely replaced by the acrylic-epoxy-SO2 system, although some nonferrous foundries still use the furan system, because core shakeout can be critical. The acrylic-epoxy-SO2 system offers a number of advantages for the foundries: specifically, no formaldehyde or isocyanates in the decomposition products, excellent shakeout properties, reduced hot tearing and veining defects, higher productivity from reduced cleaning and start-up costs, and consistent properties due to long bench life. Resin and Curing. The epoxy-SO2 process uses a two-component system. Part 1 is an epoxy resin with an organic hydroperoxide. Part 2 is a mix of acrylic resin, epoxy resin, additives, and a solvent. The sand is coated with the two components and exposed to the SO2 gas, which starts a free-radical reaction with the acrylic. The acrylic component cures rapidly, which gives the system early handling strength. The epoxy reacts slowly, allowing for good pattern release and, after 24 h, good hot strength. Resin consumption ranges from 0.5 to 1.5% based on sand and application. Sulfur dioxide (SO2) is supplied in cylinders as a liquified gas when at a pressure of 0.24 MPa (35 psi) at 25  C (77  F). Gas generators are available to provide blends of SO2 and nitrogen (N2) for curing. Consumption will normally be less than 4 kg (9 lb) per ton of cured sand. The use of SO2/N2 blends uses less SO2, and less SO2 is left in the mold or core. The SO2 must be distributed evenly and finally purged from the sand.

Purged air must be free of oil and water, and the best results are obtained by using a heated purge air. Typical temperatures are 35  C (95  F), and commercial air heaters are available for this purpose. Excess SO2 is removed from the purge air in a chemical scrubbing unit, typically a packed scrubbing tower where the SO2 gas is absorbed by reaction with 5% sodium hydroxide solution. The resultant sodium sulfate solution has a pH of 8.5 and can usually be discharged into municipal sewage systems. Due to the toxic nature of sulfur dioxide, the system must be closed to prevent leakage into the working environment. Sand mixing can be accomplished either in the batch mullers or continuous mixers. The two resins are added to the sand and mixed. At this stage, the mixed sand has a very long bench life and will only cure when the SO2 gas is introduced. Curing is then instantaneous, with very high tensile strengths being obtained. Coated sand is easily handled and does not stick to the sides of hoppers or blow magazines. The long bench life allows coated sand to be left in blow magazines, with easy startup after production delays or plant shutdowns. Losses and cleaning are thus minimized. Mold/Core Making. As with the other binder systems previously mentioned, the methods used for producing cores or molds are blowing, jolt/ squeeze, hand ramming, or mechanical vibration. For high-production rates, blowing is commonly practiced, with curing taking place within the automatic cycle of the machine. Generally, any machine designed for use with a gas-curing cold box system can be used. Where manual core blowers are employed, these machines can either be converted to cure the cores within the machine, or the filled core boxes can be gassed in a conventional curing chamber. Where jolt/squeeze, hand-ramming, or mechanical vibration methods are employed, the filled core or mold is normally placed in a suitably sized chamber for curing. To obtain the most cost-effective operation, it is most important to have good sealing. Correct sealing ensures the optimum curing of the mold/core and also prevents excessive use of the SO2 gas, thus extending the life of the solution within the scrubbing unit. As with any gas-cured cold box system, the SO2 processes are dependent on the ability of the curing gas to reach every area of the core or mold and then be purged clean in as short a time as possible. To produce a positive pressure within the cavity, the venting area must be sufficiently smaller than the gas inlet area. Although vent areas and pressure drops are important, they can only work if the vent placements within the core or mold cavity are correct. To prevent gas bypass and uncured areas within an otherwise well-cured core or mold, vents should be placed in the deepest areas of the core or flask and as far away as possible from the gas inlet, as is true for any gas-cured system.

Sand Reclamation and Reuse Strong economic incentives exist for the reclamation and reuse of foundry sands, both as a means to offset higher material costs and to avoid the sometimes exorbitant expenses of waste sand disposal, particularly if leachate from the sand includes formaldehyde, phenol, aromatics, and metallics such as lead. For this reason alone, limited reclamation may make economic sense to mitigate the high disposal costs posed by the growing use of chemically bonded sands with complex chemical binders and partially pyrolyzed binders, whose behavior in landfill sites is not environmentally attractive. Most foundries can only achieve high sand recovery rates by reclaiming green sand (although this may be complicated by the porous layers of burnt clay on the grains). Green sand is typically reused and recycled until the additions of new and core sand dictate that room be made in the sand system, typically, from 4 to 20 cycles, depending on the relative sizes of cores and molds (see the article “Green Sand Molding” in this Volume). Unused chemically bonded sands often require different reclamation procedures. Grain size and screen distribution, cleanliness and refractoriness of the sand, ADV, LOI, and moisture content must be controlled. These are critical parameters for many of the chemical binders and processes. If sands other than silica are used, mixed sands can affect not only the reclamation processes but the subsequent reuse of the reclaimed sand or sands. (Heavy sands can accumulate in fluid beds and reduce the residence times of lighter sands. Chromite reacts with silica sand at higher temperatures.) In the case of shell sand, while the process is more forgiving of impurities than most, very few foundries coat their own shell sand. There are basically three commercial systems for sand reclamation: the dry method, the wet method, and the thermal method. Some systems may be comprised of more than one of these basic methods, depending on foundry requirements as dictated by the binder systems in use. The primary requirement of any reclamation system is to remove the resin coating around the sand grains. This involves abrasion and attrition to break the bond, as well as classification to remove the fines that are generated. Selection of a system depends greatly on the type of organic binder to be removed from the sand grains. Mechanical processing is the first step in most reclamation processes. Garbage, castings, and rigging must first be separated to protect equipment from excessive damage. Cores and molds are mechanically reduced (by crushing, grinding, abrasive blasting, and so on). Most mechanical systems require the sand be reduced to near-grain size prior to processing to improve efficiency.

578 / Expendable Mold Casting Processes with Permanent Patterns Early reclaimers subjected the sand to a scrubbing action that abraded the adherent layers of binder on the sand grains by upward airflow and impingement on a plate and the counterflow as sand falls back. Scrubbing systems need dust collection equipment to remove fines, particularly because the fines contain a high proportion of the removed binder. In some cases, afterburners and recycling systems are necessary.

Wet Reclamation Systems Wet reclamation systems were used for claybonded system sands in the 1950s but are now used for silicate binder systems only. Silicate systems are very difficult to reclaim by dry processes and are impossible to reclaim in thermal systems. This is because silicate is an inorganic system that melts rather than burns in the furnace. The complete system includes lump-breaking and crushing equipment, an attrition unit, wet scrubber, dewatering system, and dryer. The systems require approximately one pound of water per pound of sand reclaimed, and, in some cases, the water can be discharged directly into municipal sewer lines. Most installations allow 100% reuse of the reclaimed sand, with makeup sand as the only new sand addition. During wet reclamation, the thoroughly crushed shakeout sand is subjected to mechanical scrubbing in the presence of water. The solvent action of the water is highly important here; thus, the wet reclamation system is not well suited for sands that are bonded with insoluble organic resins. Clay-bonded molding sands respond well to this, however, since the residual clay goes into suspension readily and is washed away. In spite of the relatively high water requirements of wet reclamation and the increasingly severe Environmental Protection Agency (EPA) regulations pertaining to waste water disposal, wet reclamation has been receiving renewed attention recently due to growing interest in the use of inorganic binders such as silicates. Silicates respond best to wet reclamation because of the need to reduce residual amounts of Na2O to low levels. The use of electrolytic polymers to precipitate salts below their saturation levels now makes the treatment and recirculation of the system water supply possible. Wet treatment of alkaline phenolics rapidly removes the alkaline material, with the bulk removed in initial small washes. The British Cast Iron Research Association also found acid wash to be beneficial. It is apparent that techniques from ore dressing will find applications in sand reclamation.

Dry Reclamation Systems Many factors determine the degree of cleanliness required in a reclaimed sand. These factors include the type of resin system used for

rebonding, the sand-to-metal ratio, the type of metal poured, the condition of the reclaimed sand, the type of new sand used, and the ratio of new sand to reclaimed sand. Attrition reclaimers break down the sand lumps to a smaller grain size. Some fines are removed, but the binder is not removed completely from the surfaces of the sand grains. In most cases, these units produce a sand that requires a higher concentration of new sand when the attritor is coupled with a sand scrubber, as described subsequently. Additional scrubbing is sometimes required, and there are basically two types of scrubbers: mechanical and pneumatic. Selection between the two types is primarily a question of wear, ease of maintenance, and energy consumption, because the units provide comparable performance in terms of scrubbing action. Pneumatic Scrubbing. Figure 4 shows one cell of a pneumatic scrubber. Sand is introduced by gravity at the top of the unit, and it flows down around the blast tube. High-volume lowpressure air from a turbine blower flows through the nozzle and lifts the sand up through the blast tube to the target plate. The sand grains undergo intense attrition in the tube by impacting on each other; further attrition occurs at the target as binder is removed from the sand grains. These fines and resin husks are then removed from the system by a classification dust collection system. Scrubbed sand falls from the target and is deflected to the next cell or is kept within the same cell for further scrubbing. The degree of cleanliness attained is determined by the retention time in the cells (controlled by the deflector plate) and the number of cells. Sand exiting the final cell should be screened to remove any foreign material that may be present in the refuse sand. A conscientious maintenance program must be followed for these scrubbers. High-wear parts are the impellers, targets, and tubes. Improperly maintained units will not yield a consistent reclaimed product. The dust collection system is of equal importance and must also be properly maintained. Excess fines in the sand increase residual binder and decrease sand permeability. Mechanical Scrubbing. There are two types of mechanical scrubbers: horizontal and vertical. Figure 5 shows a horizontal scrubber. Clean sand (crushed, with metal removed) is fed into the center of the unit and thrown against the target ring at a controlled velocity by the impeller. Some sand-on-sand attrition takes place, but the intense scrubbing occurs at the target ring. The exhaust plenum surrounds the target ring to remove dust and binder husks. These units can be arranged in sequence for additional scrubbing. Figure 6 shows a vertical mechanical scrubber. Sand enters the center of the impeller and is thrown upward at a target plate. Attrition takes place as the sand hits the target. The sand falls away and exits into an air wash separator, where the fines are exhausted from the sand.

Fig. 4

One cell of a pneumatic scrubber

For additional scrubbing, this unit can also be operated in series, as shown in Fig. 7. As with pneumatic scrubbers, the impellers and targets are high-wear parts. The units must be properly maintained to yield a consistent product. The dust collection system must also be properly maintained. Process Controls. There are a number of tests that can be performed on the reclaimed sand from the scrubber. Two of the more important tests are LOI and screen distribution. The LOI test involves firing a 50 g sand sample at approximately 980  C (1800  F) to determine the amount of carbonaceous material burned off. These tests are good measures of the operating efficiency of the classifiers. A buildup of sand on the fine screens, such as 200, 270, and 300 mesh, is usually associated with an increase in LOI, which is a problem that is most likely attributable to the classifier. An increase in LOI without the accompanying shift in screen distribution will indicate that binder removal in the scrubber is low and that maintenance may be required on the unit. Most manufacturers supply a troubleshooting guide.

No-Bake Sand Molding / 579

Fig. 5

Fig. 6

Horizontal mechanical scrubber

by three factors. The first is the temperature of the sand at shakeout. This will vary with the sand-to-metal ratio, the type of metal poured, and the amount of lump reduction from the shakeout. Secondly, heat will be generated within the reclamation system itself. All the components will gradually heat up, thus increasing the temperature of the sand. The third factor that affects the temperature is the season of the year. In the summer, for example, there are more problems with hot sand. Therefore, based on these factors, a sand heater or cooler may be a necessary addition to the total system. Blending of the sand from the scrubbers with new sand is very important. Blending is done to help replace the sand that is lost in the casting and reclamation processes and to limit the effects of residual binder on the sand grains. The sand should be measured and blended thoroughly to ensure that there are no concentrations that would cause undesirable effects on the castings. Most of the sand reclaimed from these units can be used in blends of 80% reclaimed sand to 20% new sand. Again, the exact ratio will be determined by the metal poured and the type of resin system used. Vertical mechanical scrubber

Temperature control is critical when working with chemically bonded sand. The heat of the reclaimed sand at the discharge point is affected

Thermal Reclamation Thermal reclamation of chemically bonded sand is achieved by bringing the sand to a

sufficiently high temperature over the proper time period to ensure complete combustion of the organic resin and material in the sand. If the proper temperature and atmosphere are not maintained in the unit, the organic resin will volatilize and send volatile organic carbons up the emission stack. This would be a violation of clean air standards. Currently, the two types of thermal units available for sand reclamation are rotary drums and fluidized bed furnaces. For these units to be cost effective, they must be operated 20 to 24 h per day and make maximum use of energy recuperation techniques. The rotary drum has been in use since the 1950s for the reclamation of shell and chemically bonded sands. The direct-fired rotary drum is a refractory-lined steel drum that is mounted on casters. The feed end is elevated to allow the sand to flow freely through the unit. The burners can be at either end of the unit, with direct flame impingement on the cascading sand; flow can be either with the flow of solids or counter to it. In indirect-fired units, the drum is mounted on casters in the horizontal position and is surrounded by refractory insulation. Burners line the side of the drum, with the flames in direct contact with the metal drum. The feed end is elevated to allow the sand to flow freely through the unit, and, in some cases, flights (paddles connected by chains) are welded to the inside to assist material flow. The advantage of the rotary drum is its lower capital cost. Its disadvantages are high heat losses, use of moving parts at high temperatures, short refractory life, poor control of material flow, and poor atmospheric control. Fluidized bed units have been in use for the reclamation of clay and chemically bonded sands since the 1960s. The fluidized bed calciner illustrated in Fig. 7 consists of a cylindrical, brick-lined, vertical combustion chamber. Sand that has been crushed is taken from the surge (feed) hopper by means of an adjustable, closed screw feeder and is fed into the unit. In the bottom, a hot sand bed is kept fluidized by the use of a combustion air blower. This blower controls the output of the unit and ensures the availability of sufficient air for combustion. The fluidized bed uses two sets of burner systems. The start-up burner brings the bed of sand up to operating temperature. After bed temperature is reached, the second system is energized. This system consists of gas lances positioned around the perimeter of the unit and inserted directly into the bed. It maintains the sand bed temperature to within  þ8  C (þ15  F).  The intense mixing of the sand and the hot gases provides for combustion of the hydrocarbons and residual resin at an appropriate temperature and retention time. The fluidized bed is heated by a postcombustion zone that ensures complete combustion of the waste gases without the use of an afterburner. After the postcombustion zone, an induced-draft fan pulls the dust and hot gases from the unit through the dust collector to ensure constant throughput

580 / Expendable Mold Casting Processes with Permanent Patterns
Carbon Dioxide Molding, Forging and

Casting, Vol 5, Metals Handbook, 8th ed., American Society for Metals, 1970
Casting Copper-Base Alloys, 2nd ed., American Foundry Society, 2007, p 49–65
Cold Setting Core Materials, Br. Foundryman, July 1978, p 147–170
M. Zatkoff, Sand Reclamation, Casting, Vol 15, Metals Handbook, 9th ed., ASM International, 1988, p 351

SELECTED REFERENCES
J.J. Archibald, Benchmarking the Nobake

Binder Systems, Mod. Cast., July 28, 2005


G. Galante, O. Michilli, and R. Maspero,






Fig. 7

Fluidized bed reclamation unit




requirements and efficient emission control. The sand then exits by gravity feed to the cooling and classifying systems. The disadvantage of the unit is its higher capital cost. The advantages are long refractory life, low heat losses (energy consumption), no moving parts at high temperatures, and control of material flow, temperature, and atmosphere.

ACKNOWLEDGMENTS Portions of this article were adapted from:
M. Blair and T. Stevens, Steel Castings

Handbook, 6th ed., Steel Founders’ Society of America and ASM International, 1995




Troubleshooting Casting Defects in Nobake Molding, Mod. Cast., July 28, 2005 T. J. Gilbreath, Maximizing Productivity of Nobake Molding Operations, Mod. Cast., July 28, 2005 “Reclamation of No-Bake Organic and Sodium Silicate Sands,” Special Report 16, Steel Founders’ Society of America, Dec 1978 W.D. Scott and J. Alexander, Process Control Tips for a Phenolic Urethane Nobake Foundry (includes related article on phenolic urethane binder chemistry), Mod. Cast., July 28, 2005 W.D. Scott, W.C. Easterly, P. Lodge, and C.S. Blackburn, Foundryman’s Guide to Sand Compaction, Transaction of the American Foundry Society and One Hundred Eighth Annual Metalcasting Conference (Rosemont), June 2004, p 609–621 W.D. Scott and G.J. Vingas, Control the Quality of Your Refractory Mold Coating, Mod. Cast., July 28, 2005

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 581-597 DOI: 10.1361/asmhba0005244

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Coremaking CORES are separate shapes of sand that are placed in the mold to provide castings with contours, cavities, and passages that are not otherwise practical or physically obtainable by the mold. In general, the principles that apply to sand molding from a pattern also apply to molding a sand core. The core must be freed from the core box by moving the box away from the core in a direction generally perpendicular to the face of the core box. Loose pieces that have been placed in the core box to assist in producing the shape of the core will remain with the core sand when the box is removed from the core and must be removed from the core by a movement in the appropriate direction. The most common method of forming cores is to use a core box, made of metal or wood, that contains a cavity the shape of the desired core. This cavity is rammed full of sand-binder mix to form the core. After ramming, the core box may be removed from the formed sand core, which is mounted on a support plate (core plate) (Fig. 1) for subsequent hardening by baking. Alternatively, the compacted sand-binder mix may be cured in the core box by the socalled hot box, cold box, or no-bake methods. As the name implies, the term hot box refers to binders that are cured by heating the core box. Conversely, the term cold box refers to sand binders that cure in the core box when a vapor catalyst is passed through the core box (Fig. 2). Vapor curing (cold box) methods can be used for both coremaking and molding, but they are perhaps more widely used for coremaking than molding. Thus, the term cold box often is used to refer to vapor-cured no-bake systems. The no-bake methods also include self setting with liquid catalysts (see the preceding article, “No-Bake Sand Molding,” in this Volume). For coremaking, the self-setting resin binders are simply referred to as no-bakes.

Cold box processes (vapor cured, no-bake)


Phenolic urethane/amine
Furan/SO2
(Acrylic/epoxy)/SO2 (free

radical

cure

process)
Sodium silicate/CO2
Phenolic ester Heat-cured processes Shell process Core oil (oven bake) Phenolic hot box Furan hot box Furan (furfuryl alcohol) warm box








No-bake (self-setting) processes Furan/acid Phenolic/acid Alkaline phenolic/ester Silicate/ester Phenolic urethane/amine Polyolisocyanate Alumina phosphate










The cold box and heat-cured (heat-activated) processes lend themselves exceptionally well to medium- and high-production applications. Examples of high-production operations are foundries that cast automotive, pipe fitting, air

conditioning, and refrigeration components (Fig. 3). The no-bake processes are generally used for short-run production quantities. Both cold- and heat-activated cores are cured in the core box, thus maintaining excellent dimensional accuracy. Cold processes use gases that are forced through the compacted sand mixture to cure the core. Heat-cured (hot box) processes require the core box to be heated to 175 to 290  C (350 to 550  F) prior to introduction of the prepared sand. The no-bake process uses binder systems that consist of chemicals that, when mixed together in sand, cure without the introduction of an external agent, such as heat. The process is well suited to the fabrication of larger cores. In the core-oil (oven-bake) process, sand is mixed with corn flour and/or clay and water along with the binder to give sufficient strength to the mass of sand so that it can be immediately removed from the core box onto core plates. The cores are then hardened by baking at 200 to 260  C (400 to 500  F). The oven-bake process does not require complicated curing equipment, and materials are inexpensive; however, dimensional accuracy is more difficult to maintain. The type of metal to be poured must be considered in selecting a core process. For example, steel and certain high-alloy cast irons

Binders for Sand Coremaking As noted, sand binders for coremaking are similar to those for sand molding. The three major systems used for coremaking are:

Fig. 1

Basic principles of core molding. See text for discussion.

582 / Expendable Mold Casting Processes with Permanent Patterns core boxes and machines designed for one type of heat-activated process can usually be used for alternate heat-activated processes. The same is true for cold processes. Although equipment is available for blowing or shooting core weights of up to 540 kg (1200 lb), most cores of this size and above use one of the no-bake processes. Smaller cores are usually made using cold, warm, or hot box techniques unless they have high surface-area-to-volume ratios, in which case the required equipment would be cumbersome. Elevated temperatures accelerate chemical reactions. Summer temperatures may adversely affect some of the hot systems to the point where they actually harden prematurely. In cold systems, the silicon dioxide (SO2) types possess the longest working times. The core sand will eventually become system sand. Therefore, suitability for reclamation and compatibility with the sand reclamation and sand system used must be considered.

Basic Principles of Coremaking

Fig. 2

Cold box (vapor-cured) coremaking process. The wet sand mix, prepared by mixing sand with the twocomponent liquid resin binder, is blown into the core box. The core box is then situated between an upper gas input manifold and a lower gas exhaust manifold. The catalyst gas enters the core box through the blow ports or vents and passes through the core, causing almost instantaneous hardening of the resin-coated sand. The core is ready for ejection from the core box after purging with clean air for a few seconds. After the catalyst gas passes through the core, it leaves the core box through vents into the exhaust manifold. From the gas exhaust manifold, the catalyst gas is piped to an appropriate disposal unit.

are sensitive to urea because of the tendency to form gas defects due to the amount of nitrogen present. In making aluminum castings, the pouring temperature may not be high enough to break down the binder system adequately for easy removal from the castings. Although silicate cores usually collapse sufficiently when used with aluminum and steel, their use in cast

iron may be a problem because of a secondary bond that forms at iron-pouring temperatures. Cores for magnesium castings require inhibitors that must be compatible with the binder system. Wood, plastic, and metal core boxes can be used with the cold processes, while only metal can be used with the heat-activated systems. However, once a system is established, metal

During its preparation, a core must be strong enough to retain its shape without deforming. After baking or drying, it must be strong enough to withstand handling and to resist erosion and deformation by metal during the filling of the mold. To make a true form for the casting, it must be stable, with a minimum of contraction and expansion. The core must be sufficiently low in residual gas-forming materials to prevent excess gas from entering the metal. Provision must be made for venting any gases that are produced by the core. Furthermore, the core must collapse after the metal solidifies, to minimize strains on the casting and to facilitate removal of the core from the casting during shakeout. Figure 1 illustrates various elements for coremaking by baking (perhaps by air-setting outside the core box). In the top row, core boxes are shown after compaction. In each of the middle and lower views, the core box has been inverted on a plate and lifted, leaving the core on the plate, ready for either baking or no-bake curing. The shape in Fig. 1(a) is a semicylindrical core—one of the most common of core shapes. No problem is presented in removing this core from the core box; natural draft is provided by core shape, and no obstructions exist. In Fig. 1(b), two bosses are formed by the core box. Boss A is simple to produce and adds nothing to the cost of producing the core, although it does add a small amount to the cost of the core box. Boss B, however, creates an undercut that requires a loose piece, as shown, to permit the core to be freed without destroying the desired shape. This adds to the cost of the core box. It also adds to the cost of producing the core and may increase the dimensional variations in the location of the boss. Figure 1(c) shows a core box in which two more conditions are incorporated that require

Coremaking / 583

Fig. 3

Cores and corresponding castings produced by the cold box process. (a) Large transmission case and resulting gray iron casting for an agricultural equipment manufacturer. (b) Disk brake rotor and turbine cores and castings. Courtesy of Ashland Chemical Company

Fig. 4

Details of a mold-and-core assembly for making a cast iron shell, showing the use of a core arbor

loose pieces. Loose piece A permits the side of the core to be produced without draft. Bosses or ribs could be appended to this face without adding to molding problems, provided they could be drafted to permit easy withdrawal of the loose piece. Loose piece B forms a cavity in the core that assists in producing a rib on the casting. The overhang of the core box would lock the core in the box unless a loose piece were provided as shown. Figure 1(d) shows the additional problem that arises when the loose piece at the face of the core box projects excessively into the core. When the core box is inverted and the loose piece removed, excessive overhang of the core could cause the sand to sag and lose its dimensional accuracy. To prevent this, support must be provided for the core. In the core box shown, the loose piece is removed and replaced with

bedding sand before the core box is inverted. (Bedding sand is a special sand that does not adhere readily to the core and thus can be removed cleanly after the core has been baked.) Core boxes may be simple in design or composed of several pieces to form an intricately shaped core. Where only one or a small number of cores are needed, a skeleton core box may be used to reduce the cost of the core box. For a simple core, no equipment other than the core box and core plate is required. For more complicated cores, additional equipment may be necessary for forming, supporting, reinforcing, venting, and handling. The materials and procedures used for making core boxes are essentially the same as those used for making patterns. In general, the principles that apply to molding a pattern apply also to molding a sand core. The core is freed from the core box by moving the box away from the core. Loose pieces that have been placed in the core box to assist in producing the shape of the core will remain with the core when the box is removed from the core; they must be removed from the core by a movement in the appropriate direction. Core Driers. In the core box shown in Fig. 1(d), a core drier could be the loose piece. (A core drier is a contoured metal form or “cradle” that remains with and supports an irregularly shaped core while it is being baked.) However, since a drier is required for each core being baked, the expense of producing all the necessary driers adds substantially to core cost. Although supporting an overhanging or irregular section of a core with metal core driers provides greater accuracy and production speed, the use of bedding sand is usually more economical for short runs. The tolerance specified for the cavity being cored will influence the choice. Reinforcing of Cores. Cores often must be reinforced to give them the necessary strength for handling and for withstanding the lifting force exerted on the core by metal under it in the mold. Cast iron weighs approximately 41/3 times as much as core sand. Hence, for iron castings, it is more often necessary to reinforce a core for resistance to the lifting force of the liquid metal than for resistance to handling.

Small cores can be reinforced by a rod or wires placed in the cores, but in large cores, where rods may not give sufficient strength, iron or steel frames of bolted, cast, or welded construction are used. These frames are called core arbors. Figure 4 shows a cross-sectional view of a mold for casting a cored shell. Note the core arbor, wrapped with burlap, on which the sand has been rammed to form the core. Here, the arbor is used to support the core in an upright position and to center it precisely and rigidly in the mold. Whenever practical, an arbor should be used in large cores, because it eliminates the need for rods and pipe. In a jobbing shop that has no facilities to make arbors and where the jobs change daily, it is more satisfactory to use pipes and rods than to obtain arbors. Arbors also permit the handling of cores under circumstances where it would be impossible if only rods and pipes were used. A disadvantage in the use of arbors is that they are more difficult to remove at shakeout. Core venting is necessary, and the general rule is to provide the maximum allowable venting. To understand the fact that gases must be dispersed, one need only consider the volume change that occurs in the transition of a liquid or solid to a gas. The principal gases resulting from the breakdown of the chemical binder systems have been found to be hydrogen, carbon monoxide, carbon dioxide, methane, and nitrogen. In venting, however, the volume of gas is more important than its composition. Care must be taken to lead all vents to core prints and to provide exits from the mold with minimum pressure buildup. In addition, vents must be kept far enough away from the surface to prevent metal breakthrough. As coarse a sand mix as good practice permits should be used. The permeability of sand is related to its coarseness, assuming a normal distribution. The coarser material will have the highest permeability and the most natural venting. Cut, punched, filed, or scratched vent passages can be made by removing sand during coremaking, after curing, and before assembly. Mandrels can be used to hollow out cores, thus saving sand, or small holes can be drilled into cured cores. Strands of wax and/or plastic tubing can also be used. These materials are available in a variety of sizes. In the heat-activated processes, some of these materials melt and are absorbed into the sand, leaving a passage. Permeable plastic tubes can be used in cold box processes. In very large cores, the center can be filled with dry unbonded sand, coke, slag, or gravel, thus avoiding gas generation. A secondary advantage of using these materials is that they replace the same volume of mixed sand and thus reduce cost.

Core Sands and Additives Core sands are usually silica, but zircon, olivine, and carbon are used. Coarser sands

584 / Expendable Mold Casting Processes with Permanent Patterns permit greater permeability and are preferred for cores. Sand that contains more than 5% clay cannot be used for cores, because of the large amount of additives required. In selecting a core sand, it must be considered that most of the sand from burned-out cores enters the molding-sand system and may well be a significant portion of the system sand. Under these conditions, a sand suitable for both molds and cores is desirable. Sand mixing is easily accomplished for all of the processes except the oven-bake and shell processes. Special mullers are required to mix the sand and dry ingredients adequately with water and binders to develop the necessary strength to permit baking process cores to be removed from the core boxes in the uncured state. Most shell process sands are hot coated in paddle/plow-type mixers using solid or liquid novolac resins. Cold and warm coating, both of which use powdered or liquid novolac resins, have declined. With the hot coating process, a solid flake or liquid novolac resin is dispersed into a sand that has been preheated to 130 to 175  C (265 to 350  F). Hardener (contents ranging from 10 to 17% based on the weight of the resin) is first dissolved in water, and this solution is added after the resin is dispersed, and a lubricant is added near the end of the cycle. The moisture content of the sand mix will range from 1.5 to 2.5%. Immediately after the addition of the water/hardener solution, cold air is blown into the mixer to cool and dry the sand. Whatever process is selected, the coating process is carried out in a paddle-type mixer that is usually equipped with a system (fluid bed, screen, or cyclone) to cool and complete the drying of the coated sand. Various types of blending equipment are used to disperse the binders uniformly over the sand used in the other processes. Batch or continuous mixers can be employed to coat the sand used in the cold- or heat-activated processes. If a batch mixer is used, the sand and additives are weighed or volumetrically measured before being introduced into a mix. There, the mixture is agitated by a revolving mixer or by vibratory motion for 20 s to 5 min. Prepared sand is then discharged and transferred to the coremaking station. A continuous mixer is almost a necessity for a no-bake coremaking operation but is also adequate for mixing sand for the cold- or heat-activated processes, with the exception of the shell process. Sand and additives are continuously metered into a mixing trough or chamber that contains revolving mixers of various designs, and prepared sand is discharged. In the no-bake process, the sand is discharged directly into the core box, where it is immediately compacted. Prepared sand for the cold- and heat-activated processes is generally discharged into a storage hopper above a coremaking machine. Release Agents. All core binders are sticky, but some are less so than others. It is this characteristic that requires the use of release

agents/parting compounds on core boxes, regardless of the type of core box material. The desirability of having a dense core compounds the problem of core release, because binder-coated sand is packed or blown against the surface with force, which enhances sticking. In the heat-activated processes, the binder melts or softens and migrates to the heated surface. It is necessary to start with a good core box; no amount of release agent will overcome roughness and scoring. The core box should be chromium plated if it is to be used for high production, and it should be broken in gradually. If a release agent is functioning properly, some buildup will occur; it should be removed frequently because it adversely affects core dimensions and quality. One method of removing buildup is blasting with ground walnut hulls, corn cobs, or other soft media. Solvents can also be used. Minimum amounts of release agents should be used, and when adding them during mixing, they should be added near the end of the cycle. The release agent should be chemically compatible with the binder system. Oleic acid diluted with kerosene (usually 10 to 1) is used as a release agent for core-oil cores. It is added to the sand mixture and used to keep the core boxes clean. The amounts used are 125 to 250 mL/450 kg (1 pt to 1 qt/1000 lb) of sand. Shell and other heat-cured processes use emulsified silicone release agents diluted in water. A low-surface-tension silicone should be used for spraying core boxes. Stearates, usually calcium, are added to shell sand during the coating process. Amounts recommended are 2 to 5%, based on resin solid weight. Small amounts of core oil or silicone emulsion can be added to hot box sand during mixing. The no-bake and cold box processes use a wide variety of release agents. Small-to-medium sized high-production core systems will normally use liquid types, which can be sprayed through jet nozzles and integrated into an automatic cycle or sprayed with a hand gun. No-bake systems normally use liquids, suspensions in liquids, or dry materials for larger cores. The suspension systems usually use a solvent carrier that evaporates readily. Materials can be sprayed, brushed, dusted, or hand rubbed onto the core boxes. Ingredients include (in many combinations) water, alcohol, chlorinated solvents, mica, talc, silicones, aluminum powder, graphite, and soybean oil (lecithin). Other Additives. Various additives are used to improve sand molds in both green sand molding and no-bake sand molding. Silica flour is often added to a core mixture to increase the hot strength of the core. Sawdust and wood flour are sometimes added as fillers; their main purpose is to increase collapsibility of the core.

Compaction Cores must be sufficiently dense to obtain good grain-to-grain bonding of the sand mass and to prevent metal penetration between the

sand grains. The degree of compactness necessary to produce a satisfactory core depends on the type of binder used and on the size and shape of the core. Core sands bonded with air-drying binders, such as sodium silicate or portland cement, must be rammed well to bring the sand grains into proper contact. Oil-sand core mixtures do not require hard ramming; many cores are made from oil-sand mixtures by jarring, blowing, or pressing. The size and shape of a core also dictate to a large extent the firmness to which the sand must be rammed. For example, a short core need not be rammed as firmly as a tall core. Cores with restricted sections need firm ramming to avoid spongy areas. Core-blowing machines are commonly used to compact sand into the core boxes. Other methods used to ram a core are hand ramming, jolting, squeezing, and slinging. Except for low-volume production, or for very large cores, the sand mixture is compacted into the core box by core blowing. Hand ramming into the core box is the usual procedure for no-bake cores, but mechanical methods are required for highproduction operations. Core blowing is done by filling the core box cavity with sand that is suspended in a stream of air and introduced into the cavity at high velocity through blowing holes. The sand is retained by the cavity walls, and the air is exhausted through vent holes in the core box and core plate. A large volume of compressed air at 275 to 700 kPa (40 to 100 psi) is suddenly introduced above the sand in the chamber, forcing the sand into the core box through the investment area. The carrier air is exhausted through the vents in the core box (Fig.5). The blow air is then turned off, the excess pressure in the blow chamber is exhausted, and the core boxes are disengaged from the blow ports. All these steps are generally electrically or mechanically controlled and require only seconds to complete. Core blowing is a high-speed operation; a core-blowing machine, under proper conditions, will deliver a blown core in a core box within 3½ to 6½, depending on the size of the machine. Core blowers are available in various sand capacities per blow. In the larger machines, handling of the core box is automated. In making cores in a core-blowing machine, the two halves of the core box (Fig. 6a) are mated, using pins and bushings to ensure correct alignment; then, the core box is placed in the machine under the blowplate beneath the magazine or chamber that contains the sand mixture (Fig. 6b), using guides to ensure alignment of the core box opening with the hole in the blowplate of the machine. The sand is carried into the core box cavity by air pressure (620 to 700 kPa, or 90 to 100 psi). Vents at critical positions in the core box and blowplate allow air to escape but retain the sand. The vents range from 3.2 to 38 mm (1/8 to 1½ in.) in diameter and are covered with fine screening consisting of small holes or slots to retain the sand.

Coremaking / 585

Fig. 6

Production of a blown sand core in a two-piece core box. See text for discussion.

minute or less by blowing the sand-resin mixture into a heated core box. Baking is not required. This section focuses on the curing by core baking and the hot box processes (except the shell process). The shell (Croning) process—which is one of the more prevalent binder methods—is detailed in the next article, “Shell Molding and Shell Coremaking,” in this Volume.

Fig. 5

Typical core-blowing setup

Core Baking (Oil Sand) Oil-sand mixtures are used for cores in sand molds, and by varying their composition, they can be used for almost any sand casting application. The following is a typical formulation that contains cereal and a small amount of clay (bentonite); this mixture has proved satisfactory for the casting of gray iron and several other metals:






Fig. 7

Effect of baking time and temperature on tensile strength of oil-sand cores. The dry mixture was 99% silica sand (fineness, AFS 40) and 1% cereal. Based on total weight of dry material, 1% linseed oil and 3% water were added.

Curing of Compacted Cores Curing of the compacted sand mix depends on the type of binder. No-bake systems are self-setting, and time alone is required for the binders to cure. Typically, no-bake cores are rigid enough to be removed from the boxes in times ranging from 30 s to 1 h. Additional curing may continue for several hours. Additional details on the self-setting resins and the cold box (vapor-cured) binders are covered in the preceding article, “No-Bake Sand Molding,” in this Volume. In the heat-activated process, prepared sand is blown into a heated core box. The heat releases curing agents from latent hardeners in the sand mix, and these curing agents then cure the binder. Hot box cores and shell cores can be cured in a

Sand (by weight), 95.80% Cereal flour, 1.01% Core oil, 1.17% Water, 1.86% Bentonite, 0.16%

These materials are mixed in a muller as follows:
Combine sand, flour, and bentonite; mull dry for 1 min.
Add water and mull for 1 min.
Add oil and mull for 4 min. Of course, formulations may require modification to meet specific requirements. Cores from the core-oil process are placed in ovens after stripping from the core box and are baked for 1 h or longer, depending on section thickness. They are removed from the oven when thoroughly cured and allowed to cool to ambient temperature. The baking temperature and the length of time that a core must be heated to develop optimum properties depend on the following factors:






Type of oven used Size of the core Moisture content of the core Type of binder Shape, fineness distribution, and permeability of sand grains


Thermal conductivity, heat capacity, and

density of the sand


Atmospheric humidity

Large cores obviously require longer baking times than small cores. Moisture content of cores also affects the completeness of baking for cores of a given size baked under the same conditions. The more moisture present, the slower the cores will bake. Most of the moisture must be evaporated from the core before the oil and cereal can harden and form an effective bond. Proper curing requires oxidation and polymerization of the oil. As a result, the cores baked from such a mixture not only must be heated to a suitable temperature in an atmosphere that has the necessary oxygen but also must remain at this temperature long enough for the chemical reactions to take place. Figure 7 shows the relationship between baking time and tensile strength for a mixture containing sand, raw linseed oil, and water. Note that the temperature has a pronounced effect on the rate at which tensile strength is developed and on loss of strength from overbaking. Underheated oil-bonded cores can spall and erode, particularly in those applications where the metal strikes the core in front of a gate or intermittently contacts it, as in falling along a vertical wall. For this reason, oil-bonded cores that have baked long enough at a given temperature to have just passed the peak strength will result in better castings than will cores that have been baked just long enough to attain peak strength, or have been underbaked. However, overbaking must also be avoided, because it causes the surface to become so weak that the force of the molten metal will erode it. The baking temperature for a given oil-sand mixture depends on the type of oil used. Optimum baking temperature for most oils is approximately 220  C (425  F). Baking is seldom done at 190  C (375  F) or less, because of the long baking time required. Most oilbonded cores are baked in production ovens at temperatures of 220 to 260  C (425 to 500  F). Different procedures are required for baking cores made with thermosetting resin binders that can be diluted with and are dispersible in

586 / Expendable Mold Casting Processes with Permanent Patterns water. Combinations of phenolic resin and cereal are frequently used in steel foundries; the cereal furnishes the green strength, and the resin and cereal the baked strength. The core must be heated to a temperature at which the cereal will dry and the resin will set or cure, changing from its water-soluble form at room temperature to a hardened moisture-resistant form as the water is driven off. For most phenolic resins, this curing takes place at approximately 150  C (300  F). No oxygen is required. Cores using phenolic-resin binders can be baked to maximum tensile strength at a lower temperature and in less time than oilbonded cores (Fig. 8). Thus, resin-bonded cores may be used to shorten baking cycles and yet avoid the likelihood of casting defects from underbaked cores. Although resin-bonded cores can be baked at 150  C (300  F), optimum results are obtained at 190  C (375  F). Usual production oven temperatures for phenolicresin-bonded cores range from 190 to 230  C (375 to 450  F). Urea-formaldehyde cores are baked at 165 to 175  C (325 to 350  F). Sand Grain Shape, Fineness, and Permeability. The shape of the sand grains (angular, subangular, or rounded) influences the flowability of the sand and the density of the core. The fineness of the sand grain has an effect on permeability. If water vapor or gases that are evolved as the core is heated cannot readily escape, or if oxygen cannot be made readily available to the inside of an oil-sand core, longer baking times are required. An oil-sand core can be held at temperature for a long time without being cured. Figure 9 shows the relationship between fineness of core sand and time required for baking to maximum strength, for cores

Fig. 8

Fig. 9

Baking speeds of phenolic-resin-bonded versus oil-bonded sand cores

Effect of sand fineness on time required for baking oil-sand cores to maximum strength

prepared with a linseed-oil compound. Baking times shown in Fig. 9 were established with test cores in accordance with American Foundry Society standards. Production ovens would have required considerably longer baking times. Thermal conductivity, heat capacity, and density of the core sand influence to varying degrees the rate at which the core can be baked. For instance, over twice as much baking time is required for an oil-bonded zircon sand core as for a similar silica sand core. Atmospheric humidity, if excessive, inhibits the drying of cores and increases the time required to bake them.

Hot Box Cores Although shell cores are made by a hot box process, in that a heated core box or pattern is used and the core or mold is cured in contact with that pattern, the term hot box, as applied to cores, was coined with the introduction of furan-based resins. Hot box core mixtures provide an extremely fast cure—nearly twice as fast as that of the shell-core phenols. In the hot box method of coremaking, the pH value of the sand is important and must be adjusted by a catalyst. The catalysts used in the furan systems are derivatives of furfuryl alcohol, and they react exothermically with mild acids. The heat of the core box triggers the reaction, and when the core is removed from the core box, the reaction is sustained by the residual heat in the core and the exothermic reaction still taking place. The original hot box techniques were developed using furan binders. Phenol is sometimes substituted for all or part of the furfuryl alcohol in some formulations. This has made available several binder systems for the hot box process, among which are the furan hot box system (furfuryl alcohol, urea-formaldehyde), the phenol hot box system (phenol, urea-formaldehyde), and the phenol modified hot box system (phenol, furfuryl alcohol, urea-formaldehyde). Special formulas accommodate the shakeout properties required for casting specific metals; a phenol-formaldehyde formula for steel foundries eliminates the urea phase to avoid pinhole defects. The mild acid catalysts are usually used in the ratio of 5 to 1 (resin to catalyst); the amount and type vary with the sand, temperature, and resin type. Mixing. Almost any type of mixing equipment can be used if the sand is not heated. The ideal temperature for sand leaving the mixer is 20 to 25  C (70 to 80  F). Sand temperatures of 10 to 20  C (50 to 70  F) require longer mixing of ingredients to obtain better coverage, because of the higher viscosity of resins at these temperatures. Temperatures of 25 to 30  C (80 to 90  F) require shorter mixing times, to avoid premature setting of resin. With sand temperatures of 30 to 40  C (90 to 100  F), a slower catalyst is recommended.

A typical sand mixture for hot box cores contains washed and dried sand, 2% hot box resin, and approximately 0.4% activator. The amount of activator used is based on weight of the resin; activator is 20% of resin weight. There are two recommended mixing methods:
Preferred method: Premix catalyst and resin

in an open container (mixture should be below 30  C, or 90  F, and should not be allowed to stand more than 10 min before using). Add to sand and mix for 1 to 4 min, as required.
Add catalyst to cool, dry sand and mix for complete distribution. Add resin and mull for 1 to 4 min, depending on the equipment used for mixing. Properly mulled sand may be usable for up to 3 to 5 h, but storage for more than 2 h is not recommended because core strength will decrease. Temperature and the catalyst used will also affect the bench life of the sand. The higher the temperature, the shorter the bench life. Sand Control. Sands containing appreciable amounts of alkaline-reacting materials will affect the resin-catalyst reaction. Because the activator systems used in the hot box process are only mildly acidic, the effects of these materials can be highly detrimental. The pH value of the sand is not indicative of the final acid requirements of the sand mixture. The pH value is measured by adding the sand to distilled water and measuring the pH of the water layer. If the sand contains alkaline-reacting materials that are insoluble in water, these materials have no influence on the final pH value. If these materials are reacted with acidic activators, a portion of the catalyst is no longer available for the requirements of the resin. Therefore, it is sometimes necessary to adjust the catalyst system to fit the requirements of the sand. However, a better method for determining the catalyst requirements of the sand is to measure the acid requirements of the sand. The test for acid requirement is done as follows: 1. Stir together 50 g of sand, 50 mL of 0.1 N HCl (hydrochloric acid), and 50 mL of distilled water. Let the mixture stand for 1½ h, to allow the materials that are alkaline to react with the acid. 2. Back titrate the mixture to neutralize the excess HCl with 0.1 N NaOH (sodium hydroxide), to determine the amount of HCl that was used in the reaction with the alkaline materials. For example: 50 mL HClðadded originallyÞ  40 mL NaOHðused in back titrationÞ ¼ 10 mLðamount of material reactedÞ

In this example, the sand has an acid requirement of 10. (As the acid requirement increases, the acidity requirement of the catalyst system also increases.)

Coremaking / 587 Table 1 Effect of curing time and temperature on tensile strength of hot box cores Tensile strength, kPa (psi), for a curing temperature of: Curing time, s

5 10 15 20 30 40 50 60

175  C (350  F)

205  C (400  F)

772 (112) 2234 (324) 2620 (380) 3082 (447) 3475 (504) 3675 (533) 3737 (542) 3716 (539)

1200 2648 3220 3551 3833 3661 3606 3516

(174) (384) (467) (515) (556) (531) (523) (510)

230  C (450  F)

2055 3068 3613 3640 3751 3634 3365 3323

(298) (445) (524) (528) (544) (527) (488) (482)

The following table, which compares pH values to acid requirements for three different types of sand, shows how erroneous pH values can be as indicators of acid requirements (note that bank sands show more acidity yet have a greater acid-requirement value): Type of sand

Silica Lake Bank

pH value

Acid requirement

6.9 6.7 6.5

2 4 40

Core Boxes. Cast iron is the preferred metal for core box equipment. The walls surrounding the core cavity are made uniformly to 22 mm (7/8 in.) thickness, where practical, to achieve uniform heating. Ribs are placed on the outside of the box to keep it in shape when it is exposed to heat. Copper-base alloy inserts generally are not used, because of the corrosive action of the binder ingredients. Hardened steel is used for the locating pins, bushings, and wear pads. Stripping pins have a clearance of as much as 0.20 mm (0.008 in.). This clearance also acts as a vent. Since the hot box mixture is wet and is blown at a pressure of 480 to 700 kPa (70 to 100 psi), provision must be made for exhausting the air from the core box. Venting practice is comparable to that for an oilcereal-bonded mixture. Slotted steel vents are preferred; slots usually are 0.25 to 0.45 mm (0.010 to 0.018 in.) wide. Vents are usually plugged by silicones and parting agents, rather than by hot box resin. Parting-line venting is good practice when blowing through the parting line; the vent should be a milled cavity approximately 0.13 mm deep by 25 mm wide 0.005 by 1 in.). Because furan hot box cores are usually closer to pattern dimensions than phenol shell cores, they require better ejector pin design for preventing mechanical seizure. Ejector pins should be located close to a vertical draw. Sand Introduction. There are two methods of introducing the sand mixture into the heated core box: core blowing and core shooting. Dump-box methods are not applicable for hot box cores.

Core blowing is more widely used than core shooting; the technique and blow pressures are essentially the same as for oil-sand mixtures. Core shooting is a means of extruding prepared sand by the sudden release of air at extremely high pressure. Core shooting has been adapted to some of the smaller hot box machines. With this method of sand introduction, core box venting is not quite so critical as with core blowers. Curing Temperature and Time. Core box temperatures should be 205 to 230  C (400 to 450  F)—a lower temperature can be used for a longer time and a higher temperature for a shorter time. A cycle that produces a core with good tensile strength should be chosen. Typical relations of time and temperature to cured strength are shown in Table 1. Curing time depends on core size and shape, core box temperature, and production requirements. For maximum production, the prepared sand should be introduced into the core box as quickly as possible and the semicured core should be ejected as soon as it can be handled without distortion. When this practice is followed, a secondary curing cycle usually is required; this ordinarily is 10 to 15 min through a tunnel oven or a continuous oven at 205  C (400  F). When maximum production is not required, the core is left in the box until completely cured. Most cores can be removed from the core box in 5 to 20 s. At this time, the curing reaction will have penetrated the core to a depth of 4.8 to 7.9 mm (31/16 to 51/16 in.). The surface will be hard, with no indication of thermoplasticity. Cores cured to this depth can be safely handled or set on racks or plates until needed. Additional internal curing will take place during the 10 to 20 min after the cores have been ejected. Machines have been designed that are specially adapted to the hot box process. Also, adapter kits have been designed to fit the shell machines currently in use. Equipment is available to make cores weighing from a few ounces to approximately 45 kg (100 lb). Because the mixture used for the hot box process is wet, blowing pressures from 480 to 700 kPa (70 to 100 psi) are commonly used. Also, because the mixture is sensitive to heat, the blowplates on the core blowers are water cooled to minimize heat transfer from the heated box to the sand in the blowholes and in the sand hoppers. Water flows continually through the water jacket while the machine is in operation. Blowholes for the hot box mixture are the same size as for tempered sands or for oil-sand mixtures. Blow tubes in the upper half of the boxes are tapered to facilitate the ejection of cores. Blow tips also are usually water cooled. Applications. Hot box cores have been used in the production of castings of all types of metals, although binder formulations are changed. For instance, for aluminum and magnesium castings, the furfuryl alcohol content

of the binder is decreased to aid core collapsibility, whereas for steel castings, the furfuryl alcohol content is increased to provide greater core strength. Because of the suitability of the hot box process to high-volume production, its main use has been in the automotive industry (cores for water jackets, head jackets, manifolds, and other engine castings) and in the plumbing-fixtures industry (cores for elbows, tees, valves, and faucets). Advantages of the hot box process over other methods of coremaking are:
No special mixing equipment or heating

systems are required for coating the sand.


Fewer machines and less floor space are

required.


No special sand additives are required.
The high degree of core stability prevents

warping.


Lower temperatures and shorter heating

times are required during curing cycles.


Laminations near the surfaces of cores are

eliminated.


Cores are produced in a fraction of the time

required by conventional practices.


Increased core strength reduces handling

breakage.


Core wires are eliminated.
Core driers are not required.
Shakeout efficiency is greatly improved,

because cores have high collapsibility.


Castings can be cored to closer dimensional

tolerances.


Casting costs are decreased.
Production rate is increased.

Disadvantages of the hot box process are:
Sand sections are limited to approximately

64 mm (2½ in.) thickness.


Blowplates must be water cooled.
Venting is more critical than it is with the

shell process.


Basis-sand temperatures are critical.
Sand mixture has a short bench life.
An odor of free formaldehyde is present in

the work area, so that good ventilation is mandatory.

Core Coatings (Core Washes) Cores used for ferrous castings are usually coated, and those used for casting of nonferrous metals may be coated. Core coatings (or core washes) serve to improve the casting finish, because the coating prevents metal penetration. The coating also serves as an insulator between the sand and hot metal, preventing sand from adhering to the casting and thus reducing the difficulty of cleaning the castings. This results in a reduction of casting cost. Core coatings vary widely in composition, but most are of one of two general types: carbonaceous and carbon-free. Carbonaceous coatings are largely composed of graphite or some

588 / Expendable Mold Casting Processes with Permanent Patterns other form of powdered carbon. In addition to carbon, they may contain silica flour, clay, or talc. Carbon-free coatings are composed of materials such as silica, talc, magnesite, alumina, zircon, asbestos, or mica. The carbonaceous coatings are used most widely for gray iron, malleable iron, and nonferrous castings. Cores used for steel castings are usually coated with carbon-free materials. Coatings are usually purchased as proprietary compounds, available in three forms: dry powder ready for mixing with water or other liquid, premixed paste ready for dilution by the user, and ready-to-use mixtures containing a volatile solvent to promote quick drying. Small cores are usually dipped in the coating material. Larger cores are often coated by brushing, although they may be sprayed with a siphon gun of the type used for coating molds. Coatings must be thoroughly dried before mold assembly. For baked cores, a common practice is to coat the green core prior to the baking cycle and then to recoat the core after it cools to approximately 80  C (180  F) following the baking cycle; thus, the residual heat is used to dry the second coat. Coatings may also be dried by torching or in an oven. Most coatings are formulated with five major components:






Refractory materials Carrier system Suspension system Binder system Chemical modifiers

Refractory materials for coatings can be made of any of the following materials:













Zirconium oxide Zirconium silicate Magnesium oxide Olivine Chromate Pyrophyllite Talc Carbon (many forms) Silicon dioxide Magnesium/calcium oxide Mica Iron oxide

These materials are blended in many combinations to make up a broad selection of proprietary coatings. Silica, alumina, and carbon are the most commonly used, and water, alcohol (isopropyl, methanol), naphtha (petroleum distillate), chlorinated solvents (inhibited 1,1,1trichloroethane, or chlorothene NU, and methylene chloride), and aliphatic hydrocarbon are used as liquid carriers. The carrier allows the refractory to be applied evenly to the sand mold or core. After the coating has been applied, the carrier must be removed. When the carrier is water, it can be torched or oven dried. Alcohol carriers are commonly ignited and allowed to burn off. The volatile chlorinated hydrocarbons readily

air dry and require neither torching nor oven drying. Suspension System. The function of the suspension system is to maintain the refractory material uniformly dispersed in solution. With water carriers, sodium bentonite is commonly used, while organic bentonite or bentones are generally used with alcohol and chlorinated hydrocarbon carriers. The binder system in a coating is usually an organic resin, and it behaves similarly to the resins used to bond sands. The amount of binder used depends on the density and fineness of the refractory. It is used sparingly to avoid casting surface defects. Chemical modifiers to the coating include surfactants to enhance wettability, antifoaming agents, and bactericides. Coating Benefits. Most of the reasons for using coatings center around reducing casting costs by improving sand peel and reducing mold/core reaction, reducing or eliminating metal penetration (burn-in and/or burn-on), and reducing or eliminating veining. These reasons are directly associated with casting cleaning costs. Coatings are also used to reduce machining time and tool wear by having a clean, smooth surface, to improve casting appearance, to facilitate handling (especially where low-strength cores may be required because of shakeout conditions), or to promote chill or increase hardness in metal (for which selenium or tellurium paste is used). Coatings should not be used indiscriminately. No coating will compensate for a poor-quality core. Blending and Applying Coatings. All the advantages of selecting the proper coating, or wash, can be lost if it is not properly blended. Coating ingredients are available in powder, paste, slurry forms, and ready-to-use forms. In preparing a core wash, the ingredients are mixed to a slurry with an electric or air-driven mixer. The mixing tank should be twice as high as its diameter, should be equipped with flowthrough baffles spaced at approximately 120 , and should have an impeller shaft with two blades. The slurry must be mixed until creamy and lump-free, with ingredients in suspension/ solution. The Baume´ must be kept 25 to 30 higher than application requires and further diluted prior to use. The carrier should be added in steps to avoid overdilution, which will adversely affect suspension. Coatings should be mixed a specified time before use (usually 12 to 24 h) to permit tempering of the suspension and bonding agents, and overmixing should be avoided. The methods of application will depend on a number of factors, such as the size and shape of the core, plant layout (if a continuous process is used), and production requirements. For example, small cores would not lend themselves to spraying but to dipping in groups using fixtures and/or baskets. Conversely, large cores in a production line could move continuously on a conveyor through a dip tank to a drain station and then to a drying station. Regardless of the

method of application, obtaining a smooth, even coat is of paramount importance. Water- or solvent-based washes can be used for furan, furan/phenolic, urethane cold box and no-bakes, phenolic esters, and baked coreoil cores. However, water-based washes should not be used on silicate or phosphate-bonded cores, because such washes will degrade the binder. The best practice for drying coatings is to apply coatings on cured cores and then dry them by appropriate means, such as ovens, burn-off, or warm air circulation. If cores are cold, they should be heated before or immediately after application. If cores are coated green (incomplete cure), heat should be applied immediately. Evaporation of the solvent is important for avoiding deterioration of the core binder.

Assembly and Core Setting Cores consisting of more than one piece are pasted together as a subassembly. Also, when two or more cores are set in the same mold and cannot be anchored securely with core prints, they are often pasted in place. Sometimes, two cores are pasted together in the assembly to seal the joint against metal penetration. Core pastes can be prepared by mixing talc, dextrin, flour, and molasses. However, it is usually more economical to use a ready-mixed paste from a foundry supply house. Core paste can be applied by brushing, by squeezing from a paste tube, or by the use of a pasting fixture. A pasting fixture is a metal form, usually of brass, that has the same outline as the flat surface of the core to be pasted. By means of pulleys, wire cable, a slide, and a foot pedal, it is immersed in a paste tank and then withdrawn. The core is placed against this form, thus placing the correct amount of paste at the correct location. The paste used with this method of application usually is mixed thinner than when applied by any of the other methods. Core Patches. Generally, the same coresand mixture is used for patches as for the cores. Patching is done mostly before baking. If patching is done after core assembly, the patches can be dried by baking or torching before setting the core into the mold. Talc sealers are used between molds and cores to prevent runouts. A satisfactory sealer is made by mixing 4.3 kg (9½ lb) of talc and 3.8 L (1 gal) of core oil. Blowholes caused by cores that are largely surrounded by metal can be minimized by applying sealers. Core Mudding. Proprietary mudding compounds, mixed with liquid, are used to smooth rough surfaces and fill small holes and seams. They must be dried thoroughly. In the example that follows, the mudding compound was inadequately dried, and casting defects resulted. Example: Change from a Two-Piece to a One-Piece Core to Eliminate Cold Shuts Caused by Mudding Compound (Fig. 10). Cold shuts occurred above the horizontal

Coremaking / 589

Fig. 10

Pipe fitting that developed cold shuts because of inadequately dried mudding compound in the two-piece core. Dimensions given in inches

Fig. 11

Comparison of maximum recommended core length versus core diameter for baked sand cores positioned in the mold (a) at the parting line and firmly anchored at both ends, (b) at the parting line but anchored at one end only, and (c) vertical to the parting line and anchored at both ends

Fig. 12

Two methods of anchoring a core. (a) Core float is prevented by a large core-printed section clamped between the mold halves. Venting of the cores is difficult. Core-print taper prevents the core from shifting in the mold, which could result in unequal walls. (b) By inverting the pattern, better venting of core gas is possible. A large core print is still necessary to position the core securely and to prevent shifting.

parting line of a gray iron fitting that was cast with a two-piece core (Fig. 10). Raising the pouring temperature and pouring rate had little or no effect. The difficulty was finally traced to the mudding compound used in patching the core after the two halves had been pasted together. This material had not been dried sufficiently before the casting was poured. The liberation of excessive gas when the metal came in contact with the parting caused splashing, which resulted in cold shuts. The problem was solved by replacing the two-piece core with a one-piece core, which did not require pasting and patching.

Design Considerations Cored holes should be designed as simply as the intended function of the casting permits. An effort should be made to eliminate projecting and overhanging parts, pockets, and recesses and to use simple and smoothly contoured core shapes. The casting design should take into consideration the requirements for draft necessary for successful removal of the core from the core box. Although the incorporation of loose pieces permits cores with straight sides, overhangs, and undercuts, these add to dimensional discrepancies and increase the cost of producing the core. Bosses should be extended to the parting line, if this can be done without adding impermissibly to the weight of the part. When this is a questionable procedure, the undesirability of additional weight should be balanced against

the desirability of savings in the casting cost through the design simplification. Of course, if bosses can be located on the parting line, both cost and weight are minimized. Core Size Limitations. Figure 11 presents recommended relations between core length and core diameter. These maximums can serve as guides in the design of cored holes. Recommended maximum lengths for horizontal sand cores supported at both ends are given in Fig. 11(a) for steel, aluminum, and magnesium castings. Such cores often bend upward because of the buoyant force of the molten metal. Steel causes more bending than does aluminum or magnesium because, being heavier, it exerts a greater upward pressure on the core. Also, the higher temperature at which steel is cast is more likely to break down the binder that holds the core sand together. One foundry producing a casting with a horizontally supported core reported an upward bending of 0.8 mm (1/32 in.) in a tubular casting with a 38 mm (1½ in.) outside diameter, 25 mm (1 in.) inside diameter, and an over-all length of 254 mm (10 in.). Often, one of these cores bent 1.6 mm (1/16 in.). Cantilever-supported cores (Fig. 11b) are even more likely to bend, because they are supported at only one end and are therefore less able to resist the buoyant force of the molten metal. A larger tolerance is needed on dimensions at the unsupported end of the core, because of the necessity for a small amount of side clearance between the core and the mold at the opposite end. This clearance permits a displacement of the core when the molten metal enters the mold. The displacement is amplified as the core extends into the casting and has a pronounced influence on dimensional discrepancies. A blind hole whose length is approximately equal to its diameter usually offers no cleaning problems, because excessive heat concentration

is unlikely in the core. As the length of the blind hole increases in relation to its diameter, the possibility of burnt-in sand increases, making cleaning more difficult. It is for these reasons that blind cored holes should be kept to minimum length; as indicated in Fig. 11, the maximum length recommended is less than for cores supported at both ends. For cantilever-supported cores, the absolute minimum that must be allowed for core-print lengths is 1/3l, as shown in Fig. 11(b). This minimum length is suggested for use only in dry sand molds and in shell molds, where the fit is accurate enough to hold the core firmly in place. In green sand molds, enough mass must be provided in the core-print section to balance the core so that it stays in place while the mold is being closed. Vertical cylindrical cores supported at both ends are not distorted by the buoyant action of the metal. As long as they are free to expand, they remain straighter than horizontal cores. With such vertical cores, however, it may be difficult to close the mold without either disturbing the core location or shaving sand from the mold. This sand may fall down into the mold and appear as a casting defect. Figure 11(c) shows maximum recommended lengths. Although vertical cores supported only at the bottom are not subject to distortion by the buoyant effect of the casting metal, they can float out of position if they are not anchored firmly. Furthermore, such cores vent less efficiently than horizontal cores or cores with a core print into the cope. Another method of anchoring such a core in place is illustrated in Fig. 12(a). The printed section of the core extends past the cavity and is held firmly in place by the cope half of the mold. This method requires the use of more core sand than a ram-up core (one that is placed in the pattern before the flask is filled with sand, so that as the mold is rammed-up, the core

590 / Expendable Mold Casting Processes with Permanent Patterns becomes an integral part of the mold) and is more expensive, but it is also more accurate. The insertion and removal of ram-up cores abrade the pattern and permit variation in core location from casting to casting. Because of the difficulties usually present in molding and coring vertical blind holes, the maximum length of the cores should be reduced to approximately half the values given in Fig. 11(c). Vertical cores supported from the top, as in Fig. 12(b), are not subject to distortion from the buoyant effect of the molten metal, and they are prevented from floating by the core print, which is held in place by the cope half of the mold. Such cores are easily vented, because core gas ordinarily travels upward to escape. For vertical cores supported at the top only, recommended maximum lengths with reference to diameters are the same as those given in Fig. 11(b) for cantilever-supported cores. When cores are so large or complex that they are made in pieces and are then baked and pasted together, any distortion will prevent the faces from meeting firmly. The gap between the cores will cause fins in the cored passage unless special processing is given the cores. Chaplets. It is sometimes permissible to provide the necessary support for cores by the use of chaplets, which are metal spacers placed in the mold to support a core or to space it a fixed distance from the mold wall. Chaplets (illustrated in Fig. 13) are usually made of the same metal as the casting in which they are to be used. In theory, the chaplet fuses with the casting metal, to form a continuous wall. In practice, this does not always happen, and poor fusion may create a weak spot in the wall that will leak if subjected to pressure. Fusion between aluminum or magnesium chaplets and casting metal is even more uncertain than in steel, because of the lower heat capacity of aluminum and magnesium and the fact that an oxide film is usually present on the surface of aluminum and magnesium chaplets. A long, slender core supported at one end only, such as the core shown in Fig. 14(a), requires support to prevent it from sagging or floating and thus causing distortion. Chaplets would provide this support and would also help the molder position the core properly. Two cored openings, in the same positions as the chaplets, would provide support for the core (from their core prints) and would also allow for better venting (Fig. 14b). The cored holes would be tapped and plugged later. The enlarged end of the core print (Fig. 14a) positions the core axially. The addition of the small core prints (Fig. 14b) eliminates the need for this enlarged section. In the following example, the problem of inadequate core support was solved by adding holes to the casting. Example: Addition of Holes to Provide Core Support (Fig. 15). Inadequate support was a problem with the core necessary to form the inside configuration of the oil slinger shown

in Fig. 15, which was sand cast in aluminum alloy. The original design had six core supports, three on each side, spaced as shown. After casting, the 8 mm (5/16 in.) wall thickness varied (in some instances, to as little as 2.4 mm, or 3 /32 in.), and rejection rate was high. It was necessary to add six more holes, three on each side, as shown, to provide adequate support for the core. These additional holes were later tapped and plugged, to permit the casting to function as originally intended. Molding and Coring Details. The example that follows demonstrates alternative methods for simplification of molding and coring in the production of a complex sand casting. Although there are other ways to produce this casting, the methods described show the effect of design on the cost of a casting and emphasize the close relation of design to molding and coring practice. Example: Redesign of a Sand Casting to Solve Core and Mold Problems (Fig. 16). The hypothetical casting considered in this example is shown at top-center of Fig. 16. The pattern for this casting, as originally designed, is shown in Fig. 16(a). The pattern is made in five sections and requires a special four-part flask. Although the end sections of the flask (for mold sections 1 and 4) may be made to any height that provides enough sand to retain the metal within the mold cavity, the height of the center sections must correspond to the height of the sections of the pattern involved. The molding sequence is:

For simplicity, details of gating and risering of the molding sequence have been omitted. Because two of the outside diameters of the casting are smaller than three of the inside diameters, a one-piece core cannot be placed

Fig. 13

Use of chaplets to support cores to prevent core sag before pouring and core float after

pouring


Pattern sections 2 and 3 (Fig. 16a) are







assembled, flask section 2 is set in place, and mold section 2 is produced. Pattern sections 4 and 5 are placed in position on pattern section 3. Flask section 3 is set in place, and mold section 3 is produced. An end section of the flask is set in place on flask section 3, and mold section 4 is produced. The three sections of the mold that have been produced are turned over as a unit, exposing the mold surface at parting line A and pattern section 2. Pattern section 1 is set in place on pattern section 2, and the remaining end flask section is set in position. Mold section 1 is produced.

Fig. 14

Redesign to eliminate chaplets. (a) In original design, chaplets are required for core support. (b) Improved design eliminates the need for chaplets by adding two holes, which provide adequate support by the core itself.

At this point, the four flask sections are in place. The mold is complete, but the patterns have not been removed. The steps that follow concern removal of the pattern:
Mold section 1 is removed and turned over

to expose pattern section 1. Pattern section 1 is then removed from the mold.
Pattern section 2 is removed from mold section 2. Mold section 2 is removed and turned over. Pattern section 3 is withdrawn from the mold.
Pattern section 4 is withdrawn from the mold. Mold section 3 is then removed and turned over. Pattern section 5 is withdrawn from the mold. (The mold was reassembled as the pattern sections were removed.)

Fig. 15

Casting in which six cored holes were added to provide core support needed to prevent core distortion when molten metal filled the cavity. Dimensions given in inches

Coremaking / 591

Fig. 16

Hypothetical casting showing core and mold problems if placement of cores is not considered in the design of a casting. (a) Required patterns and mold sections for producing casting by original method. (b) Cores and related mold sections for original method. (c) Core box, and (d) core drier, to produce core 2 in one piece. (e) Core 2 produced in two pieces. (f) Revision of molding and cores to simplify production. (g) Further simplification of mold and core design

Fig. 17

Casting in which core holes A and B were enlarged, and hole C was added, for additional support of the cores. Dimensions given in inches

in the mold. A three-piece core (Fig. 16b) is required. The preparation of each section of the three-piece core entails as much work as would the making of a single large core for the entire casting (if such were possible). To understand the details of preparing a core section, the reader may begin by considering the cross section of the core box used for core 2, which is shown in Fig. 16(c). To produce the required core shape, both a loose ring and a loose center plug are needed. The sand used to fill the core box must be introduced in the limited space separating the ring from the plug. Thus, packing the sand in the core is difficult and time-consuming.

After packing the sand and removing the ring and plug, it is not possible to set the core down on its upper face without sagging of the core. As shown in Fig. 16(c), this upper face or contact area consists of a relatively narrow ring beyond which a portion of the core extends. If only a few castings were to be produced, it would be feasible to prevent core sag by filling the cavities formed by the ring and the plug with bedding sand. For production quantities, however, a core drier (Fig. 16d) is required. The core drier provides greater stability when the core is removed from the box. An individual drier is required for each core placed in the baking oven.

An alternative method for making core 2 is shown in Fig. 16(e). Here, the core is produced in two halves, each having a large flat face that makes the use of core driers unnecessary. The large core box opening also simplifies packing of the sand. After baking, the two halves are pasted together to form the complete core. The possible disadvantages of this method are the chance of misaligning the halves in assembly and the effect of parting on the accuracy of dimensions measured across the parting line. Cores 1 and 3 are prepared in a manner similar to that used for core 2. Core 1 is placed in mold section 1, and the remaining cores and mold sections are assembled as shown in Fig. 16(b). Although sectional cores and molds provide a high degree of flexibility in forming complicated shapes, they have certain inherent disadvantages. Among these is the possibility of movement, which would cause dimensional variation in the casting. Also, at every junction of the loose pieces in the pattern or core box, and at joints between mold and core sections, there is a possibility of forming seams or fins, which would have to be removed at extra cost. The first of two revisions of the original design of this casting is shown in Fig. 16(f). This redesign provides a maximum core diameter smaller than the minimum mold diameter through which the core must pass. The contour of the outer midsection of the casting requires the use of loose pieces in the core box forming core 1. This design eliminates sectional cores and permits the use of a conventional copeand-drag mold. A further design improvement (Fig. 16g) simplifies the core box producing core 1 and results in a cleaner casting with less dimensional variation. Core Fragility. It is important that the smallest dimensions of the core be large enough to facilitate production of the core. Thin or small-diameter cores are extremely fragile and are a cause of both core scrap and casting scrap. In the following example, a casting was redesigned to prevent core breakage. For example, casting the pylon shown in Fig. 17 presented a problem because of core fragility. View (a) shows the as-cast shape of the part. View (b) is the machined casting cut in half. In this casting as originally designed, holes A and B, formed by cores 1 and 2, respectively, were all 13 mm (½ in.) in diameter. Additional support for core 1 was provided by the core print for the 25 mm (1 in.) diameter core at its left end. Core 2 was supported solely by the four (two on each side) 13 mm (½ in.) diameter core prints. This design was unsatisfactory because, with normal foundry handling, the 13 mm (½ in.) diameter core sections were too readily broken off the main core bodies. Also, it was difficult to remove the core sand through the small holes in the casting. The casting was redesigned. Holes A were enlarged to be 19 mm (3/4 in.) in diameter. These four holes (two on each side of the

592 / Expendable Mold Casting Processes with Permanent Patterns

Fig. 18

Door-fitting casting that was redesigned from cored to solid, to reduce rejects caused by heat retention of the core. Dimensions given in inches

Fig. 19

Sand-cast fitting with well-proportioned cored passages. See text for discussion. Dimensions given in inches

Fig. 20

Thin center section, formed by two cores, was not readily accessible to gating and risering. See text for discussion.

casting), plus the 25 mm (1 in.) diameter core print at the left end, now provided adequate support for core 1. In core 2, holes B were enlarged to 16 mm (5/8 in.) in diameter, and hole C, 19 mm (3/4 in.) in diameter, was added for additional support. Increasing the size of holes A and B not only eliminated the problem of core breakage but also increased dimensional accuracy of walls formed by the cores, through more positive placement and anchorage during the pouring and solidification of the metal. Venting of the cores was also improved, and removal of core sand was made easier. Thin Core Sections. The separation of heavy masses of metal with thin core sections should be avoided. If too much metal is adjacent to a core or surrounds it, the core will be unable to dissipate the heat from the molten metal. This will result in hot spots in the casting during cooling and solidification. Such hot spots inevitably give rise to shrinkage or porosity, which can cause rejection of the castings.

An additional problem is the inability of these thin cores to vent core gas adequately, which can result in blowholes. Also, metal penetration of the core may cause difficulty in core removal. The designer should attempt to forestall these problems by increasing the cored area or thinning the metal walls that surround it, or both. Often, the thickness of metal wall sections can be decreased, without sacrificing strength, by the use of ribs. In the following example, the problem of heat retention was solved by eliminating a thin core between two heavier sections. Example: Redesign of Casting That Eliminated Thin Core Section That Caused a Hot Spot (Fig. 18). The presence of a thin core section between two heavier masses of metal presented a problem in production of the turbine door fitting shown in Fig. 18. This sand casting was made of magnesium alloy AZ63A and, as originally designed, had two isolated walls, each 10 mm (0.38 in.) thick, separated from each other by a core section 6 mm (0.23 in.) thick. When the casting was poured, this core section was too thin to dissipate the heat from the surrounding molten metal. A hot spot developed, which resulted in a heavy shrinkage defect on each inside surface of the bosses, as shown in Fig. 18. Rejection rate was 90%. The casting was redesigned to eliminate the core section between the bosses, so that the metal section was cast solid in this region. Although this required a milling operation to remove the added metal, rejections fell to 3%. Figure 19 shows a sand-cast fitting with cored passages exemplifying good design for any process that uses expendable cores, because:
The cores are large enough in cross section

to enable them to support themselves without sagging and to withstand normal foundry handling without breaking.
The cores have sufficient mass, in relation to that of the surrounding metal, to dissipate heat rapidly enough to prevent hot spots that can cause shrinkage defects.
The cores can be easily and adequately vented, to prevent any problems arising from generation of core gas.


The presence of a core print at each of the

three areas where the core emerges from the casting assures that the core can be accurately positioned and that it can be firmly anchored to remain in place while molten metal is filling the cavity.
The straight-through shapes of the cored sections permit easy removal of core sand after the casting has cooled. Thin-Wall Casting Sections Formed by Cores. The minimum thickness of walls formed entirely by cores is governed by the same considerations as those that control the minimum thickness of walls formed by the mold. Most important is that thin sections be given access to an adequate supply of molten metal, to prevent cold shuts or misruns. This molten metal can be supplied either from adjacent heavier sections or from risers. The thin-wall magnesium sand casting shown in Fig. 20 had a center web formed by two adjacent cores. The necessity for core prints to anchor the two cores made it difficult to gate and riser this casting properly. Consequently, misruns and cold shuts, particularly in the thin center web, occurred frequently when castings were made in production. In producing the wave-guide power divider shown in Fig. 21, an aluminum alloy plaster mold casting, the 1.0 mm (0.040 in.) vertical sections, formed by cores, were difficult to fill with casting metal. The metal cooled so quickly when poured into the mold at normal casting temperature that it lost the fluidity necessary to fill the thin sections. Increasing the temperature of the metal caused shrinkage of the heavy sections and excessive formation of gas. Whenever possible, extremes in section size should be avoided, preferably by increasing the thickness of thin sections or by padding the thin sections.

Coring of Tortuous Passages For reasons of function, many castings are required to have tortuous internal passages of small diameter. Sand cores for these passages are difficult to produce. Proper ramming of the

Coremaking / 593 During recent years, however, considerable progress has been made in the development of techniques employing metal tubing for these cores. These give the designer greater freedom in the inclusion and arrangement of passages and permit highly complex design. Three different techniques are currently available:
The older method of drilled passages plus

complex, bulky external tubing. When feasible, this may still be the most economical method.
Lined passages produced by pouring aluminum or magnesium around tubing that remains in the casting
The most recent method, which involves the introduction of unlined passages into sand and permanent mold castings

Fig. 21

Wave-guide casting that was difficult to produce because the 1.0 mm (0.040 in.) sections, formed by cores, froze prematurely

The unlined passages require cores that can be dissolved or otherwise removed from the casting. These cores are usually made of steel, copper, glass, or refractory material. For practical purposes, the unsupported length of such cores depends not only on the diameter but also on the design of the core. Large-diameter cores support themselves over a longer span than do cores of smaller diameter, because increasing the diameter increases the rigidity of the core. Best design practice allows for at least a 6.4 mm (¼ in.) wall of cast metal around a passageway. In some castings, a 4.8 mm (3/16 in.) wall may be adequate, but using a wall of less than 4.8 mm (3/16 in.) invites trouble.

Green Sand versus Dry Sand Cores Fig. 22

Bearing housing that was redesigned to use green sand cores to form the inside diameter

Fig. 23

Coring limitations of permanent mold castings are illustrated by this simple casting. (a) The corner radius of the cored passage must be sacrificed if metal cores are to be used to achieve production at the most economical level. (b) Plaster or sand cores can provide the inside radius, but casting cost will be higher than for (a).

sand in the core box, in conformity with the outlines of the intended core, is often impossible. Because of the shape of the cores, they must be supported by core driers during baking. Also, such cores are easily broken in normal foundry handling. The designer of castings has been greatly restricted as to the passages that could be cast economically.

Some cavities can be cored by the use of green sand cores (cores formed from the molding sand and generally an integral part of the pattern and mold). Because these cores are formed when the mold is made around the pattern, they usually add no expense to the production of a casting and save the cost of producing and placing equivalent dry sand cores in the mold. Although green sand cores are less expensive than dry sand cores, they are at a disadvantage in several respects. Because they are less rigid than dry sand cores, they are less resistant to movement or deformation under the static pressure of the molten casting metal. This is an important consideration when wall thicknesses must be held to close tolerances. Dry sand cores can provide smoother casting surfaces, and thus, they are preferable when surface finish is a consideration. Furthermore, green sand cores usually require more draft than would be normal for the same length of draw on an outside surface of the pattern. This is a precaution to prevent the core from breaking away from the mold when the pattern is withdrawn. Redesign of a casting to employ green sand cores is described in the following example. Example: Casting Redesign That Combined Two Cores into One, Using Green

Sand Cores to Form the Inside Diameter (Fig. 22). As originally designed (Fig. 22a), a thrust bearing housing, sand cast in malleable iron, required a baked sand core and one of green sand. It had a deep circular lightener pocket on one face and a circular oil passage opening to its outside diameter. The lightener pocket was too deep to be made in green sand with normal processing, so a ram-up core was used. This core also formed a portion of the inside diameter of the casting, as shown. A ring core was required also, to form the circular oil passage. Since the pocket was merely a lightener and the function of the oil passage was noncritical, no precise dimensions or tolerances were involved. A redesign, shown in Fig. 22(b), combining the oil passage and the lightener, was no heavier and required only one baked sand core. The inside diameter of the casting was completely formed using green sand cores, half in the drag and half in the cope. They met at one of the steps and were produced as the pattern was molded. The use of green sand cores substantially reduced production costs.

Cores in Permanent Mold Castings Cores in the permanent mold process may be of gray iron, steel, sand, or plaster. Metal cores may be either movable or stationary. Stationary cores must be perpendicular to the parting line to permit removal of the casting from the mold and must be shaped so that the casting is readily freed. Metal cores not perpendicular to the parting line must be movable, so they can be withdrawn from the casting before it is removed from the mold. When baked sand or plaster cores are used in a permanent mold, the process is referred to as semipermanent molding. With sand or plaster cores, which are expendable, more complex shapes can be produced than with metal cores. The two cored passageways shown in Fig. 23 illustrate the limitations of cores in permanent mold castings. The cored passageway shown in Fig. 23(a) could be produced with metal cores. Sand or plaster cores would be needed to provide the inside radius of the passageway shown in Fig. 23(b), which would also be more expensive to produce. Cored holes in permanent mold castings can usually be held to closer tolerances, of both size and location, than in sand castings. Both movable cores and stationary cores can be machined to close dimensions and accurately located. Although sand cores in permanent molds are no more accurate dimensionally than in sand molds, in the metal molds they may be positioned and retained in place with more accuracy than in sand molds. In general, the problems of expendable cores in permanent mold castings and in sand castings are similar. The problems with metal cores are similar to those in die castings.

594 / Expendable Mold Casting Processes with Permanent Patterns Table 2

General dimensional relations for core cavities in investment castings

Length X, in.

For a typical ferrous casting 0.250 0.500 0.750 1.000 1.250 1.500 2.000 2.500 Inside diam A, in.

Diam Y, in.

Min wall, in.

0.125 0.250 0.375 0.500 0.750 1.000 1.500 1.750

0.030 0.040 0.050 0.060 0.060 0.060 0.060 0.060

As-cast roundness (TIR), in.

0.092–0.250 0.251–0.500 0.501–1.000 Over 1.000, add 0.020 in. per in. A diam, in.

0.012 0.016 0.020

B diam, in.

As-cast concentricity (TIR), in.

0.750 1.000 1.500 2.000

0.008 0.010 0.016 0.020

0.250 0.500 0.750 1.000

Minimum diameter Y Depth W, in.

Ferrous, in.

Nonferrous, in.

Min draft

0.188 0.250 0.500 0.625 0.750 0.750 1.000 1.000

0.188 0.250 0.312 0.375 0.438 0.500 0.562 0.625

0 0 0 0 0 150 0 150 0 150 0 150

0.250 0.500 0.750 1.000 1.250 1.500 2.000 2.500 H, in.

D, in.

Z, in.

Parallelism (a), in.

0.250 0.500 0.750 1.000 1.250 1.500 2.000 2.500

0.125 0.250 0.375 0.500 0.625 0.750 0.875 1.000

0.500 0.750 1.000 1.250 1.500 1.750 2.000 2.250

þ0.005  þ0.005  þ0.005  þ0.010  þ0.012  þ0.013  þ0.015  þ0.015 

T, in.

X, in.

a0 , in.

Y, in.

b0 , in.

0.125 0.250 0.125 0.250 0.500 1.000 2.000

0.250 0.500 0.750 1.000 0.500 0.500 2.500

þ0.004  þ0.005  þ0.005  þ0.005  þ0.008  þ0.010  þ0.015 

0.125 0.250 0.500 0.750 0.250 0.250 1.500

þ0.003  þ0.004  þ0.005  þ0.005  þ0.005  þ0.008  þ0.010 

X

Y, in.

Z

Y, in.

0.250 0.500 0.250 0.500 0.750

1 1

þ1  þ1   þ1 

15 30 45

X diam, in.

/64 /32 /16 1 /16 1 /16 1/16 1 /8 1

a0 min



0.250 0.500 1.000

R, in.

a0



0.500 0.750 0.500 0.750 1.500

15 30 60 90 120 and up



þ0   þ0  þ0   þ0   þ0 

300 300 300 450 600

A, in.

B, in.

C, in.

D, in.

E, in.

a0 , in.

0.500 0.750 1.000 1.250

1.000 1.500 2.000 2.500

0.250 0.500 0.750 1.000

0.750 1.000 1.500 2.000

0.375 0.500 0.750 1.000

þ0.004  þ0.006  þ0.008  þ0.010 

(a) Parallelism may be regarded as the tolerance for dimension Z.

Cores in Investment Castings Holes of various shapes can usually be formed with little or no difficulty by the investment casting process. Limiting factors are the size of the opening, the ratio of the size of opening to the length of core, and the complexity of the shape as it affects metal fluidity and wall thicknesses. Generally, in investment casting, through holes of less than 2.4 mm (3/32 in.) diameter and blind holes of less than 4.8 mm (3/16 in.) diameter require special coring techniques that add significantly to the cost of the part. Blind holes that have a length-to-diameter ratio of more than 2 to 1 are more difficult to cast, because of breaking or shifting of the unsupported end of the core. General rules for designing cored holes in investment castings are embodied in the dimensional recommendations of Table 2. Because other factors of the design can influence these values, and because some foundry methods have greater capabilities than others, these figures are useful primarily as a general guide. Seven typical examples of cored holes and cavities easily incorporated in investment castings are illustrated in Fig. 24. None of these coring situations presented any difficulty. Small Cavities. The investment casting process permits coring of cavities that would be too small to be cored in sand castings. Typical of such cavities is the internal oil slot shown in Fig. 25. When the part shown was produced as a sand casting, this slot had to be milled in, and the hole casting, this slot had to be milled in, and the hole leading to it drilled. Producing the part as an investment casting permitted the slot to be cored, and no subsequent machining on it was required. Abutting Cores. In investment mold castings, as in permanent mold castings (Fig. 23), where holes connect internally to form a sharp corner, the use of special cores is not required. In an investment casting, a special core would be required for the curved hole in Fig. 23(b), whereas that in Fig. 23(a) could be produced with regular mold equipment. In die casting, the design shown in Fig. 23(a) would be possible to manufacture, but the design shown in Fig. 23(b) would be impossible. Where sharp corners will reduce product performance (as by turbulence in fuel valves, fraying of insulation in electrical wire housings, or signal error in radar components), fillet radii should be incorporated. This requires expendable cores, which are more expensive than steel core pins. Figure 26(a) shows an example of an investment casting that required expendable cores. As designed, this casting could not be produced with steel core pins in the die used to produce the wax or plastic patterns.

Coremaking / 595

Fig. 24

Examples of cored holes and cavities easily made in investment castings. Dimensions given in inches

Fig. 25

Oil slot was easily cored when part was produced as an investment casting but had to be milled in when part was sand cast. Total savings of 41¢ per casting were realized by changing from sand to investment casting. Dimension given in inches

Fig. 26

Special coring practice involving additional cost would be required to produce this investment casting as shown in (a). Normal practice using metal core pins in the pattern die is possible for the casting as shown in (b). Design (b) is less expensive than (a).

expansion of the core, causing the core to bend. This bending moved part of the core closer to one wall and farther away from the other, thus producing a thinner wall on one side of the casting than on the other. Recommended minimum tolerances for the thickness of walls enclosing tortuous cored passages in investment castings are as follows: Length of hole, mm (in.)

Fig. 27

Because it was completely surrounded by molten metal, the core that formed the passage through this investment casting expanded more in length than the mold when the metal was poured. This caused bending of the core, which influenced the casting wall thickness.

To permit the use of steel core pins, the casting would have to be redesigned as shown in Fig. 26(b). Six core pins would then be necessary. They would be located and pulled from the pattern as shown. An added operation would be required to plug the hole at core 1 after casting. This redesign has five core-pin junctions. It would be virtually impossible to produce these castings without prominent flash at these junctions, because of the difficulty of

Fig. 28

Redesign that eliminated a core and reduced the cost of the casting

maintaining firm butt-type joints. Removing the flash would be an added expense. Unequal Expansion of Mold and Core. In investment castings, unequal heating and expansion of the mold and core can cause nonuniformity of wall thickness. This is particularly likely in castings that have tortuous passages formed by cores completely surrounded by metal. Typical of such castings is the one shown in Fig. 27. When this casting was poured, the mold was able to dissipate heat at a greater rate than the core. Consequently, the core became hotter than the mold and expanded more. The rigid mold resisted the

Under 50 (2) 50 to 100 (2 to 4) Over 100 (4)

Wall thickness tolerance, mm (in.)

þ0.127 (þ0.005)   þ0.25 (þ  0.010) þ0.30 (þ0.012)  

Designing to Eliminate Cores Useful as they are, cores add to the cost of producing castings; hence, the designer should be alert to possibilities for avoiding them. A decision often depends on cost analysis. An example is shown in Fig. 28. In the original design of this casting (Fig. 28a), a core is required to permit molding of the “hook” shape. The possible redesign shown in Fig. 28(b) would permit easy removal of the pattern from

596 / Expendable Mold Casting Processes with Permanent Patterns

Fig. 29

Improved design that eliminated one core and eight ribs from a sand casting. This resulted in a stronger, more economical part.

casting in application. As redesigned, the broader base of the tubular section provided sufficient strength to permit elimination of the ribs. Figure 30 shows a sand-cast launching frame, produced from 4330 steel, that was redesigned to eliminate cores. The original design of this casting called for the coring of two pockets approximately 100 mm (4 in.) deep, to reduce weight. In production, however, it was difficult to attain sound metal in this cored area. At the expense of adding weight, the cored pockets (and the cores) were eliminated in a redesign, as shown. This not only corrected the defects but also facilitated production. The solid center was easily risered and encouraged a beneficial freezing pattern in the entire casting. Simplified Cores. Not all designs or redesigns can eliminate cores, but often cores can be simplified. This was accomplished for the fuel cell interconnect, an aluminum casting, shown in Fig. 31. A semipermanent mold casting as originally designed, it required a contoured sand core to form the inside configuration shown. A revision to eliminate defects encouraged a further revision to eliminate defects encouraged a further revision to simplify the core. By moving the wall to coincide with the inside contour of the flanges, a simple straight-through metal core sufficed. The redesigned castings (now permanent mold, because of the metal core) were more accurate and less costly.

Coring versus Drilling

Fig. 30

By eliminating two cored pockets from this 4330 steel sand casting, a better solidification pattern was established and defects were eliminated. Dimension given in inches

the sand, eliminate the need for a core, and effect a savings in molding cost. Figure 29 shows a sand-cast malleable iron wheel hub for which redesign eliminated a ring core and, at the same time, provided a stronger casting. As originally designed (Fig. 29a), the eight ribs and eight small bosses prevented this casting from being molded with the parting line parallel to the axis of the hole. Furthermore, adjacent to the flange, the casting had a cross section smaller than either the flange or the extreme end of the casting. The undercut section that was thus formed prevented the pattern from being

withdrawn from the mold in a direction perpendicular to the mounting flange. A ring core, as shown, was necessary to produce the shape. By revising the casting as shown in Fig. 29 (b), the need for the ring core was eliminated and the shape could be withdrawn easily from the mold. By broadening the base of the tubular section, the eight ribs were also eliminated. In the original design, the small diameter of the tubular section at the junction with the flange section was unable to withstand the forces of service. Eight strengthening ribs were required, to assure satisfactory performance of the

Although for many castings coring is essential to successful production, for others it is advisable to omit cores and to remove excess metal by other means. The choice may be based on considerations of soundness, dimensional accuracy, economy, or producibility. For example, if a casting is to have one or more round holes, these may be produced with greater accuracy or economy by subsequent boring or drilling, rather than by cores. The cost that coring adds to casting must be considered as only part of the cost of producing a hole, if the hole must later be machined. The cost of coring and subsequent machining must be weighed against that of drilling in determining whether or not a part should be cast solid, if this is otherwise feasible. As a general rule, it is often cheaper to core larger holes if an appreciable amount of metal can be saved, or if machining operations can be speeded up or eliminated. It is usually more expensive to core smaller holes, unless a large amount of machining expense can be saved. Required holes in castings with heavy outside sections should be obtained by boring or drilling, rather than by coring, if the holes are round and if soundness is more important than cost. In the casting illustrated in Fig. 32, the core adjacent to the gate impeded flow of metal into the mold. This caused cold shuts at the end of

Coremaking / 597

Fig. 33

A cored hole 21 mm (0.84 in.) in diameter was incorporated in this casting to reduce machining cost. It was less expensive to core and drill than to drill through solid metal.

ACKNOWLEDGMENT Adapted and reviewed from:


Sand Cores and Coremaking, in Forging and

Fig. 31

A sand core was required to produce the center configuration of this semipermanent mold aluminum casting. By increasing the wall thickness and placing the inner face of the wall in line with the inner edge of the flanges, defects were eliminated, and a simple metal core produced the center cavity. Dimensions given in inches

Casting, Vol 5, Metals Handbook 8th Edition, American Society for Metals 1970, p 209–222
Designing for Economical Coring, Casting Design Handbook, American Society for Metals, 1962, p 9–21 SELECTED REFERENCES
Developments in Moulding and Coremaking,








Fig. 32

Better flow of molten metal was obtained by eliminating the 10.3 mm (0.406 in.) diameter core adjacent to the gate of this investment casting. This enabled the metal streams at the extreme end of the cavity to unite properly. Dimensions given in inches

the casting opposite the gate. Dispensing with this core permitted metal to flow freely to all parts of the cavity, so that the cold shuts were eliminated and sound castings produced. The previously cored hole was later drilled. In the sand casting shown in Fig. 33, it was cheaper to core and finish-drill the 21 mm (0.84 in.) diameter hole than to cast solid and produce the hole entirely by drilling. The cost

of metal saved was greater than the cost of coring. Actual savings in this instance amounted to 1½¢ for each casting. If the cored hole had been much smaller, however, this would not have been true, because the core cost would have remained relatively constant, whereas the value of the metal saved would have been reduced in proportion to the volume of the hole.







Foundry Trade Journal, Vol 180 (No. 3641), Jan-Feb 2007, p. 10–15. Improving Process Efficiencies in Coremaking, Foundry Management and Technology, Vol 133 (No. 4), Apr 2005, p 16–19. T. Lewis, K. B. Horton, Establishing New Rules for Coldbox Coremaking, Modern Casting (USA), Vol 93 (No 1), Jan 2003, p 43–45 Latest achievements in coremaking flexibility, Cast Metal Times (UK), Vol 4, (No 4), JuneJuly 2002. p 27 J. Werling, Evaluation of Sands for the Coremaking Process: A Practical Approach, Transactions of the American Foundry Society and the One Hundred Sixth Annual Casting Congress, Kansas City, MO; USA; 4–7 May 2002, p 641–650 J. Werling, Evaluation of Sands for the Coremaking Process: A Practical Approach, Transactions of the American Foundry Society, Vol 110: Part 1, 2002, p 641–650 J. Cavanaugh, D. Gilson, Examining Key Variables of a Coldbox Coremaking Operation, Modern Casting (USA), Vol 90 (No 6), June 2000, p 32–34. Recent Studies Comparing Coremaking Processes, Modern Casting, March 1996, p 39–42

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 598-616 DOI: 10.1361/asmhba0005252

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Shell Molding and Shell Coremaking Updated by Scott McIntyre, Grede Foundries, Inc

THE SHELL PROCESS was first developed in World War II Germany and gained acceptance throughout industrialized nations during the 1960s and 1970s. The most recognizable characteristics of the process are:

racy and precision, allowing metalcasters to produce castings that are nearer to net shape than attainable by some other processes. The process is also flexible and adaptable and may allow the foundry to manipulate core/mold

heat-transfer rates, mold and core strengths, dimensional stability, or other properties that impact the final qualities of the cast product. Examples of components cast in shell molds are shown in Fig. 2.


Cores and molds are made from a dry, free-

flowing, resin-coated sand.


Cores and molds are cured into useful

shapes through the application of heat.


Cores and molds are usually produced in

thin sections, often resembling shells of bonded sand (Fig. 1). Like any other core or moldmaking process, the shell process has advantages and disadvantages, usually defined by the specific metal-casting application. The most common perceived disadvantage is that the process is slow. Because the process requires heat curing, the rate of shellmold- and coremaking is often limited by the rate of heat transfer through the resin-coated sand. To an extent, production rates for shell molds and cores can be improved through materials selection, tooling design, and equipment design. Nevertheless, the process will not reach productivity rates equivalent to some high-volume molding and coremaking processes. However, the advantages of the process can outweigh its perceived disadvantages. The shell process is capable of producing cores and molds with a high level of dimensional accu-

Fig. 1

Cope side of a shell mold

Fig. 2

Shell-mold castings. (a) Shell-mold casting. (b) Wheel cylinder with O-ring grooves. (c) crankshaft

Updated from: Metal Handbook, 8th Edition, Volume 5, Forging and Casting, 1970

Shell Molding and Shell Coremaking / 599
High cost of patterns, which must be

Applicability





Shell molding is used for making production quantities of castings that range in weight from a few ounces to approximately 180 kg (400 lb) apiece, in both ferrous and nonferrous metals. Castings weighing at least 455 kg (1000 lb) apiece have been made in shell molds but not in production quantities. As compared to conventional green sand molding, the shell-mold process can produce castings with greater dimensional accuracy and thus reduce the amount of machining required for completion of the part. Shell molds impart a smoother surface to castings than molds made from green sand or baked sand, as shown by the following comparison for small steel castings (up to 2.3 kg, or 5 lb, weight) made by three processes:








Properties and Selection of Shell Sands The most common base aggregate for shell sand is silica sand. Sieve analyses and other characteristics of typical sands used for shell molding are given in Table 1. The silica and zircon contents of these sands are high, which suggests that the sands are generally low in organic material, and clay content is low. Low percentages of organic material and clay, obtained through washing, minimize resin consumption. When excessive amounts of organic material and clay are present, more resin is required to strengthen the mold. Resin consumption also increases as the sand grain size becomes finer; therefore, the coarsest finish required should be used. A specific silica sand is usually chosen on the basis of the application, economics, and

Surface finish Process

Shell mold Baked sand mold Green sand mold

mm

min.

3–6 6–12 12.5–25

125–250 250–500 500–1000

In addition to producing castings with greater accuracy and smoother surfaces, other advantages of shell molding include less sand required and fewer restrictions on casting design than for green sand molding. The limitations or disadvantages of shellmold casting are:
Maximum casting size and weight are

limited (see the aforementioned). Table 1

machined from metal High cost of resin binder Relative inflexibility in gating and risering. Gates and risers must be incorporated, at least in part, into the shell-mold pattern. Shrinkage factors vary with casting practice. (Two foundries using the same pattern may pour castings with different dimensional variations.) More equipment and control facilities are needed, such as for heated metal patterns. Manufacturing rate, relative to competing processes

characteristics of the sand, including sand grain size distribution, silica content, silica shape, and other properties relating to its packing density. The shell process, however, also works well with many other naturally occurring and manufactured base aggregates, including zircon, olivine, or mullite aggregates. These may be chosen in circumstances that require different heat-transfer or thermal expansion rates. Grain Size and Shape. In shell-mold casting, the greater strength available in the resin bond allows the use of a finer grain size for a given casting weight than is possible in sand casting. Except for aluminum castings, best results are obtained with sand having a narrow screen distribution, as is suggested by the sieve analyses for the sands given in Table 1. For aluminum castings, the mold sand should be a four-screen, subangular sand with approximately 5% pan fines of 325 mesh. Each delivery of sand should be checked for grain-size distribution. Even when sand is segregated in storage bins, periodic in-plant screen analyses may be necessary for adequate sand control. Shell-molding sands may be round, subangular, or angular. Round sands require the least resin binder and have the highest cold strength and permeability. These properties, in addition to lower hot strength, make the round sand grains particularly suitable for coremaking. Angular grains, because of their high hot strength, are preferred for mold construction. Subangular grains have properties that are intermediate between those of round grains and

Characteristics of typical shell-molding sands from various geographic locations Sieve analysis

Origin of sand

Silica sands McConnellsville, Oneida Co., NY Providence, RI Southern NJ Huntington and Mifflin Co., PA Vassar, Tuscola Co., MI Ottawa, LaSalle Co., IL

Geauga Lake, OH Sewanee and Ochlochnee, GA Overton, NV Pittsburg, CA Zircon sands Great Barrier Reef, Australia Ocean beaches, FL

Sand number

AFS fineness number(a)

Percent retained on National Bureau of Standards sieve No.: 40

60

70

100

140

200

270

On pan, %

Grain shape

Silica or zircon, %

Color

Clay, %

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

113 177 141 168 87.1 124.0 123 143 83 141 83 108 124 137 78 91 109 125 106 124 87 105

0.02 0.02 ... 0.20 ... ... ... ... ... 0.1 ... ... ... ... 0.28 0.20 ... ... ... ... ... 0.20

0.04 0.04 0.06 0.40 0.70 ... 1.20 ... 0.72 0.20 ... ... ... ... 3.74 1.00 ... ... 1.10 ... 0.72 2.15

1.56 0.06 3.00 1.60 8.80 0.30 2.72 11.50 13.88 1.05 21.00 0.04 ... 0.08 26.30 6.20 0.04 ... 5.70 ... 13.8 10.6

23.80 0.88 15.40 7.80 43.1 15.3 15.48 12.30 50.80 10.00 45.00 28.20 0.20 2.81 39.04 35.20 28.70 0.20 24.30 0.20 48.70 27.00

33.82 19.42 26.80 19.20 35.30 41.60 29.80 27.30 23.80 36.20 23.00 44.80 8.40 43.19 23.08 43.80 45.20 8.40 48.80 8.40 24.70 34.85

33.20 41.80 29.60 31.00 10.60 31.60 28.00 3.80 9.44 25.00 7.00 9.44 50.20 34.91 5.09 7.80 18.80 50.20 13.60 50.70 9.40 15.50

4.08 13.34 13.40 19.60 0.90 7.80 11.80 18.80 1.92 16.95 4.00 1.92 28.20 12.69 1.85 1.00 5.00 28.20 3.20 28.20 4.20 5.30

2.82 24.42 11.20 20.00 0.60 3.40 8.42

Subangular

96.0

Pink

1–2

Subangular

97.0

Buff

1–1.5

Subangular

99.0

Buff

0–0.5

Subangular

98.0

White

0–0.5

0.44 8.25 ... 0.44 8.80 5.55 0.62 1.20 0.34 8.80 2.80 6.70 1.70 4.50

Subangular

93.0

Tan

1.5–2

Round

98.0

White

0–0.5

Subangular

99.5

White

None

Subangular

97.0

Translucent

1–1.5

Subangular

96.5

Pink

0–0.5

Subangular

98.0

White

0–0.5

23 24 25 26

113 141 108 146

... ... ... ...

0.04 0.60 ... 1.46

1.56 3.00 0.04 7.68

23.80 26.80 28.20 14.24

33.80 26.80 44.80 20.14

36.70 29.60 18.80 24.80

2.08 13.40 5.00 16.10

2.80 11.20 2.60 15.10

Round

99.4

White

None

Subangular

99.0

White

None

(a) AFS, American Foundry Society. Source: Ref 1

600 / Expendable Mold Casting Processes with Permanent Patterns weighed in their solid form. Any change in the solids content of the resin requires an appropriate adjustment in the hexa-wax ratio. If the proper adjustment is not made, either brittle shells or soft (spongy) shells will be produced. The foundry cannot control the solids content of a resin except by diluting it with alcohol. However, resins with a range of controlled solids levels are commercially available. These resins can be used to maintain the correct solids content. Phenol-formaldehyde resins, produced with an insufficiency of formaldehyde so they develop thermoplastic characteristics, can be made to harden permanently, or to set, on heating by the addition of hexamethylenetetramine (commonly called hexamine or hexa). The storage life of these resins is limited, and care must be taken to avoid any increase in temperature during storage. Frequently, the resin is refrigerated to extend storage life. Phenol-formaldehyde resins are most commonly used in shell molding, because, when combined with sand, they have high strength, resistance to heat and moisture, and good flowing and curing properties. Generally, the phenolic resins used for shell molding are of the thermosetting, two-stage type, often referred to as novolak resins. The two-stage designation indicates that the resin was formed with a molar excess of phenol in the presence of an acid catalyst and is similar to a thermoplastic. To obtain thermosetting properties, the catalyst hexa is added during coating. When the resin is formed with a molar excess of formaldehyde with a basic catalyst, it is referred to as resol, a one-stage resin. This resin is reactive as made and does not require the addition of a catalyst during coating. Because of this, the resin will gradually age, and if it is to remain suitable for use in shell molding, it must be refrigerated. Also, the one-stage and urea-modified phenolics absorb moisture and therefore are not widely used to coat sands that will be used for shell molds and cores. Sand that is properly coated with phenolic novolak resins has excellent resistance to moisture absorption and remains free-flowing.

angular grains and are the most widely used for molds. Moisture Content. Dry sand is necessary for the cold resin-coating process. Moisture can cause dry resin to ball up, thereby decreasing the amount of resin available to coat the sand grains. In the hot coating process, moisture is generally not a problem, because it is driven off when the sand is heated to the minimum temperature of 120  C (250  F) before coating. Breakage of molds or cores during molding reflects low strength. This can be caused by the use of a finer sand screen distribution without a proportionate increase in resin content, by excessive amounts of clay or organic material, and by round sand grains. Breakage of cold molds or cores during handling can be caused by many variables in shell production. Sand can be responsible when sand properties change so that insufficient resin coating of sand grains results. This may be caused by the shape or size of the sand grains. Round grains and coarse grains have the highest strength for the least amount of resin.

Resins The synthetic resins used in shell molding are thermoplastic resins to which setting or polymerizing agents have been added to produce thermosetting characteristics. Resin powders are purchased according to the properties that are important in the formulation of resinsand mixtures: melting point, cure rate, flow rate, particle size, and hexa content. Table 2 lists properties of ten synthetic resins used in shell molding. Solids content of a resin is the amount or percentage of binder or resinous material that is available for bonding the sand. The remainder of the resin is alcohol and water, which are removed in the coating cycle. The only limit to the maximum solids content of a resin is that imposed by an excessive increase in viscosity. The highest solids content is 80%. Coating formulations are usually based on resin solids, and hexa and wax are added as Table 2

Properties of ten synthetic resins used in shell molding Melting point

Resin No.(a)

1 2 3 4 5 6 7 8 9 10



C

100–110 105–115 95–100 110–115 105–110 110–115 95–100 100–110 100–110 110–120

Flow in 17 min at 125  C (257  F)(b)

F

Curing time at 150  C (300  F), s

mm

in.

Particle size (% on 200-mesh screen)

Hexa concentration, %

210–225 220–235 205–215 225–240 220–230 225–240 205–215 210–225 210–225 230–250

25–35 35–45 35–45 40–47 45–50 50–55 55–60 55–60 65–75 80–100

10–17 16–22 10–17 10–15 16–22 22–28 16–22 22–34 22–28 24–32

0.4–0.7 0.6–0.9 0.4–0.7 0.4–0.6 0.6–0.9 0.9–1.1 0.6–0.9 0.9–1.3 0.9–1.1 0.9–1.3

0.5–1.0 0.5–1.0 0.5–1.0 8.0–10.0 1.0–1.5 5.0–10.0 0.1–0.2 0.5–1.0 0.1–0.5 0.1–0.2

13–15 11–13 13–15 13–15 13–15 11–13 11–13 13–15 8–10 8–10



(a) See Table 4 for formulations of molding mixtures in which resins are identified by some of these numbers. (b) Determined by heating a sample of resin on a level plate for 3 min in an oven at 125  C (257  F), tilting the plate at 65 , and after 17 min measuring the resin flow in millimeters. Source: Ref 1

Phenolic resins used in shell molding are:
Novolak varnishes with 60 to 70% solids
Water-borne novolaks with 75 to 80% solids
Flake and lump resins

Novolak varnishes are only slightly soluble in water and are furnished in organic solvents, mainly alcohol. Water-borne novolaks must be used either with hot sand or with air heated to 190 to 260  C (375 to 500  F); flake and lump resins require hot sand. The advantages and disadvantages of the three forms of phenolic resins used in shell molds are listed in Table 3. Viscosity of a resin varies with resin temperature; therefore, the temperature at which viscosity is measured must always be recorded along with the viscosity. If the viscosity of a resin is too high, sand coating may be incomplete, which will result in low-strength shell molds that will break and crack during pouring. The viscosity of resins depends on the solids content, as shown in the following tabulation: Solids, %

58–62 68–72 75–77

Viscosity at 25  C (75  F) centipoises

250–600 1800–4500 8000–14,000

The viscosity of a resin can be reduced by diluting the resin with an organic solvent, such as alcohol. Alcohol Content. Alcohol is the organic solvent ordinarily added to novolak varnishes, which are only slightly soluble in water. The amount of alcohol present in the resin is inversely proportional to the solids content. Excessive alcohol in the resin can result in a prolonged coating cycle, caking, mold peeling, and low mold strength. Water Content. In the manufacture of resin, water is introduced with alcohol. As the molecular weight of the resin increases, the solubility of water decreases. Typical water contents of the three types of phenolic coating resins used in shell molding are as follows:
Novolak varnishes, 2 to 8%
Water-borne novolaks, 8 to 20%
Flake and lump resins, 1% max

An increase in the water content reduces the risk of fire associated with the use of coating resins. However, water should not be added after the resin is received from the manufacturer, because this may cause precipitation of the resin. If resin with excessive water content is used, problems similar to those caused by excessive alcohol may develop. An early indication of excessive solvent, both alcohol and water, is a long mulling cycle. Melting point is an important property of solid resins used in shell molding and is roughly analogous to measuring the “stick point” of a resin-coated sand (see the section “Control Testing of Resin-Sand Properties” in this article).

Shell Molding and Shell Coremaking / 601 Table 3 Advantages and disadvantages of three types of phenolic resins used in shell molding Advantages

Disadvantages

Novolak varnishes


No heating equipment is required.
Solvent removal varies with atmospheric conditions, resulting in variable behavior of resin
Less resin is required to develop the necessary strength.
Afford good release of shell, if wax is added at end of
High incidence of sand caking
Coating cycle is long. preparation cycle
Readily handled in bulk
Solvents are generally flammable. Water-borne novolaks


Yield large volume of high-strength sand per pound of resin
Contain less alcohol (fire hazard reduced)
Coating cycle is shorter than with varnishes.
Readily handled in bulk


Slight caking tendency
High viscosity, which can cause problems in handling
Require special equipment to introduce hot air or hot sand into the muller

Flake and lump resins







Fastest coating cycle Excellent strength properties Minimum caking tendency(a) Contain wax, which eliminates addition of wax at the muller and simplifies inventory


Require hot sand
Cannot be handled in bulk

(a) Because no removal of solvent is required for flake and lump resins

Generally, flake and lump resins are divided into three melting-point groups:
Low melting point, 64 to 73.8  C (147 to

165  F)


Medium melting point, 74 to 85.5  C (166 to

186  F)


High melting point, 86 to 99

210  F)



C (187 to

To a limited degree, the melting point provides an index to the polymerization of the resin; a higher melting point indicates more advanced polymerization. Such measurements, however, can be used only to compare different lots of the same resin or different resins of the same type. Resins that have a low melting point have the fastest sand pickup and thus form shell molds rapidly and to greater thickness. Low-melting resins are slower curing, however, and are more likely to cause “peelback” (delamination of the uncured shell). A resin with a high melting point will produce shells with less bridging on the vertical walls. Sand coated with high-melting resins packs better, forming a denser shell that is less subject to metal penetration. However, if the melting point of the resin is too high, thin, weak shells will be formed. Hexamethylenetetramine (hexa) is required to develop the thermosetting characteristics needed in a resin binder. The amount of hexa used depends on the curing speed desired; a hexa content of 14 to 16% of the resin solids is considered a good average. At this hexa level, both hot and cold mold strength are adequate, and brittleness is not extreme.

For improved hot tensile strength and reduced curing time and temperature, the hexa content may be increased to as much as 20% of the resin solids. However, this will result in reduced cold strength and increased brittleness of a shell mold or core, with the added risk of thermal cracking of the mold during pouring. Increasing the hexa level above 20% of the resin solids reduces hot tensile strength, because the excess hexa acts as filler. Reducing the hexa to 10 to 12% of the resin solids results in molds with higher cold strength but with reduced hot tensile strength. The shells are more plastic at ejection and are less brittle than those that have high hexa content. Additives. Shell sands are often modified with additives. Nature-based additives include substances such as kaolin clay or red iron oxide. Some trademarked, manufactured additives include SphereOx (Chesapeake Specialty Products, Inc.) or Veinseal (IGC Technologies, LLC). Whether natural or manufactured, additives are usually chosen to address the expansion or gas evolution properties of molds and cores at metal-casting temperatures. The use of additives or alternative base aggregates can often allow the foundry to tailor the properties of the molds or cores to specific casting properties or process needs. Lubricants (usually calcium stearate, a mixture of stearates, or silicone) are added to resin-sand mixtures to permit easy release of the mold from the pattern and to improve the flowability of the sand. The tensile strength of the mold is also increased, because sand containing lubricant packs more densely. The amount of lubricant added is usually 2 to 5% of the resin solids. When there is minimum

draft on pattern projections, or there are insufficient or unevenly distributed ejector pins, or cores are unusually complex, the lubricant may be increased to approximately 9% of the resin solids content.

Preparation of Resin-Sand Mixtures A resin-sand mixture for shell molding can be prepared by either of two methods:
Mixing resin and sand according to conven-

tional dry mixing techniques


Coating the sand with resin

The dry resin-sand mixes are subject to segregation and dusting, which makes them unsuitable for mold blowing; they can be used only for dump-box production of molds. The resincoating method has largely superseded dry resin-sand mixing. Coated sand is produced in a mulling machine, by mixing together the sand, a suitable resin, a catalyst (hexa), lubricants, and additives. Typical formulations of resin-sand mixtures for use with various casting metals are presented in Table 4. Mulling conditions vary, and some of the additions to the muller may be in liquid form, but the coated sand must be completely dry when discharged from the muller to ensure that it is free-flowing and suitable for molding. Cold coating is a process in which sand and liquid resin, both at or near room temperature, are blended with other materials to produce resin-coated sand. The liquid resin can be either a one-stage or a two-stage resin, having either water or alcohol, or both, as a carrier. The carrier must be subsequently removed from the resin-sand mixture, leaving dry, free-flowing sand, each grain of which is covered with a thin coating of resin. The main advantage of the cold coating method is that the sand is used at room temperature, thus eliminating the necessity of equipment to heat the sand. Also, the melting point and tensile strength of the coated sand are easier to control by this method. The main disadvantages of the cold method of coating sand are:
The necessity of handling large amounts of

liquids that are not incorporated into the final product
The need for removing these liquids from the sand mix, which greatly increases the overall mixing cycle time
The explosion hazard associated with the evaporation of low-molecular-weight hydrocarbons In the cold coating process, all materials are near room temperature when added to the muller, and the temperatures of the sand and air are not critical. If the sand or resin becomes extremely cold, however, it is more difficult to obtain an effective blending of the materials

602 / Expendable Mold Casting Processes with Permanent Patterns Table 4 Typical formulations of resin-sand mixtures for single-investment shell molds used in the casting of various metals Typical applications Sand-resin formulation(a)

Low-carbon and alloy steels 63 parts silica sand (AFS 140) 30 parts zircon sand 5% resin 2% silica flour Medium-carbon and high-carbon 60 parts sand No. 25 40 parts sand No. 23 3% resin No. 7 0.75% manganese dioxide powder 0.5% calcium stearate

Size of castings

Types of castings

Section thickness: 6.4–51 mm (¼–2 in.) Overall linear dimension: 1219 mm (48 in.) Maximum boss diameter: 76 mm (3 in.) Maximum flat area: 12,900 mm2 (20 in.2) Weight: 0.5–135 kg (1–300 lb)

Valve bodies, pump casings, transmission casings, pintles, plates, rotary cams, pump impellers, bearing housings, crankshafts

steels Section thickness: 3.2–9.5 mm (1/ 8–3/ 8 in.) Overall linear dimension: 205–255 mm (8–10 in.) Maximum boss diameter: 13 mm (½ in.) Maximum flat area: 1935 mm2 (3 in.2) Weight: 31–373 g (1–12 oz)

Levers, bellcranks, pawls, sleeves, cover plates, gear pinions, bushings, racks, quadrants

Sheaves, pulleys, flywheels, valve seats, plungers, gear wheels, pinions, racks, connecting rods

Gray iron, classes 30, 35, 40, and 50 Section thickness: 6.4–25 mm (¼–1 in.) 60 parts sand No. 12 Overall linear dimension: 1219 mm (48 in.) 35 parts sand No. 9 Maximum boss diameter: 76 mm (3 in.) 4% resin No. 8 Maximum flat area: 12,900 mm2 (20 in.2) 2% treated bentonite 1.5% calcium stearate

Motor housings, pinions, transmission casings, gearboxes, valve bodies, cylinder heads, pipe fittings, forming dies

Ductile iron, 60–45–15 60 parts sand No. 10 35 parts sand No. 12 4% resin No. 7 1.5% iron oxide 1.5% calcium stearate

Brackets, levers, support columns, side frames, clutch plates, crankshafts, cam shafts, gear wheels, racks, quadrants

Section thickness: 6.4–19 mm (¼– /4 in.) Overall linear dimension: 610 mm (24 in.) Maximum boss diameter: 51 mm (2 in.) Maximum flat area: 5200–6500 mm2 (8–10 in.2)

Aluminum alloys 355 and 356 60 parts sand No. 12 35 parts sand No. 9 4% resin No. 7 2% treated wood flour 1.5% calcium stearate

Section thickness: 3.2–19 mm (1/ 8–3/4 in.) Overall linear dimension: 914 mm (36 in.) Maximum boss diameter: 51 mm (2 in.) Maximum flat area: 12,900–19,400 mm2 (20–30 in.2)

Levers, bellcranks, sleeves, bushings, cover plates, pistons, bucket wheels, meter cases, lamp fixtures, cylinder heads

Section thickness: 6.4–25 (¼–1 in.) Overall linear dimension: 914 mm (36 in.) Maximum boss diameter: 51 mm (2 in.) Maximum flat area: 9700–12,900 mm2 (15–20 in.2)

Levers, brackets, clamps, small frames, housings, valve bodies, guide bushings, pump impellers, bucket wheels

(a) Numbered sands are identified by origins, sieve analyses, and other characteristics in Table 1. Numbered resins are identified by melting points and other characteristics in Table 2. Percentages in each formulation are based on combined weight of the sands. Source: Ref 1

1. Add the sand, hexa, wax, and other dry materials to the muller and mix for 1 min. 2. Add the liquid resin and mix for an additional 3 min. 3. Introduce air (cold, warm, or hot) into the muller and mull until all liquid is removed from the resin-sand mixture and the desired properties of the coated sand

bonds between sand grains Molds with low hot tensile strength Blockage or caking of the coated sand in the storage hoppers and molding machines Poor or irregular buildup on the backs of the shell molds Increased curing time during making of the mold

Overdrying or overmulling may result in the following problems:
High-melting-point

Radiator fittings, pipe fittings, electric terminals, levers, brackets, frames, caps, covers

and coating of the sand grains during a normal blending cycle. If the materials are too hot, the resin begins to harden before the sand grains are thoroughly coated. Either of these conditions has an adverse effect on the tensile strength of the coated sand. A typical mixing procedure for the cold coating process is as follows:





Leaded yellow brass, alloys 6A and 6B; tin bronze, alloys 1A and 1B Section thickness: 3.2–25 mm (1/ 8–1 in.) 60 parts sand No. 12 30 parts sand No. 14 Overall linear dimension: 610 mm (24 in.) 6% resin No. 8 Maximum boss diameter: 51 mm (2 in.) 4% treated bentonite Maximum flat area: 5200–6500 mm2 1.5% calcium stearate (8–10 in.2)

Magnesium alloy AZ63A 60 parts sand No. 25 35 parts sand No. 23 4% resin No. 7 1.5% ammonium fluoborate


Low-melting-point



Gray iron, classes 20, 25, 30, and 35 Section thickness: 3.2–19 mm (1/ 8–3/4 in.) 60 parts sand No. 12 35 parts sand No. 1 Overall linear dimension: 914 mm (36 in.) 4% resin No. 5 Maximum boss diameter: 51 mm (2 in.) 2% treated bentonite Maximum flat area: 6500 mm2 (10 in.2) 1% calcium stearate

3

depending on the type of resin used and on the desired length of the mulling cycle (the hotter the air, the shorter the cycle). The time needed for completely drying the resin-sand mixture depends on the air temperature (optimum is 230  C, or 450  F) and on the amount of air introduced into the muller (optimum is 0.10 m3/s, or 200 ft3/ min, per 45 kg, or 100 lb, of sand). The rate at which the air is exhausted from the muller also has some effect on drying time. Inadequate drying of the coated sand may result in one or more of the following problems:

are obtained (as determined by meltingpoint and tensile tests). The overall time for this cycle will vary from 6 to 15 min, depending mainly on the type of resin and mixing equipment used, on the temperature of the air used for removal of liquid, and on the properties desired in the final product. The coated sand should be screened to remove lumps, which would be undesirable, after being discharged from the mixer. The cold-coated sand must be thoroughly dried before it is usable. This can be done by continuous mixing and mulling of the resinsand mixture at room temperature or by introducing large volumes of air into the mulling chamber. The air may be cold, warm, or hot,

bonds between sand grains
Molds with low hot or cold tensile strengths
Slow buildup of shell during the molding period
Weak molds, causing runouts or other casting defects Hot coating is a process in which resin and sand, in combination, are heated to 120 to 135  C (250 to 275  F). At this temperature, the thermoplastic novolak resin will coat the sand grains during mixing but will not set. The catalyst (hexa) is then added, usually in water solution, and distributed throughout the mixture, which is then cooled. There are at least three advantages of hot coating:
The tensile strength of hot-coated sand is

approximately 25% higher than that of cold-coated sand.
Hot-coated sand does not dust, as does coldcoated sand.
There is no explosion hazard in the hot coating process, as there may be in cold coating, because alcohol is not used in the hot coating process. The principal disadvantage is the necessity for heating the sand prior to mixing and mulling. Equipment for Coating and Handling. Resin coating of sand can be done at high speed on a nearly continuous basis in mullers (Fig. 3). A low-speed muller is heated to 150 to 175  C (300 to 350  F). The sand charge and granulated novolak resin are mixed in the low-speed muller until the sand is coated. The resin-coated hot sand is then discharged into a high-speed muller, which is water cooled. When all of the hot sand has been discharged, a water solution

Shell Molding and Shell Coremaking / 603

Control Testing of Resin-Sand Properties At a minimum, foundries typically characterize and select a shell sand based on the following properties:
American Foundry Society (AFS) grain fine-

Fig. 3

Two-muller system for resin coating of shell sand by the hot process. See text for discussion of hot coating.

of the required amount of hexa is sprayed onto it. The sand is then discharged immediately through the vibratory screen into the oscillating conveyor for transfer to storage bins. Depending on the size of the foundry operation, various methods can be used to transfer coated sand from the muller or storage hopper to the shell-molding machine. In small foundries, the sand is generally loaded into fiber drums or tote boxes and transferred by hand carts or fork trucks to the feed hopper on the machine. Larger plants use automated transfer systems. Because prolonged and excessive handling of coated sand causes resin to separate from the sand grains, a sand-transfer system should keep travel distance to a minimum. A conveyor-belt system should be constructed to limit the number of directional transfers, not only to minimize separation of sand and resin but also to eliminate segregation of sand grains. Automated bucket-monorail conveyors are normally used when several different sand mixtures are required or when the number of directional transfers cannot be reduced. If a pneumatic transfer system is selected, a low-pressure system is preferable to minimize abrasion of the resin and reduce the creation of resin dust. Coated sand does not pick up moisture readily unless the surrounding air is extremely humid; therefore, transfer time from the muller or coating unit is not critical. (In fact, coated sand can be stored for several months if kept in a fairly dry area.) Temperature variations that result from changes in weather have no effect on coated sand, provided it was thoroughly dry when removed from the muller.

Sand Reclamation When economically feasible, shell molding sand may be reclaimed, usually by either the dry-scrubbing or the thermal method. In the




dry-scrubbing method, the sand grains are rubbed together and then air-scrubbed. The equipment used consists of a chamber into which the sand is introduced from the bottom through a nozzle at relatively high velocities. The sand is shot up in the chamber, and as it falls back to the bottom, it is blasted by other incoming sand. This process is repeated until all of the resin has been removed. In the thermal method, the sand is heated to a temperature high enough to burn off the resin and other material, such as carbon. There are three types of thermal reclaiming units:
A rotary kiln or calciner, with which the

sand is fed into one end and heat is supplied from the other end as the sand moves through the kiln
A multiple-hearth vertical furnace, in which the sand is raked across the top hearth, then down upon the next lower hearth, and so on, until the sand is clean. Burners are at the lowest hearth.
A unit that employs a fluidized bed, through which hot air is passed to fluidize the bed of sand and burn the resin In the thermal reclaiming method, the resin on the sand acts as fuel to supply much of the heat required for burnoff. When thermally reclaimed sand is subsequently molded, less resin is required for obtaining a given level of tensile strength. It is also important to note that shell molds, due to their resin contents and shell thicknesses, may be largely in situ thermally reclaimed during the metal-casting process. Enough heat may be provided from the poured metal to burn the resin coating from the shell mold, leaving sand that has essentially been thermally reclaimed. If this sand can be recaptured and properly screened for agglomerates with sufficient dust collection to remove fine material, much of this sand could be made available to be recoated and reused in the shell process.










ness number, a calculated industry number that attempts to capture size distribution data of the base aggregate. Normally, the resin coat on a sand will cause it to screen out approximately one screen coarser than the uncoated sand. That is, sand with a grain fineness of AFS 70 when uncoated will screen to approximately AFS 50 fineness when coated with 3% resin. Excessive coarsening of a sand during coating is a sign of too much resin. Grain fineness is determined by shaking 50 g (1.6 02) of sand for 15 min in the standard AFS sieve stack. Melt point (or stick point), the temperature at which resin bridges first form between coated sand grains in a standardized industry test. It is controlled by the nature of the resin, the amount of hexa present, and the degree of preadvancement of the resin. Low stick points promote peelback, delamination, and slow cure. High stick points can indicate slow cure and usually indicate low tensile strengths caused by excessive advancement of the resin before the cure is made. Most stick points are between 80 and 100  C (180 and 210  F), although other values may be required for special purposes. Heavy resin coats, such as usually occur on zircon sands, will cause the stick point to read a few degrees lower, all other conditions being equal. Loss on ignition, a characteristic that is linked to resin content and the amount of other combustibles in the coated sand. The length of time that a shell can resist hot metal is largely a function of the amount of resin present, rather than of mold strength. A simple loss-on-ignition test will give the percentage of resin in a shell sand. Cold tensile, a property that is indicative of the strength of cured sand at room temperature. Cold tensile and surface hardness indicate the ability of a cured mold to be handled, transported, and stored without damage. Bursting from metallostatic pressure is prevented by adequate cold strength and shell thickness. Hot tensile, a property that is indicative of the strength of cured sand at shell-moldand coremaking temperatures. It depends solely on the advancement of set resins, which are a result of the resin-hexa reaction. It indicates the strength of the shell as it is ejected from the hot pattern and thus has a bearing on warpage during pasting and handling. Hot tensile strength is important in holding the dimensions and shape of the casting cavity against metallostatic pressure during solidification of the casting. The curing speed of a resin may be studied by

604 / Expendable Mold Casting Processes with Permanent Patterns making hot tensile tests at a series of curing times. Normally, coated sand will have a hot tensile strength of 60 to 85% of the cold tensile strength, although some resins produce equal hot and cold strengths.
Hot distortion testing (HDT), another less common but more informative test. This test characterizes shell sand by its four-stage, elevated-temperature distortion behavior and is useful for comparing different shell sand samples. The four stages are thermal expansion, thermoplastic relaxation, thermosetting, and the final degradation and failure stage. If HDT data are unavailable at the foundry, these data are typically available through the foundry’s resin or coated-sand supplier.
Surface (scratch) hardness, which is a better measure of scuff resistance than a tensile test. A standard core-hardness tester is used, in which the penetration of the specimen by a spring-loaded carbide knife is measured after 90 rotation. A hardness index of 100 indicates no penetration; an index of 0 signifies 0.25 mm (0.1 in.) penetration.

Patterns and Core Boxes Only metal patterns (Fig. 4) and core boxes can be used in the shell process. Gray iron is the metal most used. It is readily available and has excellent stability at shell-molding temperatures. If the gray iron is stress relieved before final machining, distortion from stress is minimal. Gray iron is easily machined, and patterns and core boxes made from gray iron require little maintenance for indefinite use. Bronze patterns and core boxes have excellent wear properties and stability during prolonged use at molding temperatures. Bronze, however, is expensive and difficult to work. Aluminum is the easiest to work of the metals used. However, aluminum patterns are difficult to maintain. Also, aluminum patterns and core boxes warp and expand during prolonged use. Steel is more difficult to work and warps during prolonged use. Pattern Construction. The principal components of shell-mold patterns are the pattern plate, the pattern itself, the sprue, risers,

runners, gates, ejector pins, and sand strips or rims (Fig. 5). Shell-mold patterns generally do not differ in design from patterns for green sand molding. However, because of slower metal solidification in shell molds (caused by the lower moisture content), it may be necessary to increase the size of the risers slightly. Risers are not always needed, however, particularly when pouring gray iron and brass. Metal used for making the sprue, riser, runner, and gates is usually the same as used for the plate and the pattern. Sometimes, beryllium copper or brass is preferred for the sprues, risers, runners, and gating, to obtain quicker heat recovery. Because the lower portion of the ejector pins is exposed to direct heat, they are preferably made of stainless steel. An example of a typical shell-mold pattern assembly is illustrated in Fig. 5. The pattern plate and pattern halves were machined from gray iron. The ejector pins were stainless steel, and the metal-feeding elements were machined from brass. The locating plugs were made of steel, and the sand strips that formed the rim were also made from steel. The mold sections, after production in a twopattern molding machine, were broken apart at the location of the break strip on the pattern assembly. A core was put in place, and the sections were aligned by means of the locating plugs and glued together. The completed mold was placed flat on the sand in a pouring box and was poured without additional support. For simple castings with no critical dimensions, patterns sometimes are cast to size and hand finished. More complicated castings, and those with close tolerances, require patterns that are completely machined. All shell-mold patterns require ejector pins, but the ejector pins for this pattern (Fig. 5) had to be precisely placed, and most of them were fairly close to the pattern to permit lifting the mold without breaking fin sections. The fins on this casting (for the production of air-cooled motorcycle cylinders) were 25 mm (1 in.) deep, and each fin had 0 draft angles. A pattern of this design must be completely machined; otherwise, it would be impossible to eject the mold.

Split core boxes, heated to the correct temperature, are used to produce hollow or solid cores. The core boxes can be equipped with electric heating elements, or they can be heated in an open gas flame. A core box with the cores cured in it may be split horizontally or vertically. If possible, the interior of the core should be contoured by a mandrel; if this cannot be done, the loose pieces and uncured sand should be dumped from the interior of the core. In the shell-mold process, more vents are required in the core boxes, because the sand-resin mix embeds itself in vents and reduces their effectiveness.

Production of Molds and Cores Shaping of shell molds and cores from resinsand mixtures is accomplished in machines. Different types of machines vary in operating details and in degree of automation, but all are built to allow close control of time and temperature cycles. Regardless of the type of machine used, or the degree of automation, there are five major steps in producing a mold or core: 1. The resin-sand mixture is placed over a heated pattern, by means of a dump box or by blowing. In the dump-box process, the amount of the mixture is considerably greater than that needed to form the shell. The resin adjacent to the pattern melts, causing the mixture to adhere to the pattern. The portion of the mixture that adheres to the pattern is called the investment, and hence, the operation is called the investment cycle. 2. When the desired shell thickness is obtained, the pattern is inverted, allowing the excess resin-sand mixture to drop back into the container. 3. While still on the heated pattern, the investment (now called the shell) is finish cured in an oven. 4. After the curing cycle, the shell is ejected from the pattern. 5. In the final operation, any cores to be used are set, and the shell halves are bonded together with glue. The mold is then ready for pouring. Heating of Patterns. Shell-molding machines are equipped to heat pattern plates either electrically or with gas manifolds and jets. Thermostatic controls are incorporated in the units. To heat the pattern plates electrically, strip or tubular elements are attached to the back of the pattern plate (Fig. 6a and b). Cartridge-type elements are inserted in patterns for risers or sprues, as shown in Fig. 6(c). Heating pattern plates with electric elements can present several problems:
To ensure that the heating elements will

Fig. 5 Fig. 4

Metal pattern with pins for shell molding

shown

Motorcycle cylinder that was cast in a shell mold made using the cope-and-drag pattern

make complete contact and provide uniform heat throughout the plate, the back of the plate must be ground flat.

Shell Molding and Shell Coremaking / 605
If the pattern detail is recessed below pattern

parting, it is difficult to provide adequate heat to this area without overheating the rest of the plate.
Replacing burned-out elements, particularly those located in sprues and risers, is timeconsuming.
Frequently, the placement of the heater element conflicts with the location of stripper pins. Gas-fired manifold heating systems are similar on most shell-molding machines. Fork-type or rectangular manifold burners (Fig. 7a and b) are used for those machines that have the pattern frame hinged to the dump box. Although both are effective, the fork-type heating unit limits the placement of ejector pins in the pattern. A machine that has the pattern carriage as a separate unit moving independently to and from the dump box and curing oven is often preferred; with this type of machine, a box-type manifold burner (Fig. 7c) is used to heat the pattern.

Copper or brass dowels heated by a gas flame can be inserted into risers or sprues (Fig. 7d) to provide additional heat. These dowels should extend approximately 50 mm (2 in.) below the bottom of the pattern plate so as to be in the line of the gas flame. Pattern temperature settings are gaged according to the intricacy of the pattern, the weight of the mold required, and the type of resin-sand mixture being used. The temperature settings may vary from a low of 205  C (400  F) to a high of 345  C (650  F). Temperatures within small patterns should be uniform within 28  C (50  F). For larger patterns, temperature uniformity within 42  C (75  F) is usually adequate. It is extremely important to maintain the proper temperature relationship between the pattern and the curing oven, to prevent warpage or distortion of shell molds. Dump-Box Molding. The dump-box method is more widely used than mold blowing for placing the resin-sand mixture on the heated pattern. The dump-box method offers the following advantages over blowing:






Simplicity of operation Lower tooling cost Less mold scrap Better control over mold hardness Minimum amount of segregation when various additives, such as iron oxide, calcium carbonate, or manganese dioxide, are mixed with the sand

Dump-box machines of various designs and sizes are available for producing molds from 356 by 457 mm to 660 by 2032 mm (14 by 18 in. to 26 by 80 in.). With some machines, the pattern is clamped in place, and then pattern and dump box are rotated 180 . Other units employ a circular track that permits continuous and more rapid rotation of the pattern and dump box to an inverted position; with a third type of unit, the pattern is raised vertically to the dump box. When a pattern is clamped onto a dump box and rotated in a 180 arc, as shown in Fig. 8(a) and (b), sand slides diagonally onto the pattern. When the pattern has sharp changes in contour, sponginess or voids may occur in the portions of the shell mold that are on the “leeward” side of the pattern, as shown in Fig. 8(c), because of the direction and slow speed of sand movement.

Fig. 6

Use of three types of electric elements for heating patterns for shell molds

Fig. 7

Four types of gas-fired units for heating patterns for shell molds

As the pattern is raised from the dump box following the investment cycle, the voids cause sections of the mold to fall away from the pattern, as shown in Fig. 8(d). This condition, called peelback, is the principal disadvantage of this technique. In machines with a circular track on which the dump box and the pattern are brought together and rotated in a continuous cycle (Fig. 9a), the movement of rotation is so rapid that the sand is suspended in the dump box until the box is directly above the pattern (Fig. 9b); the sand then drops directly on the face of the pattern. Baffles in the dump box (Fig. 9b) help to distribute the sand uniformly over the pattern. This minimizes the possibility of sponginess or voids in the mold (due to masking) and thus, together with the speed of the reverse rotation (Fig. 9c), minimizes peelback. With the third type of dump-box machine, sand is suspended in the upper portion of the dump box by means of a louver, and the pattern is raised vertically to the dump box (Fig. 10a). Then, the pattern is clamped to the lower part of the box, and the louver retracts to permit the sand to drop on the pattern (Fig. 10b). After the investment is made, and while the louver is in the “out” position, the dump box and pattern are rotated 180 . so that the excess sand is separated from the investment, as shown in Fig. 10(c). The louver is then moved to the “in” position to seal the sand in the box. The pattern and dump box are rotated to the original position, and the pattern is lowered from the box (Fig. 10d). Mold Blowing. Shell molds can also be made by blowing sand onto the heated pattern. Some molds are blown with the outer surface contoured to the pattern (Fig. 11a), some are blown in the shape of an open-back box with reinforcing ribs (Fig. 11b), and some are blown in the shape of a closed-back box (Fig. 11c). Generally, contoured molds and most openbox molds with ribbing are made on machines that have a blowhead similar to that used with conventional core blowers (Fig. 12a). The upper and lower halves of the pattern are heated by gas ovens or by manifold-heating assemblies equipped with gas jets. Because the blowhead is located above the hot pattern and is in contact with the box during the blow cycle, the blowhead and the blowplate must be water-cooled

606 / Expendable Mold Casting Processes with Permanent Patterns

Fig. 8

Steps in making a shell mold by the dump-box technique. (a) The pattern is rotated and clamped to the dump box. (b and c) The box is then rotated 180 to make the investment. (d) Pattern and shell are removed from the box. Voids (c) and resulting peelback (d) are disadvantages of this technique.

to prevent the sand from hardening in the chamber. Closed-back box molds are blown in shellcore blowers, as shown in Fig. 12(b). These molds can be blown hollow or solid, depending on the amount of pattern detail. Cycle times are faster than with the dump-box method. Box-shaped ribbed molds can be poured vertically or horizontally. If the ribbed molds are poured vertically, the two halves need not be glued together but can be held together with mechanical backup. If poured horizontally and not glued, the cope should be weighted to prevent the mold from opening. The main disadvantage of blown molds is their weight, which is two to three times that of dump-box shell molds. Other disadvantages are:
Pattern and pattern maintenance costs are

high.


Clogged vents cause an excessive amount of

downtime.


Misblows, caused by clogged vents or by

improper closure of pattern halves due to sand grains on the pattern parting lines, result in a high percentage of scrapped molds.
Additives, which may be required in the sand mixture to obtain a better casting finish or to control mold hardness, usually separate from the sand grains.

Fig. 9

Production of a shell mold by the dump-box method in which pattern and dump box are rotated at high speed on a circular track. (a) Rotation. (b) Rollover. (c) Reverse rotation

Fig. 10

Steps in the production of a shell mold by the use of a louver-type dump box. (a) Pattern raised into dump box. (b) Pattern clamped to dump box, and louver retracts to dump sand onto the pattern. (c) Pattern and dump box inverted, and excess sand falls. Louver returns. (d) Original position. Louver retains excess sand. Pattern and shell are lowered.

Fig. 11

Cross sections of blown shell molds of three different structural shapes. (a) Outer surface contoured to pattern. (b) Open-back box. (c) Closed (hollow) box

Shell thickness is a function of pattern temperature and of the time the resin-sand mix remains in contact with the heated pattern. As shown in Fig. 13, low pattern temperatures and short contact times result in thin shells. The shell thickness required for a specific application depends on the pouring temperature, the shape of the casting, and whether or not a backup material is used. As the pouring temperature increases, a thicker shell is required. Castings with heavy cross sections in one area will require greater shell thickness at that location. For a given casting, the use of backup material (such as shot, gravel, or sand) will allow the use of a thinner shell. In a preliminary evaluation of the requirements of a shell mold, molds should be prepared to various thicknesses. These molds should then be poured and the castings evaluated, to determine the minimum mold thickness practical for that application. Curing. Most shell-molding machines are supplied with units containing either electric or gas-fired ovens for curing the mold. In an electric oven, the temperature can be held to  þ0.5  C (þ 1 F), but shell patterns give good  results if the temperature is held within  þ5  C  F). (þ 10  Electric ovens ordinarily have the heating elements near the ceiling of the oven (Fig. 14a). This may result in overcuring of the upper portions of high mold details. To prevent this localized overcuring, additional heating elements may be attached to the inner side walls of the oven (Fig. 14b).

Shell Molding and Shell Coremaking / 607

Fig. 12

Two types of machines for blowing shell molds. Machine in (a) is shown as used for blowing an open-back box mold (Fig. 11b) but is also used for molds whose outside surface is contoured to pattern (Fig. 11a). Machine in (b) is used for blowing closed (hollow) box molds (Fig. 11c).

Fig. 13

Shell thickness as a function of pattern temperature and duration of contact of resin-sand mix with heated pattern

Fig. 14

Electric ovens for curing shell molds. Heating elements normally located under ceiling of oven, as shown in (a), may be augmented by elements along inner side walls (b) to prevent overcuring of high portions of

a mold.

Most gas-fired ovens are designed with manifolds and blast tips (Fig. 15a), although some have ribbon-burner manifolds. In both types, the gas heat is directed onto the molds. If a pattern has deep cavities, or if pattern details, risers, or sprues are more than 230 mm (9 in.) high, an oven that has manifolds and blast tips is preferred, because it can be provided with extension burners to direct heat into deep cavities (Fig. 15b) or around high sprues and risers.

In gas-fired ovens, temperature may be thermostatically controlled, or the gas flames may remain constant while the mold is being cured and then be shut off automatically when curing is complete. Overcured molds may break down prematurely; that is, the bond in the mold may burn out before the casting metal has completely solidified, causing the casting to have swells or runout. If a mold is seriously overcured, it

may be eroded during filling with the casting metal. Overcuring is indicated when shell molds are dark brown in color. Undercuring may result in gas holes or blows in the castings (particularly in deep pockets where a thick section of shell material may be built up and not completely cured). An undercured shell also may not have enough tensile strength to withstand handling without breakage. Shells that are adequately cured are typically dark yellow or light brown in color; ordinarily, this is achieved in 30 to 40 s. Whether making shell molds or shell cores, repeatable cycle times are very important to having a controlled shell process. This is because the properties of the mold or core are dictated by the heat transferred from the equipment to the sand. Even with a thermocouplecontrolled core box, the entire surface area of a core box is rarely at a uniform temperature. Inconsistent cycle times, or unplanned delays, can allow these temperature variations to become exaggerated. These temperature variations in the core box will affect the amount of heat transferred to the core and create variations in core properties. The best controlled coremaking process, cycle to cycle, is an uninterrupted process of repeating cycles. Cores. Dry resin-sand mixtures can be used for coremaking only when they are dumped into the core box. They cannot be blown without segregation of the resin binder. Also, the dust created would be a health hazard. For blowing cores, resin-coated sands should be used. They may be coated by either the wet or the hot process. Gating is incorporated in a vertically parted shell mold, as shown in Fig. 16(a). This gating system consists of sprue, well, runners, and gates of calculated dimensions corresponding closely to those in conventional sand practice. Except for the pouring basin, the gating system shown in Fig. 16(a) is integrally formed with the mold cavity. Although the pouring basin also may be integrally formed with the mold cavity, a separate pouring basin, such as shown in Fig. 16(a), allows better control of the metal entering the cavity. A horizontally parted mold may be provided with a sprue formed integrally with the mold cavity. This requires the development of a choke in some other element of the gating system, because of the draft needed on the sprue to allow its removal from the pattern. If a tapered sprue is desired, it must be formed separate from the mold and added, together with the pouring basin, in the mold assembly (Fig. 16b).

Mold Assembly Molds made in an automatic molding machine are usually bonded together at the parting line, although sometimes they may be held together mechanically. The bonding (gluing) is done with a resin added just before the mold halves are clamped together. Bonding

608 / Expendable Mold Casting Processes with Permanent Patterns time is approximately 20 to 40 s while clamped, usually in a machine made for this purpose. The total time required for assembly depends on whether or not cores are used. In shell molding, it is sometimes possible to produce a casting without a core that would be required if the casting were produced in a green sand mold. For example, the propeller bushing shown in Fig. 17 was shell-mold cast without a separate core; the bore was formed by mating cylindrical mold sections (section A-A in Fig. 17), which also served to align the two mold halves. Had this casting been made by conventional sand practice, a cope-and-drag mold would have been required, as well as a core to develop the inner cylindrical surface.

Fig. 15

Fig. 16

Blast-tip gas-fired manifold oven (a) for curing shell molds. Oven can be fitted with extension burners (b) for directing heat into mold cavities.

Fig. 17

After the two halves have been bonded together, shell molds often are poured flat without further support. Because of the pressure of the molten casting metal, the mold halves may separate, and this can prove troublesome. If the parting opens, backup support may be used, but the use of such support is frequently undesirable, and other means to prevent opening of the parting are sought, as in the following two examples. Example: Use of Flush-Head Ejection Pins (Fig. 18). Good dimensional control could not be maintained across the parting line of the 4340 steel pintle casting shown in Fig. 18. It was determined that parting-line spread was caused by inadequate bonding at those points where the ejection pins removed the mold from the pattern. The contoured heads of the pins left slight depressions on the bonding surface, lowering the bonding strength at these points. The contoured pins were replaced by flush-head pins, and the problem of distortion in the castings was eliminated. Processing details are given in the table that accompanies Fig. 18. Example: Bolting of Mold Halves (Fig. 19). The 4340 steel plate castings shown with attached rigging in the upper half of Fig. 19 exceeded specified thickness dimensions when poured into a conventional peripherally bonded shell mold. By adding four metal bolts at the locations shown in the sectional view of the mold in Fig. 19, using the bolts to anchor hold-downs, mold spread was prevented, and casting thickness met specifications. Processing details are given in the table with Fig. 19.

Gating systems for (a) vertical-parting and (b) horizontal-parting molds

Number of castings per mold Weight of metal poured per mold Weight of trimmed casting Pouring temperature Mold temperature Warpage Production rate Total production Cost per casting

Propeller bushing that was cast in a shell mold without a separate core

Fig. 18

Methods to Minimize Mold Spread at Parting Line

1 34 kg (05 lb) 7 kg (5 lb) 1705 ⬚C (3100 ⬚F) Room None 600 castings per month 70,000 castings $7

Pintle casting (with rigging attached) for which flush-head ejection pins were used to prevent distortion caused by mold spread at parting line

Causes and Prevention of Mold Defects The problems most frequently encountered in shell-mold casting are mold cracking, soft molds, low hot tensile strength of molds, peelback, and mold shift. Mold cracking results from thermal stress that develops when hot metal is poured into the cold molds. If the crack extends into a casting cavity, it can cause bleeding and metal

Shell Molding and Shell Coremaking / 609

Number of castings per mold Weight of metal poured per mold Weight of trimmed casting Pouring temperature Mold temperature Production rate Total production Cost per casting

2 16 kg (36 lb) 3 kg (6 lb) 1705 ⬚C (3100 ⬚F) Room 600 castings per month 35,000 castings $3

Fig. 19

Plate castings (with rigging attached) that were produced in a mold bolted together to prevent parting-line spread that resulted in out-of-tolerance casting thickness

runout, dimensional inaccuracy, and distortion of castings. Cracking can be prevented or controlled by using sands with less thermal expansion, using thermoplastic additives, or directing the crack formation away from the casting cavities by means of crack strips. The preferred methods for prevention of mold distortion and cracking are by adjusting the expansion characteristics of the sand and by using thermoplastic additives; however, the use of crack strips is common. Extension of cracks into the casting cavity can be prevented by cutting crack strips (slots) into the perimeter of a shell mold. The original method for making crack strips is illustrated in Fig. 20. A saw is used to cut the slots. The type of crack strip shown in Fig. 21 is generally preferred and more often used. These crack strips are made by mounting sharp metal wedges on the pattern plate, with the sharp edge up. Because the sharp edge remains relatively cool when the molds are made, the resin-sand buildup is very thin along this edge, creating a weak spot, which acts as a crack strip. Because the metal strips are placed in matching locations on the cope and drag, the cavities formed when the mold halves are glued together take the shape of a diamond, as shown in Fig. 21. Crack strips when used must be placed so they do not extend into the mold cavity. Soft molds and cores can result from one or more of the following variables in moldmaking:
Resin content: A low resin content gives




Fig. 20

Shell mold with crack strips in perimeter to prevent cracking of casting cavities by control of thermal expansion








Fig. 21

Shell mold for long narrow castings that has diamond-shaped crack strips at ends and along sides to prevent cracking and distortion of casting cavities

weak or thermoplastic shells and cores. Maximum and minimum resin contents must be established for each metal, casting shape, and type of sand. Hexa content: As the hexa is reduced (other variables remaining constant), the cured mold or core will exhibit increased softness. Maximum and minimum hexa contents should be established for the job and maintained. Additives content: Excess additions of resin plasticizers, iron oxide, and silica flour will produce weak and soft molds. Temperature: A pattern temperature that is too low will produce undercured, rubbery shells and cores. Most shell-mold machines operate efficiently at 220 to 250  C (425 to 475  F), and shell core machines at 250 to 275  C (475 to 525  F). Screen distribution of sand: Sand with too wide a screen distribution can be the cause of soft molds.

Softening of molds can occur during pouring. Usually, this is caused by pouring at too high a temperature or by inadequate mold venting. Mold softness results in inaccuracy of casting dimensions and rough surfaces. Low Hot Tensile Strength. Slow curing causes broken shells at the molding machine. It also results in low hot tensile strength; this

can be verified by tensile tests completed at holding times to 1 to 1.5 min at 220 to 230  C (425 to 450  F). Tensile failures may result from the combined action of a number of factors, of which the following are significant:
Resin content: A minimum tensile strength

should be established to ensure flexibility in molding. After this is established, the resin content should be held constant, unless changes in other factors, such as sand, should occur.
Type of resin: Resins can be formulated to various cure rates. Requirements in individual foundries dictate the appropriate cure rate.
Hexa content: Variations in hexa content can result in wide variations in mold properties. Good control of resin formulation is essential for uniform molding characteristics. The particle size of the hexa also should be checked. If the hexa is not dissolved in water before it is added at the muller, powdered hexa should be used to ensure even distribution. Granular hexa, unless added as a solution, will give lower strengths per pound than will hexa in powdered form.
Solvent retention: Incomplete removal of solvent will slow down the cure of resins. To eliminate this condition, the resins and mixture should be screened and aerated more effectively. Peelback, which occurs during the “drain” cycle in dump-box molding (Fig. 8d), not only ruins the shell on the pattern but (unless the nearly cured sand is removed from the dump box) also causes the next mold to be scrapped. Major causes of peelback are:
High or low resin content: High resin con-














tent may cause too heavy a buildup on the pattern, so that peel occurs from excessive weight. Too low a resin content results in weak and slow-curing shells that will peel. Resin with improper melting point may also cause peelback. Excessive release agent: Because the coated sand already contains a lubricant, spraying of the pattern should not be necessary. The amount of lubricant used in coating formulation should be checked. Insufficient removal of solvent: Solvents in the mix promote gas, slow curing, and peeling. A longer drying cycle, together with screening and aeration, is required for correcting this condition. Improper mixing: Amounts of the ingredients and the mixing procedure should be checked. The muller should be free of core oils or other contaminants. Wet sand: With moisture above 1%, sand will coat poorly and retain part of the water, thus decreasing curing rate. High pattern temperature: When pattern temperatures are too high, they cause a

610 / Expendable Mold Casting Processes with Permanent Patterns buildup of extrathick shells or cores, which may peel from the heavy weight.
Pattern-contact time: With long contact time, the shell may build up to an excessive thickness and peel from heavy weight.
Pattern design: Patterns that have high projections, which result in large temperature variations, will peel. Temperature variation can be minimized by the use of inserted heaters (Fig. 6 and 7). Mold Shift. Although seldom the cause of mold shift, the pattern mounting should be inspected first. Next, the temperature differential between the two mold halves at the time they were bonded together should be checked. This differential can be as great as 55  C (100  F) and very often is the cause of mold shift. To avoid this, the mold halves should be at the same temperature when they are bonded. Where mold shift is a continuing problem, one or more additional locators should be added to the pattern plates.

Mold Backup To obtain accurate dimensional control of castings made in shell molds, especially those of large size and thick sections, the relatively thin molds usually must be supported over the entire outer surface while the casting is being poured. Support permits the use of molds with thinner walls, thus reducing shell costs. Backup also helps to prevent runouts and casting bleeders and provides a suitable bed for mold positioning. In the casting of malleable iron, backup material may be required for control of the cooling rate of castings to prevent mottling or primary graphitization during solidification. Backup Material. Various means and materials have been used to support or back up shell molds during casting and the initial cooling period. To be successful, the backup means or material must rigidly support the shell mold against the pressure of the molten metal in the internal cavity during the casting period. It must not, however, exert a pressure on the outside of the shell mold in excess of the internal pressure, as this would cause the mold wall to move toward the molten metal and would result in casting distortion just as serious as an outward movement. Backing up a shell mold with a solid material such as cast iron or plaster, contoured to fit exactly and to support the exterior surface of the shell mold, would meet the requirements ideally. This method, however, would be expensive and difficult to handle. It also would be necessary for the exterior surface of the shell mold to be accurately controlled in order to fit with precision into the rigid backup support. The most feasible backup of shell molds for a large variety of high-production castings is a granular material having semifluid characteristics and a density approaching that of the

casting being poured. A choice between the two common backup materials (shot and gravel) depends mainly on consideration of the more effective cooling of shot against the lower cost of gravel. Cast iron shot is generally preferred for backup on high-production, mechanized shellmolding lines. It is readily available, will flow when vibrated because of the spherical shape of the particles in a manner similar to a true fluid, and will pack to maximum density around the shell molds. Even though shot approaches true fluid characteristics while being vibrated, it must not produce pressure on the outside of the shell mold in excess of the internal pressure during pouring. The fluid nature of the shot under vibration accomplishes a uniform support over all areas of the outer surface of the shell molds, especially beneath horizontal projections. This characteristic must be maintained as high and as nearly constant as possible for uniform control of casting dimensions. Backup shot is controlled by a specification of shot diameter and weight per cubic foot. Controlling the shape and soundness of the shot results in optimum fluidity and packing characteristics. Backup material must be processed and cleaned to remove fines and remnants of shells. When not removed, fines will build up in the shot or gravel and block escape of gas from the shell during solidification of the casting. Cleaning, Cooling, and Recycling of Backup Shot. Because the weight of hot backup shot shaken out per hour may be more than 15 times the weight of iron poured per hour and 30 times the weight of sand molded per hour, the cleaning, cooling, and recycling of the backup shot are important considerations in foundry engineering. It is most desirable to maintain a constant temperature of the shot used in the backup of shell molds. Very hot shot is not only a personnel hazard and discomfort but can heat the shell molds to a softening point, resulting in distortion of the mold cavity and the casting. Very hot shot also can reduce the cooling rate of poured castings enough to cause them to be softer than specifications require. With the temperature of backup shot controlled in the range of 90 to 95  C (190 to 200  F), the personnel hazard and discomfort are at reasonable levels. The cooling rate of any given casting is held consistent so that hardness can be readily controlled by iron composition and pouring temperature. Constant temperature of backup shot at bed-in and pouring will also result in consistent casting temperature at shakeout, a factor that also can affect casting hardness and machinability.

Casting Process Molds poured with the parting line in the horizontal plane may be placed on a bed of sand and weighted down, and then poured and

handled in much the same way as conventional sand molds. Molds poured with the parting line in the vertical plane usually require the use of backup material because of the greater static head. Generally, backup materials are mandatory only when pouring unsupported vertical shell molds or when casting metals that have long liquidus-to-solidus (solidification) times, such as some of the aluminum alloys. Such metals can break down the mold wall or burn through before a sufficient amount of metal has solidified, thus causing a runout. Vertical versus Horizontal Parting. Most molds having a vertical parting line are not self-supporting. Even those that could be would be subject to the metal pressure from the sprue, which could force the shell open at the parting line to develop a runout or an increase in the dimension of the casting across the parting line. Thus, most vertical molds must be supported by backup material. However, tooling is normally less expensive on vertical patterns than on horizontal patterns, because of the way in which the rigging is laid out. One pattern can make two casting cavities more easily in a vertically parted mold than in a mold having a horizontal parting line. Other advantages of vertically parted molds are that they require less space and are better adapted to automatic handling; for this reason, they are often preferred for high-production operations. Also, the aspiration of air during pouring of the metal is more easily prevented when using a vertical sprue, because of its more favorable taper. Many horizontal-parting molds can be poured unbacked. Usually, these molds are pushed into a sand bed and weighted, before pouring. Horizontal pouring thus usually eliminates the need for cleaning and processing of a backup material to remove fines and shell remnants. In jobbing foundries or in short-run operations, a horizontal mold is more manageable and is usually preferred. Aspiration of air during the pour in a horizontal sprue is sometimes a problem. However, this can usually be overcome by the use of a ceramic choke, such as a strainer, placed in the sprue or by the use of separately molded sprues that have a taper. Metal-Mold Reactions. Because of the relatively high binder content in shell molds and cores, mold gases are produced that consist primarily of hydrocarbons, hydrogen, and carbon monoxide. Varying amounts of water vapor and carbon dioxide will also be present, depending on the availability of oxygen in the mold. The products of shell-binder decomposition have little or no effect on the surface quality of many cast irons and medium-carbon or high-carbon steels. However, low-carbon ferrous materials and alloys that contain strong carbide formers (stainless steels, for example) are subject to gas-induced surface defects and carburizing action when cast in shell molds. In plain carbon and low-alloy steel castings, the

Shell Molding and Shell Coremaking / 611 gas-metal reaction is manifested by surface gas pockets and subsurface pinholing and by carburization of the surface to a concentration of approximately 0.30% for a depth of 1.3 to 2.5 mm (0.050 to 0.100 in.). The surface voids are probably caused by rapid buildup of gas pressure within the mold before casting-skin formation has proceeded sufficiently to withstand penetration or deformation by the gas. Although some surface carburization of mild steel castings is seldom significant, shell-mold casting of alloys containing appreciable amounts of strong carbide formers (chromium, tungsten, molybdenum, and vanadium) results in carbide formation at surface grain boundaries. Subsequent heat treatment of such castings may result in surface cracking. Re-solution or spheroidization of the grain-boundary carbides is not feasible, because of the high temperatures and long heating times required. As discussed subsequently, the mold-metal reaction and carbon pickup in shell-mold casting can be minimized by:
Reducing the mold gas available at the

metal-mold interface


Increasing the rate at which the casting skin

solidifies


Chemically modifying the mold material and

the casting metal Reduction of Mold Gases. Because the quantity of mold gases generated increases with binder content, an obvious means of reducing mold gases is to use only the amount of resin necessary to provide adequate mold strength. In present shell-molding practice, resin content is restricted to a lower limit of approximately 3%. Below this level, the effects are low mold strength and excessive handling breakage, although these may be partly offset by the use of a thicker shell. Although metal-mold reaction still occurs in shell molds at the 3% resin level, the amount and intensity of the casting surface defects are lower than in molds containing more resin. Another means of reducing the amount of gas available at the metal-mold interface is to prevent rapid buildup of pressure within the mold by increasing the venting of the mold cavity and the permeability of the shell material. Mold gases can also be physically isolated from the cast metal by applying a dense, impermeable refractory wash, such as zirconite wash, to the mold surface. Decreasing the gas available at the metalmold interface should, in addition to minimizing surface gas defects, also reduce carburization of casting surfaces. Although it is generally agreed that carburization of castings in shell molds is primarily a gas-metal reaction, it is possible that solid carbonized resin at the mold surface also contributes to the carburizing action. Carbon pickup from this source should be eliminated by the use of wash coats. Solidification of Casting Skin. To promote rapid chilling of the casting skin, sands having

high heat diffusivity, such as zirconite, chrome ore, magnesite, and forsterite (olivine), have been used in shell molds. Of these, zirconite and forsterite (olivine) have been most frequently used and are effective in reducing surface defects in cast low-alloy steels. Molds made of these two materials are heavier than molds made of silica sand but possess higher chilling capacity and are less sensitive to thermal shock. Calcium carbonate (limestone) has been used as a chilling agent in shell-mold mixtures, with favorable results on casting quality. The carbonate absorbs heat by endothermic decomposition at mold temperatures attained in steel casting. However, high percentages of the carbonate (20 to 50%) are required for the chilling action, and the most effective use of the materials has been in backup layers in composite shell molds. Metal-pouring variables also have a direct influence on the rate of casting-skin formation and thus on the occurrence of metal-mold reaction. In general, pouring conditions that increase the time available for effusion of mold gases into the metal aggravate the surface defect. High pouring temperatures and rapid pouring rates, both of which promote rapid heat buildup and longer skin-solidification time in the mold, promote the occurrence of metalmold reaction. Chemical Modification. Some success has been achieved in decreasing metal-mold reaction in shell molds through chemical modification of the mold atmosphere and the metal being poured. Manganese dioxide and lead dioxide, when added in small percentages to shell-mold mixtures, reportedly improve casting surface finish and reduce carbon pickup for steel castings. Small percentages of the carbonates of sodium, magnesium, and calcium have been beneficial, and magnesium silicofluoride and ammonium silicofluoride have been used to reduce surface defects in carbon steels and gray iron. All of the additives mentioned are somewhat oxidizing, and it is believed that benefits imparted are the result of a retarding effect on mold-gas evolution. However, stronger oxidizing agents, such as the nitrates and dichromates, have exerted little or no effect in improving casting surface quality. Modification of the reactivity of molten metal poured into shell molds has been confined to deoxidation. Because of the reducing nature of the atmosphere produced during decomposition of the shell-mold resin, it is generally agreed that thorough deoxidation of the metal is desirable for decreasing metal-mold reaction. Partial deoxidation of the metal, using Ca-Si-Mn compounds, has reportedly produced steel castings with surface finish superior to that obtained by more thorough deoxidation with aluminum. This effect was ascribed to hydrogen activity in the cast metal, which increases as oxygen activity decreases. With shell-molded malleable irons and some high-strength gray and ductile irons, trouble

may be encountered with “inverse chill” or gray rims, particularly in pockets and reentrant angles. This may be counteracted by treating the molten metal with an appropriate alloy and by increasing the rate of solidification. Solidification rate may be increased by the use of chills in the mold, by the use of a sand with higher thermal conductivity than silica, or by spraying the molds with water immediately after pouring. Shakeout. Because most shell-mold castings are relatively small and are cast by this method to obtain superior surface finish or close dimensional tolerances, they should not be subjected to the abuse imposed by flask shakeout. Shakeout may consist of hooking the gates of castings from a bedding pan onto the foundry floor. The most frequently used shakeout device, however, consists of an oscillating or reciprocating conveyor. If the molds have been poured on a pendant or pallet conveyor, they are dumped mechanically into the oscillating conveyor. If the molds are in flasks, with a backup material, the flasks are dumped mechanically or by using an overhead crane. It is good practice to put used sand into the conveyor trough ahead of the dumping point to cushion the fall of the castings and to insulate the conveyor from the heat of the castings. Castings, mold, bedding sand, and backup material are moved along the conveyor to a sorting area, where the castings are removed. With the exception of those castings poured from low-melting-point metals (aluminum or magnesium), the heat of the metal burns out the resin bond sufficiently for the vibration of the conveyor to cause almost all of the sand to fall away from the castings. If a sand-reclamation unit is used, a portion of the trough conveyor can be multidecked, with each deck perforated so as to pass particles of only a given size. For instance, a four-deck section could retain castings and gating on the top level, unburned broken shells on the next level, backup material (if used) on the next, and burned-out sand on the lowest, unperforated level. An oscillating or reciprocating conveyor system used for shakeout can perform several functions:
Convey castings from pouring to cleaning

area


Remove sand from castings
Provide time for cooling of the castings
For some casting metals (such as gray iron),

provide an area for breaking off gates and risers
Sort the castings
Separate the metal, mold chunks, burned-out sand, and backup material

Dimensional Accuracy As noted, castings made in shell molds are generally more dimensionally accurate than sand castings, and they can be held to closer

612 / Expendable Mold Casting Processes with Permanent Patterns tolerances. Problems concerning dimensional accuracy do arise, however, in the shell-mold casting process, and many of these are similar to dimensional problems encountered in sand casting. Processing considerations discussed in the section “Surface Finish” of the article “Sand Molding” in this Volume also apply to shell molding. Tables 5 and 6 list typical tolerance relations for shell-mold steel and gray iron castings, for two directions of measurement. The design of the casting and the capabilities of the foundry may cooperate to permit a closer control of dimensional variations, or the design may require restrictions that are too stringent. These general relations are, however, useful as guides. Table 7 lists tolerances obtained in shell-mold castings made to closer dimensional requirements than those in Tables 5 and 6 Principal causes of dimensional inaccuracy and variation in shell-mold castings are:








Scoring or wear of patterns Dimensional variation in patterns Warped shells Uneven mold surfaces Mold mismatch and shift Buildup of release agents Shrinkage allowance

Scoring or wear of patterns can occur during long production runs. Gray iron patterns have the best wear resistance. Cast aluminum patterns generally wear much faster than gray iron patterns. The surfaces of aluminum patterns may be anodized or plated to improve wear resistance. Dimensional variation in patterns can occur because of different response to temperature. Pattern metals have different specific heats, coefficients of thermal expansion, thermal conductivity, and strength at operating temperatures. Because the patterns for the shell process must be heated, problems are likely to be encountered in a changeover from the green sand process. It is preferable to use only one kind of metal for any given pattern. Gray iron is an excellent pattern material but is a poor conductor of heat. However, it may be hollowed out to reduce the heat content locally, or copper inserts may be used to improve thermal conductivity. The lower heat capacity of aluminum may be offset by using thicker patterns. The high coefficient of thermal expansion of aluminum is sometimes detrimental, particularly where there are deep cavities with little draft. Warped shells can result from insufficient curing temperature or time, the use of a resin with inadequate hot rigidity, uneven ejection from the pattern, excessive difference between oven and pattern temperatures, warped patterns, and shape of the mold cavity. Uneven mold surfaces can be caused by segregation of sand fines or resin. With the dry-resin process, proper sand grain sizes and distribution and the use of baffles in the hopper

are important. A high concentration of resin or resin balls will cause blows on casting surfaces. Mold Mismatch and Shift. Loss of close fit between alignment pins on one side of the shell mold and the socket on the other half will cause mismatch and shift. Release agents, if allowed to build up excessively on the pattern, will sometimes cause loss of surface dimensions. This can be avoided by gaging of castings. Shrinkage factors can vary with casting practice, and two foundries using the same pattern may pour castings with different dimensional variations. Shrinkage allowances may also require adjustment, if the core-box and pattern equipment is being converted from a conventional process to a shell practice for the production of a specific casting. Methods of Increasing Accuracy. If the dimensional accuracy obtained is inadequate, it may be possible to develop greater accuracy by increasing the thickness of the mold shell, by reducing the number of castings per mold, or by using additional heating elements in the Table 5 Typical tolerance relations for shell-mold steel castings Typical tolerance Across parting line

Dimension

Between points in one part of mold

mm

in.

þ mm 

þ in. 

þ mm 

þ in. 

25 50 75 100 125 150 178 205 230 255 280 305 330 355 380 405 430 455 485 510

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

0.51 0.51 0.64 0.64 0.76 0.76 0.76 0.89 0.89 1.02 1.02 1.02 1.14 1.14 1.27 1.27 1.27 1.40 1.40 1.52

0.020 0.020 0.025 0.025 0.030 0.030 0.030 0.035 0.035 0.040 0.040 0.040 0.045 0.045 0.050 0.050 0.050 0.055 0.055 0.060

0.25 0.25 0.38 0.38 0.51 0.51 0.51 0.64 0.64 0.76 0.76 0.76 0.89 0.89 1.02 1.02 1.02 1.14 1.14 1.27

0.010 0.010 0.015 0.015 0.020 0.020 0.020 0.025 0.025 0.030 0.030 0.030 0.035 0.035 0.040 0.040 0.040 0.045 0.045 0.050

pattern or core box to equalize temperature distribution and reduce distortion. Increasing mold thickness can add considerably to mold cost. Reducing the number of castings per mold will reduce piece-to-piece dimensional variation, but it will also lower productivity and increase cost. Careful design of the gating system is also important in holding close dimensional tolerances. Surface shrinks that develop at a section of critical dimensions can often be prevented by regating to assure adequate feeding or by chilling the metal at the critical location, either by the use of metal inserts in the mold or by local application of a mold material (for example, zircon sand) that will extract heat faster than the silica sand. Variation across the parting line can be reduced by:
Using higher squeeze pressure after applying

bonding glue


Blowing off both halves of the mold with an

air blast before applying the glue and clamping. (This is done to remove loose sand grains, which can be as large as 3.2 mm, or 1 / 8 in., in diameter. In one foundry, mold blowoff added 0.08 min to production time per mold but reduced tolerance across the parting line from +0.25 to +0.10 mm, or +0.010 to +0.004 in., in a 19 mm, or 3/4 in thick pad.)
Exerting closer control over consistency and placement of bonding glue
Using clamps or bolts to hold mold halves together in critical areas, particularly

Table 6 Typical tolerance relations for shell–mold gray iron castings Typical tolerance Dimension mm

in.

25–150 1–6 150–230 6–9 230–305 9–12 305–510 12–20

Across parting line

Between points in one part of mold

þ mm 

þ in. 

þ mm 

þ in. 

0.51 0.64 0.76 0.89

0.020 0.025 0.030 0.035

0.38 0.51 0.64 0.76

0.015 0.020 0.025 0.030

Table 7 Dimensional tolerances obtained in close-tolerance shell-mold castings Tolerances Normal Casting metal

Possible

þ mm/mm 

þ in./in. 

þ mm/mm 

þ in./in. 

Low-alloy steel

0.005 0.005(a) 0.010(b) 0.0035(c) 0.0020(d) 0.0013(e) 0.005(f) ...

0.005 0.005(a) 0.010(b) 0.0035(c) 0.0020(d) 0.0013(e) 0.005(f) ...

0.002 0.0015 ... ... ... ... ... 0.0025

0.002 0.0015 ... ... ... ... ... 0.0025

Zircon sand mold Alloy steels

0.005(g)

0.005(g)

0.0015

0.0015

Silica sand mold Stainless steel

4140 steel Pearlitic malleable iron

(a) þ0.76 mm (þ0.030 in.) on a 510 mm (20 in.) dimension. (b) Across parting line. (c) Section 13 mm (½ in.) thick, 130 mm (5 in.) long. (d) On a   75 mm (3 in.) diameter. (e) On a 150 mm (6 in.) diameter. (f) Across parting line on a 50 mm (2 in.) dimension. (g) Maximum in 965 mm (38 in.)

Shell Molding and Shell Coremaking / 613 when casting cavities are large and liquid metal pressure is correspondingly high (Fig. 19)
Backing up the shell mold with packed green sand or metal shot. This can add considerably to cost unless the system is adequate to handle the sand or shot. Examples of Net or Near-Net Dimensional Accuracy with Shell Molding. Shell-mold castings can be produced to a degree of dimensional accuracy not obtainable by the conventional sand-mold process. The nine components illustrated in Fig. 22 are examples of dimensional control that can be maintained in production. Several castings were accurate enough so that no subsequent machining was required. In all of these, the mold mixture was zircon sand

Fig. 22

with 2½% resin and ½% calcium stearate. The cold coating process was used:
Highway-tractor fifth wheel (Fig. 22a) dim-

ensions obtained by shell molding were so accurate that the complex surfaces of the underside of the fifth wheel required no machining. The four bracket holes were out-of-round less than 0.076 mm (0.003 in.), which was less than the variation of the forged pins that were pressed into these holes. The two 9.5 mm (3/ 8 in.) diameter holes (into which grease fittings were pressed) were held within  þ0.05 mm (þ0.002 in.) diameter. 
Front and rear jaw castings (Fig. 22b) required no machining, because they were cast to within 0.178 mm (þ 0.007 in.) of

required dimensions and within a variation of approximately 0.003 mm/mm (0.003 in./ in.) The dimensions were so accurate that the pinholes, which required critical alignment, could be used as-cast.
Fifth-wheel jaw (Fig. 22) was originally produced as a forging. When the jaw was made by shell-mold casting, the entire outside contour fit a gage, and the 31.8 mm (l¼ in) diameter hole, formerly drilled, was cast to size with a variation of only 0.102 mm (0.004 in.), which was within the specified tolerance range. No machining was required.
Lock casting (Fig. 22d): The length of the arms was critically related to the position of the holes and to the round bearing surface outside the horizontally aligned holes. By shell molding, sufficient accuracy was

Nine shell-mold castings that were produced to a high degree of dimensional accuracy. The mold mixture was zircon sand with 21/2% resin and 1/2% calcium stearate. The cold coating process was used. (a) 100 kg (220 lb) carbon steel casting (0.25% C) of highway-tractor fifth wheel (underside view). The complete shell mold, including shell cores, weighed approximately 135 kg (300 lb), compared to 2040 kg (4500 lb) for a corresponding green sand mold with flask. (b) Front and rear jaws for fifth-wheel coupling. 4.5 kg (10 lb) steel castings (0.25% C, 1.20% Mn) were poured, four to a mold. The molds were in the vertical position for pouring the front jaws, and in the horizontal position for pouring the rear jaws. (c) Jaw for fifth-wheel assembly. Steel castings (0.25% C, 1.20% Mn) were poured, four to a mold, with the molds in the vertical position. A tree-type (one common sprue) gating system was used. (d) 4 kg (8 lb) lock for fifth wheel. Castings (0.25% C, 1.20% Mn) were poured, four per mold and two per gate, with the molds in the vertical position. (e) Cast-tooth bevel gears. Steel castings (0.45% C) were poured with the molds in the horizontal position: one per mold for the larger gear, and two per mold for the smaller gear. Gating was through the hub riser. (f) 150 kg (325 lb) track roller for a crawler tractor. Steel castings (0.45% C) were poured, one per mold, with the molds in the horizontal position. Center gating was used. (g) 18 kg (40 lb) steel transmission planet carrier, cast in one piece (formerly a bolted assembly). Castings were poured, one per mold, with the molds in the horizontal position. Gating was through a center sprue and finger gates. (h) Radome hub. Castings were poured, one per mold, with the molds in the horizontal position. Gating was through the center hub. (i) Brake beam. Castings were poured, one per mold, with the molds in the vertical position. A modified-knife gating system was used.

614 / Expendable Mold Casting Processes with Permanent Patterns
















obtained to eliminate machining of the holes and bearing surface. Bevel gears (Fig. 22e): Machining was eliminated except for boring and facing the hub and cutting the key slot in each gear. Dimensional variation in diameter of a 325 mm (123/4 in.) gear was approximately 0.003 mm/mm (0.003 in./in.). The total variation in tooth thickness at pitch line was also 0.003 mm/mm (0.003 in./in.). These variations were within the allowable tolerances on the diameter and tooth thickness. Tractor track roller (Fig. 22f): A track roller for a crawler tractor was cast vertically in a shell mold. Machining on the outside of the part was completely eliminated, as was machining on some of the inside surfaces. Machining stock on other inner surfaces of the casting was reduced by 0.75 mm (0.030 in.). The out-of-roundness of the roller paths was 0.51 mm (0.002 in.) for the lower tread (smaller diameter) and 0.178 mm (0.007 in.) for the upper tread (larger diameter). The roller paths for 90% of the castings were concentric within 0.46 mm (0.018 in.) for the lower tread and 0.61 mm (0.024 in.) for the upper tread. All castings were concentric within the required 0.79 mm (0.031 in.). Transmission planet carrier (Fig. 22g) was originally made by bolting together two green sand castings. By the shell process, it was produced by casting as one piece. Although this casting still required machining on all surfaces, stock allowance for machining was only 2.3 mm (0.090 in.) per surface, approximately half that required for conventional sand castings. Radome hub (Fig. 22h) was cast with the thickness of the arms held to a tolerance of þ0.25 mm (þ  0.010 in.), which made it unnecessary to machine mated surfaces. Brake beam (Fig. 22i): The 1.8 m (6 ft) long, 54 kg (120 lb) steel brake beam shown was cast in a shell mold approximately 66 by 203 cm (26 in. wide by 80 in. long). The average section thickness of the T-shaped beam was 13 mm (½ in.). No machining was required.

Causes and Prevention of Casting Defects Most of the difficulties encountered in shellmold casting involve loss of dimensional control, metal contamination caused by metal-mold reactions, and loss of required surface finish. However, shell castings are also subject to defects such as misruns and hot tears, and it is often necessary to revise the molding technique to eliminate these defects. Misruns in a casting are surface irregularities caused by incomplete filling of the mold. An example of a misrun is shown in Fig. 23. The gear casting of aluminum bronze was poured horizontally, three to a mold, and fed through

two gates, as shown in Fig. 23. The gear teeth were used as-cast and had to be clean and sharp. A high percentage of castings made from the initial shell mold were rejected because of incomplete fill of the last few teeth at each end of the gear. When flow-offs were added (Fig. 23) at both ends of each mold cavity, the misruns did not occur, and all castings produced were acceptable. Prevention of misruns due to gas entrapment can be done by venting. For example, a thinwall 1050 steel hub casting was poured in shell molds (12 per mold) by means of an overlapping runner (or “kiss” gate), as shown in Fig. 24. The parting line was in a vertical plane; the casting was entirely on one side of the parting line, with the mold forming an integral core. With the parting line positioned vertically, a misrun, consisting of a smooth depression, developed on the inner face of the outer wall (Fig. 24), resulting in a high percentage of scrap. A combination mold-gas vent and metal flow-off was added, and the parting line of the mold was tilted slightly in the direction of the mold cavity. This vented the gas that evolved from the enclosed sand rib, thus preventing misruns and considerably decreasing the number of scrap castings. Hot tears in shell-mold castings may be caused by the following:
Improper root radii or re-entrant angles:

Fig. 23

Shell-mold-cast sector gear for which addition of flow-offs prevented misruns

Fig. 24

Thin-wall casting in which misruns were prevented by addition of a flow-off for venting trapped mold gas

Small radii at section changes cause hot spots, and hot tearing will occur at the root of such radii.
Improper joining of sections: Sections joined in a “T” are likely to tear if the ratio of the areas of the two sections joined exceeds 2 to 1.
Hindered contraction: Hot tearing caused by hindered contraction is not unusual in casting sections less than 9.5 mm (3/8 in.) thick. A casting redesign that eliminated hot tearing is illustrated in Fig. 25. A 1050 steel casting was initially cast to the shape shown in Fig. 25(a). Hot tears developed at the shoulders, as indicated, causing a high rejection rate. The shoulders and re-entrant angle sections were eliminated from the shaft design, as shown in Fig. 25(b). The redesigned castings were produced without hot tears. Chill. Casting alloys that develop a desired structure only under conditions of controlled cooling may be sensitive to a change from sand to shell molding that is made to reduce machining stock. This is particularly true in casting gray or ductile iron, wherein a reduction of section thickness may result in unacceptable hard areas caused by chill. One such problem and its solution are described in the following example. Redesign of a gray iron casting for preven tion of chill is illustrated in Fig. 26. The gray iron cylinder was initially cast in a green sand mold, to a 9.5 mm (3/ 8 in.) wall thickness

Fig. 25

Casting that was redesigned without sharpangle shoulders, to avoid hot tears

(Fig. 26a). This casting had an acceptable pearlitic matrix but required an excessive amount of machining. To obtain better dimensional control and a casting that required less machining, the cylinder was redesigned to a 356 mm (14 in.) wall thickness and was produced by shell-mold casting. However, as shown in Fig. 26(b), chills occurred at the ends of the casting. The chilled metal was not machinable, and the castings were rejected. By thickening the wall at the ends of the shell-mold casting, as shown in Fig. 26(c),

Shell Molding and Shell Coremaking / 615 Molding sands that are too coarse cause an “orange-peel” effect that is evenly distributed over the entire surface of the casting. Other causes of casting defects are cracked molds, soft molds, and shifting of the mold halves as they are being bonded together.

Comparisons with Green Sand Molding

Fig. 28

Cost of shell molding versus green sand molding for production of a chair seat bracket in ascending quantities. Pattern costs not included

Fig. 26

Redesigns to minimize machining and then to prevent chills. (a) Original sand casting developed chills at ends. (b) Redesign for shell-mold casting allowed thinner walls for less machining. (c) Thickening the walls at the ends of the cylinder prevented chills without adding appreciably to the machining required.

Fig. 27

Cost of a conduit elbow produced in various quantities in shell molds and in green sand molds. Pattern and core equipment costs not included

a machinable pearlitic matrix was obtained throughout the casting. The redesigned shellmold casting still required less machining than the original sand casting.

Fig. 29

Cost per pound (top chart) and per casting (bottom chart) for shell versus green sand molding for production of two different castings in ascending quantities

Surface Defects. The appearances of the ascast surfaces on a shell-mold casting are at their best when the molten metal solidifies in intimate contact with the mold and core surfaces. When the metal is too hot, the surface appearance deteriorates progressively from the lowest to the highest point on the casting with respect to the pouring position. If the metal is too cold, definition of detail deteriorates. Gassy metal can cause tiny irregular depressions on the cope side of thin sections. Depressions can be located in a group at a re-entrant corner where gas has become trapped. Dirt and slag can cause the surface of a casting to be marred with pits and small projections in widespread isolated areas. A pimply surface can occur where burned particles of molding sand have adhered to the pattern surface. Such areas generally occur on horizontal surfaces rather than vertical surfaces, because sand adhering to a vertical surface of a pattern would cause scratches in the mold surface.

The cost of a particular casting depends on the size and complexity of the casting, number of castings per mold, number of cores required, casting procedure, efficiency of the process, and the need for subsequent machining. The following examples provide some relative cost comparisons between shell molding and green sand molding. In some cases, the cost advantage of shell molding may be realized at relatively low production quantities. Example: Cost Comparison versus Quantity for a Cast Iron Elbow. A 0.15 kg (0.33 lb) ferritic malleable iron conduit elbow (Fig. 27) was originally cast in green sand, using an eight-piece pattern, and with solid oil-sand cores that were made in a two-cavity core box. A change was made to shell-mold casting, using a 16-piece pattern, and shell cores made in a two-cavity box. Based only on direct molding costs, the castings were cheaper to produce by the shell process when quantities exceeded 375 castings (Fig. 27). However, the cost of patterns and core equipment was 28% higher for shell molding than for green sand. If these costs were included in the comparison, the shell process would be cheaper only if quantities exceeded approximately 14,000 castings. Additional benefits from shell molding the elbow castings were improved surface finish, more consistent dimensions, and reduced scrap caused by misruns. Example: Cost Comparison versus Quantity for a Cast Iron Bracket. A cast 1.1 kg (2.38 lb) bracket for a swivel-chair seat (Fig. 28) was changed from green sand to shell molding. As the relative costs plotted in Fig. 28 show, shell molding was the lower-cost method for quantities of more than 300 brackets. These data do not, however, consider difference in pattern costs. If pattern costs had been included, the point at which the shell-mold cast brackets would have cost less than the sand-cast brackets would have been approximately 15,000 castings. Comparison with Pattern Costs. Figure 29 presents additional comparisons of relative cost versus quantity for shell molding and green sand molding in the production of castings weighing 0.57 and 1.96 kg (1¼ and 41/3 lb). The top chart in Fig. 29 shows cost per pound, not including pattern costs. The bottom chart shows cost per casting and does include pattern costs, which generally are higher for shell molding than for green sand molding, even though the two types of patterns are basically

616 / Expendable Mold Casting Processes with Permanent Patterns

Production details for shell-mold casting Production details for shell-mold casting

Production details for shell-mold casting Number of castings per mold Weight of metal poured per mold Weight of trimmed casting Pouring position Pouring temperature Mold temperature Shakeout method Surface finish Production rate Total monthly production

Fig. 30

12 39 kg (87 lb) 1.1 kg (2.5 lb) Vertical 1370 ⬚C (2500 ⬚F) min Shot backup, approx 175 ⬚C (350 ⬚F) Automatic dump 6.35 mm (250 min.) (visual) 7800 castings per month 20,000 castings

Cast section of crankshaft from green sand molding and shell-mold casting

the same. The reasons for this are that shellmold patterns must be capable of operation at temperatures in the range of 175 to 260  C (350 to 500  F) but maintain the required temperatures from one mold or coremaking cycle to the next at the required production rate. Also, shell-mold patterns must possess and maintain the required degree of accuracy demanded by the finished as-cast dimensional tolerances of the casting to be produced. Example: Reduction of Weight and Machining with Shell Molding. A small ductile iron crankshaft was cast in green sand to the shape shown in Fig. 30(a). This casting required machining on all surfaces. To reduce both the weight of the casting and the amount of machining required, production was converted to shell molding. The shell-mold casting

Number of castings per mold Weight of metal poured per mold Weight of trimmed casting Pouring position Pouring temperature Mold temperature Shakeout method Surface finish Production rate Total monthly production

6 36 kg (80 lb) 3.2 kg (7 lb) Vertical 1370 ⬚C (2500 ⬚F) min Shot backup, approx 175 ⬚C (350 ⬚F) Automatic dump 6.35 mm (250 min.) (visual) 3600 castings per month 100,000 castings

Fig. 31

Pump casting produced with minimum draft in a shell mold When casting had been poured in a green sand mold, shaft portion required a draft angle of 1 30 min and had to be machined.

(Fig. 30b) was nearly 30% lighter, and the surfaces shown without shading in Fig. 30(b) were acceptable as-cast, without machining. Production details for the shell-mold casting are presented in the table with Fig. 30. Example: Smaller Draft and Less Machining with Shell Molding. When a gray iron transmission-pump casting (Fig. 31) was poured in a green sand mold, the shaft portion required a draft angle of 1 30 min and had to be machined. However, by casting in a shell mold, it was possible to reduce the draft angle to 0 15 min, thus eliminating the machining of the shaft portion. Production details for the shell mold casting are given with Fig. 31. Example: Elimination of Machining for Valve Plate. When a gray iron refrigeratorvalve plate (Fig. 32) was produced as a green sand casting, the eight kidney-shaped slots had

Number of castings per mold Weight of metal poured per mold Weight of trimmed casting Pouring position Pouring temperature Mold temperature Shakeout method Surface finish Production rate Total monthly production

18 25 kg (55 lb) 0.7 kg (1.6 lb) Vertical 1370 ⬚C (2500 ⬚F) min Shot backup, approx 175 ⬚C (350 ⬚F) Automatic dump 6.35 mm (250 min.) (visual) 10,800 castings per month 20,000 castings

Fig. 32

Valve plate that did not require machining of slots when it was cast in a shell mold instead of a green sand mold

to be machined completely to the correct final dimensions. The slots were not cored, because they could not be cast to required dimensions and therefore would have required machining regardless of coring. When production was converted from green sand to shell-mold casting, the slots were produced by cores; no machining was required. All slots were within  þ0.25 mm (þ0.010 in.) in diameter of the circumscribed  circle (for the four slots); also, width and length of the individual slots were within  þ0.25 mm (þ 0.010 in.). Production details for the shellmold process are given with Fig. 32.

REFERENCE 1. Forging and Casting, Vol 5, Metals Handbook, 8th ed., American Society for Metals, 1970

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 617-627 DOI: 10.1361/asmhba0005249

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Slurry Molding Revised by Charles D. Nelson, Morris Bean and Company; Wayne Rasmussen, Consultant; and John Jorstad, J.L.J. Technologies, Inc.

SLURRY MOLDING encompasses two distinct processes, plaster molding and ceramic molding, each of which is presented in this article. This content was prepared by experts in the field, selected for their many years of practical experience. Both processes are used to produce castings having fine detail, smooth surfaces, and dimensional accuracy. Ceramic molds have good refractory properties and can withstand high ferrous-metal pouring temperatures, whereas plaster molds are only suitable for lower-melting-temperature nonferrous metals and alloys. Details of the metals that can be cast and mold materials are given in the article, along with details of versions of the two processes. Both processes use expendable molds, and mold materials are not fully reclaimable; thus, mold costs tend to be high. Still, slurry processes are cost-effective, especially for high-integrity parts too large to be investment cast.

castings weighing as much as 34 kg (75 lb) are being produced in substantial quantities. Advantages. Most of the advantages to be gained from the use of plaster molds have to do with casting quality:

PLASTER MOLD CASTING is a specialized casting process used to produce nonferrous castings that have greater dimensional accuracy, smoother surfaces, and more finely reproduced detail than can be obtained in sand molds or coated permanent molds. The three generally recognized plaster mold processes are:
Conventional plaster mold casting
The Antioch process
The foamed plaster process

Applications Near-net shape castings, such as compressor wheels, impellers, components of electronics gear, tire molds, aerospace fuel pump systems, and heat exchanger systems, are examples of castings produced by plaster mold casting. Approximately 90% of the plaster mold castings weigh less than 9 kg (20 lb) each, although

Plaster Mold Compositions


As-cast surface finishes of 50 to 125 root








Plaster Molding

during pouring or to provide greater permeability by a special procedure.




mean square (approximately ISO N7 or N8) are readily obtained. By using flexible rubber patterns and incorporating the advantages of lost-wax techniques, slurry molding successfully produces intricate designs. The dimensional accuracy of the castings is good (approximately equal to the accuracy of castings made in investment molds). Thin sections can be cast because the castings normally cool quite slowly due to the insulating qualities of the plaster; wall thickness limits are a function of casting design, but selected areas 0.64 to 1.02 mm (0.025 to 0.040 in.) thick are typical. Slow cooling minimizes warpage and promotes uniformity of structure and mechanical properties in the castings. Plaster molds are especially well suited to the use of chills, which permit close control of thermal gradients in the molds; this optimizes mechanical properties desired in isolated areas of a complex casting.

Disadvantages in using plaster molds are largely related to molding and casting procedures:
Cost is high, partly because of the lengthy

processing procedures and partly because the mold materials are not reusable.
Mold processing requires more equipment than is required for most other molding processes.
Because of the lengthy processing procedures needed to make plaster molds, duplicate (or multiple) pattern equipment, as well as duplicates of other processing equipment, may be required in order to meet production schedules.
With the exception of the foamed plaster process, the permeability of plaster molds is inherently low when compared to the sand-casting process; therefore, it is usually necessary to apply pressure or vacuum

In all of the processes for making plaster molds and cores, the principal mold ingredient is calcium sulfate (gypsum). Other materials are used to enhance such mold properties as green and dry strength, permeability, and castability. These include cement, ceramic talc, fiberglass, sand, clay, wollastanite, pearlite, and fly ash. Cores are typically made of the same material and by the same process as molds, but cores are sometimes made of other materials, such as shell molding sand. Cores for the match plate plaster mold process are made of sand. Specific mold and core formulations, as well as techniques for making the slurries, are discussed under each process in this article. Characteristics of Calcium Sulfate. Calcium sulfate exists as two different hydrates and in an anhydrous form. In its dihydrate form (CaSO42H2O), it is known as gypsum. (Commercially, all three forms of calcium sulfate are sometimes referred to as gypsum.) When heated above an equilibrium temperature of 128  C (262  F), gypsum loses three-fourths of its water to form the hemihydrate CaSO4H2O, which is plaster of Paris. Similarly, when the plaster of Paris is heated above an equilibrium temperature of 163  C (325  F), all of the water of crystallization is removed, and anhydrous calcium sulfate (CaSO4) is formed. The aforementioned temperatures are based on laboratory studies of pure calcium sulfate under equilibrium conditions. Temperatures at which the hydrates are formed depend greatly on vapor pressure. For practical purposes, temperatures well above equilibrium temperatures are generally used. If it is exposed to moisture at a temperature lower than 100  C (212  F), the anhydride will absorb water and will revert to the hemihydrate. Similarly, when plaster of Paris (the hemihydrate) is mixed with water, gypsum (the dihydrate) is reestablished, yielding a coherent solid in a few minutes. This reaction is termed setting. Prolonged exposure of the hemihydrate to moist

618 / Expendable Mold Casting Processes with Permanent Patterns air at temperatures below 100  C (212  F) results in a slow conversion to the dihydrate. In the production of a plaster mold, plaster of Paris is mixed with water in excess of that required to form the dihydrate to make a slurry. This slurry is immediately poured over a pattern, standard or match plate, or into a core box, where it sets, forming a solid mold or core composed of gypsum with free water distributed throughout the plaster mold. The next stage is to dry the plaster in an oven to remove the excess water. The drying temperature depends on the plaster molding process. If partings and waxes are used, the temperatures must be high enough to remove any waxes. Plaster molds have low heat capacity. Therefore, the cooling rates for castings made in such molds are low. Figure 1 shows that the freezing time for a test casting made in a conventional plaster mold is more than four times as long as that for one made in a hard-rammed green sand mold. Slow cooling has advantages and disadvantages. It permits better feeding, particularly in thin sections, as well as the replication of intricate detail that would be difficult to obtain with fast cooling. However, it slows production by lengthening the time between pouring and shake out, and for most alloys, it results in castings of lower strength. This latter disadvantage can usually be eliminated by the use of chills in specific areas.

Patterns and Core Boxes Metal and Rubber Patterns. Patterns and core boxes for plaster mold casting are commonly made of metal (aluminum alloy, brass, or zinc alloy). Flexible rubber patterns are widely used for producing intricate molds, such as the molds used for casting wheels having angular vanes. Flexible rubber patterns having as much as 30 negative draft can be withdrawn without damaging molds that have high green strength. Cast epoxy-resin patterns are useful for some purposes. For example, when several patterns are required, they can be made quickly by molding them from epoxy resin in metal master dies. Cast epoxy patterns also have greater dimensional stability and longer production life.

Stereolithography apparatus is now frequently used to make patterns for plaster molds, either as an end in itself or to make metal components in plaster molds as prototypes for die-cast parts. A choice of resins is available. Wood Patterns. Because patterns for production casting by plaster mold processes must withstand repeated exposure to liquid slurries, wood patterns usually are not adequate. Wood patterns, although not extensively used in the production of castings, are specified as patterns for match plate pattern molds. Wood patterns must be sealed, or they will swell and distort.

Mold-Drying Equipment Plaster molds made by any process must be dried. The most common drying temperatures range from 120 to 260  C (250 to 500  F), but temperatures to 870  C (1600  F) have been used under carefully controlled conditions. When the optimal temperature for a specific application has been determined, it must be closely maintained—usually within  þ6  C  (þ 10 F). After this initial requirement has  been met, the choice of equipment depends largely on the temperature to be used and the number of molds to be processed. For drying at 120 to 260  C (250 to 500  F), either batch-type or conveyor core ovens are satisfactory. Higher-temperature drying requires furnaces similar to those used for the heat treating of steel. Either batch-type or conveyor furnaces can be used, but conveyor furnaces are generally preferred, especially for drying large quantities of molds. The molds must be cooled to at least 205  C (400  F) before they are removed from the furnace.

Metals Cast in Plaster Molds Only nonferrous metals are cast in plaster molds. Ferrous alloys are not suitable, because most of them are poured at temperatures that would melt the calcium sulfate. Calcium sulfate undergoes a phase transformation at 1195  C (2180  F) and melts at 1450  C (2640  F). Aluminum. All of the aluminum alloys that can be successfully cast in sand molds are suitable for casting in plaster molds. The more readily castable alloys—B443, A344, 355, 356, and 357—are preferred. The premiumstrength compositions, such as 224 A356, C355, and A357, are frequently used for premium engineered castings. The impellers (Fig. 2) show the intricate thin-wall, highstrength sections that can be cast. Copper. Most of the coppers and copper alloys that can be successfully cast in sand molds can be cast in plaster molds. Cast alloys are designated UNS C80100–C99999. Copper alloys containing more than approximately 5.0% Pb are generally not recommended for casting in plaster molds, because the higherlead alloys react with some mold compositions, resulting in poor surfaces on the castings and defeating one of the objectives of plaster mold casting. Magnesium alloys are not recommended for plaster mold casting. A reaction is likely between the magnesium alloys and the mold material. In particular, magnesium alloys will react with any free water that remains in the mold and will cause an explosion. Zinc alloys are frequently cast in plaster molds, most often for prototype castings. The die-casting alloys AG40A and AC41A are often used, but a proprietary alloy whose coefficient of thermal expansion is very close to that of

Type of mold Aluminum slab at room temperature Green sand, 7% water (rammed hard) Dry sand Green sand, 5.8% water (rammed lightly) Antioch-process plaster Conventional plaster Freezing time, minutes

Fig. 1

0

5

10

15

Comparison of freezing times for identical test castings poured in various types of molds

Fig. 2

Aluminum alloy 224.0 impellers produced by low-pressure plaster casting. Source Ref 1

Slurry Molding / 619 aluminum alloys is frequently cast. Master patterns appropriate for this zinc alloy or for aluminum can be made according to a single shrinkage rule. Nominal compositions of the three zinc alloys mentioned are:

Alloy

AG40A AC41A ZA-12

Composition, %

UNS designation

Al

Cu

Mg

Zn

Z33520 Z35531 Z35631

4.10 4.10 11.0

0.10 max 1.00 0.85

0.035 0.055 0.022

bal bal bal

Except for allowances for shrinkage, the technique for making the molds is the same for aluminum and zinc alloys.

Conventional Plaster Mold Casting Melting practice equipment and methods for metal to be poured into plaster molds are the same as those used for preparing the metal for pouring into other types of molds. A brief outline of conventional plaster molding is given in the following sequence of operations for producing conventional plaster molds:











Mix dry ingredients Add dry ingredients to water Soak (0.5 to 4 min) Mix (0.5 to 5 min) Coat patterns (or core boxes) with parting agents Pour slurry Set at room temperature (15 min) Remove pattern Dry molds (or cores) Assemble cores and mold halves

Composition and Preparation of Dry Ingredients. Dry ingredients ready to mix with water are available as proprietary compositions. For small operations, it is usually cost-effective to purchase these ready-to-mix dry ingredients. Compositions vary considerably, but the gypsum content (typically, a blend of gray and white gypsum) is commonly 50 to 60%, by weight, of the dry mixture. The rest of the mixture consists of two or more filler materials, such as wollastanite, ceramic talc, fly ash, pearlite, and sand. Portland cement, Hydrocals (U.S. Gypsum), fiberglass, and terra alba are also sometimes used to enhance the strength of the mold and to improve setting behavior. Mixing the Slurry. Equipment requirements for mixing the slurry are very precise. The principal components of a batch-type mixer are a bucket and a propeller (Fig. 3). The bucket height should be equal to or slightly greater than the top diameter, and the bottom diameter should be approximately two-thirds of the top diameter. Preparing the Pattern. Because the slurry begins to set as soon as it is mixed, it must be poured over the pattern (or into the core box) almost immediately, and the pattern must be ready to receive the slurry.

Pouring the Slurry. When the slurry is poured into the flask, the lip of the bucket should be kept close to the pattern. The slurry should be poured at a constant rate and made to flow over the pattern rather than splash on it. Set time varies somewhat, depending on the slurry composition, but 30 min is usually the maximum set time. After the slurry has set, the pattern and flask are removed, and the drying cycle is started as soon as is practical. Drying the Molds. All conventional plaster molds must be dried enough to expel the free water and most of the water of crystallization to a depth of at least 13 mm (0.5 in.) below the surface. Some foundries prefer to have the molds completely calcined to remove all volatile materials. Oven drying should begin as soon as is practical after the mold has been removed from the flask. Molds that have partially dried by standing at room temperature are more susceptible to cracking than those that are oven dried immediately after setting. If the molds must stand at room temperature for some time (such as over a weekend or even overnight), they should be covered with a damp cloth or stored in a humid atmosphere. During drying, the mold should be uniformly supported on its edge or face. Common practice is to place the mold on a perforated flat metal plate, a rigid metal grid, or some other type of level support. If smooth plates are used for support, they should be covered with a thin layer of talc or similar material to allow movement of the mold during drying without the danger of its cracking or warping. Time and temperature cycles used for drying conventional plaster molds vary widely among foundries. Temperature may vary from approximately 175 to 870  C (350 to 1600  F), and time from 45 min to 72 h. The fact that furnaces are operated at high temperatures (760 to 870  C, or 1400 to 1600  F) does not mean that all areas

of the mold must reach this temperature range, but the interior of the mold must be at least 105  C (220  F) to ensure removal of the free water. (Mold temperatures are measured by thermocouples in the center of mold sections.) The specific time and furnace temperature required for drying a specific mold usually must be determined by experimentation. Once established, the time-temperature cycle must be rigidly controlled for best reproducibility. Mold Assembly. After the mold has been removed from the oven, cores are placed in the drag half. Depending on mold complexity, this can be done while the molds are still hot or after they have cooled to room temperature. Ceramic or plaster pins are placed in the holes provided in the drag half. The cope half is then lowered so that the pins protruding from the drag enter the matching pin holes in the cope, as shown in Fig. 4. Preheating. Following assembly, the mold is ready for preheating. Some foundries preheat all conventional plaster molds to a pre-established temperature (commonly 120  C, or 250  F) before pouring the metal. Other foundries preheat the molds only in specific applications for which preheating has proved advantageous. Preheating of the mold can help to minimize defects or to obtain better replication of fine detail in the casting. Pouring Practice. A dried plaster mold made by the conventional process has extremely low permeability—approximately 1 to 2 AFS, compared with 80 and upward for sand molds. The American Foundry Society (AFS) established a standard laboratory test for measuring the permeability of sand. Analog and electronic instruments are available to measure to this standard. Because of this low permeability, either vacuum assist or pressure is usually required for the pouring of molds. Simple gravity pouring is rarely done. Most foundries use vacuum assist for pouring conventional plaster molds. A typical setup for vacuum-assist pouring is shown schematically Sprue

20 to 30⬚

Cope 10 to 15⬚

Pin hole (1 of 4) Core

Ceramic locating pin (1 of 4)

Propeller Bucket 1 to 2 in. Drag

Fig. 3

Preferred shape of bucket and position of propeller used in the batch mixing of plaster slurries for conventional plaster molds

Fig. 4

Plaster mold and core showing location pins for matching the cope and drag sections

620 / Expendable Mold Casting Processes with Permanent Patterns in Fig. 5(a). In this setup, the mold is supported by a fixed bottom plate (plate B, Fig. 5a).

Match Plate Pattern Molding Match plate patterns are cast by a particular adaptation of the conventional method of plaster molding. Changes in details of the conventional method ensure high accuracy and smooth surface finish, which are required in metal match plate patterns. Common dimensional requirements for match plate patterns are:
Match between the cope and drag, within

0.25 mm (0.010 in.)


Parallelism between the two halves, within

0.51 mm (0.020 in.) A surface finish of 2.5 mm (100 min.) or better is obtained. Sizes and Weights of Match Plate Patterns. A match plate pattern consists of a cope section and a drag section separated by a plate. These three components are cast as an integral unit. A typical match plate pattern is shown in Fig. 6. Metal match plate patterns vary considerably in size and weight. Patterns weighing as much as 180 kg (400 lb) each have been cast, although these are unusually large. Match plate patterns seldom weigh more than 45 kg (100 lb) each, and most weigh from 8.2 to 16 kg (18 to 35 lb). Mold Composition. Materials for molds for match plate patterns are available as proprietary dry mixtures ready for mixing with water. However, some foundries prefer to make their own mixtures. In heavy sections, it is common practice to insert aluminum chills with nail-like protrusions into the plaster to make an internally cast chill. It is common to have 50 to 100 chills in a large mold with varying sections. Pouring Practice of Match Plate Molds. Because of the low permeability of a match

plate mold, some assistance is required to fill the mold quickly and completely. Pressure assist is used rather than vacuum, partially because pressure equipment is more adaptable to a wider variety of flask sizes than vacuum equipment is. However, some foundries use a combination of pressure and vacuum assist. The equipment for pressure casting is illustrated in Fig. 7. The first step in its use is to place a diaphragm of a ceramic fiber paper such as Fiberfrax (Unifrax Corp.) (approximately 3.2 mm, or 0.125 in., thick) over the sprue in the mold. A ceramic-lined cylinder is then placed on the diaphragm. A predetermined amount of metal is poured into the cylinder, and a cap is placed on the cylinder and clamped tight. The cap is attached to a source of compressed air. After the cap has been secured in position, the air valve is opened. Air pressure against the molten metal ruptures the ceramic fiber diaphragm and forces the metal into the mold cavity. A pressure of 10 to 17 kPa (1.5 to 2.5 psi) is kept on the sprue for approximately 30 min.

The Antioch Process The Antioch process (U.S. Patent 2,220,703) was developed to overcome the principal limitations of conventional plaster molds and cores without sacrificing the advantages of the plaster mold process. If undried molds are partially dehydrated and then allowed to rehydrate without being disturbed, gypsum crystals slowly recrystallize into granules approximately the size of sand grains, and the mold acquires a porous structure of relatively high permeability. Permeability is held within a range of 15 to 30 AFS; the permeability of conventional plaster molds is 1 to 2 AFS. Recrystallization does not take place at the surface of the mold, because not enough water is present. Therefore, the surface remains smooth. In addition to the greater permeability developed by the dehydration-rehydration process, the molds produced have greater heat capacity

than conventional plaster molds because they are composed of approximately 50% sand. The freezing time for a casting in an Antioch process mold is only approximately 20% longer than the freezing time for an identical casting in a lightly rammed green sand mold and is less than half the time required for a casting in a conventional plaster mold (Fig. 1). Unlike conventional molds, Antioch process molds do not shrink. In fact, they expand slightly—from 0.001 to 0.0025 (unitless, i.e., mm/mm or in./in.)—during processing. Because of their porous structure, the molds have low dry strength. This characteristic, in promoting early collapse of cores as the casting cools, minimizes hot tears in the castings. For very large molds, the low dry strength sometimes necessitates the use of internal reinforcement, which is achieved with hardware cloth or core rods such as those used in making sand cores. When possible, reinforcement is avoided because of the difference in expansion between the reinforcing metal and the molding material. After setting, but before the dehydrationrehydration treatment, Antioch process molds have relatively high green strength. When flexible Cope side Plate

Drag side

Fig. 6

A typical metal match plate pattern

Pressure Cap

Toggle clamp (I of 2)

Steel shell Ceramic pouring cup

Plate A

Hole for pouring cup (in plate A only)

Vacuum line (1 of 2)

Ceramic liner

Pressure

A

A

(b) Detail of plate A; plate B identical except as noted (a)

Sprue Plate B

Section A-A Mold half (cope)

Vacuum channel

Fig. 5

Typical setup for vacuum-assist pouring of a conventional plaster mold casting. (a) Side view of conventional plaster mold positioned between upper and lower plates for pouring with vacuum assist. (b) Details of a top plate (A) seen from the bottom showing vacuum channels

Fig. 7

Ceramic fiber paper

Cylinder used for pouring a casting with pressure assist

Slurry Molding / 621 rubber patterns are used, this high green strength permits the withdrawal of patterns having a severe back draft without damage to the mold. This makes the Antioch process particularly well suited to the production of molds for parts having angular, bladelike sections—rotor sand nozzles, for example. Mold and Core Materials. The dry mixture for Antioch process molds and cores consists of silica sand, white molding plaster, ceramic talc, and a small amount of material (such as Portland cement) for expansion control. The typical mixture given in Table 1 varies somewhat among different foundries. However, once a formulation has been established in a specific foundry, it is retained for all castings, regardless of their size and shape. Optimal results are obtained by weighing all ingredients accurately. Only by consistent use of a specific formulation is it possible to obtain maximum reproducibility. Processing. The sequence of operations for producing Antioch process molds is as follows:













Mix dry ingredients Add dry ingredients to water Soak (0.5 to 3 min) Mix (1 to 4 min) Coat patterns (or core boxes) Pour slurry Set at room temperature (15 to 20 min) Remove pattern Dehydrate in autoclave (6 to 12 h) Rehydrate in air (14 h) Dry molds (or cores) Assemble cores and mold halves

Table 1 Typical composition of dry material for Antioch process molds Weight

Washed silica sand (AFS 50 is typical) White molding plaster Ceramic talc Portland cement Sodium silicate Western benonite Terra alba

accomplished with several types of mixers. Regardless of the type of mixer used, the greater the power input, the finer the mold structure (the smaller the air cells) and the smoother the surface of the mold. The bucket should be similar to that shown in Fig. 3, but the mixing device is a round, twoply, 3 mm thick (0.125 in. thick) rubber disk attached to a shaft, as shown in Fig. 8. The diameter of the shaft is not critical. Other mixer designs that are successfully used for mixing air into foamed plaster incorporate a 127 to 152 mm (5 to 6 in.) diameter perforated disk mounted on a shaft that is tilted slightly from normal to the shaft. Processing. The following sequence of operations is typical of the routing required to produce a component by the foamed plaster molding process:
Mix foamed plaster dry mix with 32 to 43  C

Foamed Plaster Molding Process

The set time for a slurry formulated from a composition such as that shown in Table 1 will be approximately 15 to 20 min. Set time can be decreased by adding up to 3% terra alba and heating the water. For example, the minimum set time of 6 to 7 min is achieved by adding 3% terra alba and using water at 32  C (90  F). For dehydration, the molds are placed on suitable racks in a standard autoclave. The autoclave is sealed, and steam is admitted. The autoclave is operated with a steam pressure of 105 kPa (15 psi) for 6 to 8 h. For extremely large molds, it is operated for 12 h. The autoclave is then opened, and the molds are removed.

Ingredient

Rehydration. The mold is permitted to remain at room temperature for 14 h in a humid, low-temperature atmosphere. After rehydration, the mold is ready for drying. Drying temperature ranges from 175 to 230  C (350 to 450  F), and drying time from 1 to 70 h. Drying time depends mainly on the size of the mold and the temperature used. The center of the mold must reach a temperature of at least 120  C (250  F). This can be accomplished considerably more quickly at an oven temperature of 230  C (450  F) than at 175  C (350  F). Regardless of the cycle used, it is important that the same cycle be used for all molds of the same size. Maximum reproducibility (of dimensions, in particular) can be achieved only through close control of the cycle. Metals Cast. All of the aluminum alloys that can be cast in other types of plaster molds can be cast in Antioch process molds. Most copper-base alloys can be cast in Antioch process molds. Yellow brass is the copper alloy most often cast. The Antioch process is seldom used for alloys that must be poured at temperatures above approximately 1040  C (1900  F).

kg

lb

20.0 18.6 3.6 0.2 0.4 1.1 1.5

45 41 8.0 0.4 0.8 2.5 3.4

Note: Slurry is made by mixing 45 kg (100 lb) of the above dry mixture with 24 kg (54 lb) of water.

The foamed plaster process offers a means of achieving greater mold permeability than can be obtained in conventional plaster molds and at much improved dry strength over Antioch plaster molds. This gain is achieved by adding a foaming agent, such as alkyl aryl sulfonate, either to the dry ingredients before mixing or to the liquid slurry, as a separately generated foam mix. A special method of mixing foams the slurry with many fine air bubbles, thus decreasing the density and increasing the volume of the slurry. In general, with regard to the composition of the metal poured, casting size, and casting shape, the applicability of foamed plaster molds is the same as that of plaster molds made by other procedures. Characteristics. Foamed plaster molds have smooth surfaces with air cells just below the surface. During setting and subsequent drying of the molds, these air cells become interconnected, thus permitting the escape of gases as the metal is poured. The permeability of a foamed plaster mold depends mainly on the volume increase from the addition of air when the slurry is mixed. For most molds, a volume increase of 50 to 100% is recommended. This increase usually results in a mold permeability of approximately 15 to 30 AFS for dried molds. Equipment for mixing will vary somewhat with the slurry used. The type described subsequently has proved suitable for mixing a proprietary dry mixture with water. The mixer must be capable of beating air into the slurry and producing air cells no larger than approximately 0.25 mm (0.01 in.) in diameter. Large air cells are not permitted, because they break under pressure from molten metal, resulting in casting defects. Proper mixing can be












(90 to 110  F) soft water to a consistency of 80 to 100 AFS. The mixer used should have two speeds: a high speed (1750 rpm) to mix the dry material and water thoroughly and a slower speed (800 to 1000 rpm) to beat air into the slurry and to obtain a smooth, creamy slurry. Mixing time is varied to obtain 100% volume increase. Pour slurry Set time varies from 20 to 40 min. Set time is a function of water temperature, mixing time (high speed), dry material setting characteristics, and consistency. Changes in variables to decrease set time also serve to decrease working time. Strip patterns Dry in standard-type dryers at temperatures from 175 to 260  C (350 to 500  F) for 8 to 16 h, depending on mold size. Foam plaster is very insulating, and thermal gradients can set up dimensional change, thus causing mold cracking. Avoid conditions that will cause thermal shock at mold temperatures above 120  C (250  F). Assemble and cast; molds can be poured from room temperature up to the dryer temperature.

Metals Cast. For the most part, all aluminum alloys can be cast, but the aluminum-magnesium (5xx.x) alloys are the most compatible. Shaft

Rubber disk

Fig. 8

Two-ply rubber disk attached to a shaft for the high-speed mixing of foamed plaster slurries

622 / Expendable Mold Casting Processes with Permanent Patterns For aluminum alloys that are subject to solidification cracking or have other properties that are sensitive to the cooling rate, the more insulating foam plaster may have a negative effect. The needed selective chilling may be less effective, so exacting foundry techniques are required.

Ceramic Molding CERAMIC MOLDING techniques are based on processes that employ permanent patterns and fine-grained zircon and calcined, high-alumina mullite slurries for molding. Except for distribution of grain size, the zircon slurries are comparable in composition to those used in ceramic shell investment molding. Like investment molds, ceramic molds are expendable. However, unlike the monolithic molds obtained in investment molding, ceramic molds consist of a cope and a drag or, if the casting shape permits, a drag only. The Shaw process and the proprietary Unicast processes are discussed in this section. Both investment molding and ceramic molding can produce castings with fine detail, smooth surfaces, and a high degree of dimensional accuracy. The ceramic mold surface has refractory properties that enable it to withstand high metal pouring temperatures (such as those necessary for steel and heat-resistant alloys), give it excellent thermal stability, and do not permit metal penetration or burn-in. The Unicast process provides preformulated ingredients for the slurry, thus simplifying the process, but some foundries employ variations of the older Shaw process.

Applications Ceramic molding has two principal applications. The first is the production of precision castings that require patterns too large and unwieldy for molding with expendable wax or plastic patterns (investment casting). The second principal application is the production of castings in limited quantities, for which permanent wood patterns may be more economical and require less lead time to make than the metal pattern dies required for molding wax or plastic patterns. Ceramic molding is intended for the production of castings of high quality, not only in terms of their dimensional accuracy and surface finish but also in terms of soundness and freedom from nonmetallic inclusions. In general, the capabilities of ceramic shell investment molding and ceramic molding are similar, and the selection of one process over the other is largely dependent on the size of the casting, the quantities required, and the molding costs involved. In some applications, depending on casting shape, permanent patterns used in

ceramic molding may provide greater dimensional accuracy than wax patterns, primarily because wax expands during melt-out. Permanent patterns are also less susceptible to damage and distortion in handling than wax or plastic patterns. The principal disadvantage of ceramic molding in its early stages of development was the high cost of the molding materials. These materials were used in large quantities for making the slurry for all-ceramic molds and could not be reclaimed. This disadvantage was greatly diminished by the development of less expensive, composite-ceramic molds, in which chamotte (an aluminous fireclay) serves as a backup for a ceramic facing, thus greatly decreasing the amount of binder and zircon or mullite required. Furthermore, as much as 80 to 90% of the backup molding material can be reclaimed. Suitable Work Metals. Both ferrous and nonferrous alloys are cast in ceramic molds, but ferrous applications are more numerous and account for the major tonnage of castings produced. Alloys of aluminum, copper (especially beryllium-copper), nickel, and titanium are nonferrous alloys that are suitable for ceramic mold casting; suitable ferrous alloys include ductile iron, carbon and low-alloy steels, stainless steels, and tool steels. Types of Parts Cast. The range of products cast in ceramic molds is closely related to the alloy used. For example, among the products cast from tool steels (especially H12, H13, and A2) are forging dies and punches; trimming dies; die inserts; dies for hot upsetting; extrusion bridges, spiders, and cups; inserts for diecasting dies; and patterns for the shell molding process. Products cast from stainless steel include food-machinery components; valves for the chemical, pharmaceutical, and petroleum industries; glass molds; aircraft structural components; and hardware for atomic reactors and aerospace vehicles. Typical copper alloy castings include food-machinery parts, marine hardware, and decorative trim items used in architectural applications. A decorative duck head illustrates the smoothness and detail that is achieved with ceramic molds (Fig. 9). Complex castings with thin sections (1.6 mm, or 0.063 in.) can be cast in ceramic molds, using special techniques. Casting surfaces have excellent smoothness—usually 3.2 mm (125 min.) or better. Dimensional accuracy depends primarily on the accuracy of the pattern. Metal contraction can normally be predicted closely enough to provide castings within the following tolerances:


þ1.14 mm (þ0.045 in.) on dimensions over

  381 mm (15 in.)

For dimensions across the parting line, an additional tolerance of  þ0.25 to  þ0.51 mm (þ 0.010 to þ0.020 in.) must be provided.  Closer tolerances are readily obtainable by reworking the pattern after a test casting has been compared dimensionally with the pattern. Reproducibility is excellent; once the pattern has been established, subsequent castings will be extremely consistent throughout the life of the pattern.

Shaw Process The Shaw process relates to two distinctly different types of ceramic molds: a one-piece all-ceramic mold and a composite ceramic mold consisting of an inexpensive fireclay backup material with a relatively thin facing of ceramic slurry. Selection of mold type depends almost exclusively on the size of the casting and the cost of the mold material. Many small castings can be produced economically in one-piece all-ceramic molds, because the amount of expensive ceramic slurry needed for the mold is moderate, and the additional pattern and labor costs for composite molding cannot be justified. Patterns. Two sets of patterns are commonly required for the production of composite molds: a set of oversized preform patterns for molding the coarse backup material and a second set of patterns, representative of the dimensional accuracy desired in the casting, for molding the ceramic facing. Thus, for cope-and-drag molding, a total of four patterns may be required. The preform patterns are made 2.5 to 10.0 mm (0.1 to 0.4 in.) oversized to allow for the thickness of the ceramic facing. In some applications, the preform pattern can be eliminated by using the final pattern, backed with a sheet of plastic, cardboard, or felt of suitable thickness, to mold the preform impression in the coarse fireclay backup. Solid ceramic molds are made without backup material, and only the final patterns are required for molding. The pattern materials used for ceramic molding are conventional: plaster, wood, epoxy,


þ0.08 mm (þ0.003 in.) on dimensions up to

  25 mm (1 in.)


þ0.13 mm (þ0.005 in.) on dimensions over

  25 to 75 mm (1 to 3 in.)


þ0.38 mm (þ0.015 in.) on dimensions over

  75 to 203 mm (3 to 8 in.)


þ0.76 mm (þ0.030 in.) on dimensions over

  203 to 381 mm (8 to 15 in.)

Fig. 9

Details and smoothness are evident in this bulk Unicast ceramic mold of a duck head. Courtesy of Unicast Development Company

Slurry Molding / 623 aluminum, brass, tool steel, and cast iron. Wood, epoxy, and aluminum are most widely used. Prior to use, wood patterns are coated with a sealant that will not dissolve in alcohol. Ensuring a high degree of accuracy in the alignment of cope and drag is usually done by mechanically meshing certain surfaces. It is common practice in ceramic molding to incorporate parting blocks and locators in the copeand-drag halves of patterns. These are molded into both the backup and the ceramic facing, together with the casting contours. Preparing the Backup. The backup refractory commonly consists of a coarse-grained chamotte (an aluminous fireclay that has been calcined at a high temperature) and a sodium silicate binder. A typical screen analysis for the chamotte is as follows: Mesh

Cumulative percentage

6 8 12 16 30 70

Trace 5–9 27–41 60–68 94 99

The refractory and binder are mixed in a muller or an auger to the desired consistency and are rammed or vibrated over a pattern assembly. The pattern assembly consists of a preform pattern (or a final pattern covered with a suitable spacer material) mounted on a molding board and carefully located in a conventional flask, together with the required gating system. Before use, the preform pattern and other elements of the pattern assembly are coated or sprayed with a wax-silicone mixture to facilitate parting. Carbon dioxide gas is diffused into the mold material for hardening. One method of doing this consists of piercing the mold material in several locations with small-diameter rods, thus creating channels into which the gas can be introduced. More uniform gassing can be accomplished in a vacuum chamber. After the gassing and hardening operation, the preform pattern is removed from the mold. Preparing the Ceramic Facing. The principal ingredient of the ceramic facing is usually finely pulverized zircon, calcined mullite, or a mixture of both, although fused silica, magnesium oxide, and other refractory flours have also been used. Typical screen analyses for two calcined mullite flours are: Cumulative % on mesh size of: Mullite flour

A B

100

200

325

3 2

35–65(a) 17–32

... 30–45(a)

(a) Remainder passes through this mesh size.

Typical slurry contains a mixture of 75% zircon and 25% calcined mullite combined with a hydrolyzed ethyl silicate binder in proportions of approximately 0.91 kg (2 lb) of refractory blend to 100 mL (3.4 oz) of binder. To this

mixture is added a small amount of gelling agent, which causes the slurry to set in a predetermined period of time, usually 3 to 4 min. Immediately after this addition, the slurry is ready for pouring. Pouring the slurry over the pattern is always a gravity operation. However, when pattern detail is critical, the pattern and slurry are placed in a vacuum chamber, where the ungelled slurry is vacuumed to remove entrapped air bubbles. Immediately prior to use, the final pattern is coated with a wax-silicone mixture to facilitate parting, and the backup mold is placed over it. The slurry is poured through pouring gates until the level rises to invest the final pattern. In a few minutes, the chemical gelling action is complete; leaving the green ceramic with a consistency like that of vulcanized rubber, and the pattern is stripped from the mold. The rubberlike consistency is critical because it facilitates the separation of molds from patterns that have no draft, straight sides, extremely fine detail, protruding pins, or holes. Stripping is sometimes done mechanically to ensure that withdrawal of the pattern will not damage the mold. Removal of the pattern does not cause the mold to expand or contract. After stripping, the mold is ready for burn-off. All-Ceramic Mold Casting. An all-ceramic mold is made entirely from slurry of the type used for facing a composite ceramic mold. The procedure for pouring an all-ceramic mold is shown in Fig. 10. The pattern, affixed to a flat plate, is surrounded by a flask, into which the ceramic slurry is poured. After the chemical gelling action is complete, the green mold is stripped from the pattern, and the flask, which is built with a small amount of draft, is removed from the mold. The mold is then ready for burn-off. Burn-Off (Mold Stabilization). After gelling, the ceramic mold or mold facing is ignited with a torch and burns until most volatiles are consumed. During burn-off, the ceramic develops a microcrazed pattern, which is a three-dimensional network of microscopic cracks induced by the rapid evaporation of the alcohol in the slurry and by solid-phase reactions. Microcrazing is characterized by jagged ceramic particles separated by minute fissures or air gaps. These fissures are small enough to prevent molten metal from penetrating the mold surface, but they are large enough to permit the venting of air and other gases and to accommodate the expansion of the ceramic particles when they are in contact with the molten metal. Thus, microcrazing is advantageous in promoting dimensional stability without detracting from surface finish (see the section “Mold Stabilization” in this article). Baking is the final stage of processing. All remaining volatiles are removed, and the colloidal silica left by the hydrolyzed ethyl silicate binder forms a high-temperature bond of SiO2 that is stable nearly to its melting point (1710  C, or 3110  F), thus providing enough mold strength to resist washout, or erosion, by molten

metal. To perform these functions satisfactorily, baking must be done at not less than 480  C (900  F). Baking composite molds at more than 650  C (1200  F) may cause a differential expansion between the facing and backup layers due to the contraction of the soda-rich silica bond in the backing material. This can produce distortion in the mold cavity. Nevertheless, some foundries bake composite molds at temperatures as high as 815 to 980  C (1500 to 1800  F) without harmful effects. Two methods of baking are currently in use. In one method, termed skin heating, the mold is directly heated from above with electric or gas-torch heaters. This method has the advantage of fully curing the ceramic facing without overheating the backup material. It is a rapid technique, and it permits molds to be cured in conventional steel flasks. Heating the molds in an electric or gas-fired furnace is the other method of baking. Furnace baking has the advantage of removing all moisture and volatiles from the backup material as well as from the facing, thus minimizing the risk of bleeding moisture from the backup to the facing after mold assembly. Such bleedback can produce surface defects in the casting. The normal baking period for molds is 4 to 6 h at the prescribed baking temperature. After baking, molds are allowed to cool to an appropriate temperature for pouring. Pouring. Before pouring, molds are usually checked for cleanliness and temperature. Mold temperature at the start of pouring ranges from 40 to 540  C (100 to 1000  F), depending on casting shape and the alloy being poured. Cope-and-drag halves are assembled, and a layer of cement is usually troweled on to seal the mold around the parting line. For safety, fastening a steel band around the outside of the mold reinforces heavier molds. This assists in handling of the mold and prevents seepage of hot metal in the event that the mold wall cracks during pouring. Finally, ceramic pouring tubes and exothermic or insulating riser sleeves are set up in the cope. The mold is clamped firmly, and metal is poured either directly from the melting furnace or from a ladle. Pouring must proceed as rapidly as possible to avoid cold shuts and gas voids. After the metal has been poured, additional exothermic material is sprinkled on the risers to keep them fluid and to prevent shrinkage tears. Composite Ceramic Mold. This type of mold is a sodium silicate/carbon dioxide-cured and fired chamotte backing that significantly decreases the cost of refractory materials. A preform pattern and a finish pattern are needed to make composite molds (Fig. 11). Cope-anddrag preforms are made approximately 2.5 mm (0.10 in.) oversized on all mold cavity surfaces to allow for the thickness of the ceramic facing material. The cope-and-drag flasks must be accurately aligned, and at each stage of mold production, alignment holes are drilled and reamed to a jig.

624 / Expendable Mold Casting Processes with Permanent Patterns

Refractory Consists of a variety of specially blended groups of refractory powders.

Binder The liquid medium is usually based on ethyl silicate and is specifically produced to proprietary formulations.

Mixing A small percentage of gelling agent is added to the binder and mixed with the refractory powder to produce a creamy slurry.

Pattern The slurry is poured over a pattern made of wood, metal, plaster, plastic, and so on. It is then allowed to gel in about 2 to 3 min.

Step 1

Step 2

Step 3

Step 4

Stripping The gelled refractory mass is stripped from the pattern by hand or with a mechanical stripping mechanism.

Burn-off The mold is ignited. It burns until all volatiles are consumed; this sets up the microcrazed structure.

Step 5

Step 6

Fig. 10

Baking The Shaw mold, now immune to thermal shock, is placed in a high-temperature oven or skin heated with a torch until all traces of moisture are driven off. Step 7

Pouring Cope and drag mold pieces are assembled, along with any necessary cores, and the casting is poured. Step 8

Sequence of operations used in the Shaw all-ceramic mold process

As a further safeguard, a sand strip is cast into the insides of the flasks to keep the hardened preforms from shifting during handling. Preform patterns are then mounted on a pin lift molding table. The flasks are aligned by dowel pins and the alignment holes on the table before they are fastened to the patterns by quick-acting clamps. The sodium silicate refractory mixture contains a percentage of invert sugar, which aids shakeout of the mold after pouring. The refractory mixture is rammed around a preformed pattern (step 1, Fig. 11), and the rammed half-mold is gassed for 10 to 20 s with a commercial-grade carbon dioxide (step 2). While the first preform is being gassed, the operator rams the second flask with the composite mixture. By this time, the first preform is hardened. The operator repeats the gassing procedure for the second preform. The gassed preform mold halves are then stripped from the preform patterns, and each hardened preform mold is aligned with the final pattern bolted to the molding table (steps 3 and 4.) The flasks are fastened to the patterns by the quick-acting clamps.

The Shaw ceramic slurry needed for the facing of the composite mold is virtually the same as that used for the solid body mold—a specially blended formula of refractory aggregates, ethyl silicate-based binder, and a gelling agent. The thick, creamy slurry is then poured into the gap between the finished pattern and the hardened preform material (step 5). The slurry not only fills all of the details of the mold cavity but also penetrates the aggregate voids of the surface of the porous backing, creating a strong mechanical bond. Gelling of the slurry requires 2 to 3 min, and the fine ceramic facing of the composite mold is formed. The composite cope and drag are unclamped and lifted from the patterns by a foot-operated stripping plate (step 6). The molds are then placed on a conveyor for movement through a torching zone, where the volatiles are ignited and consumed (step 7). The microcrazed structure of the Shaw mold occurs during this burn-off period. The copeand-drag molds are moved directly into a high-pressure torching zone for 5 to 10 min; in this zone, the molds are brought to

incandescence. The copes and drags are assembled using the alignment pins to ensure accuracy. The molds can be assembled either hot or cold. These molds are now ready for the casting of the metal.

Unicast Process The Unicast process differs from the Shaw process principally in the method of mold stabilization employed, the wider range of ceramics used, and additional binder options. Mold stabilization refers to the treatment given the fine ceramic facing of a composite mold, or the total mass of an all-ceramic mold, shortly after it has set and while it is still green. Advantages. The Unicast process provides accurate dimensional control, and warpage or distortion is almost entirely eliminated. As a result of the stabilized molecular growth, the mold develops a cellular or spongelike structure caused by the interstitial separation of particles. This structure provides for gas venting and

Slurry Molding / 625

Refractory preform

Flask

Finish pattern

Preform pattern

Step 1

Step 4 Ceramic slurry

Fig. 12

Impossible-to-machine grain and texture are reproduced with ceramic cast molds. Courtesy of Unicast Development Company


The process is feasible for casting sizes Refractory bonded with sodium silicate gassed with carbon dioxide Step 2

Step 5

Finish pattern

Shaw composite mold

Step 3

Step 6

Ceramic facing is microcrazed and dried by torching. Step 7

Fig. 11

Diagram of Shaw composite mold process

particle expansion during casting. The advantages of this structure are metal soundness; ease of casting long, thin sections; elimination of veining; and reductions in traditional mold venting. Molds can be poured hot or at room temperature without any change in thermal characteristics. Directional solidification of the cast alloy is fully controlled, resulting in high physical properties and outstanding metal soundness. The major advantages of solid ceramic mold casting include the following:
Only a short lead time is necessary due to

simple tools and a fast mold build cycle.


A modest installation cost can be maintained

using simple equipment.


The manufacturing cost is considerably





lower than that of machining or other alternate methods. Patterns are usually inexpensive and quickly prepared. The die life of cast tooling is longer than that of dies made from wrought materials. Precision-cast tooling can frequently be used as-cast without further machining (Fig. 9). The process is suitable for impossible-tomachine grain and texture details for large injection mold bodies (Fig. 12).

ranging from small to very large. Composite Mold. Another difference between the Shaw and Unicast processes relates to the sequence followed in preparing composite molds. In the Shaw process, preparation of the coarse mold backup precedes pouring of the fine ceramic facing. In the Unicast process, this sequence is reversed (see the section “Typical Process Sequence” in this article). The castings produced by the two processes are comparable in quality and surface finish, and several foundries are licensed to use both processes. Although the Unicast process commonly uses a solid ceramic mold for smaller and short-run castings, it generally employs a compositeceramic mold for large parts and for those produced in high volume. In this case, the Unicast ceramic is used as a 6.4 to 13 mm (¼ to ½ in.) facing backed by a specially blended, coarsesand aggregate for support. This technique sharply reduces mold cost by conserving the ceramic material, and it permits the mechanized production of molds. Patterns and Flasks. The Unicast process employs conventional pattern equipment throughout. The patterns must be of good quality because the slightest flaw or defect in the pattern will be reproduced on the final casting. The reproduction of detail is such that even finger marks on a pattern coating are likely to be transferred to the casting. Most of the common pattern materials are used, including the denser hardwoods, plaster, aluminum, plastics, and hard rubber. Alloys Cast. Alloys normally poured in Unicast molds are about evenly divided between ferrous and nonferrous. Of the ferrous alloys, approximately 60% are irons, and the rest are principally stainless steels and tool steels. Of the nonferrous alloys, approximately 30% are alloys of cobalt, titanium, or uranium, and approximately 20 to 30% are alloys of aluminum, magnesium, or zinc. The rest are principally copper alloys, with beryllium-copper predominating.

626 / Expendable Mold Casting Processes with Permanent Patterns Slurries. Some of the fine ceramic slurries used in the Unicast process are comparable to those used in the Shaw process and consist of mixtures of a finely pulverized ceramic, such as zircon. Additional blends of alumino-silicates and fused silicas are also employed for larger molds and precision cores. The majority of the binders are based on an organic gelling binder, such as hydrolyzed ethyl silicate. Accordingly, the chemical reactions that take place during gelation are the same as those in the Shaw process. However, the method of hydrolysis and additional binder alternatives available in the Unicast process allow for stronger, thinner core sectional molds to be made than are typically available to the Shaw process. Backup slurries, used in composite mold preparation, are coarse and less expensive ceramic grains mixed with a refractory binder that will harden in the presence of carbon dioxide. Mold stabilization refers to the treatment given the fine ceramic slurry after it has gelled but while it is still green. The methods used for mold stabilization in the Shaw and Unicast processes are distinctly different, although the chemical reactions involving the binder are the same. When the binder becomes a hard, rubbery gel, it consists of a liquid and a solid phase. The liquid phase is largely free of alcohol, and the solid phase is a form of polysilicic acid. By itself, polysilicic acid is unstable and rapidly forms new chemical chains that are of larger mass than the original molecule. Molecular growth in the green state causes the

suspended ceramic particles to move or separate, thus producing a pattern of microscopic cracks. Unless molecular growth is arrested in a short period of time, the cracks can become excessive and form macrocracks. In the Shaw process, rapid evaporation of the alcohol resulting from hydrolysis or gelling reaction is accomplished immediately after (or very soon after) gelling by ignition or firing of the mold. This operation, referred to as burnoff, is usually performed by igniting the mold facing with a torch, as shown in Fig. 10. Burn-off follows removal of the pattern from the mold. It arrests excessive molecular growth, which may be catalyzed by moisture in the air or by other contaminants. This reaction must be avoided to ensure optimal binding properties and to avoid excessive cracking of the mold. In the Unicast process, the gelled mold is immersed in a liquid hardening bath or vapor atmosphere that permits aging and internal stabilization of the mold. Various hardening media can be used, depending on the composition of the mold binder and the final curing step. Thus, commercial ethyl alcohol can be used with an ethyl silicate binder, because the binder evolves ethyl alcohol as a result of the hydrolysis action. Alternatively, the bath can be a liquid such as acetone, kerosene, benzene, or other hydrocarbon with which the evolved alcohol is readily miscible. The hardening bath is more reactive when hot; the heat provides good results even when the bath liquid is not chemically identical to the evolved alcohol. Even boiling water can

be used; the temperature of the water drives the volatiles out of the mold to be dissolved in the water. A typical aging period for the mold in the bath is 10 to 15 min. The aging reaction is enhanced when the pH of the hardening bath or atmosphere is maintained at a level somewhat higher than that of the mold and, preferably, on the alkaline side of neutral pH. Decreasing acidity serves to accelerate setting and gelation. Organic amines or other generally alkaline ammonium compounds can be added to the bath for this purpose. Aging is enhanced when the gelled mold material is entirely immersed in a liquid or vapor that is substantially identical to the evolved alcohol, as compared to burn-off or to isolation of the mold surface from air by application of a wax or other suitable impermeable coating. Final curing, which follows the hardening bath, is usually done in an oven maintained at approximately 980  C (1800  F), but direct flame heating can be used. The cured molds are then transferred to the melting station for pouring. Typical Process Sequence. In a typical molding and casting sequence (Fig. 13), a pattern, mounted on a base plate, is enclosed within a flask. A thin (4.8 to 9.5 mm, or 3/16 to 3/8 in.) coating of fine-grained ceramic slurry is applied to exposed pattern surfaces by spraying or similar means. The coating becomes tacky almost upon contact and is ready to receive backing material. Inexpensive coarse backup slurry is poured rapidly over the facing coat until the flask is filled. The slurry is chemically controlled to set to a semirigid state within

Flask

Pattern

Baseplate Pattern positioning The pattern is mounted on the baseplate and enclosed within the flask.

Step 1

Pattern stripping The pattern is stripped from the mold, and the clamping vacuum is released. Step 5

Fig. 13

Coating A thin coating of ceramic finefacing slurry is applied to the exposed surfaces. The coating becomes tacky almost immediately and is ready to receive the backing material. Step 2

Mold hardening The stripped mold is transferred to the location at which hardening fluid is applied by spray or immersion. Step 6

Steps involved in producing a Unicast process casting

Backing A ceramic backing slurry is charged onto the facing coating until the molding box is filled.

Step 3

Mold clamping The flask is removed, the vacuum clamping plate is located in position, and the entire assembly is inverted on the stripping machine. Step 4

Mold curing

Metal pouring

The hardened mold is heat cured by direct flame or in a furnace.

A cope half is similarly prepared and assembled to the mold, and the metal is then poured to form the casting. Step 8

Step 7

Slurry Molding / 627 2 to 3 min, at which time the upper surface of the mold is leveled to the flask edges with a striking tool. The flask is removed, and a vacuum clamping plate is placed in position on top of the mold. The entire assembly is inverted on a stripping machine, the pattern is stripped from the mold, and the clamping vacuum is released. The stripped mold is transferred to the chemical hardening tank or chamber, where it is immersed in, or sprayed with, hardening fluid.

It is then cured by heating with direct flame impingement or in a furnace.

“Ceramic Molding,” Casting, Volume 15, ASM Handbook, ASM International, 1988, p 248–252.

ACKNOWLEDGMENT Portions of this article are adapted from the article “Plaster Molding,” revised by Charles D. Nelson in Casting, Volume 15, ASM Handbook, ASM International, 1988, p 242–247. Other portions are adapted from the article

REFERENCE 1. J.G. Kaufman and E.L. Rooy, Aluminum Alloy Castings: Properties, Processes, and Applications, ASM International, 2004.

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 628-633 DOI: 10.1361/asmhba0005251

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

No-Bond Sand Molding SAND MOLDING PROCESSES are classified according to the way in which the sand is held (bonded). Most sand casting employs green sand molds, which are made of sand, clay, and additives. Binders are also used to strengthen the cores, which are the most fragile part of a mold assembly. Bonded sand molds can be based on inorganic binders, such as clay or cement, or organic resinbased binders. Inorganic binders, which have long been used in the production of foundry molds and cores, include such processes as green sand molding, dry sand molding, skin-dried molds, loam molding, sodium silicate-carbon dioxide systems, and phosphate-bonded molds. Organically bonded systems include no-bake binders, heat-cured binders, and cold box binders. However, some molding processes do not use binders. Instead, the sand or mold aggregates are held together during pouring by the pattern itself (as in lost-foam casting) or by the use of an applied force (as in vacuum molding and magnetic molding described here). Unbonded (or no-bond) molding processes involve freeflowing mold particles and do not require binders, mulling equipment, or mold additives. This article describes the no-bond methods of vacuum molding and magnetic molding. Lostfoam processing, which uses expandable polystyrene patterns with unbonded sand molds, is discussed in a separate article.

molten metal. Evolving gases are drawn off through the base of the flask. The magnetic field is turned off after solidification and cooling, resulting in immediate shakeout. The free-flowing magnetic shot molding material is returned to its point of origin after cooling, dedusting, and metal splash removal. Advantages of the magnetic molding, like other no-bond methods, include the absence of a chemical binder, reductions in dust and noise levels, full mechanization or automation of the process, and the elimination of normally used molding activities (such as ramming and jolting). The increased heat conductivity of the iron or steel molding material also results in a finer grain structure in the cast metal. Another advantage is that a onepiece mold can be produced without a joint line. Early research on magnetic molding was carried out in the Soviet Union and Germany on cast irons, carbon and low-alloy steels, highchromium steels, and copper-base alloys. In the early 1980s (Ref 4), additional investigations were conducted on process variables such as

vibration time, magnetic field strength, magnetic properties of the shot, and the application time of field. A more recent study (Ref 3) does indicate finer microstructures in an aluminum-silicon casting due to the higher cooling rates with the steel mold material.

Vacuum (V-Process) Molding The vacuum molding process, or V-process, is a sand molding process that does not require the use of binders. Instead, the sand is held in place by the application of a vacuum on a flask covered with two sheets of thin plastic. The vacuum molding process was originally developed in Japan in the early 1970s (Ref 5, 6) for the production of castings with high surfacearea-to-volume ratios. In the 1970s and 1980s, Heinrich Wagner Sinto commissioned more than 30 vacuum molding plants. Since the invention of the V-process in 1971, the process is now licensed worldwide and has

Magnetic molding process

8)

Magnetic Molding 1)

Based on a concept similar to the lost-foam process using an expandable polystyrene (EPS) pattern, magnetic molding was developed by Wittmoser in the 1960s with some German foundries (Ref 1) and subsequently with a producer of polystyrene foam (Ref 2). The initial development of the magnetic molding process took place at the same time as the lost-foam process, but it has never achieved the same level of industrial development as the lost-foam method. The magnetic molding process (Fig. 1, Ref 3) involves a coated EPS pattern that is surrounded by a mold material of magnetic iron or steel shot (instead of sand as in lost foam). After the EPS pattern is positioned in the flask and encased with magnetic shot particles (between 0.1 and 1.0 mm, or 0.004 and 0.04 in., in diameter), the mold is compacted further by periodic vibrating and/or tilting. The mold is then made rigid by the application of a magnetic field prior to pouring the

2) 5)

3) Vibrations

6) 4) Electromagnet

1) Coating of expandable polystyrene (EPS) pattern by soaking 2) Drying of coated EPS pattern 3) Filling of the box with the steel shot and compaction by vibrating system 4) Casting: magnetic field on

Fig. 1

Magnetic molding process. Source: Ref 3

5) Cooling of component: magnetic field off 6) Demolding step 7) Final component 8) Reusing of short steel particles

7)

No-Bond Sand Molding / 629 been successfully used to cast all metals normally cast in conventional green sand mold mixes. It is considerd suitable for gray, ductile, malleable iron; various grades of steel; or aluminum and copper-base alloys. Application of this vacuumsealed manufacturing process for magnesium casting is a more recent development (Ref 7). V-process casting of magnesium was really challenging mainly due to low thermal heat content and high chemical reactivity of magnesium. Process application development for magnesium required development of in situ gating-runner systems, fast heat-removing sand mold systems, advanced process control sensors, proper plastics film, and sand core and coating materials. Aluminum alloys (series 200, 300, 500, and 700) have been cast by the V-process. Capabilities of the V-process for aluminum casting are compared with other casting methods in Tables 1 and 2. Although the V-process is primarily suited to the production of large castings in relatively small batches, any shape and size can be Table 1

Process

produced in a V-process mold. Cast configurations range from thin walls to thick sections for castings weighing from a few ounces to several tons. Sand thermal conductivity is lower, and metal fluidity is improved. Solidification time is slower. In addition, zero-draft designs are common, and the process can reduce cleanup and rough machining operations. Fine surface finish and excellent dimensional accuracy are two other key advantages of the V-process. The casting surfaces can be structurally superior to those of comparable processes. In terms of tooling, the V-process is costeffective for prototypes and production quantities. Tooling or patterns for the V-process are constructed of machined plastic or stereolithography apparatus (SLA) models. All of these methods of producing patterns result in unlimited life patterns capable of supporting both prototype and production requirements. Prototype quantities range from 5 to 500 units, with lead time from computer-aided design model to

V-Process Description The vacuum-sealed molding process allows molders to make complex molds using dry, unbonded, and freely flowing sand. Molds are sealed by using plastic films along the top and bottom sand surfaces of the cope and the drag molds and then vacuum applied to the sand medium of the cope and the drag (Fig. 2). The plastic film along the top of the drag mold and the bottom of the cope mold is softened by

Comparison of vacuum (V-process) molding with typical production capabilities of casting methods for aluminum Typical maximum size

Description

Minimum section thickness mm

in.


3.2

0.125

Treated sand is molded around a wood or metal pattern. Tons The mold halves are opened and the pattern removed. Metal is poured into the cavity. The mold is broken and the casting removed. To
10 kg Investment A metal mold makes wax replicas surrounded by an (20 lb) (lost wax) investment material. Wax is melted out and metal is poured into the cavity. The molds are broken and the casting removed. Permanent Molten metal is poured into a steel mold. The mold is To
45 kg mold opened and the casting is ejected. (100 lb)

1–5

635

None

Extremely fine sand is vacuum packed around pattern halves.

Up to
70 kg (150 lb)

Minimum draft required

None

V-process castings

castings in 2 to 4 weeks for large or small castings. The only difference between prototype and production tooling is that additional impressions may be added to the pattern to fully use the entire mold cavity for production runs. Patterns can be easily revised with quick turnaround using SLA- or computer numerical control-machined patterns. Other advantages of this process include reduced wear on patterns, reduced labor, and environmental benefits (Ref 8, 9).

Typical tooling costs(a)

Nominal lead times

$3000–14,000

Samples: 2–6 weeks Production: 2–6 weeks after approval

25

Prototype or production quantities of 5000– 10,000 All

$800–4000

Samples: 2–6 weeks Production: 2–6 weeks after approval

1.5

0.060

Under 1000

$3000–20,000

Samples: 8–10 weeks Production: 5–12 weeks after approval

2–5


4.8

0.1875

500+

$5000–400,000+

To
20 kg (50 lb)

½–2


2

0.070

Prototypes or up to 250 units

$3000–15,000

To
7 kg (15 lb)

1–3

0.75–1.5 0.030–0.060 2500+

Samples: 8–20 weeks Production: 10–12 weeks after approval Samples: 2–10 weeks Production: 4–8 weeks after approval Samples: 12–22 weeks Production: 8–14 weeks after approval

Sand castings

Plaster mold A plaster slurry is poured into the pattern halves. After setting, the mold is removed from the pattern, baked, assembled, and metal is poured into the cavity. The mold is broken and the casting removed. Die casting Steel dies, sometimes water cooled, are filled with molten aluminum. The metal solidifies, the die is opened, and the casting ejected.

Typical order quantities

$10,000–100,000

(a) U.S. dollars in the approximate time frame of 2003–2005. Source: Adapted from literature of Harmony Castings

Table 2 Comparison of aluminum casting tolerances with vacuum (V-process) casting and other casting methods As-cast tolerances with part dimensions of: Casting method

V-process Sand cast Investment cast Permanent mold Plaster mold Die cast

Under 75 mm (3 in.)

þ0.25 mm (0.010 in.) for the first 25 mm (1 in.),  then add þ0.002 mm/mm (in./in.)  þ0.8–150 mm (1/32–6 in.), then add  þ0.003 mm/mm (in./in.) 

75 mm (3 in.)

þ0.35 mm  (þ0.014 in.)  þ0.75 mm  (þ 0.030 in.) þ0.23 mm  (þ 0.009 in.)

þ0.075–6.35 mm (0.0003–¼ in.)  þ0.10–13 mm (0.004–½ in.)  þ0.125–75 mm (0.005–3 in.), then add þ0.003 mm/mm (in./in.)   þ0.38–25 mm (0.015–1 in.) then add  þ0.002 mm/mm (in./in.)  þ0.48 mm  (þ0.019 in.)  þ0.125–50 mm (0.005–2 in.), then add þ0.002 mm/mm (in./in.) þ0.38 mm    (þ 0.015 in.) þ0.002 mm/mm (in./in) þ0.15 mm   (þ 0.006 in.)

150 mm (6 in.)

300 mm (12 in.)

600 mm (24 in.)

Additional tolerance across parting line shift

As-cast surface finish (rms)

þ0.50 mm  (þ0.020 in.)  þ0.90 mm  (þ 0.035 in.) þ0.38 mm  (þ 0.015 in.)

þ0.8 mm  (þ0.032 in.)  þ1.5 mm  (þ 0.060 in.) þ0.685 mm  (þ 0.027 in.)

þ0.81 mm  (þ0.056 in.)  þ3.175 mm  (þ 0.125 in.) þ1.25 mm  (þ 0.05 in.)

þ0.5 mm 3–4 mm (125–150 min.)  (þ0.020 in.)  þ0.5–1.5 mm 5–13 mm (200–500 min.)  (þ 0.020–0.060 in.) N/A 1.6–3 mm (65–125 min.)

þ0.635 mm  (þ0.025 in.)  þ0.61 mm  (þ 0.024 in.) þ0.23 mm  (þ 0.009 in.)

þ0.94 mm  (þ0.037 in.)  þ1.05 mm  (þ 0.042 in.) þ0.38 mm  (þ 0.015 in.)

þ1.5 mm  (þ0.06 in.)  þ2.0 mm  (þ 0.08 in.) þ0.685 mm  (þ 0.027 in.)

þ0.25–0.75 mm 1.3–5 mm (50–200 min.)  (þ0.010–0.030 in.)  þ0.40 mm 1.6–3 mm (65–125 min.)  (þ 0.015 in.) þ0.40 mm 0.8–1.5 mm (30–60 min.)  (þ 0.015 in.)

For each process listed, cored areas or slides require increased tolerances. Flatness depends on size and geometry of the part. Source: Adapted from literature of Harmony Castings

630 / Expendable Mold Casting Processes with Permanent Patterns heating and then formed on an appropriate pattern to produce the hollow cavity for the final casting in the finished mold. The control factors of the V-process that may affect the quality of the castings are the molding sand, vibration frequency, vibrating time, degree of vacuum imposed, and pouring temperature. The overall sequence of operations in the V-process is shown in Fig. 3 and 4. The molding equipment consists of a vacuum system, film heater, vibrating table, pattern carrier, and flasks, perhaps with automation for higher production. There is no need for heavy, noisy jolt-squeeze equipment or ramming of slingers. The molding medium is clean and dry; therefore, shakeout equipment is usually nothing more than a grate to allow the sand to flow away from the casting and back into the sand system. Fume and odor control is unnecessary in most applications. The first step is pattern production (Fig. 4a). Patterns are identical to those used for green sand molding, except that vent holes must be judiciously placed around the patterns for plastic film adhesion. The pattern is then placed on a hollow carrier plate (Fig. 4b). A thin (0.05 to 0.10 mm, or 0.002 to 0.004 in.) plastic film is then heated and stretched over the pattern. Small vent holes in the pattern, the pattern board, and also the hollow construction of the pattern carrier allow a moderate vacuum (27 to 53 kPa, or 200 to 400 torr) to be applied to the pattern. This draws the plastic film tightly to the pattern (Fig. 4d) by suction through the small vent holes on the pattern, pattern board, and carrier. A mold coating is then sprayed on the plastic film to reduce penetration and burn-in. Drying of the mold coating does require caution to ensure the plastic film does not begin to adhere to the pattern. A specially constructed hollow-wall flask containing a vacuum connection is then placed around the pattern (Fig. 4e). The flask is filled with loose sand and vibrated at a low power level (Fig. 4f). A two-screen sand (70 and 270 mesh) is recommended for producing highdensity molds; however, sand with a conventional four-screen distribution also produces excellent castings. High-grade silica sand is typically used, although some improvement in surface finish of the castings has been observed when zircon sand is added to silica sand in Vprocess molds (Ref 10). After vibration, the excess sand is struck off flat, and a sheet of plastic film is placed over the top of the sand. A vacuum is then applied to the flask. The vacuum holds the sand tightly in place and produces an extremely hard (American Foundrymen’s Society mold hardness 90+), dense mold. When the vacuum is released on the pattern carrier plate, the pattern is easily withdrawn. After the mold is stripped from the pattern, the vacuum is maintained on the plastic-lined sand in the flask for the cope-and-drag portions of the complete mold. The cope-and-drag portions are then assembled (Fig. 4h) for pouring. Vacuum is maintained on

the cope-and-drag flasks for mold assembly and pouring. Slightly higher vacuum levels (40 to 53 kPa, or 300 to 400 torr) are required during pouring to prevent mold collapse. After the casting is poured, the vacuum may be removed from the flasks after 5 to 15 min. The flasks may then be stripped after the casting has cooled. The castings are then easily removed from the loose sand. This mold sand only requires cooling and screening for reuse in the molding process. Before pouring, the plastic film that covers portions of the mold cavity normally open to the atmosphere (the tops of risers, vents, sprues, etc.) is removed. This ensures a pressure differential is maintained between the mold and the mold cavity. If no air can be drawn into the mold cavity during pouring to compensate for the air drawn out by the vacuum applied to the mold, the mold will collapse. This also allows gases generated during pouring to exit to the atmosphere and not to the sand and vacuum system. Plastic Film Characteristics. The plastic films used for vacuum molding are thermoplastic compositions, while thermosetting resins are used as foundry sand binders. The types and properties of plastics that have been evaluated for their applicability as a film for vacuum sealing are listed in Table 3. The most commonly used material is an ethylene-vinylacetate co-polymer with vinylacetate contents ranging from 14 to 17%. The plastic film must have characteristics suitable for vacuum forming to ensure pattern contours will be closely replicated without film shape defects (folds and creases) or rupturing. The original sheet should also be free of pinholes, tears, air bubbles, or blemishes. The important characteristics for easy shaping to the pattern contours are a proper balance of both the elastic and plastic behavior of the plastic film. These properties are a function of film composition, temperature, thickness, stress, rate of loading, direction of loading, and the presence of any notch inducers. Detailed information on the film characteristics and test methods used to evaluate these properties is available in Ref 5.

Advantages The most obvious advantage of the V-process is that the use of vacuum to maintain the mold eliminates the requirement for a sand binder. Consequently, no sand mixing system is required, and the machinery for shakeout and sand reclamation are therefore less costly to install and operate. Additional benefits for V-process molding include reduced requirements for sand control and lower fume and dust generation. The vacuum molding process has many additional benefits. Pattern wear is eliminated in the V-process because the sand never makes contact with the pattern surface, which is covered with plastic film. The patterns are identical to those used in green sand molding, except that vent holes must be placed around the patterns to allow the plastic film to adhere. Due to the hardness and density of the V-process mold reducing mold wall

movement, riser efficiency is also improved. The V-process mold will also retain heat longer, slowing solidification, due to the presence of no moisture in the mold sand. Metal fluidity has also been observed as being higher in vacuum-process molds than that of green sand. Vacuum molding is often used in the floor molding of relatively large patterns (flask sizes 750  750 mm, or 30  30 in., and larger are typical) in jobbing foundries. Typical production rates of molds in these applications are five to ten molds per hour. The V-process is not normally used for high-production volume molding due to the number of functions that must be performed. However, V-process production rates of 60 to 100 molds per hour have been achieved. The layout for a highspeed automated vacuum molding line used in the production of medium-sized castings at a rate of 60 molds per hour is shown in Fig. 5. Dimensional Specifications. Vacuum molding also provides excellent dimensional repeatability and an excellent casting surface finish that reduces cleaning and finishing costs. General comparison with other processes for aluminum castings is given in Table 2. Typical component specifications for V-process casting are: Tolerances:  One side of parting line þ0.25 mm (þ0.010

  in.) up to 25 mm (1 in.)  Over 5 cm (2 in.), add þ0.002 cm/cm (in./in.)   Across parting line, add þ0.50 mm (þ0.020 in.)   Surface finish:  3 to 4 mm (125 to 150 min.) root mean square

on top side

 5 mm (200 min.) on bottom side

Minimum draft requirements:  0 to 1 on top side

Normal minimum section thickness:  Nonferrous: 3 mm (0.120 in.)  Ferrous: 2.5 to 3.2 mm (0.100 to 0.125 in.)

Fig. 2

Cope-and-drag assembly of vacuum-bonded sand mold. The sand in the elevated cope mold is under vacuum and is held in place by the atmospheric pressure against the sand. Courtesy of Auburn University

No-Bond Sand Molding / 631

Fig. 3

Sequence of operations in producing V-process molds

632 / Expendable Mold Casting Processes with Permanent Patterns

(a)

(b)

(c)

(d)

(e)

(f)

(g)

Fig. 4

(h)

Stages of mold production in the V-process. (a) Mold pattern. (b) Pattern placed in a hollow pattern carrier. (c) A thin sheet of plastic film is heated and vacuum fitted to the pattern. (d) Vacuum is applied to shrink wrap thin plastic film around the pattern. (e) The film-covered pattern is placed into a flask (shown here below the hollow carrier plate). (f ) The flask is filled with a fine, dry unbonded sand. A slight vibration compacts the sand, and a plastic film is placed over the sand-filled flask. A vacuum is applied to the compacted sand in the flask. (g) After the pattern is released from the flask (step 7 in Fig. 3), the sand-filled flask is kept under vacuum and placed on the pouring line, as shown here for the drag portion of the mold. (h) The cope portion (produced in similar fashion) is placed over the drag portion of the mold to form a cope-and-drag assembly with a plastic-lined cavity ready to be filled with molten metal. Courtesy of Harmony Castings

No-Bond Sand Molding / 633 Drag pattern drawing and rollover

Table 3 Plastic films used for the vacuum molding process Melting point Type of film

Low-density polyethylene High-density polyethylene Nylon Polypropylene Ionomer EVA(a) Polyvinylchloride

Density, g/cm3



C



Vacuum system

Flask closing Mold conveyor

F

0.920

88–90

190–194

0.960

94–97

201–207

1.13 0.90–0.91 0.93–0.94 0.940 1.450

215–222 160–170 72–75 58 56–90

419–432 320–338 162–167 136 133–194

Drag vibrating table Cope vibrating table Film forming for cope

(a) Ethylene-vinylacetate co-polymer

ACKNOWLEDGMENT The editors thank Tom Waring of ME Elecmetal for review and revision and the section on the V-process of vacuum molding.

Film forming for drag

Drag transfer

REFERENCES 1. A. Wittmoser and R. Hofmann, 35th International Foundry Congress, Foundry Technical Association, Kyoto, 1968 2. A. Wittsomer, K. Steinack, and R. Hofmann, Br. Foundryman, Vol 65, 1972, p 73–84 3. P.-M. Geffroy et al., Thermal and Mechanical Behavior of Al-Si Alloy Cast Using Magnetic Molding and Lost-Foam Processes, Metall. Mater. Trans. A, Vol 37, Feb 2006, p 441–447 4. E.L. Kotzin, Metalcasting and Molding Processes, American Foundrymen’s Society, 1981, p 143–147, 162 5. Y. Kubo, K. Nakata, and P.R. Gouwens, Molding Unbonded Sand with Vacuum—

Cope transfer

Flask separation

Dust hood Discharging of products

Fig. 5

High-speed automatic V-process production line. Production: 60 complete molds per hour. Flask size: 1300  1100  400 mm (50  40  15 in.). Amount of sand: 120 Mg (130 tons/h). Number of operators: 5

The V-Process, Trans. AFS, Vol 81, 1973, p 529–544 6. T. Miura, Casting Production by V-Process Equipment, Trans. AFS, Vol 84, 1976, p 233–236 7. S. Bakhtiyarov, and T. Overfelt, First V-Process Casting of Magnesium, Transactions of the American Foundry Society and the 108th Annual Metalcasting Congress, June 12–15, 2004 (Rosemont, IL), p 959–970

8. Revisiting the V-Process, Mod. Cast., Vol 85 (No. 5), May 1995, p 33–36 9. Renaissance of the V-Process, Foundry Trade J., Vol 177 (No. 3609), Nov. 2003, p 23 10. P. Kumar and R. C. Creese, Effect of Zircon Sand (Added to Silica Sand) on Surface Finish of Al-11% Si V-Process Castings, 103rd Annual Meeting of the American Foundrymen’s Society, Sept 9–10, 1998 (Rosemont, IL), p 95–98

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 637-639 DOI: 10.1361/asmhba0005253

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Introduction—Expendable Mold Processes with Expendable Patterns Brian V. Smith, General Motors

CASTING WITH EXPENDABLE MOLDS is a very versatile metal-forming process that provides tremendous freedom of design in terms of size, shape, and product quality. Adding expendable patterns to this equation increases the complexity and tolerance of the cast product. Depending on the size and application, castings manufactured with the expendable mold process and with expendable patterns increase the tolerance from 1.5 to 3.5 times that of the permanent pattern methods. The two major expendable pattern methods are lost foam and investment casting. A hybrid of these two methods is the Replicast casting process (a patented process exclusive to Casting Technology International, United Kingdom), which involves patternmaking with polystyrene (similar to lost foam) but with a ceramic shell mold (similar to investment casting). These three methods are briefly reviewed here, with more detail in separate articles in this Volume. One major difference between the permanent pattern casting methods and the expendable pattern methods is that the expendable pattern is typically always the positive shape of the part (Fig. 1). In contrast, permanent patterns are the negative or mirror image of the part to be cast. When using the expendable pattern method, the part is typically made twice: once in an

expendable form of the part (which is disposable) and then as the actual functional metal form of the part. Typically, an expendable pattern still needs to be made from some set of tooling, and although this may sound confusing, many foundries refer to this set of tooling as the pattern and/or the permanent pattern. Thus, the expendable (disposable) pattern casting process requires this extra step of making the expendable pattern. This extra set of tooling or patterns may not be needed if only one part or just a few parts are needed. If this is the case, then expendable patterns can be hand crafted or machined from some type of expendable pattern material where no tooling is required. In contrast, higher production volumes of parts require repetitive fabrication of expendable patterns. In this case, tooling is needed to make the patterns (Fig. 2). Table 1 summarizes the differences in the steps of casting a part between the permanent pattern versus the expendable pattern methods. Investment casting uses an expendable/disposable pattern, typically made of wax and then coated with some type of slurry molding medium. The wax positive shape of the part is usually made from tooling mounted in some type of wax injection machine (Fig. 3). Again, for

very few parts, the wax could be hand crafted or shaped from a billet of the material. The positive shape of the part or, more typically, multiple parts is combined with a gating system and made into a cluster (Fig. 4). This cluster is then typically coated with a ceramic slurry (Fig. 5) and dried, again and again, until a sufficiently thick shell is produced. This cluster with the wax disposable pattern is then fired in an oven until all the wax is melted out and the ceramic shell is rigid enough to support the introduction of the molten metal. The wax positive shape of the part can also be solid investment cast, but this method is not used very frequently in most engineering applications. The metal is then introduced into the empty cavity created by the wax shape that has now been melted out. This is in contrast to the lost foam casting method, where the foam pattern is not melted out before the molten metal is introduced but during the introduction of the molten metal, thus the term lost foam. The lost foam casting process also uses a disposable pattern typically made of expanded polystyrene (EPS). One previous name for this process was EPS casting. The EPS material is very similar to the material used for foam coffee cups, plates, and so on. The foam pattern

Fig. 2 Fig. 1

Cast manifold with positive pattern for lost foam casting. Courtesy of the Vulcan Group

Gluing sections for high-volume fabrication of lost-foam patterns. Courtesy of Vulcan Engineering

638 / Expendable Mold Casting Processes with Expendable Patterns Table 1 Differences in casting process steps between the permanent pattern and expendable pattern methods of the expendable molding process Expendable mold process Steps in casting a part

Permanent patterns

Design the part and Pick the casting process to be used for the part Design of tooling to make the part Make the tooling

Make the patterns for the part Make the part the first time

Assemble the pattern sections, negative or positive shape of the part, with its gating system into an expendable mold Cast (create/make) the part

Expendable patterns

Simultaneous engineering with the product designer is a key to making a quality part. Once designer and caster have reached consensus on the casting process, the design can be finalized to meet functional requirements of both the final part and the chosen foundry process. Design tooling to make the Design tooling to make the negative shape of the part positive shape of the part Permanent pattern tooling to Expendable pattern tooling to make the negative shape of make the positive shape of the part the part Negative shape of the part Positive shape of the part Not applicable If necessary, assemble the positive-shaped patterns into the part to be cast. Although pattern sections may be either a negative or positive shape, they must be assembled into a final expendable mold assembly ready for casting. Introduce molten metal to the part in the selected casting process

Fig. 7

Lost foam pattern for cylinder head. (a) Four different sections glued together. (b) Foam cluster with its gating system and the casting after base cubing

Fig. 3

Wax molding machine for investment casting

Fig. 5

Ceramic-coated wax cluster in robotic gripper

Fig. 6

Horizontally parted die, lost foam molding machine that is vertically acting. Die area is fairly large, approximately 125 by 62 cm (50 by 25 in.).

Fig. 4

Wax cluster

embodies the positive shape of the part to be cast (Fig. 1). The foam pattern can be produced by hand crafting or machining the shape from a billet of the material. This method of making the expendable foam patterns is typically not the best way to produce a good lost foam casting, but it is acceptable for prototypes or one-of-a-kind applications. More typically, especially for high-volume parts, the foam

pattern is produced from tooling installed on a foam-blowing machine (Fig. 6). The lost foam foundryman has discovered numerous EPS additives and foam molding techniques to produce the best foam pattern that results in the best casting with the least number of defects. Lost foam is especially suitable for fairly complex castings with many internal passages. Cylinder blocks and cylinder heads for internal combustion engines are fairly good candidates for this process, where internal oil, water, and gas passages can be integrally cast as one piece. Typically, these complex internal passages are made possible by gluing together numerous pattern sections and then attaching the final gating system to this positive shape of the part to be cast (Fig. 7). After the positive shape of the part is assembled into a cluster, the cluster is coated with a thin ceramic slurry, dried, and placed into a flask (Fig. 8) for investing of a casting mold consisting of unbonded sand or other molding medium. Some fairly sophisticated sand investment and compaction methods for lost foam casting are used within the lost foam casting industry. The key sand compaction attribute is that the sand flows into all the cavities of the part being cast and that it becomes fully dense behind the thin ceramic coating on the foam pattern. After the sand is fully compacted around this foam expendable pattern, the part is ready for molten metal introduction, with the foam still in the mold assembly. Now, the molten metal

Introduction—Expendable Mold Processes with Expendable Patterns / 639

Fig. 9

Dumping and extraction of casting in lost-foam operation. Courtesy of Vulcan Engineering

must not only become the shape of the final part, it must also decompose the foam pattern, thus the term lost foam. This process has also been referred to as full mold casting, because the mold cavity is not empty but rather filled with foam. A typical defect in lost foam casting is that this foam is not 100% lost; if it does not escape from the mold entirely, the liquid or gaseous remains can mix with the metal and become defects within the final cast part. A key casting attribute of the ceramic slurry is that it acts as a barrier and/or valve between the foam pattern and the sand during casting. The thin ceramic shell allows the foam decomposition materials to escape through it and into the sand at the proper rate. If it is too

Fig. 8

Lost foam cluster being (a) dipped into ceramic slurry, (b) dried, and (c) inserted into sand-cast mold. Courtesy of the Vulcan Group

impermeable for the foam decomposition materials to pass through it, the molten metal could freeze off, and the part would fail to fill out. If the coating is too permeable, it would allow the foam decompositions to escape too quickly, allowing the unsupported sand to cave in and either mix with the molten metal stream or lose the shape of the part being cast. To be successful, the lost foam foundryman must determine the key coating characteristics for making a particular part with a particular molten metal. Many of the coating/slurry suppliers to the industry have learned these lessons and can be of invaluable service to the lost foam foundryman. A major attraction of lost foam casting is the next step: shakeout of the part from the expendable mold material (Fig. 9). In lost foam casting, the unbonded sand (or other molding material) just flows out of part; no violent shaking and/or heating of the mold is needed to break up the mold, as in green sand or resin-set sand molds and cores. Replicast casting is a hybrid method between lost foam and investment (lost wax) castings. The expendable pattern is made from some type of foam and/or plastic, but unlike lost foam, the pattern is removed from the mold cavity during firing of the ceramic that surrounds the pattern (see the article “Replicast Molding” in this Section). Replicast is therefore more like the investment casting process, where the disposable pattern material is removed from the mold shape before the introduction of the molten metal. Replicast was originally developed for steel casting, especially low-carbon steels, because the pattern removal before pouring eliminated carbon introduction into the metal (since EPS is comprised mainly of carbon). The process is also especially valuable for large, one-of-a-kind parts, where the caster can hand craft and/or machine the shape from a billet of foam material and then mold it with some type of expendable molding medium that will set up and become the empty cavity for the molten metal to fill after the plastic expendable pattern has been removed, either by heat or some chemical means.

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 640-645 DOI: 10.1361/asmhba0005254

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Lost Foam Casting Raymond Donahue and Kevin Anderson, Mercury Marine

THE LOST FOAM CASTING PROCESS originated in 1958 when H.F. Shroyer was granted a patent for the cavityless casting method using a polystyrene foam pattern embedded in traditional green sand, which was not removed before pouring of the metal (patent 2,830,343). The polystyrene foam pattern left in the sand mold is decomposed by the molten metal. The metal replaces the foam pattern, exactly duplicating all of the features of the pattern. Early use of the process was limited to one-of-a-kind rough castings because the foam material used was coarse and hand fabricated and because the packed green sand mold would not allow the gases from the decomposing foam pattern to escape rapidly from the mold (the trapped gases usually resulted in porous castings). Lost foam casting is also referred to in the literature as evaporative pattern casting, evaporative foam casting, the lost pattern process, the cavityless expanded polystyrene casting process, expanded polystyrene molding, or the full mold process. The most significant breakthrough in the lost foam process came in 1964 with the issuance of a patent to T.R. Smith for the use of loose, unbonded sand as a casting medium (patent 3,157,924). With this development, it became clear to the foundry industry that the lost foam casting process was an emerging technology deserving of attention. In the 1960s and 1970s, most of the lost foam casting activity took place in automotive company research facilities. Very few production castings were produced during this time. However, use of the lost foam casting process has been increasing rapidly since the expiration of the Smith patent in 1981. Currently, many casting facilities are dedicated strictly to the lost foam process.

Process Technique Because the casting is an exact representation of the polystyrene foam pattern, the first critical step in the lost foam process is to produce a high-quality foam pattern. As described subsequently, surface quality, fusion, dimensional stability, and foam pattern density are

key control areas, but foam permeability has emerged as a most significant variable. Even more important has been the effect of bead packing on casting quality. The Lost Foam Consortium of the University of Alabama— Birmingham with funding support from the U.S. Department of Energy, has shown that by vibrating a mold venturi filled with foam beads, there is significant settling of the beads, indicating a less than “closest packing” of the original filling process. Further, it has been postulated that this less than “closest packing” of the original filling process leads to foam bead movement during the steaming process, which cannot be good. The foam bead movement during the steaming process, due to the poor packing efficiency, creates a foam pattern that is believed to be highly susceptible to producing a finger-shaped metal front during the ablation process of displacing the foam during filling, which can trap liquid styrene and produce a fold. A better, more closely packed bead configuration is believed to occur with canister filling. The best bead filling method is speculated to be with molds that can be vibrated during the bead filling process. The foam pattern is prepared for casting by attaching it to a gating system (sometimes molded as part of the pattern) of material of the same type and density. The pattern system with gating is then coated inside and out with a permeable refractory coating. Once the coating is dry, the pattern system is ready for investment into a one-piece sand flask. Investment of the pattern is achieved by positioning the pattern system in the flask, which has a 25 to 75 mm (1 to 3 in.) bed of sand in the bottom of the flask. Once the pattern system is properly positioned, loose, unbonded sand is introduced in and around it (Fig. 1). The flask is vibrated to allow the loose sand to flow and compact in and around all areas of the pattern. A pouring basin or sprue cup is usually positioned around the exposed foam downsprue. When compaction (sand densification) is complete, the flask is moved into the pouring area, and molten metal is poured into the pouring basin (sprue cup). The metal vaporizes the foam pattern, precisely duplicating all the intricacies of the

pattern (Fig. 2). The casting is then allowed to cool for approximately the same amount of time as with green sand. The flask is usually tipped over, allowing the loose sand to fall away from the casting (Fig. 3a). This sand is collected for reuse and the casting is ready for degating and cleaning (Fig. 3b). Alternatively at shakeout, the flask is again subjected to vibration (but this time to undo the compaction), and a robot reaches in and extracts the solidified metal cluster and rotates and positions the cluster so that the sand in internal passageways falls out (or drains out). Then, the robot submerges the hot metal cluster (free of internal sand) into hot water that causes the original coating on the foam to pop off the hot metal casting. It should be emphasized that this robotic quenching step is effectively the low-cost cleaning room, which eliminates manual handling of the casting and the removal of the attached coating. After the removal of the coating, the robot can either position the cluster for automatic removal of the gating system and/ or any individual castings on the sprue, or the robot can put the cluster on a moving conveyer where the individual castings are manually removed from the cluster downstream. Alternatively, when compaction is complete, the flash can be transferred to open pressure vessels where molten metal is poured into the sprue cup. During this filling event, it is possible to pull a slight vacuum from the plenum in the bottom of the flashes, if so designed, to more effectively remove the liquid styrene generated during filling that the coating may not be able to completely handle. Controlling the coating permeability actively as opposed to passively from the sand side of the coating interface with the application of a slight vacuum offers several advantages in controlling the rapid generation of liquid styrene during fast filling that results with high-silicon alloys. For example, there are strong arguments that both alloy 319, with only a 6% Si content, and 356, with only a 7% Si content, have only marginal fluid life in complex foam castings because the ablation of the foam during filling sucks a large quantity of heat energy from the molten-metal front and may cause misruns. Thus, the ideal silicon content for an

Lost Foam Casting / 641

Fig. 1

Lost foam pattern system. (a) Flask that contains a 25 to 75 mm (1 to 3 in.) sand base. (b) Positioning the pattern. (c) Flask being filled with sand, which is subsequently vibratory compacted. (d) Final compact ready

Fig. 2

Pouring of a lost foam casting

Fig. 3

aluminum-silicon alloy for lost foam may be 9% Si. However, such a higher-silicon alloy generates a quantity of liquid styrene that cannot be handled by a passive coating by itself, and thus, these higher-silicon alloys that can greatly expand the application of lost foam castings require vacuum filling of some kind to suck the liquid styrene through the coating and into the sand. After filling is complete (or nearly complete), the pressure vessel lid closes (generally within 5 s), and pressure can be applied to 1.0 for 1.5 MPa (10 or 15 atm) in approximately another 60 to 90 s. The application of pressure is most beneficial to feeding and to the complete elimination of hydrogen porosity in high-quality castings. By imagining a sphere of liquid freezing (and recognizing the specific spherical shape is not needed in this general argument), it is easy to illustrate the benefits of applying 1.0 MPa (10

Processing of completed (cooled) casting. (a) Flask is tipped, and the sand is recycled. (b) Casting is ready for degating and cleaning

atm) of pressure during solidification. First, a spherical shell forms at the outer diameter of the sphere. As the temperature drops, the liquid goes into a state of hydrostatic tension, because the contraction of the liquid is constrained by the more ridge solid shell. One of three events can occur as the hydrostatic tension increases in this nonfeeding solidification event. Either the liquid experiences a tensile-tensile liquid failure with an audible sound at failure, or the sucking action of the hydrostatic tension punctures the surface (i.e., sucks the liquid from the liquid + solid shell, producing a dry aluminum dendritic network at the surface), or the sucking action of the hydrostatic tension pulls in (or deforms) the shell (resulting in what some call solid feeding). In any case, the application of isotropic pressure during solidification in real castings has the effect of decreasing the hydrostatic tension in the liquid, and this decrease in the hydrostatic tension causes the

elimination or delay of the cited failure mechanisms due to poor feeding until later in the solidification event. Thus, lost foam with the application of pressure, which is easy to apply in the lost foam casting process and difficult in the other conventional casting processes, can completely eliminate hydrogen porosity and can decrease unfed shrinkage porosity by over an order of magnitude compared to conventional sand casting processes. Recognize that in conventional sand casting, molten aluminum shrinks approximately 6% on solidification and still leaves approximately 1% porosity in sand castings. Thus, in lost foam with pressure where the porosity is basically insignificant (i.e., less than 0.1%), if the casting design is properly fed, the shrinkage is not 6% but approximately 7%. Because the typical sand casting generally contains 1% porosity, either as hydrogen porosity and/or as unfed shrinkage porosity, to fully capitalize on the application

642 / Expendable Mold Casting Processes with Expendable Patterns of pressure during solidification, greater attention must be given to the casting design to make sure it can accept approximately 17% more feed metal than typical sand castings generally receive (17% is the difference between 7 and 6% shrinkage). Thus, it is strongly recommended that lost foam with pressure castings be modeled to ensure that isolated hot spots do not develop during solidification that cannot be fed. The Lost Foam Consortium at University of Alabama—Birmingham and Flow Science (the owners of FLOW-3D) have made quantum leaps in modeling real lost foam castings and can predict the location of porosity in the largest commercial engine blocks and heads. It is not that the other commercial modeling companies cannot model lost foam as well per se, but because of the relatively small size of the lost foam market, other commercial modeling companies indicate they cannot justify the investment in program development at this time. In any case, the modeling of lost foam with pressure does not require the prediction of when and where hydrogen porosity will occur but only the prediction of where isolated hot spots develop that may not be fed. The lost foam with pressure process is remarkably sensitive to certain aluminum-silicon alloy systems in decreasing porosity. For example, all of the cited porosity levels of less than or equal to 0.1% refer to the copper-free aluminum-silicon alloys such as A356. For the copper-containing aluminum-silicon alloys such as 319, the copper phases that precipitate out late in the solidification process clog the interdendritic passageways, and thus, porosity levels cannot be decreased as much as they can be decreased with such copper-free alloys as A356.

Processing Parameters The Foam Pattern. Expandable polystyrene (EPS) has been, and will probably continue to be, the preferred material for manufacturing lost foam patterns. There are other foam materials under development that show some promise, but to date, their use is quite limited. There are several grades of EPS, as indicated in Table 1. Grades T and X are preferred because they give the foam pattern molder the ability to produce the smoothest surfaces and the thinnest sections possible for lost foam patterns. Expandable polystyrene weighs approximately 640 kg/m3 (40 lb/ft3) in its raw state. For the EPS to be useful in the manufacture of lost foam patterns, its bulk density must be reduced to a level between 16 and 27 kg/m3 (1.0 and 1.7 lb/ft3), as indicated in Table 2. This is achieved through a process known as prexpansion. The raw EPS is introduced into a heated chamber and kept in constant motion for even distribution of heat throughout the batch of material. The plastic beads soften, and the gas within the beads expands; this

Table 1

Grades of expandable polystyrene Diameter at 24 kg/m3(1.5 lb/ft3)

Raw bead diameter Bead grades

A B C T X

Table 2

mm

in.

mm

in.

Typical use

0.83–2.0 0.58–1.2 0.33–0.71 0.25–0.51 0.20–0.33

0.033–0.078 0.023–0.047 0.013–0.028 0.010–0.020 0.008–0.013

2.5–5.9 1.7–3.5 1.0–2.1 0.74–1.5 0.61–1.0

0.097–0.231 0.068–0.138 0.040–0.082 0.029–0.058 0.024–0.040

Insulation Packaging Cups (coffee) Cups/lost foam Lost foam

Typical pattern density requirements for lost foam casting Metal pouring temperatures

Metal

Aluminum Brass/bronze Gray iron Steel



C

705–790 1040–1260 1370–1455 1595–1650

Expandable polystyrene pattern density

F

kg/m3

lb/ft3

1300–1450 1900–2300 2500–2650 2900–3000

24–27 20–21.6 20 17.6

1.5–1.7 1.25–1.35 1.25 1.10



increases the diameter of each individual bead and thus reduces the bulk density (a good analogy is popping corn). The volume ratio at a finished density of 16 kg/m3 (1.0 lb/ft3) is approximately 40 to 1. After a period of time in the preexpander, the material is discharged and weighed to check density. The density of the foam pattern for a given cast product is critical. For consistent casting results, EPS density must be controlled within  þ2% of target density. This is achieved by monitoring and adjusting time and temperature in the preexpander. Once the preexpanded EPS is discharged from the preexpander, it is placed in intermediate storage to cool and stabilize. After the stabilizing process is complete (usually 6 to 12 h, depending on the type of bead used), the preexpanded EPS is conveyed to a hopper attached to a pattern-molding press. This sequence is illustrated schematically in Fig. 4. Pattern molding for the lost foam process can be grouped into four major functions: filling, fusion, cooling (stabilization), and ejection. Filling. The preexpanded stabilized material at the desired density is fed from the hopper on the press to the mold (or pattern die). A vented fill gun is used to feed the material, as illustrated in Fig. 5. The filling of beads into any production cavity can be modeled today (2008). Arena Flow and See Foam are the two programs that are commercially available. Fusion. After the mold cavity is full, heat is added by passing steam through the material in the cavity; this reinitiates the expansion process and softens the material, as in the preexpansion process (Fig. 6a). The material expands again, filling the air voids between the individual EPS beads and fusing the beads together into a solid mass to form the desired foam pattern. Cooling. During the fusion portion of a molding cycle, the molded part exerts pressure against the cavity walls. If the part has not been cooled to reduce the internal pressure of the molded part, the part will continue to expand

after ejection. This condition is known as postexpansion. To eliminate postexpansion, it is necessary to cool the mold cavity, thus reducing the internal pressure of the molded part to a point at which it can be ejected and still maintain its dimensional integrity. Cooling is usually accomplished by spraying water on the back of the mold wall cavity (Fig. 6b). Ejection. Once the part has been cooled, the press can be opened and the part ejected (Fig. 6c). This can be done pneumatically or mechanically. Pattern Assembly. Lost foam patterns typically consist of multiple pieces that must be assembled to form a completed pattern and gating system. The most widely accepted method of assembling patterns is to glue them together with a hot melt adhesive. The adhesive used is specifically formulated for the lost foam process. The temperature at which it is applied to the foam patterns is much lower than the application temperatures of other commercial hot melt adhesives. Lost foam hot melts are generally applied between 120 and 130  C (245 and 270  F). To ensure consistent dimensional control and adhesive joint quality, an automatic or semiautomatic gluing procedure is preferred. Figure 7 illustrates a typical pattern assembly sequence. For prototypes and short run quantities, handoperated glue printing and assembly jigs can be used. Other methods of pattern assembly can be used in specific applications. These may include hand-applied rubber cement, liquid and contact adhesives, robotically applied air set adhesives, and foamed hot melt procedures. Whichever adhesive material is chosen, it must be suitable for use with lost foam patterns and must be compatible with the molten metal during the casting operation. The vaporization rate and ash content of the adhesive must be approximately the same as those of the EPS pattern. The adhesive must not adversely affect the dimensional integrity of the pattern. Care must be taken to control all aspects of pattern assembly to maintain

Lost Foam Casting / 643

Fig. 4

Sequence of operations for producing a foam pattern

Fig. 5

Fill gun used to feed preexpanded stabilized material to molding press. (a) Open cylinder with pattern material transported by air/vacuum. (b) Cylinder closed prior to fusion

Fig. 6

Schematic showing (a) fusion, (b) cooling, and (c) ejection of lost foam pattern material

consistency. Poor assembly techniques will result in increased casting scrap and dimensional control problems. Thorough inspection of the completed foam pattern system is essential before proceeding to the next step in the process. Coating Types. It is possible to produce a lost foam casting without using a refractory coating on the foam pattern, but for maximum process latitude, it is advantageous to coat the pattern system. This specialized coating serves two purposes. First, it provides a barrier between the smooth surface of the pattern and

the coarse surface of the sand. Second, it provides controlled permeability, which allows the gaseous products created by the vaporizing foam pattern to escape through the coating and into the sand away from the cast metal (Fig. 8). Recognize that with a coating on the foam, the molten metal never comes in contact with the sand. Further, under these conditions, the coating is effectively the mold, wherein the compacted sand is the backup sand holding the mold (i.e., the coating) in place. Lost foam coatings are available in various permeabilities. Lower-permeability coatings

are preferred in aluminum castings with high surface-area-to-volume ratios—for example, intake manifolds. Heavy-section aluminum castings and other nonferrous castings may require medium- or high-permeability coatings. Ferrous castings generally require higher permeability than nonferrous castings. Silica was the refractory of choice in the early days of the process, but today (2008), castings with high dimensional tolerances are delivered with various low-expansion sands. The problem with silica sand is that the material has a high coefficient of thermal expansion when compared to other sands and goes through a discontinuous phase transformation when the aluminum is still molten. Thus, the surface sand in contact with the coating goes through a phase transformation while the backup sand may not go through the phase transformation. The result is that poor dimensional tolerances are obtained with silica sand. Any other sand besides silica sand is better because other sands do not go through a discontinuous phase transformation, and other sands have a lower coefficient of thermal expansion when compared to silica sand. Synthetic mullite is a recommended popular sand for aluminum castings, while olivine sand can be used with steel castings or with cast iron castings. Although its coefficient of thermal expansion is low, zircon sand is not used because of its small particle size, reflected in the American Foundry Society (AFS) grain fineness of 50 or above, and low sand permeability. Synthetic mullite, olivine, and other low-expansion sands should have an approximate AFS grain fineness of about 40 or less. Zircon and olivine coatings are also available but are not widely used. The choice of coating carrier is limited because the carrier must be compatible with the foam pattern. Hydrocarbons and chlorinated solvents will attack the EPS. Water is the most widely used carrier, particularly in countries with emission standards, but alcohol is used in Mexico. Application of the Coating. Proper preparation and control of the coating is absolutely

644 / Expendable Mold Casting Processes with Expendable Patterns essential for the consistent production of highquality lost foam castings. The coating, when mixed properly, can be applied to the pattern system by dipping, brushing, spraying, or flow coating. Brushing, spraying, and flow coating are used on very large patterns. Dipping is the preferred method for small- and medium-sized patterns. Because of the extreme difference in density between the foam pattern system and the coating slurry, large patterns are sometimes difficult to immerse. Pattern orientation and coating fluidity must facilitate pattern coverage in blind areas. The entire pattern system must be covered inside and out with the exception of the exposed downsprue area, on which molten metal will be poured. Foam pattern cluster systems must be handled carefully during coating and drying. The weight of the wet coating may cause the patterns to distort or break. Medium- and high-production lost foam facilities find automated coating systems very helpful in controlling the consistency of the coating. The next step is to dry the coated pattern. This can be done at ambient conditions within 24 h, or the pattern can be force dried in a drying oven or heated room. Forced drying is done

Fig. 7

at oven temperatures of 50 to 65  C (120 to 150  F), along with numerous changes of air per hour and possibly dehumidification. This method of drying coated foam patterns usually takes 2 to 6 h. Microwave drying in the final stage can be used in very high production applications. Investing the Foam Pattern in Sand. As indicated in Fig 1, the coated pattern system is placed in a one-piece flask. Loose, dry, unbonded sand is introduced into the flask generally through a sand raining system. This system gently fills the flask with sand, minimizing the possibility of pattern distortion due to side movement of the sand against the pattern. During the filling of the flask, the flask is isolated from the flask conveying system and vibrated with a high-frequency compaction system either from the bottom or the side of the flask. Variable frequency and amplitude may be required to compact the sand fully in both the internal and external areas of the foam pattern system. Sand System. When sand is not reclaimed, the sand most commonly used for economic reasons is silica sand, subangular to round. However, a round synthetic mullite sand can fill a flask containing complex foam patterns 35% faster than silica sand. Different sizes are used

Pattern assembly sequence showing nine steps performed in the pattern assembly machine

for different applications. Generally, AFS grain fineness number 35-3 screen distribution is used in ferrous applications, while AFS 45-3 screen distribution is used in most nonferrous systems. Most sand systems, in addition to the compaction and sand fill stations, incorporate a pouring area, cooling area, dump-out area, sand return system, and sand cooler classifier system. To maintain consistent sand permeability, it is necessary to screen or separate the fine particles that accumulate. The sand must be cooled before being returned to the flask fill area. Foam pattern distortion and adhesive softening may occur if the temperature of the sand introduced in the flask is higher than 50  C (120  F). Pouring a lost foam casting is much like pouring metal in other sand casting methods. The pouring basin or pour cup must remain full throughout the pour. Metal feed is controlled by the vaporization rate of the foam pattern system. A positive head of metal must be maintained (Fig. 2). Failure to do so may result in a mold collapse and scrapped castings or a partial mold collapse, thus providing the opportunity for sand and coating inclusions in the casting. Metal temperature at the time of pouring is critical and must be controlled tighter than with most other sand casting methods.

Lost Foam Casting / 645

Fig. 8

Reactions taking place during a lost foam pouring operation. EPS, expandable polystyrene

Further benefits of lost foam casting result from the freedom in part design offered by the process. Assembled patterns can be used to make castings that cannot be produced by any other high-production process. Part development costs can be reduced because of the ability to prototype with the foam. Product and process development can be kept in-house. Finally, cast-in features and reduced finishing stock are usually benefits of using the lost foam process. Inserts can be cast into the metal, and bimetallic castings can be made.

Casting Quality After pouring, the castings are allowed to cool at approximately the same rate as green sand castings. The flasks are moved into the dumpout area, sand and castings are separated (Fig. 3), and the castings are ready for degating and cleaning. Conventional cleaning methods, such as sandblasting or shot blasting, can be used. Vibratory media cleaning systems may be necessary to remove the coating, which adheres to internal cast surfaces. Water quenching will greatly assist the cleaning of aluminum castings.

Advantages of Lost Foam Casting The benefits of lost foam processing are numerous, and most are obvious to those who have witnessed the simplicity of the process. In lost foam casting, there is no mold parting line, and there are no cores. One-piece flasks are readily moved. The use of untreated, unbonded sand makes the sand system economical and easy to manage. There is less maintenance on sand handling equipment. Less sand is needed in the system, and the recycling of the sand is an ecological advantage. These advantages lead to others that are equally desirable. Casting cleaning is greatly reduced and is sometimes eliminated, except for removal of the wash coating that transfers from the pattern to the casting if robotic submersion in hot water is not used (as previously described). Casting yield can be considerably increased by pouring into a three-dimensional flask with the castings gated to a center sprue. Another advantage of lost foam casting is its ability to reduce the labor and the skill required in the casting process. In addition to reduced cleaning and the elimination of the need for core setting or parting line matches, handling of the flasks and castings is more readily automated with the lost foam process because locations are predictable. A lost foam casting facility has the ability to produce a variety of castings in a continuous and timely manner. Lost foam foundries can pour diverse metals with very few changeover problems, and this adds to the versatility of the foundry. Response to customer demands can also be very rapid with this type of manufacturing system.

Dimensional Tolerances Like bonded sand molds, dimensional tolerances are strongly influenced by the type of mold medium. Lost foam casting with silica sand media yields the dimensional precision similar to that of green sand molding. Dimensional precision of silica sand molds during casting is limited because of the high thermal expansion of silica. Silica also undergoes a phase transformation 55 C (100 F) below the melting point of aluminum. Low-expansion sands (or manufactured ceramics such as mullite) are used when tighter dimensional precision is required, as in the case of engine blocks. Low-expansion sands have expansion coefficients almost three times smaller than that of silica. With low-expansion sands, dimensional tolerances of lost foam aluminum castings are even competitive with die casting. Typical mold materials for aluminum die casting (e.g., H13 steel) also have higher thermal expansion rates than the lowexpansion media of lost foam molds. Folds. Several defects are unique to the lost foam casting process, and one that has received a significant amount of attention is the fold defect (Ref 1, 2). Folds in lost foam casting occur from oxide/liquid styrene films in the molten metal. These oxide/styrene films (referred to as dry-sided oxides similar to the oxide/air interfaces of casting in mold cavities) prevent the proper “knitting” when two metal fronts come in contact with an intervening dry-sided oxide interface. Folds are discussed in more detail as follows. However, with mold-filling simulations, significant improvements in the reductions of folds has been achieved. Simulation of the metal flow is important, and computational fluid dynamics models have been developed to analyze metal flow patterns and temperature distributions and to improve pattern design to avoid trapping liquefied foam during casting (Ref 3, 4). Folds in aluminum alloys are always wholly contained within the eutectic constituent, because they occur during filling and not during solidification. The resulting cracklike discontinuity is not as critical (from a fracture mechanics standpoint) as a sharp fatigue crack, because the folds can be relatively blunt compared to a fatigue crack. Partitioning of the fold into the

eutectic structure also has the effect of pinning the fold and making it less likely to sharpen its edges and propagate during fatigue. Thus, even though folds are not desirable, they are not necessarily a critical discontinuity from the standpoint of fracture mechanics. Based solely on the portion of a fold that is visible on the casting surface, it is possible to use fracture mechanics and determine the size of folds that are acceptable in design. Controlling fold formation involves the following:  Use the appropriate glue: There is a green

glue that can be used to glue foam pattern segments together that is capable of forming these nonbonding dry-sided oxide interfaces 100% of the time at the glue joint location. This is not the glue that should be used. This example demonstrates that decomposition of glue can be so slow that the fold is at the glue interface. With other glues, the fold problem can be downstream from the glue joint even though the amount of glue is the problem.  Use the appropriate foam beads: There are various foam beads available commercially today (2008), all under the banner of EPS. These foam beads can have proprietary additives, and they can be made by different processes. They decompose differently and at different rates when molten metal is poured through the resultant molded foam segments. As a result, the probability of trapping liquid styrene in the casting is also different.  Use a filling process that does not allow an unstable metal front to fold back on itself: Flow 3D has the ability to simulate the filling of lost foam castings. This computational fluid dynamics program allows one not only to see the filling of a lost foam casting but also to track the liquid styrene after it has exceeded a critical scaler quantity. In this way, a better understanding of where a fold may occur has resulted.

REFERENCES 1. M. Sands and S. Shivkumar, EPS Bead Fusion Effects on Fold Defect Formation in Lost Foam Casting of Aluminum Alloys, J. Mater. Sci., Vol 41, 2006, p 2373–2379 2. S. Jagoo, C. Ravidran, and D. Nolan, Fold Defects in Aluminum A356 Lost Foam Casting, Advanced Materials Research, Vol 15–17, Trans Tech Publications, 2007, p 1–6 3. “CFD Modeling for Lost Foam White Side,” Industrial Technologies Program, Metal Casting Project Fact Sheet, U.S. Department of Energy, 2000, http://www.eere.energy. gov/, accessed March 2008 4. C.W. Hirt, “Predicting Defects in Lost Foam Castings,” Flow Science, Inc., Dec 13, 2001, www.Flow-3D.com, accessed March 2008

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 646-661 DOI: 10.1361/asmhba0005255

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Investment Casting Revised by Robert A. Horton, Precision Metalsmiths, Inc.

IN INVESTMENT CASTING, a ceramic slurry is applied around a disposable pattern, usually wax, and allowed to harden to form a disposable casting mold. The term disposable means that the pattern is destroyed during its removal from the mold and that the mold is destroyed to recover the casting. Related processes, in which ceramic slurries are poured against permanent patterns to make cope and drag molds, are covered in the article “Slurry Molding” in this Volume. There are two distinct processes for making investment casting molds: the solid investment (solid mold) process and the ceramic shell process. The ceramic shell process (Fig. 1a) has become the predominant technique for engineering applications, displacing the solid investment process (Fig. 1b). By 1985, fewer than 20% of non-airfoil investment castings and practically no airfoil castings (the largest single application of investment casting) were being made by the solid investment process (Ref 2). Today (2008), the solid investment process is primarily used to produce dental and jewelry castings and has only a small role in engineering applications, mostly for nonferrous alloys. Because of its declining role, the solid investment process is not covered in this article except to indicate important innovations in mold materials (Ref 3–5). Only the ceramic shell process is described. The basic steps of the shell process are illustrated in Fig. 2.

only waxes are seldom used. Waxes are usually modified to improve their properties through the addition of such materials as resins, plastics, fillers, plasticizers, antioxidants, and dyes (Ref 6).* The most widely used waxes for patterns are paraffins and microcrystalline waxes. These two are often used in combination because their properties tend to be complementary. Paraffin waxes are available in closely controlled grades with melting points varying by 2.8  C (5  F) increments; melting points ranging from 52 to 68  C (126 to 156  F) are the most common. The low cost of these waxes, combined with their ready availability, convenient choice of grades, high lubricity, and low melt viscosity, accounts for their wide use. However, applications are limited somewhat by their brittleness and high shrinkage. Grades designated fully refined should be selected for pattern waxes. Microcrystalline waxes tend to be highly plastic and lend toughness to wax blends. Hard, nontacky grades

and soft, adhesive grades are available. Microcrystalline waxes are available with higher melting points than the paraffins and are often used in combination with paraffin. Ozocerite is an imported mineral wax that can serve a similar function. Fisher-Tropsch waxes are synthetic hydrocarbon waxes resembling paraffins but are available in grades that are much harder and higher melting. They improve hardness and rigidity and promote faster setup in pattern-making operations. Polyethylene waxes (of much lower molecular weight than the usual polyethylene plastics) function in much the same manner as the Fisher-Tropsch waxes but are available in even higher melting grades (to over 132  C, or 270  F). Candelilla is an imported vegetable wax that is moderately hard and slightly tacky. It has relatively low thermal expansion and somewhat less solidification shrinkage than the hydrocarbon waxes. Candelilla is useful for hardening

Mold

Pattern Materials Casting

Pattern materials for investment casting can be loosely grouped into waxes and plastics. Waxes are more commonly used, plastic patterns much less so, although foamed polystyrene patterns are frequently used in conjunction with relatively thin ceramic shell molds (such as the Replicast process).

Waxes Wax is the preferred base material for most investment casting patterns, but blends containing

(a)

Fig. 1

(b) (a) Shell and (b) solid mold investment casting. Source: Ref 1

The terms resins and plastics are used in accordance with the definitions proposed for pattern wax technology in Ref 6

Investment Casting / 647

Pattern

Firing

Fig. 2

Assembly

Pouring

Investing

Stuccoing

Knockout

Finishing

Dewaxing

Inspection

Steps in the shell investment casting process

paraffin and raising its softening point. Unfortunately, it has been subject to shortages and wide price fluctuations. Carnauba is another imported vegetable wax. It has a fairly high melting point (83  C, or 181  F) and a low coefficient of thermal expansion. Carnauba is very hard, nontacky, and brittle. It is rather costly, which tends to limit its use. Beeswax is a natural insect wax that melts at approximately 64  C (147  F). It is widely used for modeling and is also useful in pattern blends, contributing some of the same characteristics as comparable microcrystalline waxes. It is believed to be the original wax used for lost-wax casting, but today (2008) its use is limited by its cost relative to other waxes. Palmitic and stearic acids are fatty acids that have good wax properties. They have low viscosity when melted, relatively low thermal expansion in the solid state, relatively low solidification shrinkage, and good compatibility with a wide range of materials. Still other waxes that have been used are montan wax (both acid and ester types), fatty acid amides, ceresin, stearone, and hydrogenated castor oil. The waxes listed represent those most commonly used for pattern blends. Of these, the paraffins and microcrystalline waxes are the most popular. Many other waxes are available and are at least potentially useful. In many cases, a particular wax is not used simply because a less expensive wax offers similar properties. Waxes are reasonably priced, yet they provide a good balance of properties not readily obtainable in other materials. They are easily blended to suit different requirements and are compatible with other materials that can improve their properties still further. Their low melting points and low melt viscosities make them easy to compound, inject, assemble into clusters, and melt out without cracking the thin ceramic shell molds. These properties allow waxes to be

injected at low temperatures and pressures, and this, combined with their lack of abrasiveness, leads to lower tooling costs. Additives. Plain waxes possess many useful properties, but there are two important areas in which they are deficient:  Strength and rigidity, especially where very

fragile patterns need to be made

 Dimensional control, especially with regard to

surface cavitation resulting from solidification shrinkage during and after pattern injection Improvements can be made in these areas with nonwaxy additives. The strength and toughness of waxes are improved by the addition of well-known highmolecular-weight plastics such as polyethylene, ethyl cellulose, nylon, ethylene vinyl acetate, and ethylene vinyl acrylate. These materials are highly viscous at wax-working temperatures, which tends to limit the amounts that can be used. Nevertheless, these amounts are sufficient to make significant improvements in the handling characteristics of patterns and clusters. Polyethylene is widely used because it is economical and compatible with a wide spectrum of waxes. Ethyl cellulose has had some use, but it is much more limited in terms of compatibility and is more expensive. Polystyrene is rarely used because of its incompatibility with the commonly used waxes. Ethylene vinyl acetate and ethylene vinyl acrylate are newer materials that are finding increased application. Nylon has had only small use. Solidification shrinkage, which causes surface cavitation, is reduced somewhat by the plastics mentioned previously. The effect, however, is limited by the low amounts that can be used before viscosity becomes excessive. Greater effects can be obtained by adding resins and fillers. Resins are used in most present-day

pattern waxes. Fillers are used more selectively, and this leads to the description of pattern waxes as being either filled or unfilled. The resins used for shrinkage reduction are of lower molecular weight than the plastics used for toughening. The most useful resins soften gradually and continuously with increasing temperature, and they do not exhibit the large solid-to-liquid expansions during heating, or the reverse contractions during cooling, that characterize the solid-to-liquid transformation in waxes. Thus, resins reduce this effect in direct proportion to the amount used. The amount used can be quite large because resins yield relatively low viscosity melts at wax-processing temperatures. Suitable resins include the numerous rosin derivatives (esterified, polymerized, hydrogenated, and dehydrogenated) as well as such tree-derived resins as Burgundy Pitch, dammar, and the terpene resins. Additional resins include coumarone-indene, various hydrocarbon resins from petroleum, and a number of coal tar resins. The resins listed encompass a wide range of softening points and viscosity-versus-temperature relationships. They vary at room temperature from soft and tacky to hard and brittle, and they may have widely different compatibilities with particular waxes and plastics. All these factors must be considered and properly balanced in using resins in pattern waxes. Fillers. The solidification shrinkage of waxes can also be reduced by mixing in powdered solid materials called fillers. These are insoluble in, and higher melting than, the base wax, and they produce an injectable suspension when the mixture is molten. Because they do not melt, fillers do not contribute to the solidification shrinkage of the mixture, which is reduced in proportion to the amount used. Figure 3 shows the manner in which fillers reduce expansion during heating (as well as corresponding shrinkage upon cooling). Fillers that have been used in pattern waxes include polystyrene, various dicarboxamides and related compounds, isophthalic acid, pentaerythritol, and hexamethylenetetramine. The filler should preferably be in the form of small, equally sized spheres (Ref 9). Several such fillers have been developed, including spherical polystyrene, hollow carbon microspheres, and spherical particles of thermosetting plastic. In a variation of the filler concept, water has been emulsified into molten wax to serve the same function as the solid fillers (Ref 10). This type of wax is commercially available. Several other additives can be used in pattern waxes. Colors in the form of oil-soluble dyes are used to enhance appearance, to provide identification, and to facilitate inspection of molded patterns. Antioxidants can be used to protect waxes and resins subject to thermal deterioration. Oils and plasticizers have been used to alter injection properties and, in certain cases, to reduce shrinkage. As a result of the aforementioal considerations, the compositions of present-day pattern

648 / Expendable Mold Casting Processes with Expendable Patterns Temperature, °F 60 10

80

100

120

140

160

180

9

8

No filler

Volumetric expansion, %

7

6

5

4

3

2 With filler Decachlorobiphenyl (60%)

1

Cyanuric acid (60%) 0 16

27

38

49

60

71

82

Temperature, °C

Fig. 3

Effects of filler materials on the volumetric thermal expansion of a pattern wax. Source: Ref 7, 8

waxes (unfilled) typically fall into the ranges given as follows: Ingredients

Waxes (usually more than one) Resins (one or two) Plastic (one) Other

Composition, %

30–70 20–60 0–20 0–5

Filled waxes usually have similar base materials, with 15 to 45% filler added. Wax Selection. In formulating wax pattern materials, a number of processing areas and other concerns must be addressed. These are listed as follows, along with the properties or considerations appropriate to each:  Injection: Softening point, freezing range,    

rheological properties, ability to duplicate detail, surface, and setup time Removal, handling, and assembly: Lubricity, strength, hardness, rigidity, impact resistance, stability, and weldability Dimensional control: Thermal expansion/ shrinkage, solidification shrinkage, cavitation tendency, distortion, and stability Mold making: Strength, wettability, and resistance to binders and solvents Mold dewaxing and burnout: Softening point, viscosity, thermal expansion, thermal diffusivity, and ash content

 Miscellaneous: Cost, availability, ease of

recycling, toxicity, and environmental factors Assistance in formulating pattern waxes can be found in Ref 6, 11, and 12. Reference 12 provides an extensive listing of pattern ingredients and their properties. A number of standard tests are available to facilitate the effort (Ref 13–15). The application of thermal analysis methods to wax testing is being investigated (Ref 11, 16), and this approach warrants extension for both control and development work. The Investment Casting Institute has provided a test die for evaluating waxes and wax injection presses (Ref 17, 18). No simple laboratory tests are available that can evaluate the suitability of a pattern wax. The best approach usually consists of laboratory tests for screening purposes combined with practical evaluation under such production-like conditions as injection behavior, handling characteristics, dimensional consistency, and dewaxability.

Plastics Next to wax, plastic is the most widely used pattern material. A general-purpose grade of polystyrene is usually used. The principal advantages of polystyrene and other plastics are their ability to be molded at high production rates on automatic equipment and their resistance to handling damage even in

extremely thin sections. In addition, polystyrene is very economical and very stable, so patterns can be stored indefinitely without deterioration. Most wax patterns deteriorate with age and eventually must be discarded. A disadvantage that limits the use of polystyrene is its tendency to cause mold cracking during pattern removal, a condition that is worse with ceramic shell molds than with solid investment molds. Other common plastics, such as polyethylene, nylon, ethyl cellulose, and cellulose acetate, are similar in this regard. In addition, tooling and injection equipment for polystyrene is more expensive than for wax. As a result, polystyrene finds only limited use. A survey of 72 investment casting plants in the United States revealed that fewer than 20% had machines for injecting polystyrene and that wax machines outnumbered plastics machines by almost 13 to 1 (Ref 19). Perhaps the most important application for polystyrene is for small, delicate airfoils. Patterns for such castings are incorporated into composite wax/plastic assemblies for integral rotor and nozzle patterns. These airfoils are extremely thin and delicate; they would be too fragile if molded in wax because even a single chip or crack will cause an expensive casting to be rejected. Polystyrene solves this problem, and the use of wax for the rest of the assembly makes it feasible to process such large patterns without excessive mold cracking. The other important use for polystyrene is for small patterns running in large quantities. However, this application has declined as ceramic shell molds have replaced solid investment molds.

Other Pattern Materials Urea-based patterns were developed in Europe and have found some application there (Ref 20). They resemble plastics in that they are very hard and strong and require high-pressure injection machines. An advantage of urea patterns is that they can be easily removed without stressing ceramic shell molds; the pattern is simply dissolved out in water or an aqueous solution. Another water-soluble pattern material is based on mixtures of low-melting salts (Ref 21). Patterns based on paradichlorobenzene and naphthalene have found some application in Japan (Ref 22). Foamed polystyrene has long been used for gating system components. It is also used for patterns in the Replicast process but does not produce the smooth surfaces obtained with wax. A recent innovation is a new foam developed just for investment casting and intended to have advantages over both foamed polystyrene and wax (Ref 23).

Patternmaking Patterns for investment casting are made by injecting the pattern material into metal molds of the desired shape. Small quantities of patterns

Investment Casting / 649 can also be produced by machining (see the section “Machining of Patterns” in this article).

Injection of Wax Patterns Wax patterns are generally injected at relatively low temperatures and pressures in split dies using equipment specifically designed for this purpose. Patterns can be injected in the liquid, slushy, pastelike, or solid condition. Injection in the solid condition is often referred to as wax extrusion. Temperatures usually range from approximately 43 to 77  C (110 to 170  F), and pressures from approximately 275 kPa to 10.3 MPa (40 to 1500 psi). Liquid waxes are injected at higher temperatures and lower pressures; solid waxes, at lower temperatures and higher pressures. In some cases, the same wax can be injected under some, or all, of these conditions, but it is often beneficial to tailor a wax for one particular condition. Equipment for wax injection ranges from simple and inexpensive to sophisticated and costly. It can be as simple as a pneumatic unit with a closed, heated reservoir tank that is equipped with a thermostat, pressure regulator, heated valve, and nozzle and is connected to the shop air line for pressurization. A small die is held against the nozzle with one hand while the valve is operated with the other. Such machines are limited by available air pressure (usually <690 kPa, or 100 psi) and are therefore generally used to inject liquid wax. Such equipment is satisfactory for a large variety of small hardware parts that are commonly made as investment castings. Parts that are more demanding in terms of size, complexity, or dimensional requirements are made on hydraulic machines. These machines provide higher pressures for improved injections as well as high clamping pressures to accommodate large dies and high injection pressures. They can also be operated with very low pressures, as is often necessary when injecting around thin ceramic cores. Hydraulic wax injection machines are sized according to their clamping force capability, with models ranging in capacity from 9.1 to 363 Mgf (10 to 400 tonf ) or more. Machines for manual or semiautomatic operation generally have horizontal platens that clamp the die closed (one stationary, one movable). Injection is horizontal along the parting line; a second vertical nozzle is sometimes available. Models are available with C-frame construction and with two-, three-, and four-post H-frame construction. Machines for automatic injection have the platens mounted vertically, with injection through the stationary platen and one-half of the die or at the parting line. A reservoir tank is usually provided for liquid and slushy wax, having either slow agitation or recirculation to keep the wax uniform and to retain solid material (wax or filler) in suspension. Reservoir tanks are sometimes omitted, and the machines are connected directly to a central wax supply.

Machines for paste wax injection require preconditioned wax in a metal cylinder. The cylinder is inserted into the machine, and the wax is pushed out by a piston. Solid injection (extrusion) machines require pretempered wax billets. Paste and solid injection both require a separate unit (oven or tempering bath) for tempering the cylinders or billets. Liquid injection usually requires separate units for melting and tempering the wax before it is introduced into the reservoir tank. The parameters customarily controlled include wax and nozzle temperature, pressure, flow rate, and dwell time. Other available features, which may or may not be considered standard, include control of flow rate acceleration and deceleration, purging of the nozzle with fresh wax before each shot, nozzle drool control, unlimited shot capacity, die handling means, nozzle position adjustment, watercooled platens, and multistation operation. Reference publications that are available include a guide for troubleshooting wax injection operations (Ref 24) and an atlas of pattern defects (Ref 25). Other recommended publications are given in Ref 26 to 31.

Injection of Plastic Patterns Polystyrene patterns are generally injected at temperatures of 177 to 260  C (350 to 500  F) and pressures of 27.6 to 138 MPa (4 to 20 ksi) on standard plastic injection machines. These are hydraulic machines with vertical watercooled platens that carry the die halves, and horizontal injection takes place through the stationary platen. Polystyrene granules are loaded into a hopper, from which they are fed into a plasticizing chamber (barrel). Modern machines have a rotating screw that reciprocates within the heated barrel to prepare a shot of material to the proper consistency and then inject it into the die. Some older machines having plungers instead of screws are still in use. The use of multicavity dies, as well as the running of a number of dies at once, is commonplace. Coupled with automatic or semiautomatic operation, this results in extremely high productivity. The same parameters are controlled as for wax injection. The development of wax-based compositions that perform well in plastic injection machines has permitted the advantages of wax patterns to be combined with the advantages of plastic injection for certain applications (Ref 32).

Machining of Patterns When only one or a few patterns are required, as for prototype and experimental work, it is expedient to machine them directly from wax or polystyrene. This avoids the time and expense involved in making pattern tooling. Special waxes have been specifically developed for this application (Ref 33). Although machining yields

the best pattern surfaces and closest dimensional control, its use for prototyping has declined since the introduction of the various rapid prototyping (RP) methods in the late 1980s. Because of the convenience and lower cost of RP, its use has grown spectacularly until it now ranks as the dominant method, it has served to generate a much higher interest in prototyping for investment casters and their customers.

Pattern Tooling for Investment Casting Investment casting permits various potential tooling options that are made possible by the low melting point, good fluidity, and lack of abrasiveness of waxes. This often represents an important competitive advantage. For a given part configuration, anticipated production requirements, choice of pattern material, and available patternmaking equipment, the selection is based on a consideration of cost, tool life, delivery time, pattern quality, and production efficacy. The methods in use can be grouped into three basic categories: machining, forming against a positive model (using a variety of methods), and casting into a suitable foundry mold. Many materials can be used, including soft metals (lead-bismuth-tin alloys), zinc alloys, brass, bronze, beryllium-copper, nickel-plate, steel, rubber, plastic, metal-filled plastic, plaster, and combinations of these. All of these are suitable for wax patterns, but plastic patterns usually require steel or beryllium-copper tooling. Soft metals and nonmetallics are often used for temporary tooling, but soft metals containing lead are excluded now because of its toxicity by ingestion or inhalation of fumes or dust. Machined Tooling. Most production tooling is made by machining. The early investment casting industry favored tooling made from master models, but as parts became larger and more complex and production methods more demanding, machined tooling became dominant. Computer-aided design, electric discharge machining, and computer numerical controlled machine tools are commonly used. Aluminum is preferred for most wax tooling, steel, for plastic tooling. Aluminum is economical to machine, has good thermal conductivity, and is conveniently light. Brass or steel inserts can be used in areas subject to wear. The higher strength of steel is not usually needed for wax injection; instead, steel is primarily used for plastic injection for small parts that run in sufficiently large quantities to justify its higher cost. For polystyrene, this quantity is usually approximately 10,000 pieces or more; for very small parts, perhaps as low as 5000. Tool steels are most frequently used for plastic dies. AISI type P20 is extensively used because it can be purchased prehardened, then machined, and finally put into use without heat treatment. AISI A2 and O1 are also used. Both are machined in the annealed state, then hardened and tempered. Carbon steels such as 1020 are sometimes used for intermediate quantities.

650 / Expendable Mold Casting Processes with Expendable Patterns Holder blocks, shoes, and mold bases used to hold the plastic injection molds can be made of prehardened alloy steel (for example, 4140). Tooling made against a positive model includes a variety of methods for forming metal or metal-faced tooling, such as spraying (Ref 34), pressure casting (Ref 35), cold and warm hobbing (Ref 36, 37), electroforming (Ref 38, 39), and (potentially) gas plating (Ref 40). It also includes the casting of such nonmetallic materials as plaster (usually used to back up a spray metal facing), plastics (Ref 41), and rubber (Ref 42), as well as the vulcanizing of solid rubber under pressure (Ref 43). All these methods begin with a positive model in the shape of the final investment casting; the model is machined oversize to include the appropriate shrink factors. The methods are economical because it is generally (although not always) less expensive to machine a positive model than to machine a negative cavity of identical shape. Furthermore, with the exception of hobbing, the model can be made of a material that is easier to machine. The various methods of making dies from the master model are relatively inexpensive, so that the combined cost of machining the model and making the die is usually less than that for full machining. When multiple cavities are needed or where molds must be replaced, the same master can be used, resulting in further economies. Cast Tooling. Steel and beryllium-copper are frequently used for cast tooling. Aluminum and zinc alloys have also been used. Wax can be cast against a master model to produce a pattern, which is then used to make an investment-cast cavity. Better as-cast accuracy can be achieved if the model is coated with only a thin layer of wax. This is used to investment cast a steel shell, which is then backed up with cast aluminum (Ref 44). The ceramic mold process can also be used to cast injection cavities (Ref 45). Both casting and hobbing can produce excellent tooling that is suitable for the wax or plastic injection of relatively simple parts, such as many small hardware castings. Machined tooling is also used for many of the same applications, as well as for complex parts, such as airfoils, and for large structural parts. Soft metal tooling and nonmetallic tooling are primarily used for short runs and experimental work or for temporary tooling. However, under favorable conditions, runs of up to 40,000 parts have been reported for cast bismuth-tin tooling (Ref 46). Cast, hobbed, and machined dies used for wax injection often last indefinitely. Rubber molds are regularly used for making patterns for jewelry casting.

Pattern and Cluster Assembly Large patterns for investment casting are set up and processed individually, but small-tomedium ones are assembled into multipattern clusters for economy in processing. Clusters of aircraft turbine blades, for example, may range from 6 to 30 parts. For small hardware parts, the number may run into the dozens or even

the hundreds. Most patterns are injected complete, including their casting gates. However, very large or complex parts can be injected in segments, which are assembled into final form. The capacity of injection machines and the cost of tooling are important considerations. Gating components, including pour cups, are produced separately, and patterns and gating are assembled to produce the pattern cluster. Standard extruded wax shapes are often used for gating, especially for mock-up work. Preformed ceramic pour cups are often used in place of wax ones. Most assembly is done manually.

Pattern Assembly Wax components are readily assembled by wax welding using a hot iron or spatula or a small gas flame. The wax at the interface between two components is quickly melted, and the components are pressed together until the wax resolidifies. The joint is then smoothed over. A hot melt adhesive can be used instead of, or in addition to, wax welding. A laser welding system has been described for this purpose (Ref 47). Manual wax welding requires a fair degree of skill and considerable attention to detail. Fixtures are essential to ensure accurate alignment in assembling patterns, and they are often useful for cluster assembly. Patterns must be properly spaced and aligned. Joints must be strong and completely sealed with no undercuts. Care must be taken to avoid damaging patterns or splattering drops of molten wax on them. Polystyrene pattern segments can be assembled by solvent welding. The plastic at the interface is softened with the solvent, and the parts are pressed together until bonded. The polystyrene becomes very tacky when wet with solvent (such as methyl ethyl ketone) and readily adheres to itself. Frequently, only one of the two halves needs to be wet. The assembly of polystyrene to wax is done by welding, with only the wax being melted. Automation. Most assembly and setup operations are still performed manually, but some automation is being introduced. In one application, a robot is used to apply sealing compound in the assembly of patterns for integrally cast nozzles having 52 to 120 airfoils apiece. In the area of cluster setup, a few units for automatic assembly are being offered commercially, one of which has been described in great detail (Ref 48). Clusters are assembled automatically by welding pairs of wax patterns on opposite sides along the length of a wax runner bar, using heated copper blades and spring pressure. Standard gate designs in a range of size increments eliminate the need for special fixturing. The unit is suitable for most smallto-medium sized castings. A fully automated wax cell is described in Ref 49.

Cluster Design The design of the cluster is a critical factor because it affects almost every aspect of the process. Factors to be considered include ease of

assembly, number of pieces processed at a time, ratio of metal poured to castings shipped, handling strength, ease of mold forming and drying, wax removal, liquid metal flow, filling of thin sections, feeding of shrinkage, control of grain size and shape (when specified), shell removal, ease of cutoff and finishing, and available equipment and processes. Three requirements are essential:  Providing a cluster that is properly sized and

strong enough to be handled throughout the process  Meeting metallurgical requirements  Providing separate specimens for chemical or mechanical testing (when required) Once these essentials are satisfied, other factors are adjusted to maximize profitability. The process is very flexible, and foundries approach this goal in various ways. Some tailor the cluster design to each individual part to maximize parts per cluster and metal use. Others prefer standardized clusters to facilitate handling and processing. Where close control of the grain is required, whether for equiaxed, columnar (directionally solidified), or single-crystal casting, circular clusters are often used to provide thermal uniformity. The specific casting process used, such as countergravity casting or directional solidification, may dictate the basic features of the cluster design (see, for example, the article “Centrifugal Casting” in this Volume). The critical aspects of cluster design are gating and risering. Basic concepts of risering borrowed from sand casting, such as progressive solidification toward the riser, Chvorinov’s rule and its extensions, solidification mode, and feeding distance as a function of alloy and section size , also apply to investment casting (Ref 50). However, feeding distances tend to be longer in hot investment molds than in sand molds (Ref 51, 52). Separate risers are sometimes used, but more often the gating system also serves the risering function, especially for the myriad of small parts that are commonly investment cast. The use of wax clusters permits great flexibility in the design of feeding systems. Process development clusters are readily mocked up for trial. Extruded wax shapes are easily bent into feeders that can be attached to any isolated sections of the part that are prone to shrinkage. Once proved, they can be incorporated into tooling if this is cost-effective. If not, they can be applied manually during cluster assembly. This capability makes it practical to cast very complex parts with high quality and makes it feasible to convert fabrications assembled from large numbers of individual components into singlepiece investment castings at great cost-savings. Considerable effort has been expended to provide simulation models of value in investment casting production, and these are being applied on the shop floor. They include the design of gating and feeding systems, effect of solidification conditions on metal structure, accurate prediction of tooling dimensions, and

Investment Casting / 651 unstuccoed layer is sometimes referred to as a seal coat.

Mold Refractories

Ceramic Shell Mold Manufacture In 1999, the American Ceramic Society commemorated its 100th anniversory by publishing a book (Ref 59) describing the most significant innovations in ceramics during that century. Two of these were developments within the investment casting industry: ceramic shell investment casting and ceramic cores for investment casting. Manufacture of ceramic shell molds is discussed in this section. Ceramic cores are disscussed in the next section. Investment shell molds are made by applying a series of ceramic coatings to the pattern clusters. Each coating consists of a fine ceramic layer with coarse ceramic particles embedded in its outer surface. A cluster is first dipped into a ceramic slurry bath. The cluster is then withdrawn from the slurry and manipulated to drain off excess slurry and to produce a uniform layer. The wet layer is immediately stuccoed with relatively coarse ceramic particles either by immersing it into a fluidized bed of the particles or by sprinkling the particles on it from above. The fine ceramic layer forms the inner face of the mold and reproduces every detail of the pattern, including its smooth surface. It also contains the bonding agent, which provides strength to the structure. The coarse stucco particles serve to arrest further runoff of the slurry, help to prevent it from cracking or pulling away, provide keying (bonding) between individual coating layers, and build up thickness faster. Each coating is allowed to harden or set before the next one is applied. This is accomplished by drying, chemical gelling, or a combination of these. The operations of coating, stuccoing, and hardening are repeated a number of times until the required mold thickness is achieved. The final coat is usually left unstuccoed in order to avoid the occurrence of loose particles on the mold surface. This final,

Table 1

The most common refractories for ceramic shell molds are siliceous, for example, silica itself, zircon, and various aluminum silicates composed of mullite and (usually) free silica. These three types in various combinations are used for most applications. Alumina has had some use for superalloy casting, and this application has increased with the growth of directional solidification processes. Alumina is generally considered too expensive and unnecessary for commercial hardware casting. Silica, zircon, aluminum silicates, and alumina find use for both slurry refractories and stuccos. Table 1 lists the typical properties of these materials, and Fig. 4 shows thermal expansion curves. Other refractories, such as graphite, zirconia (ZrO2), and yttria (Y2O3), have been suggested for use with reactive alloys. Still other materials have been proposed for specific purposes. These specialized applications are summarized in Ref 61. Silica is generally used in the form of silica glass (fused silica), which is made by melting natural quartz sand and then solidifying it to form a glass. It is crushed and screened to produce stucco particles, and it is ground to a powder for use in slurries. Its extremely low coefficient of thermal expansion imparts thermal shock resistance to molds, and its ready solubility in molten caustic and caustic solutions provides a means of removing shell material chemically from areas of castings that are difficult to clean by other methods. Silica is also used as naturally occurring quartz. This is the least expensive material used to any extent. However, its utility is limited by its high coefficient of thermal expansion and by the high, abrupt expansion at 573  C (1063  F) accompanying its a-to-b-phase transition. As a result, shells containing quartz must be fired slowly, a practice most foundries find inconvenient. Zircon occurs naturally as a sand, and it is used in this form as a stucco. It is generally

limited to use with prime coats because it does not occur in sizes coarse enough for stuccoing backup coats. It is also ground to powder (and sometimes calcined) for use in slurries, often in conjunction with fused silica and/or aluminum silicates. Its principal advantages are high refractoriness, resistance to wetting by molten metals, and round particle shape. Unfortunately, this excellent refractory is in very short supply now (2008). Unlike two other zircon shortages of the past, this one has no obvious solution. It appears that zircon sources are simply running out. The industry is striving to minimize its zircon use while seeking a substitute material or a process that does not require zircon.

Temperature, °F 600

1.4

1000

1400

1800

2200

1.2

Quartz 1.0

Linear expansion, %

so on. So far, this work has been primarily used by the larger foundries. Pertinent references include Ref 53 to 58.

0.8

Alumina

0.6

70% 60% Aluminum silicates

0.4

42% 47% 73% 0.2

Zircon Fused silica 0.0 200

400

800

600

1000

1200

Temperature, °C

Fig. 4

Linear thermal expansion of some refractories common to investment casting. Source: Ref 60

Nominal compositions and typical properties of common refractories for investment casting

Data are for comparison only, are not specifications, and may not describe commercial products. Approximate melting point

Approximate theoretical density Material

3

3



C



PCE temperature (b) 

F



Nominal composition, %

Crystalline form

g/cm

lb/in.

Relative leachability(a)

C

Aluminosilicates 42% 47% 60% 70% 73% Alumina Fused silica

F

pH

Color

Al2O3-53SiO2 Al2O3-49SiO2 Al2O3-36SiO2 Al2O3-25SiO2 Al2O3-22SiO2 99% + Al2O3 99.5% SiO2

2.4–2.5 2.5–2.6 2.7–2.8 2.8–2.9 2.8–2.9 4.0 2.2

0.086–0.090 0.090–0.094 0.097–0.10 0.10–0.104 0.10–0.104 0.144 0.079

Poor Poor Poor Poor Poor Poor Good

... ... ... ... ... 2040 1710

... ... ... ... ... 3700 3110

1750 1760 1820 1865 1820 ... ...

3180 3200 3310 3390 3310 ... ...

6.5–7.8 6.5–7.8 6.5–7.8 6.5–7.8 6.5–7.8 8.5–8.9 6.0–7.5

Gray to Gray to Gray to Gray to Gray to White White

Sillica-quartz Zircon

99.5% SiO2 97% + ZrSiO4

Mixture Mixture Mixture Mixture Mixture Trigonal Typically 97% + amorphous Hexagonal Tetragonal

2.6 4.5

0.094 0.162

Good Moderate

1710 2550

3110 4620

... ...

... ...

6.4–7.5 4.7–7.0

White to tan White to tan

tan tan tan tan tan

(a) Poor: slight reaction in hot concentrated alkali; Good: soluble to very soluble in hot concentrated alkali or hydrofluoric acid; Moderate: reacts with hot concentrated alkali solutions. (b) PCE, pyrometric cone equivalent. Source: Ref 60

652 / Expendable Mold Casting Processes with Expendable Patterns Aluminum silicates for investment casting are made by calcining fireclays or other suitable materials to produce a series of products ranging in alumina content from approximately 42 to 72%, with the remainder being silica plus impurities. Refractoriness and cost increase with alumina content. The only stable compound between alumina and silica at elevated temperatures is mullite (3Al2O3
2SiO2), which contains 72% alumina. Mixtures containing less than 72% alumina produce mullite plus free silica. The latter is usually in the form of silica glass, although some crystalline silica in the cristobalite form may be present. As the alumina content increases, the amount of mullite increases and free silica decreases until, at approximately 72% alumina, the material contains only mullite. Fired pellets of these materials are crushed or ground and carefully sized to produce a range of powder sizes for use in slurries, and granular materials for use as stuccos. Alumina is produced from bauxite ore by the Bayer process. It is more refractory than silica or mullite and is less reactive toward many alloys than the siliceous refractories. Its use is primarily confined to superalloy casting, in which these properties can be used to advantage. The usual grades are tabular, which has been calcined just below the melting point, and fused, which has been electrically melted. The latter is slightly denser and slightly purer. Yttria is now being used in forecoats for casting titanium and certain nickel-base superalloys that contain yttrium.

Binders The commonly used binders are also siliceous and include colloidal silica, hydrolyzed ethyl silicate, and sodium silicate. Hybrid binders have also been developed, and alumina or zirconia binders are used for some processes. Colloidal silica is the most widely used. It is manufactured by removing sodium ions from sodium silicate by ion exchange. The product consists of a colloidal dispersion of virtually spherical silica particles in water. The dispersion is stabilized by an ionic charge, which causes the particles to repel one another, thus preventing agglomeration. The stabilizing ion is usually sodium (up to 0.6%), although ammonia can also be used. In either case, the product is alkaline. Colloidal silica can also be stabilized at an acid pH, but such products are not widely used. The most popular grades are sodium-stabilized with a silica content of 30% and an average particle size of either 7 or 12 nm. They are either used at this 30% level or diluted with water to reduce the silica content to 18 to 30%. Coherent gels having excellent bonding properties are formed by adding ionic salts that neutralize the ionic charge or by concentrating the sol (as by drying a coating). Colloidal silica is an excellent general-purpose binder. Its main disadvantage is that its water base makes it slow drying, especially in inaccessible pockets or cores.

Ethyl silicate is produced by the reaction of silicon tetrachloride with ethyl alcohol. The basic compound formed is tetraethylor-thosilicate (Si(OC2H5)4). This corresponds to a theoretical silica content of 28.8%. The grade used for investment applications is designated ethyl silicate 40, and it consists of a mixture of ethyl polysilicates averaging five silicon atoms per molecule and having a silica content of approximately 40%. By itself, ethyl silicate has no bonding properties. It is converted to ethyl silicate binder by reaction with water (hydrolysis). The reaction is usually carried out in ethyl alcohol, which serves as a mutual solvent, using an acid catalyst such as hydrochloric acid. The reaction is fairly rapid and highly exothermic. Its progress can be monitored with temperature measurements. The reaction forms complex silicic acids that are capable of condensing to form coherent gels having good bonding properties. This process is promoted by drying (concentrating) or by adding an alkali such as ammonia. The result is similar in many ways to that obtained using colloidal silica. Prehydrolyzed grades of ethyl silicate having reasonable shelf life are also available, making it unnecessary for foundries to perform this chemical operation themselves. Ethyl silicate, with its alcohol base, dries much faster than colloidal silica. It is, however, more expensive and poses fire and environmental hazards. It is best used in applications involving rapid dipping cycles. Ethyl silicate slurries are readily gelled by exposure to an ammonia atmosphere; this permits dips to be applied very quickly, yet still provides proper drying because of the high volatility of the alcohol. In recent years, the use of ethyl silicatealcohol binders for ceramic shells has declined rapidly because it is now classified as a volatile organic substance having fumes that cannot be released to the atmosphere. It is technically feasible to capture these fumes, but this is quite expensive, causing a good number of former users to switch to aqueous colloidal silica. Hybrid binders have been developed in an effort to combine the advantages of colloidal silica and ethyl silicate. The water required for hydrolysis of the ethyl silicate is supplied by using colloidal silica, providing an additional source of silica for improved strength. Less flammable solvents can be substituted for alcohol. The resulting product presents a desirable combination of properties (Ref 62). Liquid sodium silicate solutions are used where a very inexpensive binder is desired. Upon evaporation, these binders form a strong, glassy bond. Sodium silicate binders have poor refractoriness, which greatly limits their sphere of application. In addition, they are not resistant to the steam atmosphere of dewaxing autoclaves. Nevertheless, they have found some use, sometimes in conjunction with colloidal silica or ethyl silicate. Other Binder Materials. The operation of directional solidification processes, which subject the mold to high temperatures for relatively

long times, along with the introduction of even more reactive superalloys, has led to interest in more refractory binders. Colloidal alumina and colloidal zirconia binders have been made available for this purpose (Ref 63, 64). Both, however, are inferior to colloidal silica in room-temperature bonding properties.

Other Ingredients Wetting Agents. In addition to the binder and refractory, slurries generally contain anionic or nonionic wetting agents, such as sodium alkyl aryl sulfonates, sodium alkyl sulfates, or octylphenoxy polyethoxyethanol, to promote wetting of the pattern or prior slurry coats. These are generally used in amounts of 0.03 to 0.3% by weight of the liquid. Wetting agents are sometimes omitted from ethyl silicate/alcohol slurries and from water-based backup slurries. Antifoam Compounds. Where wetting agents are used, especially in prime coats, an antifoam compound may be included to suppress foam formation and to permit air bubbles to escape. Commonly used defoamers are aqueous silicone emulsions and liquid fatty alcohols such as n-octyl alcohol and 2-ethyl hexyl alcohol. Depending on type, these formulations are effective in very low concentrations of 0.002 to 0.10%, based on the liquid weight. Miscellaneous Constituents. Organic film formers are sometimes used to improve the green strength and resilience of the dried coating or to enhance the coating ability of the slurry. Aqueous polyvinyl acetate emulsions, polyvinyl alcohol, and ammonium alginate have been used for this purpose. Small additions of clay have also been used to improve coating characteristics. Where close control of grain size is required, as for equiaxed superalloy casting, a nucleating agent (grain refiner) is added to the prime slurry in amounts ranging from approximately 0.5 to 10% by weight of the slurry refractory (Ref 65–67). Preferred nucleating agents are refractory cobalt compounds such as aluminates, silicates, titanates, and oxides.

Slurry Formulation The actual percentage composition of ceramic shell slurries depends on the particular refractory powder, type and concentration of binder, liquid vehicle, and desired slurry viscosity. Composition generally falls in the following broad range by weight: Ingredient

Binder solids Liquid (from binder or added) Refractory powder

Composition, %

5–10 15–30 60–80

Other optional ingredients, when present, are used in the amounts already given. Slurry compositions are usually proprietary, but three published formulas for zircon slurries are given in Table 2.

Investment Casting / 653 Table 2

Formulations and properties of three types of zircon slurries Constituents Colloidal silica (30%)

Slurry

1(b) 2(c) 3(d)

Water

Zircon powder

Properties Wetting agent

Silica

Density

L

gal

L

gal

kg

lb

cm3

Fl oz

kg

lb

g/cm3

lb/in.3

Viscosity, s(a)

9.5 11.4 9.5

2.5 3 2.5

3.8 ... 3.8

1.0 ... 1.0

45 45 41

100 100 90

10 10 10

0.34 0.34 0.34

... ... 4.5

... ... 10

2.7–2.75 2.9–2.95 2.65–2.7

0.097–0.099 0.104–0.106 0.95–0.097

8–10 9–11 9–11

(a) No. 4 Zahn cup. (b) A general-purpose slurry. (c) A high-strength slurry used for ceramic shell molds for castings with heavy sections and for applications in which high wax pressure during meltout and high metal pressure are used. (d) Modified with silica; used with castings with small cored holes from which cores are removed by leaching with molten alkali salts

Two extremely important properties of ceramic shell molds are green strength and fired strength. A fundamental study of the effects of composition and processing parameters on these properties is described in Ref 68 and 69, which also provide additional examples of slurry compositions. This work was further amplified by another researcher (Ref 70). The effects of the usual refractory powders and stuccos on hot strength were investigated, and it was found that strength after firing is not a good indication of hot strength (Ref 71). However, it is still an important property because it affects the shell removal operation.

Slurry Preparation and Control Preparation. Slurries are prepared by adding the refractory powder to the binder liquid, using sufficient agitation to break up agglomerates and thoroughly wet and disperse the powder. Viscosity initially tests excessively high because of air entrainment and lack of particle wetting; therefore, mixing is continued until the viscosity falls to its final level before the slurry is put into use. Continued stirring is required in production to keep the powder from settling out of suspension. Either rotating tanks with baffles or propeller mixers are used for this purpose—but generally at a lower agitation level than that used for the initial dispersion. Control procedures for slurries vary considerably among foundries, reflecting in part the wide range of specifications that different shops work to, depending on their product line. The most prevalent controls are the measurement of the initial ingredients and the viscosity of the slurry. The latter is measured by a flow cup, such as a No. 4 or 5 Zahn cup, or a rotating viscometer of the Brookfield type. Other parameters that are often controlled include slurry temperature, density, and pH, all of which are easily determined. More involved and time-consuming is the determination of the actual composition of the slurry in terms of water content, silica binder, and refractory powder. This type of analysis, therefore, is performed less often and is sometimes reserved for troubleshooting rather than control. In the ceramic retention test, which is used by relatively few foundries, the actual weight of slurry adhering to a standard test plate under controlled conditions is determined. Some properties of the finished ceramic shells that can be monitored include weight, modulus of rupture (green and/or fired), and

permeability. Equipment for determining the hot strength of shells is relatively rare in the industry and is largely used for research. Test procedures for raw materials, slurries, molds, and cores are described in Ref 60, and the problems involved in slurry control are discussed in Ref 72 to 74.

Cluster Preparation Before dipping, pattern clusters are usually cleaned to remove injection lubricant, loose pieces of wax, or dirt. Cleaning is accomplished by rinsing the pattern clusters in an appropriate solution, such as a water solution of a wetting agent, a solvent that does not attack the wax, or a solvent mixture capable of attacking the wax in a controlled manner to produce a fine uniform etching action that promotes slurry adhesion without affecting the cast surface. An additional rinse can be used to remove the cleaning agent. Another procedure involves dipping the cluster into a liquid that deposits an ultrathin refractory oxide coating that renders the surface hydrophilic; thus, the cluster is readily wetted by the ceramic slurry (Ref 75). These different solutions can sometimes be used in combination. The clusters are usually allowed to dry before dipping. Drying produces a chilling effect, which causes unwanted contraction in pattern dimensions; therefore, the clusters are generally allowed to stand until they return to room temperature.

Coating and Drying Dipping, draining, and stuccoing of clusters are carried out manually, robotically, or mechanically. Companies are increasingly using robots in order to heighten productivity, to process larger parts and clusters, and to produce more uniform coatings. When robots are introduced, they are often programmed to reproduce the actions of skilled operators. Dedicated mechanical equipment can sometimes operate faster, especially with standardized clusters, and finds some application. Most dipping is done in air. Dipping under vacuum is very effective for coating narrow passageways and for eliminating air bubbles, but it is not widely practiced. The cleaned and conditioned cluster is dipped into the prime slurry and rotated; it is then withdrawn and drained over the slurry tank with suitable manipulation to produce a uniform

coating. Next, the stucco particles are applied by placing the cluster in a stream of particles falling from an overhead screen in a rainfall sander or by plunging the cluster into a fluidized bed of the particles. In the fluidized bed, the particles behave as a boiling liquid because of the action of pressurized air passing through a porous plate in the bottom of the bed. Generally, prime slurries contain finer refractory powder, are used at a higher viscosity, and are stuccoed with finer particles than backup coats. These characteristics provide a smoothsurfaced mold capable of resisting metal penetration. Backup coats are formulated to coat readily over the prime coats (which may be somewhat porous and absorbent), to provide high strength, and to build up the required thickness with a minimum number of coats. One or two additional prime coats are sometimes used before starting the backup coats. The number of coats required is related to the size of the clusters and the metal weight to be poured and may range from as few as 5 for small clusters to as many as 15 or more for large ones. For most applications, the number ranges from 6 to 9. Between coats, the slurries are hardened by drying or gelling. Air drying at room temperature with circulating air of controlled temperature and humidity is the most common method. Drying is usually carried out on open racks or conveyors, but cabinets or tunnels are sometimes used. Drying is complicated by the high thermal expansion/contraction characteristics of waxes. If drying is too rapid, the chilling effect causes the pattern to contract while the coating is wet and unbonded. Then, as the coating is developing strength and even shrinking somewhat, the wax begins to expand as the drying rate declines and it regains temperature. This can actually crack the coating. Therefore, relative humidity is generally kept above 40%. For normal conditions, a relative humidity of 50% has been recommended as ideal (Ref 76). The essential point is that the maximum temperature differential between the wax and the drying air should not be excessive. Shop experience indicates that 4 to 6  C (7.2 to 10.8  F) is a practical maximum (Ref 77). The slurry does not have to be thoroughly dried between coats, but it must be dry enough so that the next coat can be applied without washing off the previous one. An alternative technique is to harden the slurry by chemical

654 / Expendable Mold Casting Processes with Expendable Patterns gelation, which can be accomplished without drying. This is most successful with ethyl silicate binders, which can be surface gelled by exposure to an ammonia atmosphere. Hardening by chemical gelation permits the minimum time between coats. Gelling can also be accomplished by adding gelling agents to the stucco or by alternating alkaline and acidic slurries or negatively charged and positively charged slurries in the dipping sequence. Shells that are gelled without drying are generally weaker even when subsequently dried. In addition, the weight of the unevaporated solvent can be excessive and can cause the wax cluster to yield or even break in some cases.

Manufacture of Ceramic Cores Ceramic coring is widely used in investment casting to produce internal passageways in castings. Investment casting cores are either self-formed (produced in place during the mold-building operation) or preformed (made separately by an appropriate ceramic forming process).

Self-Formed Cores Self-forming requires that the wax patterns already have the openings corresponding to the passageways desired in the castings. For simple shapes, this is accomplished by the use of metal pull cores in the pattern tooling. For shapes in which a simple pull core cannot be extracted, soluble wax cores are made, placed in the pattern tooling, and the pattern injected around them. The soluble core is then dissolved out in a solution that does not affect the wax pattern, such as an aqueous acid. Soluble cores are generally made from a solid polyethylene glycol, with a powdered filler such as sodium bicarbonate or calcium carbonate. The fillers dissolve in the acid with vigorous gas evolution, which provides agitation to speed the dissolution process. Where openings are large enough, self-formed cores are made in the normal course of shelling, but where openings are deep or narrow, the shelling process must be modified to accommodate the special requirements of the cores. This may require special attention to slurry viscosity, stucco particle size, and drying. Vacuum dipping is an excellent tool in this case. Ethyl silicate/ alcohol slurries reduce drying problems. The self-forming of cores is extensively practiced for small hardware castings, for which the cost of a preformed core would be prohibitive.

Preformed Cores Self-formed cores have severe limitations, which are overcome by using preformed cores. Preformed cores require separate tooling and can be produced by a number of ceramic forming processes. Simple tubes and rods are commonly extruded from silica glass. Many cores, especially larger ones, are produced from ethyl

silicate slurries that are cast by gravity or injected under pressure into aluminum dies and then gelled. The technology is similar to ceramic molding processes (see the section “Ceramic Molding” in the article “Slurry Molding” in this Volume). The use of ethyl silicate here is not a problem because the core slurry gels quickly and is then ignited to burn up the fuming compounds). Many cores are made by injection molding; a smaller number, by transfer molding. Ceramic injection molding is similar to the plastics injection molding described earlier, except that a mixture of fine ceramic powder in an organic vehicle is injected, usually into hardened tool steel dies. The organic vehicle, which also functions as a green binder, employs such thermoplastic materials as polyethylene, ethyl cellulose, shellac, resins, waxes, and subliming organic compounds such as naphthalene and paradichlorobenzene. After forming, the cores are subjected to a two-stage heat treatment. In the first stage, the organics are removed without disrupting the core (which is sometimes a lengthy process), and in the second stage, the core is sintered to its final strength and dimensions. Transfer molding is similar to injection molding. The primary difference is that a preform of thermosetting plastic and ceramic powder is softened and injected into a die, where it is cured (thermally set) under heat and pressure. Although a considerable number of materials have been proposed for ceramic cores, they are usually made of fused silica, sometimes with additions of zircon. Fused silica cores provide satisfactory refractoriness and are readily leached from the castings in molten caustic baths, aqueous solutions of caustic in open pots or autoclaves, and hydrofluoric acid solutions. Preformed cores are usually used by placing them in the pattern die and injecting wax around them. For some simple shapes, the cores are inserted into the patterns immediately after the patterns are injected. Cores expand differently from the shell molds in which they are used because of differences in composition and heating rates; therefore, cores of any size must be provided with slip joints in the mold (Ref 78). The factors to be considered in using ceramic cores are discussed in Ref 78.

Pattern Removal Pattern removal is often the operation that subjects the shell mold to the most stresses, and it is a frequent source of problems. These arise from the fact that the thermal expansions of pattern waxes are many times those of the refractories used for molds (compare Fig. 3 and 4). When the mold is heated to liquify the pattern, this expansion differential leads to enormous pressure that is capable of cracking and even destroying the mold. In practice, this problem can be effectively circumvented by heating the mold extremely rapidly from the outside in. This causes the surface layer of

wax to melt very quickly before the rest of the pattern can heat up appreciably. This molten layer either runs out of the mold or soaks into it, thus providing space to accommodate expansion as the remainder of the wax is heated. Melt-out tips open to the outside are sometimes provided or holes are drilled in the shell to relieve wax pressure. Even with these techniques, the shell is still subjected to high stress; therefore, it should be as strong as possible. It should also be thoroughly dried before dewaxing. To achieve thorough drying, shells are subjected to 16 to 48 h of extended drying after the last coat; this drying is sometimes enhanced by the application of vacuum or extremely low humidity. A number of methods have been developed to implement the surface melting concept (Ref 79), but only two have achieved widespread use: autoclave dewaxing and high-temperature flash dewaxing. In addition, hot liquid dewaxing has found some use among smaller companies seeking to minimize capital investment. Autoclave dewaxing is the most widely used method. Saturated steam is used in a jacketed vessel, which is generally equipped with a steam accumulator to ensure rapid pressurization, because rapid heating is the key to success. Autoclaves are equipped with a sliding tray to accommodate a number of molds, a fast-acting door with a safety lock, and an automatic wax drain valve. Operating pressures of approximately 550 to 620 kPa (80 to 90 psig) are reached in 4 to 7 s. Molds are dewaxed in approximately 15 min or less. Wax recovery is good. Polystyrene patterns cannot be melted out in the autoclave. Flash dewaxing is carried out by inserting the shell into a hot furnace at 870 to 1095  C (1600 to 2000  F). The furnace is usually equipped with an open bottom so that wax can fall out of the furnace as soon as it melts. Some of the wax begins to burn as it falls, and even though it is quickly extinguished, there is greater potential for deterioration than with an autoclave. Nevertheless, wax from this operation can be reclaimed satisfactorily. Flash dewaxing furnaces must be equipped with an afterburner in the flue or some other means to prevent atmospheric pollution. Polystyrene patterns are readily burned out in flash dewaxing, but polystyrene often can cause extensive mold cracking unless it is embedded in wax in the pattern (as in integral nozzle patterns) or unless the polystyrene patterns are very small. Liquid dewaxing requires the minimum investment in equipment. Hot wax at 177  C (350  F) is often used as the medium. Other liquids can also be used. Cycles are somewhat longer than for autoclave and flash dewaxing, and there is a potential fire hazard.

Mold Firing and Burnout Ceramic shell molds are fired to remove moisture (free and chemically combined), to burn off residual pattern material and any

Investment Casting / 655 organics used in the shell slurry, to sinter the ceramic, and to preheat the mold to the temperature required for casting. In some cases, these are accomplished in a single firing. Other times, preheating is performed in a second firing. This permits the mold to be cooled down, inspected, and repaired if necessary. Cracked molds can be repaired with ceramic slurry or special cements. This will not heal cracks but will seal them and reinforce the shell so that it can be successfully cast. Many molds are wrapped with a ceramic-fiber blanket at this time to minimize the temperature drop that occurs between the preheat furnace and the casting operation or to provide better feeding by insulating selected areas of the mold. Gas-fired furnaces are used for mold firing and preheating, except for molds for directional solidification processes, which are preheated in the casting furnace with induction or resistance heating. Batch and continuous pusher-type furnaces are most common, but some rotary furnaces are also in use. Most furnaces have conventional firebrick insulation, but furnaces with ceramic-fiber insulation and luminous wall furnaces (Ref 80) are also in use. The latter two provide very fast heatup and cooldown as well as good fuel economy. Conventional and ceramic-fiber furnaces can be equipped with ceramic recuperators for greater fuel economy and reduced smoke emissions (Ref 81). Good circulation is essential, especially in preheat furnaces, to provide uniform temperatures. Approximately 10% excess air is provided in burnout furnaces to ensure the complete combustion of organic materials. Burnout is commonly conducted between 870 and 1095  C (1600 and 2000  F). Many molds can be placed directly into the hot furnace, but others may have to be loaded at a low temperature and heated gradually. This includes molds made with crystalline silica and molds in which some parts are shielded from radiation. Preheat temperatures vary over a large range above and below burnout temperatures, depending on part configuration and the alloy to be cast. Common ranges are 150 to 540  C (300 to 1000  F) for aluminum alloys, 425 to 650  C (800 to 1200  F) for many copper-base alloys, and 870 to 1095  C (1600 to 2000  F) for steels and superalloys. Molds for the directional solidification process are preheated above the liquidus temperature of the alloy being cast.

Melting and Casting Melting Equipment. In the past, the small carbon arc rollover furnace was the workhorse of the early investment casting industry (Ref 82), but this furnace has been gradually displaced by the induction furnace, which is more flexible and economical to operate. Induction furnaces are used in conjunction with various casting methods developed for the specific requirements of investment casting. Today

(2008), most investment casting uses induction melting. The furnaces used are of the coreless type and generally employ preformed crucibles for melting (or monolithic linings for large sizes). They consist of a water-cooled copper coil that surrounds the melting crucible, a power supply to energize the coil with alternating current, and the appropriate electrical controls. The alternating current flowing through the coil sets up an alternating magnetic field that induces eddy currents in the metal charge. The heating effect is self-generated within the charge as a result of its resistance to the current flow. Earlier models used spark gap converters or motor-generator sets as power supplies, but today (2008), solid-state power supplies are standard. Most induction furnaces for investment casting have capacities ranging from 7.7 to 340 kg (17 to 750 lb). They are usually tilting models, although some small rollover models are used where very rapid mold filling is required. Melting rates of 1.36 kg/min (3 lb/min) are common, but this depends on the relative size of the power supply and the melt. Induction furnaces can be employed for melting in air, inert atmosphere, or vacuum. They are extensively used for melting steel, iron, cobalt and nickel alloys, and sometimes copper and aluminum alloys. The crucibles typically used are magnesia, alumina, and zirconia, which are made by slip casting, thixotropic casting, dry pressing, or isostatic pressing. For vacuum casting, there is some use of disposable liners, which are made by conventional ceramic shell techniques or by the slip casting of fused silica. Gas-fired crucible furnaces and, more recently, electrical resistance furnaces are used for aluminum casting. Gas-fired furnaces using glazed clay-bonded graphite crucibles are satisfactory for aluminum and copper alloy castings, and these furnaces are inexpensive. Silicon carbide crucibles are sometimes used for copper-base alloys. Resistance furnaces eliminate combustion products and help reduce hydrogen porosity. Magnesium can be melted in gas-fired furnaces using low-carbon steel crucibles. Consumable-electrode vacuum arc skull furnaces are used for melting and casting titanium in a process originally developed at the U.S. Bureau of Mines (Ref 83). A billet of the alloy to be cast serves as the consumable electrode and is arc melted under vacuum into a water-cooled copper crucible. The crucible has a permanent, thin skull of solidified alloy on its inner face, and this skull prevents the crucible from contaminating the melt. Molten alloy accumulates in the crucible until there is a sufficient quantity to fill the mold, at which time it is poured. Furnaces of various sizes are available and are capable of melting up to 522 kg (1150 lb). The number of such installations is small because only a few companies make titanium investment castings, but this reflects the technical difficulties involved rather than any lack of a market.

Electron beam melting has been used in Europe as an alternative to vacuum arc melting for the casting of titanium (Ref 84). It is being used by an aircraft engine manufacturer in the United States for melting superalloys for directionally solidified and single-crystal casting (Ref 85). The arrangement is basically similar to that used for vacuum arc skull melting except that the electron beam is used in place of the vacuum arc. Casting Methods. Both air and vacuum casting are important in investment casting. Most vacuum casting is investment casting, although there is some use of rammed graphite molds in vacuum arc furnaces for casting titanium. Most castings are gravity poured. Air casting is used for such commonly investment-cast alloys as aluminum, magnesium, copper, gold, silver, platinum, all types of steel, ductile iron, most cobalt alloys, and nickel-base alloys that do not contain reactive elements. Zinc alloys, gray iron, and malleable iron are usually not investment cast for economic reasons, but if they were, they would be air cast. Vacuum casting provides cleaner metal and often superior properties, and sometimes this is an incentive to cast some of the normally air-melted alloys in vacuum. However, its major use is for alloys that cannot be satisfactorily cast in air, such as the g’-strengthened nickel-base alloys, some cobalt alloys, titanium, and the refractory metals. Batch and semicontinuous interlock furnaces are satisfactory for vacuum casting, but the latter are clearly preferred. A major advantage of investment casting is its ability to cast very thin walls. This results from the use of a hot mold but is further enhanced by specific casting methods, such as vacuum-assist casting, pressurized casting, centrifugal casting, and countergravity casting. In vacuum-assist casting, the mold is placed inside an open chamber, which is then sealed with a plate and gaskets, leaving only the mold opening exposed to the atmosphere. A partial vacuum is drawn within the chamber and around the mold. The metal is poured into the exposed mold opening, and the vacuum serves to evacuate air through the porous mold wall and to create a pressure differential on the molten metal, both of which help to fill delicate detail and thin sections. Rollover furnaces are pressurized for the same purpose. The hot mold is clamped to the top with its opening in register with the furnace opening, and the furnace is quickly inverted to dump the metal into the mold while pressure is applied using compressed air or inert gas. Centrifugal casting uses the centrifugal forces generated by rotating the mold to propel the metal and to facilitate filling. Vacuum arc skull furnaces discharge titanium alloy at a temperature just above its melting point, and centrifugal casting is usually needed to ensure good filling. Dental and jewelry casting use centrifugal casting to fill thin sections and fine

656 / Expendable Mold Casting Processes with Expendable Patterns detail. These three applications represent the major uses for centrifugal casting in investment casting. Countergravity casting is also an excellent method of filling thin sections, and it is not limited to use with investment molds.

Postcasting Operations Postcasting operations represent a significant portion (often 40 to 55%) of the cost of producing investment castings. A standard shop routing is provided for each part, and important savings can be realized by specifying the most cost-efficient routing. For example, it is often cost-effective to detect and eliminate scrap early to avoid wasting finishing time, even if this means including an extra inspection operation. Alternative methods may be available for performing the same operation, and the most efficient one should be selected. The actual sequence in which operations are performed can be important. Some specifications require verification of alloy type, and this is done before parts are removed from the cluster. Knockout. Some of the shell material may spall off during cooling, but a good portion usually remains on the casting and is knocked off with a vibrating pneumatic hammer or by hand. Brittle alloys require special attention. Part of the prime coat sometimes remains adhered to the casting surface, and bulk shell material may remain lodged in pockets or between parts. This is removed in a separate operation, usually shotblasting. Clusters are hung on a spinner hanger inside a blasting cabinet or are placed on a blasting table. If cores are to be removed in a molten caustic bath, the entire cluster can be hung in the bath at this point, and the remaining refractory can be removed along with the cores. High-pressure water blasting (69 MPa, or 10 ksi) is sometimes used instead of mechanical knockout. Cutoff. Aluminum, magnesium, and some copper alloys are cut off with band saws. Other copper alloys, steel, ductile iron, and superalloys are cut off with abrasive wheels operating at approximately 3500 rpm. Torch cutting is sometimes used for gates that are inaccessible to the cutting wheel. Some brittle alloys can be readily tapped off with a mallet if the gates are properly notched. Shear dies have also been used to remove castings from standardized clusters. Following cutoff, gate stubs are ground flush and smooth using abrasive wheels or belts. These are also used for other finishing operations, along with small hand grinders equipped with mounted stones. Core Removal. Where the opening size permits, cores can be removed by abrasive or water blasting. If abrasive or water blasting cannot be used, the cores can be dissolved out. This can be accomplished in a molten caustic bath (sodium hydroxide) at 480 to 540  C (900 to 1000  F), in a boiling solution of 20 to 30% sodium hydroxide or potassium

hydroxide in an open pot, or in the same solutions in a high-pressure autoclave. Hydrofluoric acid can also be used with alloys that are inert to it, such as platinum and many cobalt superalloys. Heat Treatment. Both air and vacuum heat treatments are extensively performed as needed to meet property requirements. The vacuum heat treatment of stainless steels (instead of in air) is sometimes practiced to achieve a clean, bright surface and to avoid the need for any further cleaning. Before they are heat treated, single-crystal castings must be handled very carefully to avoid recrystallization during subsequent heat treatment. They must not be dropped, allowed to hit one another, or subjected to a blasting operation until heat treated. Abrasive Cleaning. Blast cleaning is also used to remove scale resulting from core removal or heat treatment. Both pneumatic and airless (centrifugal) blasting machines are employed. Metallic abrasives (steel or iron grit or shot) and ceramic abrasives (silica sand, aluminum oxide, garnet, and staurolite) are commonly used, depending on the application. The hardness and angular shape of alumina produce rapid cleaning, which is especially cost-effective when work is being blasted by hand. Alumina blasting is very effective at opening up and exposing surface defects, thus facilitating subsequent finishing or inspection operations. However, it is rather expensive, and less costly materials such as garnet, staurolite, or silica sand are often used instead. Metal shot and glass beads provide good peening action and produce shiny surfaces, both of which are sometimes desired. Equipment used for blasting includes hand cabinets as well as automatic units with rotating tables, spinner hangers, or rotating tumblers to hold the work. Another type of abrasive finishing process is mass finishing. In this process, castings are treated with chemical finishing compounds and synthetic abrasives of specific geometric shapes, such as cones, triangles, and stars, to produce deburring, fine finishing, and radii blending in vibratory, tumbling, or centrifugal machines. Miscellaneous Operations. Hot isostatic pressing is becoming increasingly important for the densification of castings to eliminate porosity; to improve fatigue, ductility, and other properties; and to reduce property scatter (see the article “Hot Isostatic Pressing of Castings” in this Volume). Its most important application in investment casting is for titanium. Applications for steel, superalloys, and aluminum are more selective but still very important. Machining is often performed on investment castings, although the amount may be minimal, and there are many applications in which no machining is required. Machining is generally confined to selected areas requiring closer dimensions than can be achieved by casting, and it is occasionally used to incorporate some detail that is less expensive to machine in than to cast. The machining and welding of investment castings are discussed in Ref 86. Welding

can be used to repair and join castings—primarily large, structural castings. Other methods used to improve dimensional accuracy are broaching and coining, as well as abrasive grinding. Straightening operations are performed when required, either manually or using hydraulic presses, with suitable fixtures. Castings can be heated to facilitate straightening. Chemical finishing treatments are also used. The various treatments include acid pickling to remove scale, passivation treatments for stainless steel, chemical milling to remove the a case on titanium, and chemical treatment to apply an attractive satin finish to aluminum or to polish stainless steel.

Inspection and Testing Alloy Type Test. The first inspection operation performed is often an alloy verification test. This test is conducted on the cluster before cutoff. A spectrometer or x-ray analyzer is used to verify that the correct alloy has in fact been poured. Visual Inspection. An early visual inspection is essential so that obvious scrap does not get passed on to expensive finishing or inspection operations. Some commercial parts require only visual inspection. Liquid fluorescent penetrant inspection is extensively used for nonmagnetic alloys. It can also be used for magnetic alloys, but a magnetic penetrant is generally specified instead. The liquid fluorescent penetrant detects defects on, or open to, the surface, such as porosity, shrinkage, cold shuts, some inclusions, dross, and cracks of any origin (hot tears, knockout, grinding, heat treat, straightening). In this technique, the surface is properly cleaned, and a liquid penetrant with low surface tension and low viscosity is applied and is drawn into the defects by capillary action. Excess liquid is wiped away, a developer is applied that functions as a blotter to draw the liquid out, and the area is examined visually in a dark enclosure under black (near ultraviolet) light, which reveals defects that cannot be detected visually. Applicable specifications include ASTM E 165 and MIL-1-6866. Turbine blades, which represent one of the most demanding applications for investment casting, are often subjected to a thermal cycle before the final fluorescent penetrant inspection; this is done to open very small defects that may otherwise go undetected. Because the defects detected are often only surface defects, they can sometimes be removed by various finishing operations, and such rework operations are commonly attempted. They are generally successful if the defect is not so deep that dimensional or surface finish requirements are violated in removing it. Magnetic particle inspection is used to detect the same types of defects as fluorescent penetrant inspection but is preferred for use on

Investment Casting / 657 ferromagnetic alloys. The method involves surface preparation, magnetization of the casting, and application of either a liquid suspension of magnetic particles (wet method) or fine magnetic iron particles (dry method). The presence of a defect causes a leakage field that attracts the magnetic particles and causes them to cling to the defect and define its outline. Colored particles and fluorescent particles (for viewing under black light) are available. Applicable specifications include ASTM E 125 (Reference Photographs), ASTM E 109 (Dry Particle), ASTM E 138 (Wet Particle), and MIL-STD-1949. X-ray radiography is used to detect differences in material density or thickness to reveal such internal defects as dross, shrinkage, gas holes, inclusions, broken cores, and core shift. Many parts receive 100% inspection; others are inspected according to an appropriate sampling plan. Even when x-ray inspection is not required, it can be used as a foundry control tool to aid in establishing a satisfactory gating system, to determine the effect of process changes, to monitor process reliability, and to troubleshoot foundry problems. X-ray inspection is sometimes used to examine wax patterns containing delicate ceramic cores to ensure that the cores were not broken during the pattern injection operation. Applicable specifications for castings are ASTM E 192, MIL-C-2175, and MIL-STD-453. Real-time x-ray inspection is beginning to be used in the larger investment foundries serving the aircraft industry. Formerly, defects within the casting detected by x-ray radiography could be repaired only by grinding and welding the affected area, but this procedure was not always cost-effective and could not be applied to all alloys or applications. Currently, many castings, including those that cannot be welded, are being repaired by hot isostatic pressing. Miscellaneous Inspection Methods. Hardness testing is widely used to verify the response of castings to heat treatment. Chemical analysis is generally controlled through the use of certified master heats. Mechanical properties are determined on separately cast test bars or test specimens mounted on production clusters. Specimens machined from castings are used for process development and periodic audits. Grain size is regularly checked on many equiaxed castings, following chemical or electrolytic etching. Even where grain inspection is not specified, it is often used as part of the process development effort. Electrolytic etching is also used for examining and detecting grain defects in directionally solidified and single-crystal castings. The orientation of single-crystal castings is determined by Laue back-reflection x-ray diffraction. Pressure tightness tests are conducted for a variety of applications. Dimensional inspection ranges from manual checks with a micrometer or simple go/no-go gages to the use of coordinatemeasuring machines and automatic three-dimensional inspection stations capable of checking a sculptured surface in a continuous sweep. Wall thickness on many cored turbine blade castings is

determined ultrasonically. Nodularity is checked metallographically on the first and last cluster poured from heats of ductile iron. Metallography is an essential part of process development for high-performance castings.

Design Advantages of Investment Castings The challenge in designing for investment casting is to make full use of the enormous capability and flexibility inherent in the process to produce parts that are truly functional and cost-effective. Often such parts will also be more aesthetically pleasing. The principal advantages of investment casting that permit this challenge to be met are discussed in this section. Complexity. Almost any degree of external complexity, as well as a wide range of internal complexity, can be achieved, and in certain cases, the only limitation is the state of the art in ceramic core manufacturing. As a result, parts previously manufactured by assembling many individual components are currently being made as integral castings at much lower costs and often with improved functionality. This ability to produce complexity with ease can even benefit simple parts, which can be redesigned to save weight without loss of strength by providing I- or H-sections, or thin walls with ribs. Thus, many parts from competitive processes can be converted to investment casting. Freedom of Alloy Selection. Any castable alloy can be used, including ones that are impossible to forge or are too difficult to machine. Further, the cost of the alloy is less important in the final price of an investment casting than it is in many other metal-forming processes; therefore, an upgraded alloy can often be specified (especially if the part is redesigned to save weight) at little or no increase in price. Close Dimensional Tolerances. The absence of parting lines and the elimination of substantial amounts of machining by producing parts very close to final size give investment casting an enormous advantage over sand casting and conventional forging. The availability of prototype and temporary tooling is a major advantage in the design and evaluation of parts. The direct machining of wax patterns or the use of any of the quick, inexpensive tooling methods described earlier facilitates timely collaboration between the designer and the foundry to produce parts that are functional and manufacturable. This capability is simply not found in such competitive processes as die casting, powder metallurgy, or forging. Further, temporary tooling used in the design phase can often serve for production while the market is tested or permanent tooling is constructed. Reliability. The long-standing use of investment castings in aircraft engines for the most demanding applications has fully demonstrated

their ability to be manufactured to the highest standards. Wide Range of Applications. In addition to complex parts and parts that meet the most severe requirements, investment casting also produces many very simple parts competitively. This capability is often made possible by the low tooling costs associated with investment casting. Investment castings are competitively produced in sizes ranging from a few grams to more than 300 kg (660 lb), and the upper limits continue to increase.

Design Recommendations The following recommendations provide a useful guide to the design of investment castings:  Look for ways to implement the advantages

mentioned previously.

 Focus on final component cost rather than

casting cost.

 Design parts to eliminate unnecessary hot

    

spots through changes in section sizes , use of uniform sections, location of intersections, and judicious use of fillets, radii, and ribs. Promote good communication between foundry and customer to resolve questions of functionality versus producibility. Use prototype castings to resolve questions of functionality, producibility, and cost. Do not overspecify; permit broader-thanusual tolerances wherever possible. Use the data in Ref 87 as a guide to the dimensional tolerances attainable. Indicate datum planes and tooling points on drawings; follow ANSI Y14.5M for dimensioning and tolerancing.

More detailed information is available in Ref 87 and 88.

Applications Applications for investment castings exist in most manufacturing industries. A partial list is given in Table 3. The largest applications are in the aircraft and aerospace industries, especially turbine blades and vanes cast in cobalt- and nickel-base superalloys as well as structural components cast in superalloys, titanium, and 17-4PH stainless steel. Examples of current applications for investment castings are shown in Fig. 5 to 7. All were cast in ceramic shell molds using wax patterns unless otherwise noted.

Special Investment Casting Processes The investment casting process is highly flexible and can handle a great variety of parts with the same basic method. However,

658 / Expendable Mold Casting Processes with Expendable Patterns Table 3 casting

Some applications of investment

Aircraft engines, air frames, fuel systems Aerospace, missiles, ground support systems Agricultural equipment Automotive Baling and strapping equipment Bicycles and motorcyles Cameras Computers and data processing Communications Construction equipment Dentistry and dental tools Electrical equipment Electronics, radar Guns and small armaments Hand tools Jewelry Machine tools Materials handling equipment Metal working equipment Oil well drilling and auxiliary equipment Optical equipment Packaging equipment Pneumatic and hydraulic systems Prosthetic appliances Pumps Sports gear and recreational equipment Stationary turbines Textile equipment Transportation, diesel engines Valves Wire processing equipment

specialized versions of the process can be highly effective in reducing costs on particular types of parts. Two variations that are being increasingly applied are the Shellvest system and the Replicast process. The Shellvest system provides a unique method of manufacturing small parts in high quantities at low cost. To provide large numbers of patterns at high production rates and low cost, automatic plastic injection machines are used, along with a special plastic/wax pattern material. Wax patterns are used when production requirements do not justify plastic tooling. Pattern assemblies are made by attaching the patterns to large-diameter horizontal cardboard drums instead of conventional wax sprues. As shown in Fig. 8, the drum has a through handle for handling and for mounting in the moldmaking equipment. Each end of the drum is closed with an end plate, one of which is removable. The drum is wrapped with a thin layer of cardboard having corrugations on its inner surface and a smooth outer surface. The corrugated cardboard is attached with masking tape. Over the cardboard is a layer of thin, perforated paper called a gate master. As shown in Fig. 8, the gate master has markings printed on it that indicate the cross section of each pattern gate and exactly where it is to be mounted on the drum. This greatly facilitates the operation of mounting patterns on the drum and ensures proper spacing and orientation. The completed drum assembly is rotated through a bath of molten wax to apply a thin wax coating to which the patterns are attached.

(a)

(b)

(c)

(d)

Fig. 5

Some aircraft and aerospace applications for investment castings. (a) Single-crystal turbine blades investment cast using complex ceramic cores. Courtesy of Pratt and Whitney Aircraft. (b) 17-4-PH stainless steel fan exit case; weight: 96 kg (212 lb). Courtesy of Precision Castparts Corporation. (c) Aircraft fuel sensor strut cast in 17-4-PH stainless steel. Both ceramic and soluble cores were used to produce the complex internal passages. Courtesy of Northern Precision Casting Company. (d) Aircraft combustion chamber floatwall. The large number of small posts required high mold and pouring temperatures. Courtesy of Pratt and Whitney Aircraft

Investment Casting / 659

(a)

Fig. 6

(b)

(c)

Biomedical applications for investment castings. (a) Whiteside hip-femoral prosthesis. (b) Whiteside II-C knee-tibial base. (c) London elbow-humeral prosthesis. All cast in ASTM F75 cobalt-chromium-molybdenum alloy; all courtesy of Dow Corning Wright

(a)

(b)

(c)

Fig. 7

Miscellaneous applications for investment castings. (a) Nosepiece for nailgun cast in 8620 alloy steel. The part is used as-cast. Courtesy of Northern Precision Casting Company. (b) Ni-Resist Type II cast iron inducer for deep well oil drilling. Courtesy of Bimac Corporation. (c) Small 17-4-PH turbine vanes; the smaller weighs 71 g (2.5 oz); the larger, 185 g (6.5 oz). Courtesy of K.W. Thompson Tool Company, Inc.

Cylinder assembly

Corrugated cardboard

16

Detail A

Pattern location (1 of 80)

Cardboard cylinder 25

1 8

End plate End plate

(b) Placement of pattern layout on cylinder assembly

(a) Cylinder assembly

Detail A (c) Flat pattern layout (1 - in.grid) 8

1 8

(d) Sprue assembly (cylinder assembly and wax-coated pattern layout) Pattern layout ( 1 -in.grid 8 not shown)

Wax pattern (1 of 80)

Detail B

7 16

1 by 45° 16 chamfer End plate

11 16

1 4

19 132

Wax pattern (1 of 80)

17 diam 64 29 132 (e) Detail of wax pattern

Fig. 8

Pattern layout

(f) Setup (wax patterns attached to sprue assembly)

Cardboard cylinder Corrugated cardboard Detail B

Schematic showing the Shellvest system for investment casting of small parts. Dimensions given in inches

Individual patterns are attached by wax welding. Splitters can be used that divide the drum into several segments so that up to four molds can be formed simultaneously (Ref 89). The completed pattern assembly is mounted horizontally in a vacuum slurry tank. The

assembly is lowered into the tank so that its lower surface is immersed in the ceramic slurry. Here, it rotates through the slurry while the chamber is under sufficiently high vacuum to boil the water in the slurry. The combination of rotation and vacuuming flushes air out of

the system and enables the slurry to enter and coat the finest detail on the patterns, including core openings. After the coating operation is complete, the assembly is raised out of the slurry and rotated in air to drain excess slurry back into the bath.

660 / Expendable Mold Casting Processes with Expendable Patterns It is then spun to provide a very uniform coating, after which it is immediately transferred to a fluidized bed for stuccoing. Here, it is again mounted horizontally and rotated so that its bottom surface passes through the fluidized stucco particles. After stuccoing, the assembly is dried, and the sequence is then repeated four to six times to build up the required shell thickness. When shelling is complete, one end plate is removed, and the cardboard tube is pulled out. Next the corrugated cardboard liner and the gate master paper are removed, and the setup readily separates into individual molds according to the number of splitters used. Each mold includes a cylinder with a large number of individual pattern cavities gated into its surface. The molds are dewaxed in a conventional steam autoclave. This operation is greatly facilitated by the fact that the gate of each pattern is immediately accessible to the steam. This provides very rapid melting and release of pressure, so that the plastic/wax material is readily removed without cracking the shell. The ceramic shell mold is prepared for casting by firing it at the required preheat temperature, thus combining burnout and preheating. The mold is then placed over a ceramic-coated resin-bonded sand core in a vacuum-assist casting furnace (Ref 90). The diameter of the core is selected so that the annular space between the mold and the core will contain sufficient metal to feed all of the castings that are gated into this area. This is called a hollow-sprue gating system. It permits the use of large-diameter sprues carrying large numbers of parts, while still maintaining a very favorable ratio of gating metal to castings. Casting is carried out by the vacuum-assist method described previously. After casting, the resin-bonded sand core disintegrates from the heat of the metal and falls out. The casting then proceeds through the conventional finishing and inspection operations. The Replicast process uses conventional investment molding techniques to form a thin ceramic shell mold around a foamed polystyrene pattern of the type used in the lost-foam casting process. ACKNOWLEDGMENT Revised from R.A. Horton, Investment Casting, Casting, Volume 15, ASM Handbook, ASM International, 1988, p 253–269 REFERENCES 1. Forging and Casting, Vol 5, Metals Handbook, 8th ed., American Society for Metals, 1970, p 238 2. C.H. Schwartz, Coated Stucco Solves Ceramic Shell Drying Problems, Mod. Cast., June 1987, p 31 3. C.H. Watts, Refractory Containing Investment and Method of Making, U.S Patent 5,310, 420, May 1994

4. T.W. Snyder, Fit for a Casting King, MJSA J., Dec 2007, p 36 5. R.A. Horton and C.H. Watts, High Temperature Investment Material and Method for Making Solid Investment Molds, U.S. Patent 6,746,528, June 2004 6. R.A. Horton, “Formulating Pattern Waxes,” Paper presented at the 31st Annual Meeting, Oct 1983, Investment Casting Institute 7. P. Solomon, Disposable Pattern, Composition for Investment Casting, U.S. Patent 3,887,382, 1975 8. P. Solomon, Disposable Pattern, Composition for Making Same and Method of Investment Casting, U.S. Patent 3,754,943, 1973 9. J. Booth, Pattern Waxes for Investment Casting, Foundry Trade J., Dec 6, 1962, p 707 10. E. Faulkner, F. Hocking, and J. AherneHeron, Wax Emulsion for Use in Precision Casting, U.S. Patent 3,266,915, Aug 1966 11. D. Mills, “Second Thoughts on Wax Testing and Its Relationship to the Injection of Investment Patterns,” Paper presented at the 11th Annual Conference, May 1971 (Bournemouth, England), British Investment Casters’ Technical Association 12. S. Rau, “Formulation, Preparation and Properties of Waxes for Wax Patterns,” Paper presented at the Investment Casting Institute Advanced Technical Seminar (Cleveland, OH), Case Institute of Technology, 1963 13. Standard Test Procedures—Pattern Materials, Investment Casting Institute, 1979 14. Acceptance Tests for Wax Pattern Materials, British Investment Casters’ Technical Association, 1979 15. H. Bennet, Test Methods Applicable to Various Waxes (ASTM, Amerwax, USP), Industrial Waxes, Vol II, Chemical Publishing, 1975, p 297 16. M. Koenig, “Quality Control of Investment Casting Waxes by Thermal Analysis,” Paper presented at the 25th Annual Meeting, Investment Casting Institute, 1977 17. J. Roberts, “Standardized Wax Testing Die,” Paper presented at the 28th Annual Meeting, Investment Casting Institute, 1980 18. R. Bird, R. Brown, H. Eck, and W. Wurster, “Reports on the Use of the Institute’s Test Die,” Paper presented at the 30th Annual Meeting, Investment Casting Institute, 1982 19. State of the Investment Casting Industry, Precis. Met., Nov 1982, p 39 20. H.S. Strache, “In-Line Production of Precision Castings up to 100 kg in Weight,” Paper presented at the 12th European Investment Casters’ Conference, Sept 1967 (Eindhoven, Holland) 21. K. Suzuki and O. Hiraiwa, U.S. Patent 3,131,999, 1964 22. K. Ugatta, Y. Morita, and Y. Mine, Investment Casting Method, U.S. Patent 3,996,991, 1976

23. R. Harrington and B.Drugan, “FOPAT— The Next Generation Pattern Material,” Paper presented at the 54th annual Meeting, Oct 2006, Investment Casting Institute 24. Wax Injection Trouble Shooting Guide, Investment Casting Institute, 1979 25. Atlas of Pattern Defects, Investment Casting Institute, 1983 26. C.S. Treacy, Some Notes on Waxes for Investment Castings, Investment Casting Institute, 1964 27. V. Stanciu, Quality of Wax Patterns as a Function of Wax Preparation and Injection, European Investment Casters’ Federation, 1970 28. I. Malkin, The Effect of Die Temperature and Wax Pressure on Cavitation of Wax Patterns, Investment Casting Institute, 1969 29. P.L. Wilkerson, Improved Patternmaking Techniques in Investment Casting, Precis. Met. Mold., Oct 1974, p 40 30. J. Hockin, “Wax Pattern Dimensional Variation in Automatic Injection Machines,” Paper presented at the 27th Annual Meeting, Investment Casting Institute, 1979 31. Wax Injection Technology: A Guide to the Influence Injection Variables Have on Pattern Contraction, Blayson-Olefines, 1967 32. C.H. Watts and M. Daskivich, Process and Material for Precision Investment Casting, U.S. Patent 3,263,286, 1966 33. R.A. Horton and E.M. Yaichner, Pattern Material Composition, U.S. Patent 4,064,083, 1977 34. Moulds for Prototype Production, Foundry Trade J., Aug 6, 1964, p 167 35. W.F. Davenport and A. Strott, How to Set Up a Precision Casting Foundry, Iron Age, April 1951, p 90 36. I. Thomas and E. Spitzig, When and How to Hob, Mod. Plast., Feb 1955, p 117 37. S. Richards, “Die Design for Productivity: Multi-Cavity Dies,” Paper presented at the 26th Annual Meeting, Oct 1978, Investment Casting Institute 38. F.L. Siegrist, Have You Considered Electroforming?, Met. Prog., Oct 1964 39. F.L. Siegrist, Electroforming with Nickel— A Versatile Production Technique, Met. Prog., Nov 1964 40. W.C. Jenkin, Gas Plating for Molds and Dies, Plast. World, March 1967, p 82 41. N. Wood, Epoxies Key to Complex, Accurate Patterns, Foundry, Nov 1968, p 149 42. H.T. Bidwell, Patternmaking, Investment Casting, Machinery Publishing, 1969 43. Rubber Molds Reduce Wax Pattern Tooling Costs, Precis. Met., March 1979, p 66 44. R.A. Horton, J.H. Simmons, and T.R. Bauer, Method of Making Tooling, U.S. Patent 4,220,190, 1980 45. R. Schoeller, Building Injection Molds by Casting, Plast. Des. Process., Aug 1975, p 13 46. Cast Dies for Investment Patterns, Foundry, April 1974, p 28 47. D. Draper, T.L. Cloniger, J.M. Hunt, J.F. Holmes, T. Mersereau, and M. Hosler,

Investment Casting / 661

48.

49.

50. 51. 52.

53.

54.

55.

56.

57.

58.

59. 60. 61.

Laser Welding of Wax Patterns for Investment Casting, Mod. Cast., Oct 1987, p 24 F. Ellin, “Automatic Pattern Assembly Machine,” Paper presented at the 33rd Annual Meeting, Investment Casting Institute, 1985 B. Phipps, “Rethinking the Investment Casting as One Automated System,” Paper presented at the 51st Annual Meeting, Nov 2003, Investment Casting Institute T.V. Rama Prasad and V. Kondic, “Basic Elements of Feeding Investment Castings,” American Foundarymen’s Society, 1994 H. Present and H. Rosenthal, Feeding Distance of Bars in Investment Molds, Trans. AFS, Vol 69, 1961, p 138 E.M. Broard et al., “Feeding Distance of 410 Stainless Steel Cast in Phosphate Bonded Solid Investment Molds,” Subcommittee Report on Casting and Solidification, May 1963, Investment Casting Institute H.D. Brody, “Numerical Analysis of Heat and Fluid Flow in Sand and Investment Castings,” Paper presented at the Solidification Processing Seminar, June 1983 (Dedham, MA), U.S.-Japan Cooperative Program A.A. Badawy, P.S. Raghupathi, and A.J. Goldman, Improving Foundry Productivity with CAD/CAM, Comput. Mech. Eng., July/Aug 1987, p 38 D.M. Wate and M.T. Samonds, “Finite Element Solidification Analyses of Investment Casting,” 38th Annual Technical Meeting, Investment Casting Institute, 1990 T.S. Piwonka, “Modeling Investment Casting Solidifications—Opportunities for Concurrent Engineering,” Paper 20, Eighth World Conference on Investment Costing, 1993 A.S. Sabau, “Numerical Simulation of the Investment Casting Process,” Paper 05–160, Transactions of the American Foundary Society, Vol 113, 2005 B. Dzugan and A.S. Sabau, “Pattern Tooling and Casting Dimensions for Investment Casting,” 54th Technical Conference, Investment Casting Institute, 2006 J. Wachtman, Jr, Ed., “Ceramic Innovation in the 20th Century, The American Ceramic Society, 1999 Ceramic Test Procedures, Investment Casting Institute, 1979 R.C. Feagin, “Casting of Reactive Metals into Ceramic Molds,” Paper presented at the Sixth World Conference on Investment Casting (Washington), 1984

62. D.J. Ulaskas, “Hybrid Binder Systems,” Paper presented at the 29th Annual Meeting (Scottsdale, AZ), Investment Casting Institute, 1981 63. K.A. Buntrock, “Development Work in Aluminosilica, Colloidal Alumina and Polysilicate Binders,” Paper presented at the 29th Annual Meeting (Scottsdale, AZ), Investment Casting Institute, 1981 64. R.C. Feagin, “Alumina and Zirconia Binders,” Paper presented at the 29th Annual Meeting (Scottsdale, AZ), Investment Casting Institute, 1981 65. R.A. Horton, R.L. Ashbrook, and R.C. Feagin, Making Fine Grained Castings, U.S. Patent 3,019,497, 1962 66. R.A. Horton, R.L. Ashbrook, and R.C. Feagin, Making Fine Grained Castings, U.S. Patent 3,157,926, 1964 67. “Mason Nucleating Compounds,” Data sheet, Engineered Materials, Inc. 68. R.L. Rusher, Strength Factors of Ceramic Shell Molds, Part I, AFS Cast Met. Res. J., Dec 1974, p 149 69. R.L. Rusher, Strength Factors of Ceramic Shell Molds, Part II, AFS Cast Met. Res. J., March 1975 70. M.O. Roberts, “Factors Affecting Shell Strength,” Paper presented at the 25th Annual Meeting, Oct 1977, Investment Casting Institute 71. J.D. Jackson, “Evaluation of Various Mold Systems at High Temperature,” Paper presented at the 26th Annual Meeting Oct 1978, Investment Casting Institute 72. P.R. Taylor, “A Rationalized Approach to Slurry Composition Control for Ceramic Shell Moulding,” Paper presented at the Fourth World Conference on Investment Casting (Amsterdam), Investment Casting Institute, British Investment Casters’ Technical Association, European Investment Casters’ Federation, 1976 73. P.R. Taylor, “Quality Control of Slurry Materials for Ceramic Shell Moulding,” Paper presented at the 26th Annual Meeting (Phoenix, AZ), Investment Casting Institute, 1978 74. P.R. Taylor, “Problems of Primary Coat Quality for Control of Ceramic Shell Molds,” Paper presented at the 27th Annual Meeting, Investment Casting Institute, 1979 75. R.A. Horton, Methods and Materials for Treating Investment Casting Patterns, U.S. Patent 3,836,372, 1974

76. L.M. Dahlin, Colloidal Silica Systems for Investment Shells, Precis. Met., Jan 1969, p 81 77. M.G. Perry and A.J. Shipstone, “Research into the Drying of Ceramic Shell Moulds,” Paper presented at the Second World Conference of Investment Casting, June 1969, Investment Casting Institute, British Investment Casters’ Technical Association, European Investment Casters’ Federation 78. S. Uram, “Commercial Applications of Ceramic Cores,” Paper presented at the 26th Annual Meeting, Oct 1978, Investment Casting Institute 79. P. Breen, “The Dewaxing and Firing of Ceramic Shell Moulds,” Paper presented at the 12th Annual Conference, May 1973 (Eastbourne, England), British Investment Casters’ Technical Association 80. J.R. Keough, “An Innovative Approach to Mold Firing,” Paper presented at the 33rd Annual Meeting, Investment Casting Institute, 1985 81. L.R. Ceriotti and J.J. Przyborowski, “Use of Ceramic Recuperators on Investment Casting Mold Burnout Furnaces,” Paper presented at the 33rd Annual Meeting, Investment Casting Institute, 1985 82. A.W. Merrick, Founding Apparatus and Method, U.S. Patent 2,125,080, 1938 83. R.E. Beal, F.W. Wood, J.O. Borg, and H.L. Gilbert, “Production of Titanium Castings,” Report RI 5625, U.S. Bureau of Mines, Aug 1956 84. P. Magnier, “Titanium Investment Castings for Aircraft Applications,” Paper presented at the 12th Annual Conference, May 1973 (Eastbourne, England), British Investment Casters’ Technical Association 85. J.D. Jackson and M.H. Fassler, “Developments in Investment Casting,” Paper presented at the 33rd Annual Meeting, Investment Casting Institute, 1985 86. H.T. Bidwell, Machining Investment Castings, Machinery, Jan 1970, p 60 87. H.T. Bidwell, Design and Application of Investment Castings, Investment Casting, Machinery Publishing, 1969 88. R.H. Herrmann, Ed., How to Design and Buy Investment Castings, Investment Casting Institute, 1959 89. C.H. Watts and R.A. Horton, Method of Investment Casting, U.S. Patent 3,424,227, 1966 90. C.H. Watts and R.A. Horton, Methods of Casting, U.S. Patent 3,336,970, 1987

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 662-664 DOI: 10.1361/asmhba0005256

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Replicast Molding THE REPLICAST PROCESS is a patented process exclusive to Casting Technology International (United Kingdom). The process can be best characterized as a hybrid of the investment casting process in terms of molding but with an expendable pattern made from highquality expanded polystyrene (EPS) as in lost foam (instead of the lost wax pattern of investment casting). The EPS pattern is coated in ceramic slurry and then fired to produce the ceramic mold. The firing process also evaporates the EPS pattern from the mold encasement before casting. This prevents carbon pickup from the EPS pattern during solidification and allows a wider range of alloys to be cast in the mold. The Replicast process was originally developed by the Steel Castings Research and Trade Association (Ref 1) to overcome several shortcomings of the early evaporative pattern (lost foam) casting processes—primarily the formation of lustrous carbon defects in the castings and carbon pickup in steel castings. Because the foam is 92% C by weight, the lost foam process is unsuitable for the majority of steel alloys. By removing the foam pattern before casting, the Replicast process system can be used for even ultralow-carbon stainless steel. The process is commonly used to produce carbon, low-alloy, and stainless steel castings (Fig. 1). This process also offers a higher weight range than is available from investment casting. Other advantages include:
Air emissions are easier to control than with

lost foam.


The application of a vacuum during casting

allows improved fill-out of the mold.


The support provided by the ceramic shell dur





ing casting allows large, thin shells to be easily poured. Sand inclusions and other sand mold-related defects can be virtually eliminated. As with investment and lost foam casting, there are no cores or parting lines, high dimensional accuracy, and excellent surface finish. The ceramic shell does not have to be as thick as for shell casting. The technique minimizes dust emissions from molding and finishing, as compared to sand molding.

Limitations are:
It is not as suitable for thin sections as the

lost wax process.


It is more expensive than the lost foam


Stop the steam and run water through cool-

ing lines


Open the mold box
Eject pattern from cavity

process.

Process Details The Replicast process has two variations:
Replicast FM (full mold), which is nearly

identical to the lost foam process


Replicast CS (ceramic shell), which is simi-

lar to investment casting in that it employs a ceramic shell to surround the pattern The ceramic shell (2 to 3 mm, or 0.08 to 0.12 in., thickness), based on ethyl silicate and refractory sand, is hardened using ammonia and sintered at 1000  C (1830  F). Figure 2 shows a flow chart of the Replicast CS process, and Table 1 (Ref 2) compares the essential features of the Replicast CS and investment casting processes. An integrated Replicast system is shown in Fig. 3. Pattern Production. Aluminum tooling must be developed in order to produce the EPS pattern. The tooling is made from cast aluminum and incorporates risers, cored passages, and so on, wherever possible. The necessity of incorporating Styrofoam fill openings, vents, steam, air, and water lines makes the tooling complex and more expensive than metal green sand equipment. However, the use of insert-type tooling can significantly reduce tooling costs. Patterns are produced from size “T” high-density polystyrene beads. The beads are preexpanded using steam or vacuum, passed into the aluminum tooling, and expanded again to fill the tool cavity and to bond with each other. The high-density EPS used in the Replicast process yields a better casting surface finish than the lowdensity material used for lost foam patterns. The following steps are required for the production of EPS patterns:
Close and hydraulically clamp molding box

halves


Blow in preexpanded Styrofoam beads
Close off Styrofoam fill guns, then inject

steam to expand and bond the beads

Fig. 1

Two applications of Replicast steel parts. (a) CF8M stainless steel 150 mm (6 in.) butterfly valve body. The part is approximately 305 mm (12 in.) in outside diameter, 64 mm (2½ in.) thick, and weighs 11 kg (25 lb). Note as-cast bolt holes and O-ring groove. (b) 8640 steel drive sprocket for armored personnel carrier. The part is 510 mm (20 in.) in outside diameter, 114 mm (4½ in.) thick, and weighs 28 kg (62 lb). Sprocket teeth are used as-cast, resulting in a significant savings in machining costs. Courtesy of Missouri Precision Castings

Replicast Molding / 663 Table 1 Essential features of the Replicast ceramic shell (CS) and investment casting processes

Design and manufacture aluminum tooling for pattern molding

Feature

Pattern manufacture

Mold EPS pattern

Shell manufacture

Coat pattern with primary and secondary ceramics

Pouring

Fire to remove EPS and to strengthen ceramic shell

Replicast CS

Investment casting

Partially expanded EPS beads are blown into aluminum tooling and completely expanded. Finished patterns are lightweight, have high density, and provide good surface finish and excellent dimensional accuracy. Successive coats of refractory slurry and stucco are applied. Three or four coats are required, resulting in a relatively light and easy-to-handle shell. Firing at 925–1000  C (1700–1830  F) for 5 min removes the EPS pattern and hardens the shell. Thin ceramic shell is surrounded by loose sand vibrated to maximum bulk density, and the vacuum is applied during pouring to prevent shell breakage.

Softened wax is injected at high pressure into a metal tool. The wax is subject to shrinkage and deformation, and it is expensive and heavy. It is reclaimable to some degree. Successive coats of refractory slurry and stucco are applied. Five to ten coats are required; completed shells are often heavy and difficult to handle. Firing at 1000  C (1830  F) for 20 min removes the residual wax and hardens the shell. Metal is frequently poured into hot, unsupported shells; breakage is possible.

EPS, expanded polystyrene. Source: Ref 2

Compact sand around shell in vacuum flask

Apply vacuum throughout flask

Pour

Remove vacuum, shake out casting

Clean Nondestructive testing and inspection if required Ship

Fig. 2

Flow chart of the Replicast ceramic shell process. EPS, expanded polystyrene. Source:

Ref 2

The finished EPS pattern has a density of 40 to 48 kg/cm3 (2.5 to 3.0 lb/ft3) and a smooth, hard surface. Minimal shrinkage of the EPS pattern occurs with age. Coating the EPS pattern with ceramic immediately after fabrication produces no detrimental effects. Pattern Assembly. In some cases, the EPS pattern is complete as-molded and is sent to the ceramic coating area. More frequently, additional work is needed to produce a complete pattern assembly. Risers can be glued to the pattern, and ceramic pouring cones can be added. Gluing is done by spray, brush, or glue gun. A low-ash-content hot-melt glue is used for brush or gun application. A low-melting wax is used to fill and smooth the glue joints.

Figure 4 shows completed pattern assemblies ready for the ceramic coating area. Ceramic Coating and Firing. The next step in the Replicast process is the formation of a thin ceramic shell over the EPS pattern assembly. This is done by using techniques similar to those found in investment casting. The pattern assembly is first dipped into a ceramic slurry and then stuccoed using a granular refractory in a fluidized bed or rain sander. The shell is allowed to air dry for a minimum of 1 h. This process is repeated until the desired shell thickness is achieved. Thickness depends on product size, shape, and section thickness; shell thicknesses of 3.2 to 4.8 mm (1/8 to 3/16 in.) are common. Completed shells are then fired at 925 to 1000  C (1700 to 1830  F) to remove the EPS pattern material completely and to harden the ceramic. The shell is then embedded in an unbonded sand mold to support the thin ceramic shell and to prevent breakage. This allows the use of shells thinner than those employed for investment casting. Pouring. The completed ceramic shells are transported to the foundry, where several shells are placed into a molding flask. The flask incorporates a plenum chamber through which air can be extracted to create a vacuum. The flasks are filled with unbonded loose sand, which is vibrated to achieve maximum bulk density. A vacuum is applied to the flask immediately before pouring. Pouring is then accomplished using conventional techniques. The vacuum can be switched off a few minutes after pouring. Cleaning. After cooling, the castings are shaken out and moved to a cleaning area. The brittle ceramic shell fractures easily and tends to break off the casting surface. After initial

abrasive blast cleaning, conventional cleaning techniques are implemented.

Process Capabilities Outstanding features of the Replicast process include:
Surface finish comparable to that obtainable

in conventional investment casting


Elimination of cores through the use of core

inserts in patternmaking tooling; patterns can often be produced in one piece, including risers
Improved casting yields because of the elimination of downsprues, runners, and so on
High as-cast quality levels with regard to casting integrity and dimensional accuracy Dimensional Accuracy. The dimensional accuracy of Replicast parts is comparable to that of investment castings. Based on measurements of Replicast parts of various sizes, the following tolerances on linear dimensions are recommended: Dimension(a), mm (in.)

2.5–102 (0.1–4)

104–305 (4.1–12) 307–610 (12.1–24)

Tolerance, mm (in.)

þ0.25 (þ0.010) or 0.75%   of dimension, whichever is greater þ0.76 (þ0.030)   þ1.52 (þ  0.060)

(a) Nonrisered surfaces

With no wear occurring on tooling, long-term dimensional reproducibility is readily obtainable. The smooth surface finish of Replicast parts, combined with near-zero draft molding capabilities, can result in the elimination of

664 / Expendable Mold Casting Processes with Expendable Patterns

Fig. 4

Completed expanded polystyrene pattern assemblies ready for ceramic coating. Courtesy of Missouri Precision Castings

considerable machining. Cast bolt holes, virtually flat flange faces, and blind and through holes as small as 9.6 mm (3/8 in.) in diameter can be attained. For those surfaces requiring finish machining, an allowance of 1.6 to 3.2 mm (1/16 to 1/8 in.) is usually sufficient. ACKNOWLEDGMENT Revised from R.E. Grote and T.S. Piwonka, Replicast Process, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 270–272 REFERENCES 1. M.C. Ashton, S.G. Sharman, and A.J. Brookes, The Replicast FM (Full Mold) and CS (Ceramic Shell) Process, Trans. AFS, Vol 92, 1984, p 271–280 2. R.E. Grote, Replicast Ceramic Shell Molding: The Process and Its Capabilities, Trans. AFS, Vol 94, 1986, p 181–186

Fig. 3

Integrated Replicast casting system

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 667-673 DOI: 10.1361/asmhba0005257

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Centrifugal Casting Sufei Wei, The Centrifugal Casting Machine Company, and Steve Lampman, ASM International

CENTRIFUGAL CASTING is one of the largest casting branches in the casting industry, accounting for 15% of the total casting output of the world in terms of tonnage. Centrifugal casting was invented in 1918 by the Brazilian Dimitri Sensaud deLavaud, after whom the process was named. DeLavaud’s invention eliminated the need for a central core in the pipe mold, and the mold was water cooled, allowing for a high rate of repeated use. The technique uses the centrifugal force generated by a rotating cylindrical mold to throw molten metal against a mold wall to form the desired shape. Therefore, a centrifugal casting machine must be able to spin a mold, receive molten metal, and let the metal solidify and cool in the mold in a carefully controlled manner. All metals that can be cast by static casting can be cast by the centrifugal casting process, including carbon and alloy steels, high-alloy corrosion- and heat-resistant steels, gray iron, ductile and nodular iron, high-alloy irons, stainless steels, nickel steels, aluminum alloys, copper alloys, magnesium alloys, nickel- and cobalt-base alloys, and titanium alloys. Nonmetals can also be cast by centrifugal casting, including ceramics, glasses, plastics, and virtually any material that can be made into liquid or pourable slurries. Centrifugal castings can be best described as isotropic, that is, having equal properties in all directions. This is not true of a forging, rolling, or extrusion. The centrifugal technique is used primarily for the production of hollow components, but centrifugal casting is used to create solid parts. The centrifugal casting process is generally preferred for producing a superior-quality tubular or cylindrical casting, because the process is economical with regard to casting yield, cleaning room cost, and mold cost. The centrifugal force causes high pressures to develop in the metal, and it contributes to the feeding of the metal, with separation from nonmetallic inclusions and evolved gases. In centrifugal casting of hollow sections, nonmetallic inclusions and evolved gases tend toward the inner surface of the hollow casting. By using the outstanding advantage created by the centrifugal force of rotating molds, castings of high quality and integrity can be produced because of their high

density and freedom from oxides, gases, and other nonmetallic inclusions. When casting solid parts, the pressure from rotation allows thinner details to be cast, making surface details of the metal-cast components more prominent. Another advantage of centrifugal casting is the elimination or minimization of gates and risers.

Centrifugal Casting Methods Centrifugal casting machines are categorized into three basic types based on the direction of the spinning axis: horizontal, vertical, or inclined. Centrifugal casting processes also have three types (Fig. 1):
True centrifugal casting (horizontal, vertical,

or inclined)


Semicentrifugal (centrifugal mold) casting
Centrifuge mold (centrifugal die) casting

The latter two methods are only done with vertical spinning. Horizontal centrifugal casting is mainly used to cast pieces with a high length-to-diameter ratio or with a uniform internal diameter. Products include pipe, tubes, bushings, cylinder sleeves (liners), and cylindrical or tubular castings that are simple in shape. On the other hand, vertical centrifugal casting is mainly for castings with a low length-to-diameter ratio (except vertically cast extralong rolls) or with a conical diameter. The product range for vertical centrifugal casting machines is wider, because noncylindrical (or even nonsymmetrical) parts can be made using vertical centrifugal casting. All vertical centrifugal castings have more or less taper on their inside diameters, depending on the gravitational (g) force applied to the mold and the casting size. The inclined centrifugal casting machine bears advantages and disadvantages of both horizontal and vertical castings and can be very useful in certain applications. Although both vertical and horizontal methods employ centrifugal force, there are some differences in how the force is applied with respect to the axis of the mold rotation and the speed of the molten metal relative to

the rotating mold. For example, with a vertical mold axis, the resultant force on the liquid is constant. This is not the case in a horizontal mold. The other difference between horizontal and vertical mold orientation is the speed obtained by the molten metal as it spins around the mold. When metal is poured into the horizontally rotating mold, considerable slip occurs between the metal and the mold such that the metal does not move as fast as the rotating mold. To overcome this inertia, the metal must be accelerated to reach the mold rotation speed. This is not a problem in the vertical centrifugal process, where the molten metal reaches the speed of the mold soon after pouring. However, with a vertical mold axis, there is a tendency for the molten metal to form a parabolic shape due to the competing gravitational and centrifugal forces.

True Centrifugal Casting This method, also referred to as just centrifugal casting, is characterized by an outer cylindrical mold with no cores. The process can be vertical, horizontal, or inclined (Fig. 1). The permanent mold is rotated about its axis at high speeds (300 to 3000 rpm), so that the molten metal is forced to the inside mold wall, where it solidifies. The casting is usually very fine grained on the outer diameter, while the inside diameter has more impurities and inclusions that can be machined away. Centrifugal casting is used to produce cylindrical, tubular, or ring-shaped castings. The need for a center core is completely eliminated. Castings produced by this method will always have a true cylindrical bore or inside diameter, regardless of shape or configuration. The bore of the casting will be straight or tapered, depending on the horizontal or vertical spinning axis used. Castings produced in metal molds by this method have true directional heat flow, facilitating a planar solidification front move from the outside of the casting toward the axis of rotation. This method results in the production of high-quality, defect-free castings without shrinkage, which is the largest single cause of defective sand castings.

668 / Centrifugal Casting Horizontal true centrifugal casting


Dimensional flexibility: Horizontal centrifu-

Vertical true centrifugal casting

Molten metal

Gate

Mold Mold

Wheel Drive Vertical semi-centrifugal (centritugal-mold) casting

Inclined true centrifugal casting Mold

Integral tube Flasks

Cope

Drag

Centrifuge casting Molten metal Mold

Casting

Fig. 1

Methods of casting with rotating molds. True centrifugal casting (top row) can be horizontal, vertical, or at an incline. Semicentrifugal and centrifuge die casting are vertical methods with molds that can be designed to produce solid parts.

Centrifugal casting is also the preferred method for Babbitting medium- and thick-wall, half-shell or full-round (nonsplit) journal bearings, because it virtually eliminates porosity and allows close control of the cooling process to promote a strong bond. A disadvantage is the need for more extensive equipment and tooling than for static casting. Minor segregation of the intermetallics can occur across the thickness of the Babbitt, but segregation along the axial length of a statically cast bearing can be more serious and is more difficult to detect. The spinning axis is usually horizontal, but vertical orientation is sometimes employed for unusual sizes (e.g., large diameters or short lengths).

Advantages of centrifugal casting include:
Flexibility in casting composition: Centrifu-

gal casting is applicable to nearly all compositions, with the exception of high-carbon steels (0.40 to 0.85% C). Carbon segregation can be a problem in this composition range.
Wide range of available product characteristics: The metallurgical characteristics of a tubular product are mainly characterized by its soundness, texture, structure, and mechanical properties. Centrifugal castings can be manufactured with a wide range of microstructures tailored to meet the demands of specific applications.

gal casting allows the manufacture of pipes with maximum outside diameters close to 1.6 m (63 in.) and wall thicknesses to 200 mm (8 in.). Tolerances depend on part size and the type of mold used.
Quality: The centrifugal action removes unwanted inclusions, dross, cleaner casting, and material that contains shrinkage, which can be machined away. Class I castings can be produced without the need for upgrading and costly weld repairs.
Properties: Mechanical properties are often superior to those of static castings due to the finer grains resulting from the process, which are of constant size in circumferential and axial directions. Due to cleanliness and finer grain size, good weldability is achieved.

Semicentrifugal (Centrifugal Mold) Casting In semicentrifugal casting (Fig. 1), a mold is rotated around its axis of symmetry. Cast configurations may be complex, determined by the shape of the mold. Molds for semicentrifugal casting often contain cores for production of internal surfaces. Directional solidification is obtained only by proper gating, as in static casting. Castings that are difficult to produce statically can often be economically produced by this method, because centrifugal force feeds the molten metal under pressure many times higher than that in static casting. This improves casting yield significantly (85 to 95%), completely fills mold cavities, and results in a high-quality casting free of voids and porosity. Thinner casting sections can be produced with this method than with static casting. Typical castings of this type include gear blanks, pulley sheaves, wheels, impellers, cogwheels, and electric motor rotors. The centrifugal force is used for slag separation and refilling of melt.

Centrifuge (Centrifugal Die) Casting Centrifugal mold or die casting, also referred to as centrifuge casting, has the widest field of application. This casting method is typically used to produce valve bodies and bonnets, plugs, yokes, brackets, and a wide variety of various industrial castings. In this method (Fig. 1), the casting cavities are arranged about the center axis of rotation like the spokes of a wheel, thus permitting the production of multiple castings. Centrifugal force provides the necessary pressure on the molten metal in the same manner as in semicentrifugal casting. The centrifugal force contributes to metal feed for casting thinner sections and making surface details. Centrifuge casting is often used in conjunction with investment casting. It is also used with

Centrifugal Casting / 669 rubber molds made of silicone rubber with sufficiently higher temperature resistance for repeated use in casting without mold deterioration. Centrifuge casting with rubber molds is also referred to as spin casting. Suitable materials for spin casting include zinc-base alloys, lead-base alloys, tin-base alloys, aluminum, and plastics.

Equipment A centrifugal casting machine must be able to perform six operations accurately and with repeatability:
The machine must be able to accelerate












the mold to a predetermined speed, maintain smooth spinning, and decelerate to a stop in a reasonable time frame. The machine also should be able to change speeds during pour and solidification for some special applications. There must be a way to heat and coat the mold before pouring the molten metal (except the cold mold method for ductile iron pipe). There must be a means to pour the molten metal safely into the rotating mold at a controlled rate, position, and orientation. Once the metal is poured, a proper solidification and cooling rate must be established in the mold to obtain a desired casting microstructure, protect the mold, and achieve required productivity. There must be a means of adding inoculants or fluxes for some special applications. There must be a means of extracting the solidified casting quickly from the mold at elevated temperatures without deforming the casting.

Centrifugal casting machines have realized mechanization and automation basically in all of the aforementioned six tooling areas in many applications. An example of an automated horizontal machine and extractor is shown in Fig. 2. New casting and machine products are still evolving, such as metal-matrix composite castings and magnetic stirring solidification. Centrifugal casting and equipment engineers

Fig. 2

continue to use computers to collect, store, and process the key production, equipment, and product information for high-tech products, so that the production parameters can be optimized and automatically set up in production. Centrifugal casting will continue to be prosperous in the casting industry. Spinners. Spinning of the mold is realized by spinners. The achievable spinning speed and, sometimes, the ability to change speeds during the pour and solidification process is an important consideration for determining whether a casting geometry or a casting microstructure can be achieved. Spinning speeds are chosen based on the centrifugal force requirement, which is measured by the multiple of the gravitational force (called g force). Horizontal casting machines are typically spun at 45 to 60 g force on the casting inside diameter, and vertical casting machines are typically spun at 75 g force. For some special applications, the spinning speed can be as high as 100 to 200 g force (e.g., horizontal roll castings and heat-resistant steel tube castings); in other cases, it can be as low as a fraction of a g to a few g’s (e.g., when vertically pouring a roll core, vertically pouring solid bars, or horizontal two-speed pouring). Figure 6 in the article “Vertical Centrifugal Casting” in this Volume provides a convenient chart to check the centrifugal force against casting diameters and spinning speeds. In order to achieve the correct g force, the proper motor size is required. Too small a motor cannot meet process requirements for the spinning speed, and too big a motor can have an excessive acceleration rate that may cause machine vibration. For large horizontal machines, hydraulic motors are preferred over electric motors, because the former offer smoother spinning. Vibration monitors are installed for some large spinners to ensure smooth spinning. The spinners also need to be stopped at a predetermined time, which can be realized by the proper electric, hydraulic, or mechanical means. Most vertical machines use a single spinner, and the mold is spun on top of the spinner. These vertical machines will have less stability than horizontal machines when the mold length-to-diameter ratio is greater than 1.

Automated horizontal casting machine. Courtesy of the Centrifugal Casting Machine Company

Vertical machines with a long mold need trunnion wheels to support the mold around the circumference at the upper section. Small spinners and molds can be clustered to form a turntable, an over-and-under, or a Ferris wheel-type casting machine, which shares the same pouring, heating, coating, and extracting systems, forming a very efficient production cell. Pour equipment and pour rate can be essential pieces of the process for successful production of certain types of castings. Pouring is a critical factor in the following applications, to name a few. The centrifugal ductile iron pipe casting process requires a long, moving trough and quadrant pour ladle to deliver the molten iron along the mold to make 6 to 8 m (20 to 26 ft) long pipe. The centrifugal soil pipe casting process needs a proper chute that can be cleaned and coated in a few seconds after each pour and a consistent pour rate to dash the molten metal to the far end of the mold to form 3 m (10 ft) long pipes with even thickness. Centrifugal heat-resistant steel tube castings rely on a fast pour rate and sufficient pour height to help form the long thin-wall tube before the molten metal gets too cold. Vertical roll casting uses a special-shaped down sprue to prevent the metal flow from whirling. Casting Cooling. Centrifugal castings should be cooled unidirectionally from outside to inside. Any two-way solidification will increase the chance of shrinkage porosity and machining allowance, which should be avoided or minimized in thick-wall castings. Cooling rate can affect the microstructure, casting hardness, circumferential and axial cracks, machine productivity, as well as mold life. In most cases, the early cooling rate of the castings is mainly controlled by the coating thickness, coating texture, coating materials, as well as mold thickness and mold materials; however, the later cooling rate is mainly controlled by water cooling (unless the mold is not cooled by water). Water-cooling methods include water submerge, waterjet spray, and water sleeve. Ductile iron pipe production uses all three water-cooling methods mentioned previously, but other centrifugal castings mainly require waterjet spray, because of its simple setup and easy control of mold temperature. Roll production usually does not require water cooling, because the chill wall thickness is sufficient to absorb heat from the molten metal. Water cooling needs to be consistent, but sometimes, it needs intensity variation along the casting length. For example, the middle section of a long tube mold usually needs more water for cooling. As far as the grain structure of centrifugal castings (e.g., columnar versus equiaxed) is concerned, the pour temperature or the variable spinning speed plays a much more important role in obtaining the equiaxed grains than the water-cooling rate or mold temperature. Water cooling can be critical in certain products, such as high leaded bronze bushings and Babbitt bearings, where intensive water cooling assures that the lead segregation is suppressed.

670 / Centrifugal Casting Some applications require continuous monitoring of casting and mold temperatures by a thermal couple probe inserted into the mold cavity or infrared sensors aimed at the casting inside diameter and mold outside diameter, so that the current process can be stopped and the next process can be started. This is critical, for example, to obtain a good bimetal bonding for roll production. Casting Inoculation and Fluxing. Some castings necessitate the use of inoculants to obtain the desired microstructure (such as ductile iron and gray iron) or grain sizes (such as thick-wall chilled iron or steel tubes). In some applications, fluxes are needed to protect the casting inside diameter in order to obtain a sound metallurgical bimetal bonding (such as in the production of rolls and brake drums) or to prevent premature solidification or oxidation (such as in thick-wall steel castings) of the casting inside diameter. Still others may use fluxes to purify the molten metal during pouring and solidification. All of these would require some means to deliver the inoculants or fluxes into the metal stream during pouring or into the mold cavity before or during solidification. Casting Extraction. Castings need to be extracted from the permanent mold. Small and low-production machines use manual extractors and manpower to extract. Large castings, however, rely on mechanized extractors to do the job. In extracting from horizontal machines, the extractor must overcome the friction acting on the whole interface between the casting and the mold. For vertical machines, the extractor must overcome friction plus casting weight. For some applications, the castings are kept rotating during and/or after extraction to prevent the casting from warping.

Molds Sand molds, semipermanent molds, and permanent molds can be used for the centrifugal casting process. Centrifugal action can also be combined with other molding methods such as investment casting. Selection of the type of mold is determined by the shape of the casting, the degree of quality needed, material to be cast, and the production (number of castings) required. In addition to the processing parameters, proper mold design and selection is vital to producing quality castings. More details on mold design are provided in the articles “Horizontal Centrifugal Casting” and Vertical Centrifugal Casting” in this Volume. Molds consist basically of four parts: the mold body, track grooves or roller tracks on the body (these can be absent if the machine has thrust wheels to hold the mold axially), endplates attached to the mold body, and endplate swing locks or taper pins/wedges. A vertical mold needs fixtures to fasten it onto the adapter table, and the table is bolted onto the spinner shaft. A mold used on a dual-faceplate

horizontal machine can have a mold body without other mold parts. Mold materials include metallic permanent molds, refractory-lined metal molds, sand-lined metal molds, and other materials such as graphite and rubbers. The metallic permanent molds are most widely used because of their reusability, accurate casting geometry, and high productivity. Mold inside diameters are subject to thermal fatigue no matter what mold material is used. The low-carbon and low-alloyed forged steel molds have much longer fatigue lives than cast steel or cast iron molds, and they are also safer to use. However, forged steel molds are more expensive to make than molds made of cast steel and iron. Copper alloys, tool steels, and superalloys are sometimes used for small castings that require high pour temperatures or high cooling rates. Hardened rubber molds are used for some metal jewelry with low melting points. Both horizontal and vertical molds must meet certain geometrical precisions (straightness, smoothness, roundness, concentricity, freedom from internal pores) to minimize machine vibration, which can be critical to safety, bearing life, machine life, and product quality. The mold should also be maintained or repaired during the life of its service. For example, the inside diameter of a ductile iron pipe mold is periodically peened to eliminate tensile stress caused by thermal fatigue. Large grooves or cracks on the mold inside diameter can be repaired by subarc welding and grinding. The mold geometric accuracy can be maintained through remachining and grinding. Mold endplates should be properly designed to prevent molten metal leakage. It cannot be overemphasized that the mold endplate locks and mold fasteners are designed with a safety factor greater than 10 to prevent the endplates from coming off the mold or the molds from coming off the machine, which can cause catastrophic molten metal spilling. The push force (F) on the endplates generated by the spinning metal is expressed by the following equation: 

F ¼ 5:375


2 N 2 2 do  d2i 100

where r is the molten metal density (kg/m3), N is the mold spinning speed (rpm), do is the casting outside diameter (m), and di is the casting inside diameter (m). This equation illustrates that the pushing force increases rapidly with the spinning speed (squared), casting diameter (squared), casting thickness (squared), as well as the metal density. Therefore, whenever a large or thick casting is to be poured, the pushing force on the endplates must be calculated so that the stresses of the endplate locks, bolts, pins, and fasteners can be calculated and properly designed. Mold Heating and Coating Techniques. The centrifugal casting mold must be heated and coated with ceramic mold wash, mold powder, sand, resin sand, or graphite. The most widely used mold coating is water-based mold

wash, which is consistently replacing the other coating materials. Sand and resin sand linings are quickly fading out, and graphite coating is mainly used for small nonferrous castings. The dominance of water-based mold wash is established on the fact that it offers significant advantages over the other types. Some of these advantages are:
Water-based mold wash is insulating enough

for almost all applications.


It allows immediate metal pouring as soon







as the coating application is finished, which greatly increases the casting productivity. It offers a better surface finish to the castings. Coating thickness can be easily used to control the microstructure. It greatly reduces the friction between the casting and the mold. It can be easily cleaned off the casting and mold. Its consumption is very small compared to sand linings. The mold has a long life under its protection.

The mold can be heated from the outside or inside diameter on the machine with the mold spinning slowly by a row, or rows, of gas burners or off the machine in an oven at 180 to 320  C (355 to 610  F). The mold temperature can be critical for the coating adhesion to the mold, coating strength, as well as casting surface quality. Once the mold is heated sufficiently, mold wash is applied on the mold inside diameter by spraying or flooding. The mold wash thickness is usually between 0.5 and 3 mm (0.02 and 0.12 in.), depending on the application. The spraying method is preferred for most processes because it gives more uniform coating thickness, smoother coating finish, and more consistent coating quality. Spraying can also reach the areas that flooding cannot. The spraying method works for both horizontal and vertical molds, but the flooding method works only for horizontal molds. Figure 3 shows a patented automatic spray lance system that offers automatic control of the forward and backward movement, moving velocity, and spray on and off operations of the spray lance. It can also adjust height and tilt angle to align the lance with all customer mold sizes.

Defects in Centrifugal Castings The three most common defects observed in centrifugal castings are segregation banding, raining, and vibration defects. Segregation banding occurs only in true centrifugal casting, generally where the casting wall thickness exceeds 50 to 75 mm (2 to 3 in.). It rarely occurs in thinner-wall castings. Banding can occur in both horizontal and vertical centrifugal castings. Banding is more prevalent in alloys with a wide solidification range and greater solidification shrinkage.

Centrifugal Casting / 671

Spray stop

Lance stop

Clamp spring


Surface smoothness and dimensional accu-

Reverse switch

Spray wash in

Spray lance Spray tip Tilt adjustment Compressed air Vertical clamps

Traverse drive Spray tank

Anchor bolts

Fig. 3

Hydraulic height adjustment

Patented automatic mold wash spray assembly. Courtesy of the Centrifugal Casting Machine Company

Bands are annular segregated zones of lowmelting constituents, such as eutectic phases and oxide or sulfide inclusions. They are characterized by a hard demarcation line at the outside edge of the band that usually merges into the base metal of the casting. Most alloys are susceptible to banding, but the wider the solidification range and the greater the solidification shrinkage, the more pronounced the effects may be. Banding has been found when some critical level of rotational speed is attained, and it has been associated with very low speeds, which can produce sporadic surging of molten metal. Therefore, both mechanisms may be involved. Minor adjustments to casting operation variables, such as rotational speed, pouring rate, and metal and mold temperatures, will usually reduce or eliminate banding. Raining is a phenomenon that occurs in horizontal centrifugal castings. If the mold is rotated at too low a speed or if the metal is poured into the mold too fast, the metal actually rains or falls from the top of the mold to the bottom. Proper process control can eliminate raining. Vibration Defects. Vibration can cause a laminated casting. It can be held to a minimum by proper mounting, careful balancing of the molds, and frequent inspection of rollers, bearings, and other equipment.

Applications Typical products produced by horizontal centrifugal casting machines include pipe and tubes of all different metals; alloyed iron, steel, and high-speed steel rolls for steel mills and the food-processing, papermaking, and printing industries; iron and steel cylinder liners, gray iron piston rings, gray iron brake drums, and

alloy and superalloy seal rings and valve seats for the auto industry; heat-resistant steel cracking tubes for petroleum refinery and radiant heating tubes and furnace rolls for heat treatment furnaces; and copper alloys and Babbitt alloys for bimetal bearings and bushings. See the article “Horizontal Centrifugal Casting” in this Volume for more information. Vertical machines are mainly used for shorter parts, such as bushings, gear blanks, rings, short rolls, wheels, aluminum and copper electric motor rotors, jewelries, and vacuum titanium alloy parts, and many different small round parts, such as balls, valve bodies, and rings for the machinery industry. With proper support, vertical casting machines can also produce extralong rolls. Slightly inclined machines are used for producing ductile iron pipe for water and gas mains, and other inclined machines are used for rolls and conical bushings. See the article “Vertical Centrifugal Casting” in this Volume for more information. Centrifugal casting is also used to ensure good filling in investment casting, as described subsequently. Another related technique is spin casting, which uses rubber molds for casting of zinc, tin, and other alloys with low melting points. It is also used in conjunction with nonmetallic materials such as glass and plastics. A more advanced materials application is the combustion synthesis of functionally graded materials. Investment Casting with Centrifugal Force (Ref 1). When the action of centrifugal force is combined with investment casting, the main benefit is an improvement in casting soundness with subsequent improvements in mechanical properties. The influence of centrifugal force in aluminum investment castings (Ref 2) found that:
A down-tapered sprue was more effective in

eliminating turbulence of the metal stream and nonmetallic incisions.

racy of aluminum alloy centrifugal castings were much better than the gravity investment castings, but they were much more affected by the fineness of investment material than centrifugal force.
The soundness and mechanical properties were improved with increased centrifugal force. The liquidus-solidus interval is important in centrifugally cast investment mold castings, and final casting soundness depends on the gating system for castings of alloys with narrow and wide freezing ranges (Ref 3). Factors that influence soundness were identified to be (Ref 3):








Ingate diameter Sprue diameter Metal pour temperature Mold temperature prior to pour Mold speed Spinning period Casting weight (expressed as a casting length, l)
Number of castings in each ring on the sprue
Distance (L) from the axis of rotation to the castings on the sprue (which characterizes the sprue length)
Liquidus-solidus interval (freezing range) For a wide freezing range (e.g., 100 K in bronze), the casting soundness was improved by increasing mold speed, ingate diameter, sprue diameter, and the product (l  L) of specimen size and sprue length. Mold speed, n, was found to be the most significant variable. For wide-interval alloys, optimal conditions (Ref 3) are “those under which each portion of molten metal newly arriving in the mold cavity should spread out at the appropriate level and freeze onto the solidification front without building up a substantial surplus of molten metal. Thus, the optimum conditions for pouring intricate castings in wide-interval alloys are those which produce layer-by-layer freezing.” “Casting soundness decreased as the number of castings at the same level on the sprue increased. Thus, as fewer castings are made, their density will increase. Increasing the number of ingates surrounding the sprue at the same level probably leads to the development of a hot spot, heat storage around the spot, and departure from conditions required for layerwise solidification. It is therefore best to assemble the patterns in a helical array on the sprue for intricate castings in wide-interval alloys.” “Density comparisons between the centrifugal castings at the two extreme levels on the sprue have shown that the sprue length has no significant influence on density.” For narrowinterval alloys, there is a sprue-length effect. Narrow-interval alloys require much greater ingate and sprue cross sections than wide-interval alloys, and narrow-interval alloy “castings should always be made on short sprues.”

672 / Centrifugal Casting “Narrow-interval alloys require heat accumulation around the sprue and ingates. Sound castings can be made by accumulating surplus molten metal ahead of the solidification front. In this case, the optimum conditions correspond to wide-section ingates and sprues, mold preheating, the use of short sprues, and some reduction in the relative rate of molten metal supply to the mold cavities.” “Intricate castings in wide-interval alloys must be made under conditions which will prevent excessive heat accumulation around the gating system and mold cavities, and minimize the volume of molten metal ahead of the solidification front. Sound metal can be ensured by directional solidification. In this case, mold preheating, commensurate metal weights and sprue lengths, helical mold assemblies around the sprue, and normal ingate and sprue cross sections should lead to a certain’freezing’ action, while rapid mold cavity filling with metal in this condition should lead to layerwise solidification” (Ref 3). Centrifugal Casting with Combustion Synthesis. Combustion synthesis (CS) is an attractive technique for synthesizing a wide variety of advanced materials, including powders and near-net-shape products of ceramics, intermetallics, composites, and functionally graded materials. One type of CS is self-propagating high-temperature synthesis (SHS), as described in Ref 4. Combustion synthesis of highly exothermic reactions (typically reduction type) often results in completely molten products, which may be processed using common metallurgical methods. Casting of CS products under inert gas pressure or centrifugal casting has been used to synthesize cermet ingots, corrosion- and wear-resistant coatings, and ceramic-lined pipes. Casting under gas pressure is similar to the conventional SHS production, while centrifugal casting allows for greater control of the distribution of phases by controlling the time of separation. For radial centrifuges (Fig. 4a), the sample is placed at a fixed radial position from the axis of rotation, and the applied centrifugal force is parallel to the direction of propagation. The influence of centrifugal acceleration characterizes the phase distribution in the final product (Fig. 5). The second type of centrifugal casting apparatus, called an axial centrifuge (Fig. 4b), is used for production of ceramic, cermet, or ceramic-lined pipes. The axial centrifuge casting method was developed further for production of long pipes with multilayer ceramic inner coatings. In this process, a thermite mixture (e.g., Fe2O3/2Al) is placed inside a rotating pipe and ignited locally. A reduction-type combustion reaction propagates through the mixture, and the centrifugal force results in separate layers of metal and ceramic oxide, with the latter forming the innermost layer (Fig. 6). The process is carried out in air under normal pressure, and pipes with diameters up to 30 cm (12 in.) and 5.5 m (18 ft) long have been obtained.

Spin casting typically refers to a special type of centrifuge casting with rubber molds. In the early 1970s, heat-cured silicone rubber (General Electric Silcast) was introduced with a sufficiently higher temperature resistance for molds in casting plastics and metals with low melting points, such as zinc alloys, tin alloys, and aluminum. Heat-cured silicone rubber molds can withstand temperatures up to approximately 550  C (1020  F) at a rate of 50 to 60 cycles per hour for hundreds of cycles. However, spin casting of metals with rubber molds is generally limited to the low-melting-point alloys (below 400  C, or 750  F), due to the degradation of the silicone rubber. Mold life is on the order of 200 to 250 shots at 400  C (750  F). During pouring, the mold is placed on a centrifuge wheel to achieve adequate metal filling

into intricate areas of the mold. Depending on the size of the mold and the material being cast, spinning speeds range from 100 to 900 rpm. Metal products can range from less than 100 g to 1.2 kg (3.5 oz to 2.6 lb). Maximum product size depends on mold diameter and thickness. Mold diameters can range up to approximately 60 cm (24 in.), with thicknesses of 10 cm (4 in.) or more. In mold making, metal patterns are laid on uncured silicone rubber sheets, which are packed into a circular metal frame. The circular metal frame is placed into a hydraulic vulcanizer press and electrically heated for curing. The curing temperature and duration depends on the type of rubber and the mold thickness. Pattern materials, such as pewter, can be used as long as they can withstand a temperature of

5

4

3

2 1

(a)

5

2

4

(b)

Fig. 4

1 3 Schematic of self-propagating high-temperature synthesis plus centrifugal casting. (a) Radial centrifuge. (b) Axial centrifuge. 1, sample container; 2, reactant mixture; 3, ignitor; 4, axle; 5, reactor

Centrifugal Casting / 673

Fig. 5

Degree of phase separation as a function of centrifugal acceleration, a, where g is acceleration due to gravity. Source: Ref 5

3MeO + 2Al

Al2O3 + 3Me + Q (1)

3MeO + 2Al + 3C

Al2O3 + 3MeC + Q⬘(2)

Fig. 6

Concept of centrifugal process for production of ceramic-lined steel pipes. Source: Ref 6

190  C (375  F) under several thousands of pounds pressure without distorting, melting, breaking, or outgassing. The heat melts the rubber, which is pressed against the surface of the pattern before the rubber sets. After vulcanization, the

mold is separated from the metal, and the patterns are removed. Special-shaped tools are used to cut gates and runners on the rubber. Air vents are also cut for air to escape from the mold cavity. Modifications to gates and vents are usually required after a few shots to obtain the best results. Making of the mold takes approximately 1 to 3 days, depending on the complexity of the product shape. While the mold is spinning, the liquid metal or plastic is melted in a suitable electric or gasfired crucible furnace and then poured into the center sprue of the mold. After the metal solidifies (the plastic parts set up), the parts are quickly removed from the mold. With metal, 50 to 60 cycles per hour can readily be made; with plastic, 10 to 15 cycles are typical. Parts with undercuts can be demolded from the rubber mold with no difficulty. Thin-walled parts can be molded comparable to die casting. The rubber mold is not as precise as the metal mold, but the surface finish is generally good. However, blisters (very small pinpoint holes) were often found on the surface of the thick section due to the shrinkage of metals. The blisters are the defects for products with a plated glossy surface. Metal Products. Spin casting has the capability to produce intricate designs for low- or highvolume quantities. Metal products include:










Cams and levers in automotive control panels Bushings Hub, clamps, and screws Triggers, firing mechanisms, and cartridges Medals, pins, figurines, key chains, name plates, and costume jewelry Door hardware, locks, hooks, and rings Handles Belt buckles, buttons, and watch cases Gears, motor housings, and pump impellers

Plastic Products. Spin casting of thermoplastics and thermoset plastics includes:
Plumbing fittings, valves, and couplings
Electrical switch plates, condensers, and





connector plugs Toy cars, wheels, and gears Pen holders, clips, and fasteners Electronic computer parts Calculator casing keys

Wax patterns for investment casting are also produced by spin casting.

REFERENCES 1. E. Roberston, LA-12370-MS, Los Alamos National Laboratory, Sept 1993 2. J.-N. Lee and G.-C. Yun, Effect of Centrifugal Force Applied During Investment Casting on the Soundness of Cast Al Alloys, J. Korean Inst. Met., Vol 15 (No. 2), 1977, p 155–163 3. V.I. Osinskii, M.V. Solov’ev, and A.V. Morozov, Soundness of Intricate Centrifugally Cast Investment-Mold Castings (Translation), Sov. Cast. Technol., Vol 7, 1987, p 50–54 4. A. Varma and A.S. Mukasyan, Combustion Synthesis of Advanced Materials, Powder Metal Technologies and Applications, Vol 7, ASM Handbook, ASM International, 1998 5. A.G. Merzhanov and V.I. Yukhvid, in Proc. First U.S.-Japanese Workshop or Combustion Synthesis, Vol 1, National Research Institute for Metals, Japan, 1990 6. O. Odawara and J. Ikeuchi, J. Am. Ceram. Soc., Vol 69, 1986, p C80

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 674-679 DOI: 10.1361/asmhba0005258

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Horizontal Centrifugal Casting Revised by Y.V. Murty, CMI Inc.*

HORIZONTAL CENTRIFUGAL CASTING is used to cast parts having an axis of revolution. The technique uses the centrifugal force generated by a rotating cylindrical mold to force the molten metal against the mold wall to form the desired shape upon solidification. The first patent on a centrifugal casting process was obtained in England in 1809. The first industrial use of the process was in 1848 in Baltimore, when centrifugal casting was used to produce cast iron pipes. In the 1890s, the principles already known and proved for liquids in rotation about an axis were extended to liquid metals, and the mathematical theory of centrifugal casting was developed in the early 1920s. Horizontal centrifugal casting was first used mainly to manufacture thin-wall gray iron, ductile iron, and brass tubes. Improvements in equipment and casting alloys made possible the development of a flexible and reliable process that is both economical and capable of meeting stringent metallurgical and dimensional requirements. Cylindrical pieces produced by horizontal centrifugal casting are now used in many industries. Of particular importance are large-diameter thick-wall bimetallic and specialty steel tubes used in the chemical processing, pulp and paper, steel, automotive, and offshore petroleum production industries.

Equipment


There must be a means to pour the molten

metal safely into the rotating mold at a controlled rate, position, and orientation.
Once the metal is poured, a proper solidification and cooling rate must be established in the mold to obtain a desired casting microstructure, protect the mold, and achieve required productivity.
There must be a means of adding inoculants or fluxes for some special applications.
There must be a means of extracting the solidified casting quickly from the mold at elevated temperatures without deforming the casting. Spinning of the mold is realized by spinners. Horizontal machines of all sizes usually use four trunnion wheel spinners to support and spin the mold. A short horizontal mold can be cantilevered (attached at only one end) and spun with a single spinner, or squeezed between two faceplates and spun by two spinners. The trunnion wheels must have certain hardness, precision, and alignment to spin the mold smoothly, which is a critical operation impacting the quality of castings and the life of bearings, wheels, and molds. The achievable spinning speed and, sometimes, the ability to change speeds during the pour and solidification process is an important consideration for determining whether a casting geometry or a casting microstructure can be achieved. Spinning speeds are chosen based on the centrifugal force requirement, which is measured by the multiple of the gravita-

tional force (called g force). Horizontal casting machines are typically spun at 45 to 60 g force on the casting inside diameter, while vertical casting machines are typically spun at 75 g force. For some special applications, the spinning speed can be as high as 100 to 200 g force (e.g., horizontal roll castings and heat-resistant steel tube castings); in other cases, it can be as low as a fraction of a g to a few g’s (e.g., when vertically pouring a roll core, vertically pouring solid bars, or horizontal two-speed pouring). See Fig. 6 in the article “Vertical Centrifugal Casting” in this Volume for a chart to check the centrifugal force against casting diameters and spinning speeds.

Machines Horizontal casting machines are of three types (Ref 1):
Flanged shaft machine
Horizontal roller-type machine
Double-face plate machine

The flanged shaft machine has a horizontal mold, which is supported overhead on a flange (Fig. 1). Horizontal mold orientations are usually used for castings that are sufficiently long, that is, length/diameter ratios greater than 1.5. The flanged shaft machine is used to make all forms of flanges, ring-type castings, and general cylindrical shapes but is most often used to make short tubes for cylinder liners, bushes,

A centrifugal casting machine must be able to perform six operations accurately and with repeatability:
The machine must be able to accelerate the

mold to a predetermined speed, maintain smooth spinning, and decelerate to a stop in a reasonable time frame. The machine also should be able to change speeds during pour and solidification for some special applications.
There must be a way to heat and coat the mold before pouring the molten metal (except the cold mold method for ductile iron pipe).

Fig. 1

Flanged shaft-type horizontal casting machine. Source: Ref 2

* This article is revised from “Centrifugal Casting” by A. Royer and S. Vassuer in Casting, Volume 15, Metals Handbook, 9th ed., ASM International, 1988, p 296–307.

Horizontal Centrifugal Casting / 675 sleeves, tubes to be cut into rings, gearwheel blanks, and piston and bearing cage rings. Casting weight is limited to less than 1Mg (1 ton). Roller-type machines (Fig. 2) are used to make tubular castings, usually when production quantities are high. It is not the same as the pipe casting machine. The horizontally rotated mold is composed of four parts:
The shell
The casting spout (the end of the mold into

which the molten metal is poured)


The roller tracks
The two end heads

The mold usually lies on four interchangeable carrying rollers that are capable of providing good contact between different sizes of mold diameters and the roller tracks. As the mold is rotated, it is also cooled with a controlled water spray system. With the horizontal centrifugation technique, castings up to 8m (26 ft) long, 27Mg (30 tons) in weight, and up to 1000mm (40 in.) in diameter, as well as tubes up to 5m (16 ft) long for all diameters down to 75mm (3 in.), in most alloys have been made. Double-face plate machines (Fig. 3) were originally developed to line bearing shells with white metal. In this technique, simple cylindrical molds of steel or cast iron are clamped between the two face plates, and the pouring operation takes place through one of the bearings supporting the face plate. When the casting operation is finished, the cylindrical mold, together with the casting, is removed from the machine for casting extraction.

Molds Molds consist of four parts: the shell, the casting spout, roller tracks, and end plates. The mold assembly is placed on interchangeable carrying rollers that enable the use of different mold diameters and fine adjustments. Molds are cooled by a water spray, which can be divided into several streams for selective

cooling. There are several features common to the mold used for most centrifugal casting methods. Since it is critical to successful casting operation, the mold features are covered in this article as well as in the article “Vertical Centrifugal Casting” in this Volume. Different types of molds are generally used according to the geometry and quantity of castings needed and the characteristics of the metal or alloy being cast. Molds can be either expendable (a relatively thin case lined with sand) or permanent. Different molds used for centrifugal casting are described in the article “Vertical Centrifugal Casting” in this Volume. Expendable molds lined with sand are widely used in centrifugal casting, especially for producing relatively few castings. A single mold case can be used with different thicknesses of sand linings to produce tubes of various diameters within a limited range. Green sand is commonly used as the liner in expendable molds. Various mixtures and binders are used, for example, a mixture of silica sand and binder resin. Phenolic binders are also used with silica sand. One proprietary process uses a mixture of sand, silica flour, bentonite, and water. Dry sand molds can also be employed; in this case, the sand is pressed down around a pattern having the same dimensions as the casting. Hardening is sometimes accomplished with carbon dioxide. Mold washes of various compositions are used with sand molds. The wash hardens the mold surface and minimizes erosion of the mold by molten metal. The mold wash coatings also help establish a smoother cast surface. Permanent Molds. The most common materials for permanent molds are steel, copper, and graphite. Steel molds are used to cast large quantities and for some casting alloys that require specific solidification conditions. Steel molds are sensitive to thermal shock; alumina- or zirconia-base mold sprays are used to lessen thermal shocks to the mold and to improve the mold surface.

Mold coatings are also important in regulating the solidification rates of some casting materials. Other ceramic coating materials are beginning to be employed. Copper molds are sometimes used for their high thermal conductivity. Their relatively high cost and the difficulty of calculating the correct dimensions of these molds limit their field of application. Graphite Molds. Because of their relatively low cost, graphite molds can be an economical alternative to sand in the production of small quantities of parts. Graphite is the mold material of choice in the casting of 80% Cu bronzes, high-phosphorus brasses, and other copper alloys. Graphite has excellent thermal conductivity and resistance to thermal shock, and it is easily machined. Care must be taken, however, to maintain the mold well below the oxidation temperature of graphite.

Casting Process Pouring. Molten metal can be introduced into the mold at one end, at both ends, or through a channel of variable length. Pouring rates vary widely according to the size of the casting being produced and the metal being poured. For example, a pouring rate of 1 to 2 Mg/min (1.1 to 2.2 tons/min) has been used to cast low-alloy steel tubes 5m (200 in.) long and 500mm (20 in.) in outside diameter with 50mm (2 in.) thick walls. Pouring rates that are too slow can result in the formation of laps and gas porosity, while excessively high rates slow solidification and are one of the main causes of longitudinal cracking. Casting Temperatures. The degree of superheat required to produce a casting is a function of the metal or alloy being poured, mold size, and physical properties of the mold material. The following empirical formula has been suggested as a general guideline to determine the degree of superheat needed: L ¼ 2:4
T þ 110

Mold Mold preheating

Pour car

Water guard

Mold cooling

Fig. 2

(Eq 1)

where L is the length of spiral fluidity (in millimeters), and DT is the degree of superheat (in degrees centigrade). The use of Eq 1 for ferrous alloys results in casting temperatures that are 50 to 100  C (120 to 212  F) above the liquidus temperature. In practice, casting temperatures are kept as low as possible without the formation of defects resulting from too low a temperature.

Trunnion wheel spinners

Automated horizontal casting machine. Courtesy of Centrifugal Casting Machine Company

Fig. 3

Double-face plate centrifugal casting machine. Source: Ref 2

676 / Centrifugal Casting A high casting temperature requires higher speeds of rotation to avoid sliding; low casting temperatures can cause laps and gas porosity. Casting temperature also influences solidification rates and therefore affects the amount of segregation that takes place. Mold Temperature. Numerous investigators have studied the relationship between initial mold temperature and the structure of the resultant casting. Initial mold temperatures vary over a wide range according to the metal being cast, the mold thickness, and the wall thickness of the tube being cast. Initial mold temperature does not affect the structure of the resultant casting as greatly as the other process parameters discussed previously.

Fig. 4

Fig. 5

Typical cycle of rotation centrifugal casting

in horizontal

Speed of Rotation. Generally, the mold is rotated at a speed that creates a centrifugal force ranging from 75 to 120 g (75 to 120 times the force of gravity). Speed of rotation is varied during the casting process; Fig. 4 illustrates a typical cycle of rotation. The cycle can be divided into three parts:
At the time of pouring, the mold is rotating

at a speed sufficient to throw the molten metal against the mold wall.
As the metal reaches the opposite end of the mold, the speed of rotation is increased.
Speed of rotation is held constant for a time after pouring; the time at constant speed varies with mold type, metal being cast, and required wall thickness. The ideal speed of rotation causes rapid adhesion of the molten metal to the mold wall, creating a good contact with the mold. Such conditions tend to result in a casting with a uniform structure. As the molten metal enters the mold, a pressure gradient is established across the tube thickness by centrifugal acceleration. This causes alloy constituents of various densities to separate, with lighter particles such as slags and nonmetallic impurities gathering at the inner diameter. The thickness of these impurity bands is usually limited to a few millimeters, and they are easily removable by machining. This density separation principle was successfully used to produce selectively graded

Effect of (a) mold coating thickness and (b) molten metal temperature on solidification in horizontal centrifugal casting. Numbers 1 and 2 indicate liquidus and solidus curves, respectively.

composite castings such as graphite or silicon carbide particulates segregating the inner surfaces of copper alloy bearings and bushings to provide lubrication and/or wear resistance of the inner surface while maintaining high strength and toughness of the outer shell. Similar concepts were used to produce copper/tungsten and aluminum composite centrifugal castings. Too low a speed of rotation can cause sliding and result in poor surface finish. Too high a speed of rotation can generate vibrations, which can result in circumferential segregation. Very high speeds of rotation may give rise to circumferential stresses high enough to cause radial cleavage or circular cracks when the metal shrinks during solidification. Solidification. In horizontal centrifugal casting, heat is removed from the solidifying casting only through the water-cooled mold wall. Solidification begins at the outside diameter of the casting, which is in contact with the mold, and continues inward toward the casting inside diameter. Several parameters influence solidification:
The mold, including the mold material, its

thickness, and initial mold temperature


The thickness and thermal conductivity of

the mold wash used


Casting conditions, including degree of

superheat, pouring rate, and speed of rotation


Any vibrations present in the casting system

Thermal Aspects of Solidification. It appears that the mold-related parameters listed previously have relatively little influence on solidification. Large variations in mold thickness, however, could become significant. The parameters with the greatest effect are the degree of superheat in the molten metal and the heat-transfer conditions that exist at the mold/ metal interface, which are influenced by the thickness of the mold wash employed. Both of these process variables affect local solidification conditions and therefore modify the structure of the casting. Figure 5 illustrates the general effects of mold wash thickness and degree of superheat on solidification rates. Charts such as Fig. 5 can be used to predict total solidification time. They are especially useful in determining the optimal casting conditions for bimetallic tubes based on the type of bond required. Metallurgical Aspects of Solidification. The as-cast structures obtained in the horizontal centrifugal casting of steels vary according to composition. Regardless of the phase or phases that solidify first, certain features are common to the structures of centrifugally cast ferrous alloys (Fig. 6a):
Very thin, fine columnar skin
Well-oriented columnar structure adjacent to

the skin

Fig. 6

Three types of as-cast structures seen in centrifugally cast ferrous alloys. (a) Fine columnar skin, large welloriented columnar grains, and equiaxed area. (b) Completely equiaxed structure sometimes observed in ferritic steels. (c) Equiaxed bands of varying grain size. This type of structure is thought to be caused by machine vibrations.


More or less fine equiaxed structure

In the case of steels that solidify as ferrite, the columnar areas may be nonexistent if

Horizontal Centrifugal Casting / 677 Table 2

Melting and pouring temperatures of tin-base Babbitts

ASTM B 23 alloy No.

Specific gravity

Cu

Sn

Sb

Pb

7.34 7.39 7.46

4.56 3.1 8.3

90.9 39.2 83.4

4.52 7.6 8.3

None 0.03 0.03

1 2 3

Fig. 7

Loading of a large steam turbine generator bearing into a centrifugal Babbitting machine. Courtesy of Pioneer Motor Bearing Company

Table 1 Horizontal centrifugal casting speeds for tin-base Babbitt bearings Optimal speed is slowest speed that will produce sound (porosity-free) castings with minimum segregation. Inside diameter of bearing shell mm

in.

Centrifugal casting speed, rev/min

51 76 102 127 152 178 203 229 254 279 305 330 356 381 406 432 457 483 508 533 559 584 610 635 660 686 711 737 762 787 813 838 864 889 914 965

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 38

850–975 700–800 600–700 535–610 485–550 450–525 420–500 395–470 375–450 360–430 345–410 330–395 320–370 305–360 295–350 285–340 280–330 275–325 265–315 260–310 250–300 245–295 240–290 235–285 230–280 230–275 225–270 220–265 215–260 215–255 210–250 205–245 200–245 200–240 195–235 190–230

superheat and mold wash thickness are low (Fig. 6b). In steels that solidify as austenite, it is relatively easy to obtain well-oriented 100% columnar structures. A phenomenon specific to horizontal centrifugal casting is the formation of equiaxed bands through the entire thickness of the casting (Fig. 6c). A plausible explanation for this phenomenon is linked to machine vibrations, which

Composition, %

Melting point 



C

223 241 240

Complete liquefaction 

F

433 466 464



C

371 354 423

700 669 792

Table 3

Melting and pouring temperatures of lead-base Babbitts

ASTM B 23 alloy No.

Specific gravity

7 8 15

Composition, %

9.73 10.04 10.05

Cu

Sn

Sb

Pb

0.50 10 15 75 0.50 5 15 80 0.5 1 15 82

Melting point

As (max)



0.60 0.20 1.40

240 237 249

may cause recirculation of the molten metal during solidification. As in static casting, the rejection of solute ahead of the solidification front leads to microsegregation and a progressive enrichment of the solute-enriched liquid. Carbon steels are particularly sensitive to this effect; carbon, sulfur, and phosphorus contents must be limited to avoid local precipitation of carbides and sulfides. Babbitting is another important process that uses centrifugal casting. Centrifugal Babbitting requires a machine expressly designed or modified for this purpose (Fig. 7). A variable-speed centrifugal casting (spinning) machine is fitted with a safe means for supporting and rotating a reasonably well-balanced workpiece clamped between recessed sealing plates. With the shell rotating, molten Babbitt is fed to the inside of the prepared bearing through a hole in the outboard spinning plate, then solidified by air and/or water sprayed on the outside diameter of the shell while it is spinning. Speed is selected to produce a centrifugal force that is high enough to eliminate porosity but low enough to minimize metal segregation. Too low a speed causes metal “tumbling,” while too high a speed causes segregation. Table 1 shows a range of spinning speeds for various bearing sizes (inside shell diameter), based on minimum centrifugal forces of 20 g for tin-base alloys. The minimum for lead-base Babbitts is 16 g in horizontal centrifugal casting. To promote directional solidification after the Babbitt is poured, the tooling plates must be preheated to 200  C (390  F) minimum and faced with gaskets cut from sheets of environmentally approved insulating materials, such as Inswool, to seal and insulate both ends of the bearing when it is clamped between the support plates by hydraulic pressure. The centrifugal casting machine must be of fail-safe design in the event of a loss of power and/or clamping pressure. After tinning, full-round (nonsplit) shells require no further preparation and can be immediately loaded into the casting machine. Splittype bearings (in halves) or segmented (tilting

C



F

464 459 469

F

Complete liquefaction 

C

268 272 281



F

514 522 538

Proper pouring temperature 



C

440 425 490

F

825 795 915

Proper pouring temperature 

C

338 340 350



F

640 645 662

pad)-style journal bearings require fixturing after the pretinning of each component to make up an assembly ready for spinning. The fixturing consists of laminated spacers for each parting face. These spacers must prevent both radial and axial leakage without unduly restricting the smooth flow of the molten rotating metal. Spacers can be cut from steel sheet with a minimum thickness of 1.0 mm (0.04 in.) and faced on both sides with 3mm (0.12 in.) gaskets that are cut to fit. Total spacer thickness should be at least 7mm (0.28 in.) before assembly to allow for the later separation of the halves or segments by saw cutting through the Babbitt without damage to the steel parting faces. When the spacers are in place and the halves or segments are retained by bolting or other means of safe clamping, the assembly must be reheated in the tinning bath, if necessary, to ensure a minimum temperature of 300  C (570  F) before loading in the casting machine. Alternatively, the cold assembly can be mounted in the machine and slowly rotated while gas torches are applied along the outside of the bearing until it reaches tinning temperatures. This latter method, however, does not allow visual in-process inspection to ensure complete wetting (tinning) before casting commences. In any case, there must be little time lost after tinning temperatures are reached, and before Babbitt metal is poured, to achieve a satisfactory bond; tinning must be molten when the Babbitt is poured. The Babbitt alloy should be melted in a temperature-controlled cast iron pot that is held at its recommended pouring temperature (Tables 2 and 3) and stirred with a vertical motion to promote uniform metal temperature and to avoid metal segregation. Dross on the surface of the pot is skimmed aside while a preheated ladle, preferably of the bottom-pour style, is filled with a predetermined volume of metal. The volume of metal poured should be sufficient to provide a minimum of 4mm (0.16 in.) machining stock per side after Babbitting so that impurities, which float to the inside during centrifugal casting, can be removed.

678 / Centrifugal Casting With the bearing shell rotating at a preselected speed, the metal is ladled in one continuous pour directly, or through a preheated trough, into the spinning shell. Water cooling should be started immediately after pouring is completed, and the bearing rotation should continue until the assembly temperature drops to approximately 150  C (300  F). At this point, the water should be shut off and the bearing allowed to air cool more slowly while spinning. This cooling regime minimizes the stress imposed on the bond by the difference in thermal coefficients between typical Babbitts and backing materials. (The ratio is almost 2 to 1 between Babbitt and steel, and 1.5 to 1 between Babbitt and bronze.) A controlled flow of coolant along the length of the bearing ensures directional solidification. That is, the Babbitt freezes (solidifies) uniformly along the bond line first, then progressively toward the inside of the casting. This ensures that molten metal under pressure is available to feed the shrinkage and to squeeze out trapped gases and dross. When a shell has a nonuniform cross section, the coolant flow should be adjusted to promote uniform solidification from end to end. Nonsplit journal bearings follow the same basic procedures as the aforementioned without the need for assembling half shells or tilting pad segments and reheating them before they are loaded into the casting machine.

Fig. 8

Three applications for centrifugally cast parts in the iron and steel industry. (a) Continuous casting roller. (b) Winding spool. (c) Annealing furnace rollers

Fig. 9

Offshore petroleum production applications for centrifugally cast parts. (a) Jackup leg. (b) Risers. (c) Bucklecrack arrestor

Applications The flexibility of the horizontal centrifugal casting process, in terms of both materials and the wide range of part sizes that can be produced, has led to the application of centrifugally cast parts in many industries. In the iron and steel industry, centrifugally cast parts are used in the production of iron and steel for continuous casting rollers, rolling mill rolls, furnace rollers, special pipelines, winding spools, and other applications. Some of these uses are shown in Fig. 8. Other applications for centrifugally cast tubes include hydraulic cylinders, rollers for glass production, pipelines for the transport of abrasive materials, rollers in the pulp and paper industry, tubes for the chemical processing industry, foundation piles, and building columns. Centrifugal casting is sometimes used to produce nonrotating gas turbine components of heat-resistant materials. Offshore petroleum production platforms in the oil and gas industry use centrifugally cast tubes in various applications (Fig. 9). Hot extruded bimetallic tubes are used in pipelines and gathering systems. Any material that can be statically cast can also be centrifugally cast. Materials that are currently being processed by horizontal centrifugal casting include high-strength low-alloy steels, duplex stainless steels, and chromium-molybdenum alloy steels. Also of importance are bimetallic tubes. High-strength low-alloy steels are important in offshore structures that must withstand extreme climatic conditions, for example,

Horizontal Centrifugal Casting / 679 Table 4

Common material combinations used in bimetallic centrifugally cast tubes

Outer material

Chilled iron 27% Cr cast iron Low-alloy steel Low-alloy steel Chromium-nickel white iron Chromium white iron Stainless steel Low-alloy steel

Inner material

Typical applications

Gray iron Ni-Resist cast iron Manganese-molybdenum steel Ni-Hard cast iron, martensitic or 27% Cr white iron Gray ductile iron Gray iron Gray iron Pearlitic gray iron, stainless steel, or superalloy

Bearing rollers Mill rollers Continuous casting rollers Wear-resistant applications Mill rollers Mill rollers Rollers for pulp and paper industry Liners, pipelines for corrosives

offshore drilling equipment in the North Sea. These materials must be weldable, with ductile-to-brittle transition temperatures below 40  C (40  F). Likely candidate materials for use under such conditions are manganesemolybdenum steels with microalloying additions of vanadium, nickel, or niobium. Heavywall weldable pipes with good mechanical properties can be produced from such materials by horizontal centrifugal casting. Several proprietary alloys have been approved for use in offshore oil applications in the North Sea. Duplex stainless steel tubes can be readily produced by horizontal centrifugal casting in

any section size. Castings retain strength at temperatures to 600  C (1110  F), have good ductility, and are weldable without special precautions. Chromium-molybdenum alloy steel tubes produced by horizontal centrifugal casting have homogeneous structures and are metallurgically sound. They are resistant to thermal fatigue and wear and have good toughness. Bimetallic tubes with metallurgical (rather than mechanical) bonds can be readily produced by horizontal centrifugal casting. They are most commonly produced by successively casting one alloy inside the other. Bimetallic tubes are used for two primary reasons: to

reduce cost by using an exotic material bonded to a less expensive backing material, and to obtain combinations of properties that could not be achieved by other methods. There are no general rules regarding what materials can be combined in centrifugally cast bimetallic tubes, although it may be beneficial to cast the inner layer of such tubes from a material that is more fusible than the outer material. In addition, relatively thin inner layers should be manufactured from alloys with coefficients of thermal expansion smaller than that of the outer alloy. In this way, the thin inner layer is put into compression, making it more resistant to cracking. Table 4 lists materials that are commonly combined in bimetallic tubes, as well as their applications. Babbitting, described earlier, is another application. REFERENCES 1. E. Roberston, LA-12370-MS, Los Alamos National Laboratory, Sept 1993 2. D.S. Gibson, Basic Designs of Centrifugal Casting Machines, Foundry Trade J. Vol 156 (No. 3278), Feb 2, 1984, p 61–62

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 680-686 DOI: 10.1361/asmhba0005259

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Vertical Centrifugal Casting* Revised by Y.V. Murty, CMI Inc.

VERTICAL CENTRIFUGAL CASTING MACHINES are used for producing bushings and castings that are relatively large in diameter and short in length. The usual maximum length of the casting is two times the outside diameter of the casting. Vertical axis machines are also used for producing castings of odd or asymmetrical configurations. Equipment modification over the years incorporated several safety features into the equipment. For this and other productivity considerations, vertical casting machines are installed below the ground level for maximum operator safety. Figure 1 illustrates such an installation with all the vital machine components and features. Vertical centrifugal casting also can be used to form conical-shaped products. An example is illustrated in Fig. 2 for molten glass poured into a rotating mold. By gradually increasing the rotation speed, the glass “walks” up the sides of the mold, forming a piece with relatively uniform wall thickness (Fig. 2). An interior profiling tool is sometimes used to improve the uniformity of the wall thickness, especially for pieces with a large height/diameter ratio.

thinner mold sections into heavier mold sections much more easily than in static casting. Certain casting configurations that do not inherently produce directional solidification can be made to directionally solidify by certain

molding practices. Pads, auxiliary gates, chills, or blind heads can be used. However, a casting of nonuniform section thickness may require such elaborate mold methods that molding and cleaning costs become too high.

Mold Design The practicality of casting a part using the semicentrifugal or centrifugal process is determined by casting configuration during the vertical centrifugal casting process. The gating system of centrifugal castings usually employs a single gate, which combines the function of gate and riser. Centrifugal force greatly magnifies the feeding action of the riser and produces a greater metal density in the casting than would otherwise result. Figure 3 shows that a centrifugal casting mold spinning at a peripheral velocity of 305 m/min (1000 ft/min) at the outer diameter of the casting(s) will have an equivalent head of 9 m (30 ft). This is equivalent to a pressure of 703 kPa (102 psi). A typical static casting has a head less than 0.3 m (1 ft) high. The centrifugal force acting on the molten metal will provide better feeding action than a static cast head; therefore, it is possible to feed molten metal into and through lighter and

Fig. 1

Typical installation of a vertical centrifugal casting machine. The equipment is controlled from a remote console (not shown).

*This article is revised from “Vertical Centrifugal Casting” by R.L. Dobson in Casting, Volume 15, Metals Handbook, 9th ed., ASM International, 1988, p 296–307.

Vertical Centrifugal Casting / 681

Distributor

Ejector value Grooving disk Mold

Glass

Centrifugal development

Charging

Fig. 2

Trimming

Centrifugal casting of a parabolic glassware product

Equivalent riser, feet of steel 10

20

30

40

50

60

70

80

90

610

2000

457

1500

304

1000

152

500

0

Peripheral speed, ft/min

Peripheral speed, m/min

0

0 0

3

6

9

12

15

18

21

24

27

Equivalent riser, meters of steel

Fig. 3

Relationship between peripheral speed in centrifugal casting and equivalent riser

Sand Molds When sand molds are used, particularly green sand, it is usually necessary to begin pouring the molten metal at a slow mold-spinning speed. When the mold is partially or wholly poured, the spinning speed is increased to that required to prevent or reduce erosion of the mold cavities from the molten metal. Molds should almost always be poured while rotating, even if the speed of rotation is only 5 rpm. This ensures the proper distribution of hot and cold metal in the mold for the optimal feeding action. Green Sand Molds. Centrifugal castings can be made in green sand or dry sand molds. When green sand molds are used, flasks (preferably round) are required. Three methods can be used to fasten the green sand mold to the table of the centrifugal casting machine. In the first method, two pins are fastened to the table over which the flask is lowered. The cope and drag are held together by clamps in the usual manner of static casting. In the second method, a device similar to a lathe chuck is fastened to the table. The cope

and drag are clamped together as in static casting. The green sand mold is lowered onto the table, and the clamps are tightened to secure the mold. In the third method, the green sand mold is transported on a roller conveyor into a flask body, which is fastened to the table. Rollers are also used in the casting machine to facilitate movement of the green sand mold. The flask body opens in half on a hinge. A cover, which holds a runner cup, serves to distribute equally the pressure from the clamping arrangement to the mold. The clamping arrangement, which is part of the flask body, keeps the cope and drag in firm contact. Dry Sand Molds. Flasks are not required when dry sand molds are used. Two methods are available for handling dry sand molds on the casting machine. In the first method, the dry sand molds are placed in a jacket somewhat similar to that used in conjunction with snap flasks. The jacketed mold is then carried to the casting machine and lowered to the table. Clamps are used to hold the cover onto the mold and to fasten the mold firmly to the table. In the second method, the dry sand molds are transported on a roller conveyor into the casting machine, in which rollers are incorporated. Using a flask body of this type, both dry sand and green sand molds can be used interchangeably in the same casting machine. Molding Costs. A green sand mold prepared for static casting generally costs the same as a green sand mold prepared for centrifugal casting. If dry sand molds are used for centrifugal casting rather than green sand, the molding cost may be higher, resulting in a reduced cost savings. There is an appreciable cost savings in

the cleaning room because risers need not be removed from castings. This reduces the amount of cutting, chipping, and grinding required to clean the castings. The additional cleaning room savings may be offset by operating costs, depreciation of the centrifugal casting equipment, and possible increased molding costs due to the use of dry sand molds. An increase in yield primarily due to the elimination of riser heads contributes to further cost savings. Cast yield improvements up to a factor of 2 can be achieved with centrifugal casting compared to static casting. Other Sand Molding Considerations. The requirements for a centrifugal sand mold or flask are more stringent than those for static cast molds because concentricity with the spinning axis must be more exact to prevent an imbalance that could cause vibration. However, most casting speeds required for centrifugal sand molding are relatively slow compared to true centrifugal casting in permanent molds. Molds made up of sections using dry sand cores are also used for centrifugal casting. The sand used can be sodium silicate, dry sand, chromite sand, or any molding material with sufficient strength to withstand the forces imposed by the spinning speed. It is desirable to use a suitable mold wash with most sand molds to reduce or minimize mold erosion from the molten metal. Excessive mold erosion generally does not occur, because of the relatively slow spinning speeds used.

Permanent Molds Two basic types of permanent molds are in use: metal molds and graphite or carbon molds. Because of the faster heat extraction from permanent molds, there is usually an increase in the quality (especially properties) of the castings produced in this type of mold. Metal Molds. Steel molds are recommended for centrifugal casting. It is very important that the molds be perfect and free of any defects. The molds are machined all over to obtain a smooth surface and to ensure dynamic balance. A machined surface finish of 3 mm (125 min.) rms is recommended on the mold bore and outside diameter. If there are any defects on the portion of the mold that contacts the casting, it will be difficult, if not impossible, to remove the casting from the mold. Steel molds should always be thoroughly stress relieved (minimum temperature of 620
C, or 1150
F ) before the finish-machining operations. In the case of small, simple molds machined from heavy-wall seamless steel tubing, it has been standard practice not to stress relieve, because there is no apparent difference in the operational life of the mold. Stress relieving effectively prevents warpage in most molds and therefore prolongs the service life of the mold. A molybdenum disulfide lubricant is recommended for all threaded fasteners used on a mold, such as mold lock clamps, end plate bolts, mold attaching bolts, and mold centering bosses.

682 / Centrifugal Casting

Fig. 4

Six end plate designs used in centrifugal casting. (a) through (c) are the most common.

Various mold end plate designs (Fig. 4) can be used, depending on individual preference and the size of the mold being used. These techniques are useful for both vertical and horizontal casting. It is recommended that low-carbon steels be used for centrifugal molds. Alloy steels and steels with more than 0.30% C should not be used, because they are subject to cracking due to thermal shock and heat stresses from the casting of the molten metal. Satisfactory mold materials are 1018, 1020, or ASTM A106 grade A steel. It is usually most convenient to make small molds from either hot rolled solid bar stock or from heavy-wall seamless steel tubing. Larger-diameter molds can be made from forgings or hot rolled bar stock with the center trepanned or from centrifugally cast heavy-wall steel tubing. Metal Mold Design. Machining allowances for castings depend on melting practices, the condition and thickness of the mold insulation coating, the volumetric shrinkage of the metal being cast, and the susceptibility of the metal to casting defects such as dross formation. The suggested machining allowances given in Table 1 can be used as a basis for mold design. With these minimum allowances, the mold can be remachined to allow for more machining allowance as determined by experience and the specific melting practices used.

Table 1 Minimum machining allowances for design of centrifugal casting molds Allowance on casting inside diameter (each side) Casting size

Allowance on casting outside diameter (each side)

mm

in.

50–200 200–300 300–500 500–700 700–900 900–1100

2–8 8–12 12–20 20–28 28–353 35–43

Irons

Bronzes

Steels

mm

in.

mm

in.

mm

in.

mm

1.6 2.5 4 6 8 10

/ 3 /32 5 /32 7 /32 5 /16 3 /8

1.6 2.5 4 6 8 10

/ 3 /32 5 /32 7 /32 5 /16 3 /8

3.2 5 8 12 16 20

/ 3 /16 5 /16 7 /16 5 /8 3 /4

4.8 7.5 12 18 24 30

16

Patternmaker’s shrinkage is the shrinkage allowance built into the pattern to compensate for the change in dimensions caused by the contraction of the cast metal as it cools. This shrinkage allowance is the factor with which the moldmaker is concerned. Different casting alloys have different shrinkage rates. Although average values for unrestricted cooling conditions are available (Table 2), it must be remembered that these can vary considerably according to the resistance offered by molds and cores. Mold Wall Thickness. Figure 5 can be used as a general guideline for determining the proper mold wall thickness for water-cooled steel molds. Wall thicknesses may vary depending on the specific equipment and application involved. Mold wall thickness is independent of casting wall thickness for water-cooled steel

16

18

in.

/ 9 /32 15 /32 21 /32 15 /16 1 /18 3 16

molds. Casting wall thicknesses exceeding 305 mm (12 in.) have been made using these recommended mold wall thicknesses. End plate thicknesses depend on casting wall thickness and end plate material. Commonly used end plate materials are ASTM A36 steel, 1015 steel, or 1018 steel plate. In cases of extremely large-diameter plates (>450 mm, or 18 in.), the end plate material can be pressure-vessel-quality ASTM A285 grade C or A515 grade 70. For maximum adhesion of the mold wash to flat surfaces of the mold top and bottom end plates, it is advisable to have a machined surface finish of approximately 9 to 13 mm (350 to 500 min.) rms, with the lay approximately circular relative to the center of the plate. The opening in the top plate should be at least

Vertical Centrifugal Casting / 683

Aluminum alloys Aluminum bronze Yellow brass (thick sections) Yellow brass (thin sections) Gray cast iron(a) White cast iron Tin bronze Gun metal Lead Magnesium Magnesium alloys (25%) Manganese bronze Copper{-}nickel Nickel Phosphor bronze Carbon steel Chromium steel Manganese steel Tin

mm/m

13 21 13 16 8–13 21 16 11–16 26 21 16 21 21 21 11–16 16–21 21 26 21

in./ft

/32

5

¼

/32 3 /16 1 /10–5/32 5

¼

/16 1 /8–3/16 5 /16 3

¼

3

6

9

12

15

18

21

24

Mold inside diameter, in. 27 30 33 36

39

42

45

48

51

54

57

4

89

3½

76

3

64

2½

51

2

38 76

152

229

305

381

457

533

610

686

762

838

914

991

1067

1143

1219

1295

Mold wall thickness, in.

Shrinkage allowance Metal or alloy

102

Mold wall thickness, mm

Table 2 Guidelines for shrinkage allowances for various metals and alloys

1½ 1372 1448

Mold inside diameter, mm

Fig. 5

Mold wall thickness as a function of mold inside diameter. Values are recommendations from one manufacturer for steel molds.

/16

3

¼ ¼ ¼ 1 /8–3/16 3 /16–¼ ¼

/16

5

¼

(a) Shrinkage in cast iron depends mainly on the speed with which the casting cools. Greater shrinkage results in more rapid cooling, less shrinkage from slower cooling.

13 mm (½ in.) smaller in diameter than the casting inside diameter but large enough to permit the molten metal to enter the mold. The end plate recess in the mold provides an effective seal to prevent leakage of the molten metal from the mold. The amount of end plate recess used depends on mold inside diameter and can be as high as the mold inside diameter plus 25 mm (½ in.) for mold inside diameters of 150 mm (6 in.) or more. End Plate Mold Lock Design. Various removable end plates are shown in Fig. 4. These end plate types are recommended for all vertical molds. Preferred end plate types are those shown in Fig. 4(a) to (c). Some end plates can be turned over to prolong their service lives, and it is recommended that they be turned over between each cast. Mold end plate diametral clearance fit into the mold recess diameter should be a minimum of 1.2 mm (3/64 in.). This clearance is effective for end plates up to approximately 380 mm (15 in.) in diameter. On larger end plates, clearances as large as 2.5 mm (3/32 in.) may be necessary for easy removal and assembly. The standard draft angle of 3
will permit easy removal and assembly. The number of fasteners or mold clamps or wedges used depends on mold diameter. Recommended numbers of fasteners are:  Three for molds up to 250 mm (10 in.) bolt

circle or outside diameter

 Four for molds from 250 to 380 mm (10 to

15 in.) bolt circle or outside diameter

 Five for molds from 380 to 480 mm (15 to

19 in.) bolt circle or outside diameter Large-diameter molds or molds for extremely heavy wall castings (>50 mm, or 2 in., in wall

section) require additional fasteners for safe operation. End Plate Wedge and Pin Design. Wedges for holding the end plate secure in the mold are principally of two different types: the tapered wedge and the tapered pin. The maximum allowable stress for tapered wedges and pins is 69 MPa (10 ksi). Wrought steel bar stock, such as cold rolled carbon steel or stainless steel, can be used; cast pins or wedges should not be used. Mold Adapter Tables. A mold adapter table is usually furnished or required with a vertical centrifugal casting machine to facilitate attachment of the molds to the machine. Also furnished or required with the mold adapter table is a mold centering boss or index boss/plug for aligning and centering the mold so that it is concentric with the spinner shaft of the machine. Two methods can be used to secure the mold to the adapter table. One method is to allow for a flanged extension of the bottom mold plate with holes for bolting directly into the table. The other method is to use a flanged extension of the mold bottom plate with dog clamps fastened to the table in the same manner in which a part is held on a mill table or vertical lathe. The vertical centrifugal casting machine is available with a water-cooled bottom mold plate. Water cooling requires the machining of radial grooves or slots in the bottom of the bottom mold plate for water passage. Water cooling is sometimes advantageous for extremely heavy wall castings, which can transfer excessive heat into the bottom of the mold. Pretreatment for Metal Molds. All molds, end plates, and surfaces to which mold wash is to be applied should be treated before use in the following manner:  Preheat mold to approximately 150
C (300
F).  Swab inside surface of mold with concen-

trated aqueous ammonium persulfate.

 Flush mold with water to remove all con-

taminants. The mold can be put into service after this treatment. For new molds, the following treatment is recommended before use:  Preheat mold to 205 to 260
C (400 to 500
F ).  Spray with a coating of mold wash approxi-

mately 0.8 mm (1/32 in.) thick.

 Brush out mold wash and repeat spray pro-

cedure. The mold should then be brushed out and resprayed a third time. The mold is ready for use after this treatment. If the wash still does not adhere, the mold should be cleaned again with ammonium persulfate as described previously. Mold Wash for Permanent Molds. The mold wash, when used on a permanent mold, serves as a refractory insulating material. When applied in sufficient thickness, the mold wash insulates the mold, thus reducing the surface temperature of the mold and increasing its useful life. A wide variety of centrifugal casting mold washes are commercially available, including silica, zirconia, and alumina washes. The centrifugal mold wash must be inert to the molten metal being cast. The insulating characteristic of the mold wash is necessary to retard or slow the initial solidification rate in order to eliminate the formation of cold shuts, laps, droplets, and so on, and to produce a high-quality homogeneous outside surface on the casting. In most cases, a mold wash coating thickness of 0.8 mm (1/32 in.) is desired to obtain satisfactory castings. Spraying equipment is available for applying centrifugal casting mold washes. Water Cooling of Permanent Steel Molds. The temperature of the mold increases with each casting poured. The mold can be operated at temperatures to 370
C (700
F ). However, the usual operating temperatures of the mold should range from 150 to 260
C (300 to 500
F ). To maintain this mold temperature for successive casting production, water cooling of the outside surface of the mold is generally employed. The high velocity of the water impinging on the mold surface will prevent the formation of an insulating steam barrier, which would actually inhibit the extraction of heat from the mold. Quick-opening valves mus be used to prevent warpage of the mold, and the spinning of the mold should be interlocked with the water valves to prevent spraying water on a mold that is not spinning and ultimately warping an expensive mold. In practice, the water cooling is usually activated immediately upon completion of the pouring and allowed to remain on long enough to permit

684 / Centrifugal Casting Total cycle time between pours should be as long as possible to allow the graphite mold to cool to 95 to 150
C (200 to 300
F), thus reducing the time that the graphite will be above its oxidizing temperature of 425 to 480
C (800 to 900
F). Permissible rebores are very influential in determining total mold life. If a casting size is of such tolerance that rebored molds cannot be considered and if there is no larger size of bushing or casting for which the mold can be used, it is obvious that total mold life will be considerably shortened.

Process Details Casting Inside Diameters. When making castings on a vertical centrifugal casting machine, the inside diameter (bore) of the

casting will be tapered in accordance with the following formula: sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi h r21  r22

n ¼ 264

(Eq 1)

where n is the speed of rotation (in revolutions per minute), r1 is the inside radius at the top of the casting (in inches), r2 is the inside radius at the bottom of the casting (in inches), and h is the casting height (in inches). Actually, if the length of the casting does not exceed approximately twice its inside diameter, the amount of taper will be negligible. The optimal speed of rotation results in a centrifugal force of 75 g (75 times the force of gravity) on the inside diameter. It can be seen from Eq 1 that too slow a speed of rotation will result in excessive taper on the inside diameter of the casting.

Casting inside diameter, mm 4000

00 000 2

25

00 300 000 1 1

15

0

80

0

50

0

40

0

30

0 5 0 5 0 20 17 15 12 10

75

50

0 800

0 0) 60 00 (2

00 00 1

16

3000

)

0 ) ) 0 00 ) 20 ) 00 00 50 15 000 1 000 (35 (30 (2 (4 (5

40

25

2000 1500

1000 800 0g

600

30 g 200 g 160 g g 120

500 Mold speed, rpm

400 300

140 g 100 g 75 60 g

45

g

g 250 g 180

15

g

5g

30

g

200 1g

5 (1 00 50 0)

150

100 80 70 60 50

30 0 2 (7 00 (1 50 00 ) 0) 15 (5 0 00 )

the mold to be sprayed with the mold wash at the proper temperature range for the next casting cycle, after extracting the solidified casting. Graphite and Carbon Molds. The choice between using a graphite or a carbon mold depends primarily on the availability to the user. Graphite has a higher rate of heat conductivity than carbon and is therefore sometimes used because of the desired metallurgical properties of the finished casting. Graphite can be easily machined into a variety of intricate mold forms with a very good surface finish. Graphite molds have excellent chill characteristics, with thermal conductivity three times that of iron and a specific heat approximately double that of iron. The chilling ability of a material is roughly equal to the product of its mass multiplied by its specific heat. Graphite is nonreactive with most molten metals. In casting phosphorus bronze, graphite molds are not burned into as iron molds are. Carbon pickup is negligible in casting low-carbon stainless steel because of the quick chilling ability of the graphite, which almost instantaneously solidifies a skin layer of steel and therefore makes carbon pickup impossible. In certain cases where rapid chilling is not desired, an insulating mold wash is used. Graphite molds are extremely resistant to thermal shock, with a thermal conductivity approximately three times that of steel and a low Young’s modulus. The strength of graphite molds increases with temperature. Graphite molds are generally designed with two considerations: minimum wall thickness and weight ratio of mold to casting. A steel sleeve or master die holder is usually employed with a heat shrink-fit over the graphite mold. The steel is preheated to 425
C (800
F) for this shrink-fit; therefore, there is considerable compression on the graphite mold. The graphite mold wall thickness must be sufficient to withstand the stress of the shrink-fit without breaking. For heat shrink-fit applications, the graphite mold wall should not be thinner than 19 mm (3/4 in.). The second factor is the minimum weight ratio of mold to casting to reduce the effects of graphite oxidation and to obtain the proper chilling effect. There are many factors involved in arriving at this ratio, but in general, the ratio should be a minimum of 0.75. The factors that would influence a larger ratio include the heavy load imposed because of rapid succession of casts, greater chill depth desired, availability of graphite mold stock, and general flexibility in machining. The graphite mold can be machined before or after the metal jacket or steel sleeve has been encased around the graphite. The expected life of a graphite mold depends on three major factors: stripping time, total cycle time, and permissible rebores. Stripping time should be as long as possible to permit the casting to shrink away from the mold, particularly if the casting diameter is less than 150 to 200 mm (6 to 8 in.). With this precaution, the casting, with burrs, will not score the graphite mold wall nearly as much as when extracting the casting immediately after solidification when there is only negligible casting shrinkage.

Centrifugal force, multiples of the force of gravity, g Peripheral speed, m/min (ft /min)

40 30

20 100 80

60 50

40

30

20

15

10

8 7

6

5

4

3

2

1.5

1

Casting inside diameter, in.

Fig. 6

Nomograph for determining mold speed based on the inside diameter of the casting and the required centrifugal force. See text for example of use.

Vertical Centrifugal Casting / 685 There are castings for which it is desirable to cast the inside diameter with a predictable taper. Using Eq 1, the exact speed can be calculated to produce an approximate given taper on the inside diameter of the casting. Speed of Rotation. To establish a temperature gradient of the molten metal from the outside diameter toward the center of rotation (that is, directional solidification), it is usually necessary for the mold to be spinning when the metal is poured. In some cases, in order to eliminate defects such as erosion and dirt in sand molds, it is desirable to pour at a slow speed of rotation. However, true centrifugal castings having a wall section of 12.7 mm (½ in.) or less must be poured at spinning speed because the metal in this thin section solidifies quickly. Nomographs are available for determining the proper speed of rotation for centrifugal casting. However, Eq 2 can be used to calculate spinning speed: g ¼ 0:0000142Dn

Crane

Mold lifting tong

Casting

Mold

(Eq 2)

where g is the centrifugal force (in pounds per pound or number of times gravity), D is the inside diameter of the casting (in inches), and n is the speed of rotation (in revolutions per minute). Equation 2 can be easily manipulated to solve for speed. Mold Speed Curves. Mold speeds are determined by the inside diameter of the castings to be made. The mold speed curve shown in Fig. 6 is based on the inside diameter of the casting. The length of the casting is not considered in determining mold speed. For example, the mold speed for producing a casting 100 mm (4 in.) in outside diameter by 75 mm (3 in.) in inside diameter at a centrifugal force of 60 g is calculated as follows. Find the 3 in. diameter at the bottom of the curve. Move vertically from this point until the 3 in. line intersects the diagonal line marked 60 g. From this intersection, move directly to the right-hand edge of the curve; the speed of rotation of the mold in this case should be 1150 rpm. From the standpoint of safety for vertical centrifugal casting, it is highly recommended that the g force acting on the outside diameter of the mold be considered. It is safe practice to limit this force to approximately 200 g on the outside diameter of the mold. Pouring Techniques. For permanent molds, the metal is generally poured approximately 40
C (100
F) higher than the temperature used for the same casting if poured statically in a sand mold. This is because of the more rapid chilling effect of permanent molds. The pouring rates required for successful permanent mold centrifugal casting are quite high compared to those for static casting in sand molds. It is particularly important that the rate of pour at the beginning be very high to prevent cold laps and cold shuts. For most types of centrifugal castings weighing less than 45 kg (100 lb), a pour rate of approximately 9 kg/s (20 lb/s) is recommended. For castings weighing up to

Machine table

Fig. 8

Fig. 7

(a) Typical pouring spout design and (b) pouring spout positioning during casting. ID, inside diameter; OD, outside diameter

450 kg (1000 lb), an initial pour rate of 9 to 23 kg/s (20 to 50 lb/s) is recommended. For castings weighing more than 450 kg (1000 lb), pour rates of 45 to 90 kg/s (100 to 200 lb/s) are recommended. When pouring into a vertically spinning mold, it is important to introduce the molten metal into the mold in such a way as to prevent or minimize turbulence of the molten metal, which can cause splashing, spraying, or droplets and can result in undesirable casting defects. Although many vertical centrifugal castings can be poured directly into the mold from the ladle to produce a quality centrifugal casting, it is more often desirable to use a pouring funnel. With a pouring funnel, the nozzle can

Lifting tong assembly used to extract centrifugal castings from the mold

be lined to the required diameter so that, with a certain riser height of molten metal in the funnel, a controlled pour rate can be obtained for a particular casting weight. In addition, with a pouring funnel, the entry of molten metal into the mold can be made to impinge on the body of the mold with initial metal flow in the direction of mold rotation. This type of pouring will provide superior casting quality by minimizing or eliminating any upsetting turbulence in the flow of molten metal that may cause defects. Figure 7 shows a pouring funnel and funnel position. Extraction of Castings. Commercially available casting pulling tongs (Fig. 8) are recommended for extracting vertical centrifugally cast castings. These pullers engage onto the inside diameter of the casting and are used to lift the casting from the mold. Centrifugal force increases directly as the square of the speed of rotation and directly as the length of the radius from the axis of rotation. Centrifugal force can be tremendous and very destructive. Speeds of rotation should never exceed those required to produce the casting. Molds must be centered on the spinning axis as accurately as possible and must be statically or dynamically balanced, if necessary. The method of attachment of both bottom and top cover plates to the mold is of great importance for withstanding and containing

686 / Centrifugal Casting the force of the molten fluid metal while spinning. All clamping arrangements should be designed to tighten with, or to be unaffected by, centrifugal force. Molds should be firmly clamped to the table, because unbalanced molds may fly off the table during operation. Adequate safety guards with interlocks should be used around all machines to protect workers

from molten metal, which can spray from the mold if too much metal is poured. Heat expansion can occur suddenly and rapidly in the mold body, top and bottom mold plates, and even into the centrifugal casting machine. This sudden expansion can put bending and shearing stresses into fasteners and other retention and clamping devices. A

thorough understanding of the forces involved in centrifugal casting is necessary to ensure the utmost in safety for all concerned. Moisture in a sand mold or moisture in the mold wash can turn into steam when the molten metal contacts it, and the resulting forces would be incalculable.

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 689-699 DOI: 10.1361/asmhba0005260

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Permanent Mold Casting Revised by Don Tyler, General Aluminum Manufacturing Co., and Robert Pischel, FOSECO Metallurgical, Inc.

THE PERMANENT MOLD CASTING PROCESS involves the production of castings by pouring molten metal into permanent metal molds using gravity, low pressure, vacuum, or centrifugal pressure. Simple reusable cores are usually made of metal. More complex cores are made of sand, plaster, ceramics, or even salt. When nonmetal destructible cores are used, the process is called semipermanent mold casting. Basically, the process involves the following steps: 1. A refractory wash or coating is applied to the surfaces of the preheated mold that will be in direct contact with the molten metal alloy. 2. Cores, if applicable, are inserted, and the mold is closed either manually or mechanically. 3. The alloy, heated to above its melting temperature, is introduced into the mold through the gating system. 4. After the casting has solidified, metal cores and loose mold members are withdrawn, the mold is opened, and the casting removed. 5. Steps 2 to 4 are repeated until repair of the refractory coating is required, at which time step 1 is repeated to the extent necessary. 6. The usual foundry practice is followed for trimming gates and risers from the castings. A basic difference between sand casting and permanent mold casting is the metal molds for the latter process. These metal molds are usually made of iron or steel and have a production life of 10,000 to 120,000 or more castings as opposed to sand casting, which requires a new mold for each casting produced. Permanent mold casting is particularly suitable for the high-volume production of castings with fairly uniform wall thickness and limited undercuts or intricate internal coring. The process can also be used to produce complex castings, but production quantities should be high enough to justify the cost of the molds. Compared to sand casting, permanent mold casting permits the production of more uniform castings, with

closer dimensional tolerances, superior surface finish, and improved mechanical properties. In many applications, the higher mechanical properties can justify the cost of a permanent mold when production volume does not. Figure 1 shows castings made by the permanent mold process (Fig. 1a) and the semipermanent mold process (Fig. 1b). Metals that can be cast in permanent molds include the aluminum, magnesium, zinc, and copper alloys and hypereutectic gray iron. Aluminum. The vast majority of permanent mold castings are made in aluminum. Therefore, the major content of this text is devoted to aluminum. High-production permanent mold castings weighing up to 70 kg (150 lb) have been made from aluminum alloys. However, much larger castings can be produced. For example, aluminum alloy engine blocks with a trimmed weight of 354 kg (780 lb) have been produced in a four-section permanent mold having a vertical parting line. Magnesium alloys, despite their comparatively low castability, have been cast in permanent or semipermanent molds to produce relatively large and complex castings. For

example, an 8 kg (17.7 lb) housing for an emergency power unit was poured from alloy AZ91C (UNS M11914) in a semipermanent mold. In another application, 24 kg (53 lb) spoolhead castings 760 mm (30 in.) in diameter were produced from alloy AZ92A (UNS M11920) in a two-segment permanent mold with vertical parting. Copper casting alloys from the silicon bronze, aluminum bronze, and high-copper families are often cast in permanent molds. Alloys with a narrow solidification range are preferred. Conventional materials for these permanent molds are H13 and beryllium-copper, and new materials are being investigated (Ref 1). Permanent mold casting of copper alloys has been more prevalent in Europe and Asia than in North America, but it is gaining wider acceptance. Castings up to 13.6 kg (30 lb) are cast routinely. Larger castings are increasingly difficult and above 22.7 kg (50 lb) are considered impossible. Gray Iron. The production of gray iron castings in permanent molds is seldom practical when the castings weigh more than 13.6 kg (30 lb).

(a)

(b)

Fig. 1

(a) Examples of permanent mold applications: control arm, brake master cylinders and calipers, steering knuckles, valve housing, and engine mounting brackets. (b) Examples of semipermanent mold applications: intake manifolds, thermostat housing, transmission pump housing, engine mounting bracket, and oil pan component. Courtesy of General Aluminum Manufacturing Co.

690 / Permanent Mold and Semipermanent Mold Processes

Gravity Casting Methods Manually operated permanent molds for low production may consist of a simple book-type mold arrangement (Fig. 2a). For castings with high ribs or walls that require mold retraction without rotation, the manually operated device shown in Fig. 2(b) can be used. With either type of device, the mold halves are separated manually after releasing the eccentric mold clamps. Semiautomatic Devices. For high-volume production, manual drives are replaced by two-way air or hydraulic mechanisms. These units can be programmed to open and close in a preset cycle. Therefore, the operation is automatic except for pouring of the metal and removal of the castings. Automatic Tilt Pouring. The mold parting for the static casting devices shown in Fig. 2 is in a vertical plane, which is often the preferred position for mold opening and casting removal. Many castings, however, benefit greatly from tilt pouring, which provides significant advantages over statically poured castings in a vertical plane (Fig. 3). Tilt pouring offers advantages in less pouring turbulence, more consistent filling speeds, less reliance on operator skills, and it facilitates full automation. Also, gating systems can be minimized in many applications; that is pouring gates can also serve as feed risers, and many castings can be poured directly through the riser into the mold cavity without excessive turbulence. An example of a standard commercially available tilt device is shown in Fig. 4. These machines are available in a number of sizes and configurations that are suitable to a large percentage of permanent mold castings. They can accommodate front and back ejection and side, bottom, and top core pull mechanisms. They can be used as individual stand-alone machines as well as in multistation, turntable setups. Figure 5 shows a model of an eight-station automated turntable with A-frame tilt machines. Such a turntable can be outfitted with robotic pouring, core setting, ejection, and casting removal. Steps in the casting process are completed progressively as the casting devices pass through the various work stations. This work cell is suitable for very high production with a minimum of manual labor. Such a setup can easily be run by one operator with assistance for die coating touchup. Identical castings can be run in all stations or a mixture of castings with similar solidification times.

Pouring basin

Web gate

Runner

Vent Core plate

Mold half

Rack Pinion

Sprue

Eccentric mold clamp

Gating system

Core pin (I of 5) Mold cavity Riser

Mold half

(a) Mold half

Gearbox for mold mover

Alignment pin (I of 2)

Pouring basin

Gating system Ejector plate

Eccentric mold clamp

Gearbox

Ejector pin (I of 4) Mold half Base

Core

Alignment key

(b)

Fig. 2

Two types of manually operated permanent mold casting machines. (a) Simple book-type mold for shallowcavity castings. (b) Device with straight line retraction for deep-cavity molds

Molten aluminum Mold cavity Note: Insert may be needed to produce bottom pour ladle effect

Plane parallel to floor

Advantages and Disadvantages There are numerous factors that must be considered when choosing a casting process. Therefore, the decision should be based on thorough engineering analysis and production cost studies for each casting. Due to more rapid heat transfer and solidification times, castings

Fig. 3

Cross section of a mold and pouring basin of a tilt-pouring machine. Pouring basin is filled when mold is horizontal. Mold is tilted to upright position to fill the cavity. Source: Adapted from Ref 2

Permanent Mold Casting / 691

Front ejection Ejection platen

Casting catcher

Fig. 4

Tilt device with casting catcher and front and back ejection mechanisms. Courtesy of CMH Manufacturing Co., Lubbock, Texas

contain entrapped gas such as is possible in high-pressure die castings. Therefore, they are superior to die castings in soundness and pressure tightness and are sometimes chosen for superior quality benefits. However, die castings achieve nearer net shape, require less draft, and can be produced at a lower cost. The lower cost is due to shorter cycle times and reduced degate and finishing costs. Degating is the removal of gates, runners, and risers (feeders). Die casters use trim dies and presses to accomplish this. In permanent mold casting, trim dies and presses are used more frequently for parting line flash removal, but in some cases, where gating contact thickness will allow it, they can also work for degating. The more common methods would be belt sanding, sawing, turning, and milling or a combination of these coupled with robotics for high-production applications. These operations are obviously slower than trim press operations. If part complexity, surface finish, and dimensional accuracy are critical, an investment casting may be more suitable. Extremely complex or irregular shapes may be difficult or even impossible to cast in permanent molds. For very complex shapes with intricate internal coring, sand casting may offer the best solution.

Applications Core set station Eject station

Pouring station

Clockwise table rotation

Fig. 5

Drawing of an eight-station turntable or carousel for automatic permanent mold casting with robotic pouring and ejection. Courtesy of CMH Manufacturing Co., Lubbock, Texa

produced in permanent molds have finer dendrite arm spacing, grain structure, and higher mechanical properties than those cast from a similar alloy in sand. This advantage can translate into thinner walls and significantly lower-weight designs. In addition, cast surfaces

are generally smoother than sand castings, and closer dimensional tolerances can be maintained. When properly produced, permanent mold castings contain less shrinkage and gas porosity than sand castings, and they do not ordinarily

Aluminum permanent mold castings are used widely throughout industry. Oil pans and engine cradles, pistons, cylinder heads, intake manifolds, and other functional parts of internal combustion engines for the automotive, trucking, diesel, and marine industries are typical applications. Other major uses for permanent mold castings include internal and accessory parts for reciprocating and jet-type aviation engines; missiles; forms for concrete; textile machine parts; electric motor housings; portable and hand tool components; support members for outdoor light standards; electric griddles and kitchen pots and pans; and a host of other commercial applications. Since the 1990s, there has been major growth in automotive applications. Prompted by the need to improve gas mileage, aluminum has replaced iron and steel for weight-savings benefits. Uses include cross members, subframes, wheels, and other safety-critical applications such as front and rear steering knuckles and control arms, brake calipers, and master cylinders.

Casting Design Good permanent mold casting designs generally follow the same guidelines as for any casting. The designer needs to have a good knowledge of the characteristics of the many available casting alloys and the various heat treatments and their effect on the production and performance of the casting. There are many sources of information available, including

692 / Permanent Mold and Semipermanent Mold Processes ASM International (Ref 3), the American Foundry Society (Ref 4), and SAE International (Ref 5). For specific materials, the Aluminum Association (Ref 6) and the Copper Development Association (Ref 7) provide information and resources. Among the basic rules for good casting designs are:  Limit drastic changes of thickness to mini-

mize stresses and aid in the casting process.

 Taper sections as liberally as possible    

toward risers to help control directional solidification. Do not interpose thin sections between thick sections and access to risers. Avoid extensive horizontal, flat surfaces. Avoid isolated thick sections that create hot spots that are difficult to feed. Determine casting properties and use commensurate safety factors.

Mold Design A number of factors must be considered when designing and building molds for sound permanent mold castings to be produced efficiently and cost-effectively. These factors are:

Multiple-Cavity Molds. The number of castings per mold is a major consideration in designing the mold; the objective is to have the optimal number of cavities per mold that will yield acceptable castings at the lowest cost. Except for very small and thin castings, the machine cycle time increases as the weight of the metal being cast per mold increases. However, these increases are not directly proportional. A mold with the maximum number of cavities will often produce more castings per unit of time than a mold with a smaller number of cavities that was designed to operate on a shorter cycle. This is because there is a minimum solidification time for every casting, regardless of the number of cavities in the mold. The number of rejects sometimes increases as the number of cavities is increased, but this is usually offset by the greater productivity. Venting. The gap that exists between the mold halves after closing is sometimes large enough to permit air to escape and thus prevent misruns and cold shuts. Frequently, however, vents must be added to allow air to escape as the mold fills. Ejector pins provide an excellent opportunity for venting air. Figure 8 illustrates how ejector pins can be ground with flats for venting purposes, along with two other vent

back (outside) of the mold so that the exterior conforms roughly to the cavity. This permits a more even temperature distribution and heat dissipation. The maintenance of a uniform equilibrium temperature during operation is critical to consistent production of high-integrity castings and is a function of consistent cycle times as well as the contoured mold thickness. Inconsistent or interrupted cycle times result in a loss of progressive temperature gradients and a nondirectional solidification pattern. For castings with heavy sections, the adjacent mold sections are generally heavier. For aluminum castings, a ratio of three or four mold wall thicknesses to one casting wall thickness is often used, but a mold wall this thin cannot always be used in making thin-wall castings without jeopardizing mold stability. Thin-wall castings are sensitive to temperature changes and will misrun readily. Consistent, relatively high mold temperatures are required; this necessitates the use of thicker, more massive molds. Suggested thickness based on cast wall thickness and weight is given for aluminum in Fig. 6. Mold temperature will vary with mold thickness and the weight and section thickness of the casting. Figure 7 gives the relative effect of mold thickness on operating temperature and solidification time.

 Cavity dimensions must compensate for the

shrinkage that occurs as the casting cools.

 Undercuts on the outside of a casting com-

 

 

Mold Wall Thickness. In designing a permanent mold, the part is laid out in the desired orientation, and the mold is designed around it, allowing sufficient space for gating and sealing to prevent metal leakage and for coring and inserts. It is common practice to contour the

Aluminum (water cooled mold for 1.3 mm, or 0.50 in. and over) 0 6 13 4

Weight of part being cast, kg

19

25

100

3

75

2

50

1

25

0 0

0.25

0.50

0.75

1.00

Wall thickness of casting, in.

Fig. 6

500

0

0

Mold wall thickness, mm

Mold wall thickness, in.



Wall thickness of casting, mm

0

0

Entire cavity in one mold sector 45 90

50% of cavity in one mold sector 135 180

100 200 300 Weight of part being cast, LB

400

Suggested mold wall thickness based on casting wall thickness and casting weight. Source: Adapted from Ref 2

13

Mold thickness, mm 25 38 51 64

Mold thickness, mm 13 25 38 51 64

0

0

Mold thickness, mm 13 25 38 51 64 38 mm (1.50 in.) casting

1.9 mm (0.075 in.) casting

0.63 mm (0.025 in.) casting

Solidification time, s



plicate mold design and increase casting cost because additional mold parts or expendable cores are needed. Complicated and undercut internal sections are usually made more easily with destructible cores than with metal cores, although collapsible steel cores or loose metal pieces can sometimes be used instead of expendable cores. Proper selection of mold material and wall thickness is necessary for economy and consistent mold temperature profiles. The gating system carries the metal into the mold cavity at the proper location and fill rate to promote directional solidification. The riser (feeding) system contributes to proper temperature gradients and availability of molten metal to feed the solidifying metal front. Venting must be placed advantageously to allow air and gas to escape ahead of the poured metal. Chills, which are sometimes inserted in the mold to initiate and accelerate solidification in desired locations and directions, must be properly sized and located. Antichills are occasionally needed. Use of both must be judicious.

400 6

300 200

4.5 Cycle time 1.5 min

8

3.5

6

2

10

6

100 10

0 0

Fig. 7

0.5 1.0 1.5 2.0 2.5 Mold thickness, in.

0

0.5 1.0 1.5 2.0 2.5 Mold thickness, in.

0

0.5 1.0 1.5 2.0 2.5 Mold thickness, in.

Relationship of solidification time to mold thickness for various casting section thicknesses for aluminum. Source: Adapted from Ref 2

Permanent Mold Casting / 693

Ejector pin with flat ground to allow venting

Grade 5 bolt

Weld in place

Hex stock

Screen mesh Brass body Slots

Steel body

Sintered vent

Fig. 8

Various vent types used in permanent molds. Courtesy of General Aluminum Manufacturing Co.

Table 1

Recommended permanent mold materials for aluminum Number of pours

Casting size

1,000

10,000

100,000

Small castings 25 mm (1 in.) max dimension

Gray iron(a)

Gray iron with H 14 inserts, H 11, H 13(a)

Medium and large castings

Gray iron(a)

Gray iron(a) H11 die steel SAE 1020 SAE 4140 Gray iron(a) SAE 1020 SAE 4140

Gray iron, H 11, H 13(a)

(a) Meehanite types of gray iron are also suitable. Source: Adapted from Ref 2

Direction of core pull

(a)

Lower cost

Direction of core pull

(b)

Higher cost

Fig. 9

Two types of cavities requiring different coring methods. (a) Corner radii must be sacrificed if metal cores are used. (b) Corner radii can be obtained with sand, plaster, or other expendable core material. Casting cost is higher for option (b) due to the additional operations of core making and removal.

designs that are common for metal molds. Mold coatings with significant texture also provide venting action. However, for chronic areas of misruns, a vented mold works best and prevents the overuse of coatings for this purpose. During operation, coatings lose some of their texture, and thereby, the venting is lost unless the coating is repaired. This practice exposes the mold surface to coating buildup and overspray, which can result in areas of higher insulation, hot spots, and interrupted solidification patterns.

Mold Materials Mold materials are usually chosen on the basis of the number of castings to be produced and the cost of the material and the cavity machining. As shown for aluminum casting (Table 1), gray iron is usually the material of choice for low-production applications. Gray irons also offer more rapid heat-transfer properties than tool steel, which can translate into

shorter cycle times and faster solidification rates and more even die heating than tool steel. Gray iron is also less prone to attack by molten aluminum, can be machined for a fraction of the cost of tool steel, and can be readily cast close to size with the desired contoured wall thickness. However, H13 tool steel is by far the most popular mold material for high-production applications. H13 is the most wearresistant material and has proven to be the most cost-effective material on a per casting basis. It provides the best dimensional accuracy, resistance to thermal stresses, and repairability. By contrast, gray iron is more susceptible to thermal stresses and is very difficult to weld repair. Cores. The cores used in the permanent mold process can be of gray iron, steel, sand, or plaster. Metal cores can be movable or stationary. Stationary cores must be perpendicular to the parting line to permit removal of the casting from the mold, and they must be shaped such that the casting is readily freed. Metal cores not perpendicular to the parting line must be movable so that they can be withdrawn from the casting before it is removed from the mold. When sand or plaster cores are used in a permanent mold, the process is referred to as semipermanent molding. With sand or plaster cores, which are expendable, more complex shapes can be produced than with metal cores. The two cored passageways shown in Fig. 9 illustrate the limitations of cores in permanent mold castings. The cored passageway shown in Fig. 9(a) could be produced with metal cores. Sand or plaster cores would be needed to provide the radius of the passageway shown in Fig. 9(b), which would also be more expensive to produce but would provide the desired design characteristics. Sand cores permit the casting of tortuous passages or of chambers or passages that are larger in section than the opening through the wall of the casting. Because sand cores are expendable, their removal presents no major problem but does require an added operation that impacts production costs. Plaster cores are used in semipermanent molds to provide a surface finish better than that obtainable with sand or coated-metal cores. In rare instances, ceramic cores or cores made from molten salt have been used to provide superior surface smoothness. Plaster and ceramic also have excellent insulating qualities and do not prematurely chill the metal in thin sections. Cored holes in permanent mold castings can usually be held to closer tolerances, on both size and location, than in sand castings. Movable cores and stationary cores can both be machined to close dimensions and can be accurately located. Draft requirements for holes formed by steel cores are 3 to 5 per side. Steel cores require the same coating as that applied to the mold; the dimensional accuracy of cavities made from coated steel cores is affected by the same factors that affect the

694 / Permanent Mold and Semipermanent Mold Processes accuracy of dimensions formed by coated-metal molds. Unless a core is stationary, the clearance that must be provided to permit its withdrawal from the mold permits core movement, which may affect dimensions. In terms of dimensional accuracy, sand cores have approximately the same limitations in permanent molds as in sand molds. Because a permanent mold is more rigid than a sand mold, it provides a more accurate seat for locating a core. Collapsible Cores. It is preferable to avoid designs that require the use of collapsible metal cores (multiple-piece cores) in permanent molds. These cores add to the cost of the castings, and their assembly and removal increase production time. Furthermore, dimensional variations can result from the use of collapsible cores, because they cannot be positioned as securely in the mold as single-piece cores or because of movement of core segments when the casting metal is poured. When castings must be designed to be made with collapsible cores, the designer should allow the loosest tolerances possible. Despite the disadvantages of collapsible cores, they are extensively used in making certain castings. For example, nearly all of the aluminum pistons made for the automobile industry in the United States are cast in permanent molds using fivepiece collapsible metal cores. Casting Ejection. Only the most simple permanent mold castings can be ejected from the mold with no mechanical help. Most castings are ejected by well-distributed ejector pins or are confined in one mold half during opening of the mold and then ejected by the withdrawal of the retaining core or cores. It is important that the casting remain in the correct mold half until ready for ejection. For castings that have similar geometry in both mold halves, there may be a need for a spring-loaded, push-off mechanism in one mold half in order to keep the casting on the ejection side during the opening sequence. Figure 10 illustrates a typical tilt pour, permanent mold rigging system with

Return pin

contoured mold back, mounting spacers, ejection plate and pins, and pouring cup. Gating and Riser System. A well-designed gating and riser system accomplishes the following objectives:  Provides rapid filling to prevent premature

metal stream

 Establishes suitable thermal gradients to



promote directional solidification

 Avoids erosion of the mold  Provides a hot reservoir of metal to feed

solidifying sections



finishing costs



 Maximizes casting yield and minimizes

The design of gating systems still remains as somewhat of an art that depends largely on the experience of the foundry design engineer and a set of internally developed trial-and-error rules and practices. One such commonly used beginning approach is modulus based. The volume-to-surface area ratios (modulus) of various parts of the casting are calculated and then used to establish cavity orientation and to design gate and riser size and placement. This is a very effective method that properly assumes solidification patterns that are dependent on the geometry of the part. Considering this information, the following is a set of “rules of thumb” that can be applied along with local foundry knowledge and practice to design a basic, vertically parted, permanent mold gating system:  Orient the thickest casting sections (high

modulus) toward the top of the mold for easy access to risers (feeders) and to provide hydrostatic pressure to thinner (low-modulus) sections.  Whenever possible, gate (fill the cavity) through side risers, thereby producing the Bumper bolt Ejector plate

Ejector pin

Mount spacer Male half

Leader pin

Female half

Fig. 10



freezing with a minimum of turbulence

 Keeps dross and inclusions out of the casting  Prevents air and gas entrapments in the

Carrier bolt

Return pin



Typical tilt pour permanent mold rigging system. Courtesy of General Aluminum Manufacturing Co.



hottest metal and highest mold temperature at these locations. Minimize the amount of metal flowing over any isolated local areas, which may cause hot spots and erosion of the mold material. Orient the low-modulus sections away from the top and side riser to promote directional solidification from that point toward the side and top, where the molten reservoir is available for feeding. Use surface-area-to-volume (SA/V) ratios when designing risers. Round or spherical risers have the lowest SA/V ratios and therefore remain molten longer for maximum feeding. Riser diameter should be at least 15 to 20% larger than the section being fed. Riser height rarely needs to be greater than 1.5 times the diameter. Side and top riser necks (in-gates) should be as close as possible to the casting cavity and be at least 70% of the section thickness.

These items represent general guidelines to begin the gating/riser design. Much more detailed information can be obtained from research papers and training literature from The American Foundry Society publications and their Cast Metal Institute. As expressed in an earlier section, tilt pouring provides a number of advantages over static methods and is the preferred gravity-pouring method for most successful aluminum foundries. The primary advantage is in the reduction of pouring turbulence, which minimizes potential air and oxide entrapments. Tilt pouring allows the metal to enter the cavity in a tranquil manner and can eliminate the need to slow down the metal flow by the use of down-sprue and choke systems. Filters. To avoid oxide contamination, it may be necessary to incorporate dross traps or filters into the gating system (Fig.11). Perforated steel screens or woven fiberglass have been commonly used for filtration and reduction of metal velocity. However, reticulated ceramic foam filters have proven to be much more effective in not only entrapping potential inclusions but also in reducing metal velocity and pouring turbulence. When properly applied, gating system filters of ceramic foam are the most effective means of controlling pouring turbulence and removing oxides, hard spots, and inclusions. The resulting benefits are improved fluidity, reduced solidification shrinkage, and improved mechanical and fatigue properties. However, it may be difficult to justify the per-piece casting cost for high-volume production. The next-best alternative is in-furnace filtration. Ceramic foam or bonded particle filters have been developed for use in dip wells or launder systems and offer many of the same benefits, except for pouring turbulence control. Design Tools. The advent of computer simulation modeling for the permanent mold process has provided an indispensable tool for the foundry engineer to optimize and prove out his gating and riser designs before the tooling

Riser

Sprue

Permanent Mold Casting / 695

Casting

Dross trap

Fig. 11

Examples of a dross trap (left) and a foam filter (right) designed into the gating system. Source: Adapted from Ref 8

is built. Obviously, this can reduce cost and lead times dramatically, problems are analyzed and resolved more quickly, and process optimization is achieved much sooner than traditional methods of trial and error. There are four basic steps to the modeling process:  Specification of the properties and initial

conditions of all materials used in the process  Development of a geometric model of the part to be cast, along with its rigging system and tooling intended for production  Computer simulation of the process  Evaluation of results and, if necessary, modification of the process to improve some aspect, such as part quality, cycle time, or gating optimization Computer modeling of the permanent mold process has progressed to the point that every aspect of the process can be simulated with very good accuracy. Considering the number and price range of computer modeling systems and services available, simulation work is economically feasible for virtually every permanent mold application. Directional Solidification. Alloys should be cast so that solidification takes place directionally and progressively toward the risers, which are generally to the side or on top of the casting. To achieve this solidification pattern, thinner sections of the casting should be away from the gating system, and heavy sections should be adjacent to it. Directional solidification can also be facilitated by proper attention to mold coating, that is, thinner, less insulating applications away from the risers and thicker, more insulating near and in the riser cavity (see the section “Mold Coatings” in this article). Appropriate use of metal chills, that is, high-thermal-conductive materials such as copper-beryllium, Anviloy, or even aluminum, can be inserted into the mold at strategic locations to reduce or eliminate hot spots that interrupt the solidification pattern. Water or air cascades and circulating water channels can be used in conjunction with metal chills or by themselves

to accelerate local heat transfer to eliminate hot spots and direct the solidification sequence.

Mold Coatings Mold coatings are applied to metal mold and core surfaces to serve as a barrier between the molten metal being cast and the surface of the mold during mold filling and solidification. The primary purposes of mold coatings are:  Protect the base mold metal from the liquid    

metal being cast Control metal flow Control heat transfer Control casting surface finish Influence casting release from the mold

Mold Coating Families. There are two basic families of mold coatings: insulating and lubricating. Combination coatings are sometimes used, but the properties of each are somewhat diminished when mixed. The control of heat transfer through all sections of the mold is the most important characteristic of mold coatings, because it allows for control of both the rate and direction of casting solidification. The relative degree of insulation depends on the physical properties of the raw materials used, the method of application of the coating to the mold, and the coating layer thickness. Mold coatings are composed of three basic components:  Refractory fillers: These are typically com-

posed of refractory powders such as titania, talc, mica, vermiculite, alumina, iron oxide, silica, and graphite.  Binder: The vast majority of mold coatings use sodium silicate with an appropriate SiO2/Na2O ratio. Alternate bonding materials may include potassium silicate or clays.  Carrier: Water of a controlled trace mineral content is the universal carrier. The mixture of fillers and their ratios is the first step to govern the relative degree of heat

transfer during solidification. However, all potential properties of mold coating cannot always be obtained from one coating formulation. Often times, certain desired performance characteristics are mutually exclusive, and it is impossible to formulate one coating to meet all casting demands. Differing casting property demands will often require the use of different coatings. Application. The compactness of the sprayed coating layer, which is governed substantially by the application method, influences the heat-transfer property of the as-applied coating. If the coating layer is not very compact, the contact between the particles is slight, and therefore, the permeability and insulation properties will be greater, but the durability will be reduced. Conversely, if the coating layer is very tight, it will be substantially more durable but less insulating. Most commercial mold coatings are supplied as a heavy suspension or paste, allowing for significant flexibility in customizing the required properties based on the rate of dilution in the foundry. Most silicate-bonded coatings can be stored indefinitely as long as they are covered and not allowed to either dry out or freeze. Exposure to freezing temperatures renders the bonding properties of the silicate useless, and they cannot be reconstituted. Surface Preparation. As in any coating procedure, surface preparation is key. If starting with a mold that has previously been in service, the spent mold coating (including residual silicate) must be removed. Various blasting media are available, including sand, metal shot, glass beads, calcined rice hulls, or nut shells. It is important that the blasting not be overly aggressive so as not to distort mold details or alter dimensions. Dry ice blasting, although suitable for touchup or light removal of damaged coatings while in service, is not aggressive enough for full recoat. The mold surface must be clean and free of grease and oil, including fingerprints. Typically, the mold is preheated to between 205 and 370  C (400 and 700  F). A light mist of water (or mild water/surfactant solution) is sprayed on the mold surface. This slightly oxidizes the mold surface and provides a better anchor for subsequent layers to adhere. Coating preparation for spraying can easily be done in the foundry. It is preferred to use deionized water because trace elements in local water supplies can result in coating inconsistency. Coating manufacturers will have recommended starting points for dilution. These can be varied to suit the specific needs of the foundry or casting. Typically, the more dilute the coating, the more durable it will be, while being slightly less insulating. Conversely, more concentrated coatings will be more insulating but less durable. Once diluted, most mold coatings will require some mechanical suspension through slow, continuous mixing. It is important to avoid any high-shear mixing that may trap air and potentially damage softer refractory particles.

696 / Permanent Mold and Semipermanent Mold Processes When the mold and the coating(s) have been prepared, the application can begin. Spraying is the preferred method of application. Although brushing can result in a thicker coating layer, it does little to actually reduce the rate of heat transfer. Furthermore, due to the physical nature of a brushed layer, it is far more susceptible to spalling, chipping, and flaking, which can actually create casting defects. When the mold has reached the appropriate temperature and preheat burners are removed, an optical pyrometer should be used to determine that the temperature of the mold has peaked and is actually beginning to decline. Primer coatings are often applied as underlayers to improve coating life. Specifically designed primers of high durability are usually recommended, although very dilute main insulating coatings can also be used. Lubricating coating should never be used as primers. Spraying should be done with the spray nozzle as square to the mold surface as possible. The spray application should entail the slow building up of numerous thin coating layers to the desired thickness rather than just a few heavy coats. This will ensure both maximum durability as well as insulation. The objective is to atomize the coating so that it will dry almost instantly on contact with the hot mold to develop the strongest possible bond. Typical coating thickness for maximum insulation in this type of thermal system is between 200 and 300 mm (0.008 and 0.012 in.). Excessive coating thickness will not significantly improve the rate of heat transfer. Excessive coating thickness may potentially be detrimental, producing a layer that will be more susceptible to increased flaking, cracking, or catastrophic failure. In gates, runners, and risers where high insulation is desired, it is recommended that a compositionally higher insulating coating be used in preference to spraying or brushing a heavier thickness of a less insulating coating. Spray pattern and angle of the spray gun/nozzle can also influence the deposition of the coating on the mold surface. Wherever practical, the spray gun should be held perpendicular to the surface being sprayed. Contact of the atomized spray at angles to the mold should be minimized. Optimal spray distance should be 60 to 90 cm (24 to 36 in.). Distances too close to the mold may result in “damp” coating or coating that bounces off the mold surface. Spray distances that are too great may result in a coating spray that dries too much as it travels through the air and is too dry to bond to the mold surface. The way the spray gun is held by the operator and moved across the mold surface also plays a role in the coating deposition pattern. To obtain optimal results, the spray gun should be moved with a constant velocity, that is, with a motion of translation imparted to the gun, wrist, and forearm alike, rather than just a rotating wrist motion around a vertical axis. The drawing on the left in Fig. 12 represents the coating distribution that would result from a perfect sinusoidal rotation of the spray

Rotation of spray gun Horizontal passes

Crossed

80

120 100

160 120

100

100

120

160

120

120

100 80

120

Thickness distribution of coating (average = 100)

Fig. 12

These drawings illustrate the differences in coating distribution across the mold surface as a function of spray gun motion. The drawing on the right illustrates a much more even coat as a result of crossed passes and constant velocity. Source: Adapted from Ref 9

gun by the wrist and shows a heavy buildup of up to 160% of the average deposit, formed near the extremities of the moving spray halo course. Even in the best conditions of the translating movement recommended previously, a similar, although less dramatic, buildup is certain to occur. It is advisable that the operator change direction of the spraying motion intermittently. Also, generously overlapping the passes by 50% or more will considerably improve the evenness at right angles to the travel of the spray gun. The sketch on the right in Fig. 12 shows how crossing the passes can improve coating evenness. This practice should be the rule when possible (on flat surfaces, for example). Coating thickness on hot molds can be measured by commercially available magnetic or ultrasonic measuring devices. Some of these devices may have limited use in high-temperature applications; however, adapting them to permanent mold coating may still be preferable to guessing as to actual coating thickness. When the mold is coated, it can be put into service by reheating it again to the desired operational preheat temperature. If the mold is allowed to cool to room temperature before it is put in service, the subsequent preheat to operating temperature should be done slowly to avoid thermal stress and shock to the coating layer. Coating life may vary considerably, not only from foundry to foundry but from casting to casting. Mold complexity and geometry, pouring temperature, cycle time, pouring rate, alloy, and nature of the coating play a part in the potential life of a coating. Catastrophic coating failure or wear are not the only limiting factors of coating life. Even if a coating remains intact on the mold, over time the coating layer will begin to compress due to thermal cycling and pressure. This results in the loss of internal coating porosity and begins to adversely affect the rate of heat transfer. Coatings that have lost a significant degree of their insulating capability over time while remaining intact on the mold are considered “dead.” In such a case,

the mold must be taken out of service, completely cleaned, and recoated. In other cases, only a section of the mold may experience excessive wear. In those cases, in-process touchup may be possible. There are two critical differences in touchup versus initial mold coating. First, the operating temperature of the mold is often greater for touchup than for initial coating, so as not to adversely affect production. This requires a more dilute version of the initial coating, since the atomized coating will now have to travel through a hotter environment before depositing on the mold surface. If the coating is too concentrated, it may completely dry before contact with the mold, leaving a powdery coat that will not adhere to the mold. Secondly, using a mask over the surrounding area to be touched up is often essential. This will ensure that overspray does not overbuild in areas that do not require the additional coating. In addition, other areas of the mold may be coated with alternate coatings for additional property requirements. Masking will ensure that those areas will not be compromised. Surface Quality. Coatings also improve the surface quality of the casting. However, there is often a compromise between surface quality (smoothness) and the ability to move metal over long, flat, thin sections, especially in aluminum and magnesium castings. For any skinforming alloy, such as most aluminum and magnesium alloys, a slightly rough coating surface texture promotes metal flow by constantly rupturing the initial metal skin during flow. Very smooth coatings often promote miss-runs in long, flat, thin castings where the solidifying metal skin can grow uninterrupted. Release. Lubricating coatings can be used over a primer or over an insulating coating to promote the release of the casting from the mold, especially in flat, deep-draw sections or areas with minimal draft. Insulating coatings should never be placed on top of lubricating coatings. By their very nature, the durability of lubricating coatings is less than for the vast majority of insulating coatings, so touchup on

Permanent Mold Casting / 697 a mold in service is quite common and necessary. In extreme cases, touchup between every casting may be required. Coating application is not necessarily a black art but rather a comprehensive set of applied physical parameters that can and should be closely controlled and monitored for best-practice reproducible results. Mold coating, however, still remains an issue of compromise, with casting and mold requirements that often compete against each other. It is always best to prioritize casting requirements, fulfilling each as possible in order, understanding that not all issues can be successfully satisfied with one coating. One measure of a good permanent mold foundry is its ability to control this element of production. It can spell success or failure just as much as other factors.

Mold Life Mold life can vary from as few as 100 pours to more than 250,000 pours, depending on the variables discussed later in this section. With proper care and maintenance, a mold for an aluminum piston made in H13 tool steel, for example, can be expected to produce 250,000 castings before requiring repair. After the production of 250,000 more castings, the repaired mold will require a major overhaul. With repeated repairs and overhauls, the mold can produce as many as 3.5 million castings before being discarded. However, a piston mold, with its relatively simple design, will have a much longer life than a mold that requires elaborate internal coring and external inserts, such as a cylinder-head mold. Mold life is likely to be longer in the casting of magnesium alloys than in the casting of aluminum alloys of similar size and shape; this is because molten magnesium does not attack ferrous metal molds. However, the difference in mold life for magnesium alloys depends to a great extent on the effectiveness of the mold coating used. In the casting of gray iron, mold life is expected to be short compared to the casting of similar shapes from aluminum alloys. Major variables that affect the life of permanent molds are:  Pouring temperature: The hotter the casting

metal is poured, the hotter the mold is operated, which leads to rapid weakening of the mold metal.  Weight of casting: Mold life decreases as casting weight increases.  Casting shape: Mold walls are required to dissipate more heat from castings having thick sections than from those having thin sections. When there is a significant variation in the section thickness of a casting, a temperature differential is set up among different portions of the mold. As the temperature differential increases, mold life decreases.  Cooling methods: Water cooling is more effective than air cooling, but it substantially decreases mold life due to thermal stresses.

 Heating cycles: Generally, a continuous run,





  







in which the mold is maintained at a uniform temperature, provides maximum mold life. Repeated heating and cooling over a wide temperature range will shorten mold life. Preheating the mold: This is done to operating temperature with a gas flame or electric heaters, and it greatly increases mold life. Thermal shock from starting a mold from well below operating temperature is one of the principal causes of mold failure. Mold coating: This protects the mold from erosion and soldering by preventing the metal from contacting or overheating mold surfaces, thus increasing mold life. Mold materials: See Table 1. Storage: Improper storage can lead to excessive rusting and pitting of mold surfaces, which will reduce mold life. Cleaning: The common practices for cleaning molds are abrasive blasting, dipping in caustic solution, and wire brushing. Dipping in caustic can be hazardous to the operator. Wire brushing and abrasive blasting can cause excessive mold wear if not carefully controlled. Glass beads are the safest abrasive blast material; their use minimizes dimensional changes due to erosion from the abrasive blast. The use of dry ice pellets has come into prominence as a blasting medium and is by far the least abrasive material. However, it may be periodically necessary to use a more abrasive medium on certain coatings with the best adherence properties. Gating: A poor gating system can greatly reduce mold life by causing excessive turbulence, erosion, and excessive localized hot spots at the gate areas. Method of mold operation: Although the same materials are used to make molds and cores for both automatically operated equipment and hand-operated equipment, the life of the tool materials on hand-operated equipment is shorter because of the abuse the tooling must withstand. Tools for automatic equipment may last up to twice as long as for hand-operated equipment. End use of casting: If the structural function of a casting is more important than its appearance and dimensional accuracy, a mold can be used longer before being discarded.

Influence of Mold Design. In addition to the aforementioned factors, mold design has a marked effect on mold life. Variation in mold wall thickness causes excessive stress to develop during heating and cooling, which in turn causes premature mold failure from cracking. Abrupt changes in thickness without generous fillets also cause premature mold failure. Small fillets and radii lead to reduced mold life because checking and cracking, as well as ultimate failure, often start at these points. Usually, less draft is required on external mold surfaces than on internal mold surfaces, because of the shrinkage in the casting. A 5

draft is desirable, but 2 on external and 3 on internal mold surfaces can be used. Lower draft angles, however, decrease the number of castings that can be made between mold repairs. Projections in the mold cavities contribute greatly to reduced mold life. These projections become extremely hot, which increases the possibility of extrusion, deformation, and mutilation when the casting is removed. Whenever possible, mold life can be extended by using inserts to replace worn or broken projections.

Mold Temperature If the mold temperature is too high, excess flash develops, castings are too weak to be extracted undamaged, and mechanical properties and casting finish are impaired. When mold temperature is too low, cold shuts and misruns are likely to occur, and feeding is inhibited, which generally results in shrinkage, hot tears, and sticking of the casting to molds and cores. The variables that determine mold temperature include:  Pouring temperature: The higher the pour  

  

ing temperature, the higher the temperature of the mold. Cycle frequency: The faster the operating cycle, the hotter the mold. Casting weight: Mold temperature increases as the weight of molten metal increases. Casting shape: Isolated heavy sections, cored pockets, and sharp corners not only increase overall mold temperature but also set up undesirable thermal gradients. Casting wall thickness: Mold temperature increases as the wall thickness of the casting increases. Mold wall thickness: Mold temperature decreases as the thickness of the mold wall increases. Thickness of mold coating: Mold temperature decreases as the thickness of the mold coating increases.

After the processing procedure has been established for a given casting operation, mold coating, cycle frequency, chills, and antichills have significant effects on mold temperature. Mold coating is difficult to maintain at an optimal thickness, primarily because the coating wears during each casting cycle and because it is difficult to measure coating thickness during production. The most widely used method for controlling coating thickness is periodic inspection of the castings. Improper coating thickness is reflected by objectionable surface finish and loss of dimensional accuracy. Preheating of Molds. In many casting operations, molds are preheated to their approximate operating temperature before the operation begins. This practice minimizes the number of unacceptable castings produced during establishment of the operating temperature.

698 / Permanent Mold and Semipermanent Mold Processes Molds can be preheated by exposure to direct flame, although this method can be detrimental to the molds because of the severity and nonuniformity of heat distribution. Customized heaters are often built for molds. Preheating of the mold in an oven or by infrared heating are the best methods because the thermal gradients are of smaller magnitude. Unfortunately, this may be impractical for larger molds. Final mold operating temperatures are achieved after the first few production cycles.

Control of Mold Temperature Optimal mold temperature is the temperature that will produce a sound casting in the shortest time. For an established process cycle, temperature control is largely achieved through the use of auxiliary cooling or heating, control of coating thickness, and cycle-time consistency. Auxiliary cooling is often achieved by forcing air or water through passages in mold sections adjacent to the heavy sections of the casting. Circulating water is by far the more effective method of auxiliary cooling, particularly when coupled with chill inserts, as illustrated in Fig. 13. Scale buildup in cooling water passages must be minimized in order to achieve optimal results. When cooling water is maintained free of solid particulate, scaling is minimized and heat transfer can be managed accurately and effectively to facilitate consistent, predictable, and precise solidification patterns and shorten cycle times for greater productivity. The problem of scale formation from particulate-contaminated water can be solved by the use of recirculating systems containing demineralized, purified water. Water treatment systems such as reverse osmosis and deionization are readily available and simple to maintain for high-production operations. These systems can be coupled with particle counters and automatic controls that can eliminate scaling problems entirely. Water flow is regulated manually to each mold section with the aid of a flow meter. Shutoff valves are used to stop the water flow when the casting process is interrupted, thereby minimizing the loss of mold operating temperature. Strategically placed Rib (I of 4)

Casting

Iron mold Copper antichill

thermocouples or timers can be used to signal on and off water cooling. In addition to the control of water flow, the temperature of the inlet water affects the performance of the mold cooling system and must be controlled to within a suitable maximum range. If water or another liquid coolant is used, it must never be allowed to contact the metal being poured, or a steam explosion will result. The intensity of a steam explosion increases as metal temperature increases. In addition, water will react chemically with molten magnesium. A mold coating of controlled thickness can equalize solidification rates between thin and heavy sections. Chills and antichills can be used to adjust solidification rates further, so that freezing proceeds rapidly from thin to intermediate sections and then into heavy sections and finally into the feeding system. Chills are used to accelerate solidification in a segment of a mold. This can be done by directing cooling air jets against a chill inserted in the mold (Fig. 13) or, more simply, by using a metal insert without auxiliary cooling. Chilling can also be achieved by removing some or all of the mold coating in a specific area to increase thermal conductivity. Chills can be used to increase production rate, to improve metal soundness, and to increase mechanical properties. Antichills serve to slow the cooling in a specific area. Heat loss in a segment of a permanent mold can be reduced by directing an external heating device, such as a gas burner, against an antichill inserted in the mold (Fig. 13). This is an effective but cumbersome method that adds cost and can significantly reduce mold life due to thermal stresses. A simpler, less costly method to slow cooling in a local area would be to cover the exterior mold surface with insulating materials and/or apply high-insulating mold coatings to the cavity surface.

Dimensional Accuracy The dimensional accuracy of permanent mold castings is affected by short-term and long-term variables. Short-term variables are those that prevail regardless of the length of the run:  Cycle-to-cycle variation in mold closure or 

A

Gas burner

A

Fig. 13

Copper chill Section A-A

  

in the position of other moving elements of the mold Variations in mold closure caused by foreign material on mold faces or by distortion of the mold elements Variations in thickness of the mold coating Variations in temperature distribution in the mold Variations in casting removal temperature

Air jet

Air- or water-cooled chills and flame-heated antichills can be used to equalize cooling rates in casting sections of varying thickness.

Long-term variables that occur over the life of the mold are caused by:  Gradual and progressive mold distortion

resulting from stress relief, growth, and creep

 Progressive wear of mold surfaces primarily

due to cleaning Given proper attention to the aforementioned factors, recommended permanent mold dimensional standards are illustrated in Fig. 14. Chart E1 represents basic tolerances for features contained on one side of the mold parting line that reflect variation caused by expansion and contraction of the mold and the metal during solidification. For dimensions crossing the parting line or involving cores or surfaces formed by moving mold members, charts E2 and E3 represent the additional tolerances that apply for such features. These features exhibit additional variation due to the hydrostatic pressure of the liquid metal. The amount of additional variation across the parting line is therefore related to the projected area of the casting at the parting line. Dimensional variations can be minimized to within those illustrated in Fig. 14 by keeping heating and cooling rates constant, by operating on a fixed cycle, and by maintaining clean parting faces. It is particularly important to select mold cleaning procedures that remove a minimum of mold material. Mold Design. Both the mold thickness and the design of the supporting ribs affect the degree of mold warpage at operating temperatures. Supporting ribs on the back of a thin mold will warp the mold face into a concave form. This mold design error can alter casting dimensions across the parting line by as much as 1.6 mm (0.06 in.). Adequate mold lockup will contribute to the control of otherwise severe warpage problems. Mold erosion resulting from metal impingement and cavitation due to improper gating design contributes to rapid weakening of the mold metal and to heat checking. These mold design errors contribute to rapid mold deterioration and resulting dimensional variation during a long run. Mechanical abrasion due to insufficient draft or to improperly designed ejection systems also contributes to the rapid variation of casting dimensions. Sliding mold segments require clearance of up to 0.38 mm (0.015 in.) to function under varying mold temperatures. This clearance and other mechanical problems associated with sliding mold segments contribute to variations in casting dimensions. Sand cores further aggravate the problem. Mold Operation. Metal buildup from flash can prevent the mold halves from coming together and can cause wide variations in dimensions across the parting line, even in a short run. Mold coatings on the cavity face are normally applied in thicknesses from 0.076 to 0.15 mm (0.003 to 0.006 in.). Poor mold maintenance can allow these coatings to build to more than 1.5 mm (0.060 in.) thick, causing extreme variation in casting dimensions. Inadequate lubrication of sliding mold segments and ejector mechanisms will contribute to improper mold lockup and consequent variation in casting dimensions. Variation in the casting cycle

Permanent Mold Casting / 699 Standards for aluminum permanent mold castings E1 E1

E2

E2

P/L

Slide core

P/L

E2 E2

E3

E3

E1

(E1) Tolerance

(E2) Additional tolerance

Fully machined mold cavites

Cast to size molds

Permanent mold

mm

in.

Basic tolerance up to 25 mm (1 in.)

±0.38

±0.015

Additional tolerance for each additional 25 mm (1 in.)

±0.050

Fig. 14

mm ±0.8

in. ±0.030

±0.002 ± 0.080 ±0.003

Permanent mold

Projected area 2

cm Up to 65 65_320 320_650 650_1600 1600_3200

(E3) Additional tolerance

2

Permanent mold to metal core

Projected area 2

cm

in. Up to 10 10_50 50_100

mm ±0.25 ±0.38

in. ±0.010 ±0.015

±0.50

±0.020

100_250 250_500

±0.6

±0.025

Up to 65 65_320 320_650 650_1600

±0.8

±0.030

1600_6500

2

in.

mm

Up to 10 10_50 50_100 100_250

±0.25 ±0.38 ±0.38

in. ±0.010

±0.6

±0.022

250_1000

±0.8

±0.032

±0.015 ±0.015

Recommended dimensional standards for permanent mold castings. Source: Ref 10

and in metal temperature will contribute to dimensional variations. Wear Rates. The dimensions of many mold and core components change at a relatively uniform rate; therefore, it is possible to estimate when rework or replacement will be required. To maintain castings within tolerances, it is sometimes necessary to select mold component materials on the basis of their wear resistance; for example, H13 tool steel has better wear resistance than mild steel or cast iron.

Surface Finish The surface finish on permanent mold castings depends mainly on:  Surface of the mold cavities: The surface fin-

ish of the casting will be no better than that of the mold cavity. Heat checks and other imperfections will be reproduced on the casting surface.  Mold coating: Excessively thick coatings, uneven coatings, or flaked coatings will degrade the casting finish.  Mold design: Enough draft must be provided to prevent galling or cracking of casting surfaces. The location of the parting line can also affect the surface finish of the casting.

 Gating design and size: These factors have

Portions of this article are adapted from “Permanent Mold Casting,” revised by Charles E. West and Thomas E. Grubach, in Casting, Volume 15, ASM Handbook, ASM International, 1988, p 275–285.

3. ASM International, www.asminternational. org 4. American Foundry Society, www.afsinc. org 5. SAE International, www.sae.org 6. Aluminum Association, www.aluminum. org 7. Copper Development Association, www. copper.org 8. R. Fuoco and E. Correˆa, “The Effect of Gating System Design on the Quality of Aluminum Gravity Castings,” IPT-Technological Research Institute, Sao Paulo, Brazil 9. F. Chiesa, Controlling Permanent Mold Coating Application Parameters, Mod. Cast., Oct 1996 10. Standards for Aluminum Permanent Mold Casting, 14th ed., Aluminum Association, March 2000

REFERENCES

SELECTED REFERENCE

a marked effect on casting finish because of the influence on the rate and smoothness of molten metal flow.  Venting: The removal of air trapped in mold cavities is important to ensure smooth and complete filling.  Casting design: Surface finish is adversely affected by severe changes of section, complexity, requirements for change in direction of metal flow, and large, flat areas.

ACKNOWLEDGMENT

1. M. Sadayappan, J. Thomson, F. Fasoyinu, and M. Sahoo, Establishing Process, Design Parameters for Permanent Mold Cast LeadFree Copper Alloys, Mod. Cast., Feb 2002 2. Aluminum Casting Technology, 2nd ed., AFS Inc., 1993

 J.G. Kaufman and E.L. Rooy, Aluminum

Alloy Casting: Properties, Processes, and Application, ASM International and AFS, 2004

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 700-708 DOI: 10.1361/asmhba0005261

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Low-Pressure Metal Casting Gregory G. Woycik, CMI Equipment and Engineering, Inc. Gordon Peters, Intermet-PCPC

LOW-PRESSURE METAL CASTING has been used since the 1950s to produce highvolume, high-integrity castings in alloys ranging from aluminum to zinc. In recent history, aluminum has become the most widely used alloy in this process (Ref 1), due in large part to the increased popularity of aluminum among automotive designers who desire lightweight alternatives for today’s vehicles. Low-pressure casting, as it is most widely practiced, is a variation of permanent mold casting (Ref 2) (although it is used to a limited degree to fill sand or plaster molds, too); thus, it is frequently called low-pressure permanent mold. It may also be referred to as low-pressure die casting, perhaps deriving from the European practice of referring to all metal mold processes as die casting (low-pressure die casting, gravity die casting, high-pressure die casting, etc.). Low-pressure casting is a process where molten metal is introduced to the mold by the application of pressure to a hermetically-sealed metal bath, forcing the molten metal up through a narrow-diameter fill (stalk) tube from a furnace usually residing below the casting machine (although there is a version using electromagnetic forces to lift metal into the mold, and then the furnace may be an open hearth located beside the casting machine). The process can be considered for low to high volumes of castings from 5 to 100 kg (11 to 220 lb) and usually incorporates the use of iron or steel permanent molds. Recent developments in sand molding technology have made precision sand molds a viable choice for high-volume lowpressure casting as well. A wide range of casting core options, such as expendable sand and shell cores and mechanical single- or multipiece permanent cores, are successfully used in the low-pressure process. The large majority of casting produced in low pressure are aluminum; however iron-, magnesium-, and copper-base alloys are successfully cast with this process also. The mechanical and fatigue properties of low-pressure cast components are typically 5 to 6% greater than those of gravity cast components of a similar alloy. Controlled application of the pressure yields a smooth, nonturbulent cavity fill from the bottom to the top,

which is highly desirable when casting light alloys such as aluminum and magnesium. The single-point entry of metal to the casting and minimal or nonexistent risers minimizes subsequent processing to remove these items, reducing overall manufacturing costs. For these reasons, the vast majority of light-alloy road wheels (Fig. 1a, b) for the automotive sector are produced using low pressure. The process is also well adapted for the production of high-integrity

Fig. 1

As-cast aluminum wheels low-pressure casting

produced

by

components such as cylinder blocks, cylinder heads (Fig. 2), pistons, automotive suspension components, brackets, and housings.

Conventional Low-Pressure Casting The low-pressure process incorporates the use of a hydraulically operated machine to vertically open and close a casting mold that is situated over a ready supply of molten metal. A basic schematic of a low-pressure machine is given in Fig. 3. The machine consists of an upper (moving) and lower (stationary) platen, to which the upper (cope) and lower (drag) mold halves are attached, respectively. The moving platen is raised and lowered hydraulically to allow for casting and extraction of a component. Standard mold design allows the casting to remain in the upper mold as it is opened and subsequently ejected from the mold for manual or automatic removal. The stationary platen is typically positioned directly over a sealed, airtight furnace with a supply of

Fig. 2

Low-pressure cast motorcycle cylinder head. Courtesy of Progress Casting, an ATEK company, Minneapolis, MN

Low-Pressure Metal Casting / 701

Fig. 3

Schematic of low-pressure casting machine with an electric resistance crucible furnace

Fig. 4

Low-pressure wheel casting cell with two casting machines and robotic casting extraction/quench. Courtesy of CMI Equipment and Engineering, Inc., AuGres, MI

molten metal. Figure 4 shows a typical low-pressure wheel casting cell comprised of two low-pressure machines and automated casting extraction robot. A fill/stalk tube is immersed in the molten metal through the top of the sealed furnace and brought in contact with the lower mold by either lowering the machine platen to the furnace or by raising the furnace to meet the platen. In former times, the fill/stalk tubes were typically constructed of steel or cast iron to which a protective refractory coating was applied to prevent deterioration in the molten metal. High-strength, nonwetting refractory materials are now the norm in most lowpressure applications, and, although initially more expensive, they typically last many weeks instead of days and are thus very cost-effective. An intermediate tube is sometimes used between the furnace fill tube and the lower mold when there may be interference of the machine platen and the top of the furnace. When an intermediate tube is used, it should be kept as short as possible, and it is imperative to incorporate some external heating to avoid metal freezing in the tube. Fiber gaskets are normally used between any junctions in the stalk tube, intermediate tube, and lower mold to maintain an airtight, leakproof seal. The fill tube extends below the metal surface between 305 and 760 mm (12 and 30 in.), depending on the style, size, and depth of the casting furnace. This gives the low-pressure process a distinct advantage in that the surface of the molten bath remains undisturbed during the casting process, thus avoiding the entrainment of surface oxides into the melt, as occurs when ladles or dippers are used to deliver metal to a casting mold. The entrainment of oxides in the melt has been routinely shown to reduce the mechanical properties of aluminum castings. For a complete discussion of this topic, see Ref 3 as well as the articles “Casting Practice: Guidelines for Effective Production of Reliable Castings” and “Filling and Feeding System Concepts” in this Volume. To provide optimum performance and consistency to the process, almost all low-pressure machines are controlled by a programmable logistic controller (PLC). The PLC will control all moving functions of the casting machine, pressurization of the casting furnace, mold cooling/heating circuits (oil, water, or air), automatic casting extraction, and manipulation of casting furnace for refilling. The PLC can also track important historical data of the casting cycles, such as mold temperature, cycle time, metal temperature, pressure profiles, and cooling cycles, which may be used to gage productivity or troubleshoot casting problems. Most low-pressure machines include automatic casting extraction due to the size and number of castings produced by the process. A casting extractor can be a simple shuttle pan onto which the casting is ejected or a robotic arm that extracts the casting and takes it to a subsequent operation.

702 / Permanent Mold and Semipermanent Mold Processes A machine operator is required to begin the casting cycle, monitor the machine performance, visually inspect castings, and make process adjustments as necessary. The operator will also inspect the mold between cycles to verify that it is ready for the next casting cycle. The sequence of operation begins with a properly coated and preheated mold in the “open” position. The lower platen is situated above the casting furnace, with the fill tube sealed in position and extending down into a full bath of molten metal. When the cycle is initiated, the machine moves the upper mold down to seal the mold envelope with the lower half. Guide pins in the mold are typically used to accurately position the two mold halves as they are brought together. Vertical position sensors confirm that the mold is indeed closed, at which point pressure in the furnace begins to increase along a preprogrammed pressure or filling profile. The filling profile is established on a partby-part basis to provide a smooth, constant metal filling front as the metal travels up the fill tube and through the mold. Complex molds may require several different “ramps” in the profile to account for changes in the crosssectional area of the mold cavity as the metal fills. Innovations in predictive pressure control can now provide very accurate and repeatable control of a filling profile, which is paramount to casting quality. The pressure is programmed to reach a maximum at the point where the mold is completely full. The maximum pressure is dictated by the casting requirements and is limited to the capability of the furnace seals but seldom exceeds 150 kPa (1.5 bar) in most applications. The pressure is held during casting solidification to keep the mold full and to provide feed metal to the in-gate or sprue. Solidification under pressure also maintains intimate contact of the casting to the mold wall, promoting heat transfer and rapid solidification. Rapid solidification results in a finer microstructure, which increases the mechanical properties of the casting. Hydrostatic pressure through the molten metal also aids in minimizing the occurrence of visible gas porosity in the casting. This effect is only realized until the sprue freezes and pressure is no longer transferred through the molten metal. As solidification of the casting progresses to the sprue, the pressure in the furnace is released, allowing the metal in the fill tube to drop back into the furnace. It is important that the casting does not solidify below the sprue choke point and become “die locked” in the mold. Figure 5 shows the sprue area of a lowpressure mold and the choke point in the sprue bushing. This condition often makes the casting impossible to remove from the mold and causes significant interruption to the casting process. Often, the tooling must be removed from the machine to extract the casting by drilling out the frozen sprue area. Some processes only reduce the pressure enough to allow the liquid metal front to drop

away from the solidifying sprue enough to avoid die lock but keep the majority of the fill tube full. This prevents potential oxide buildup on the internal surface of the fill tube as the metal drops back into the furnace, which could be carried into subsequent castings. Many lowpressure molds incorporate the use of a disposable filter screen that is placed just above the sprue before each shot to prevent large oxide skins from being carried into the casting during the fill. Typical low-pressure aluminum permanent molds incorporate the use of a replaceable sprue bushing that is made from an insulating ceramic material, which aids in retaining heat in the critical in-gate of the casting. If a casting does freeze up in the sprue bushing, it can be knocked out and replaced, usually without removing the tooling from the machine, thus reducing any production downtime for this problem. After a predetermined time to allow the casting to cool sufficiently, the machine will open the mold. By design, the casting remains in the upper mold half and travels up with the mold. The casting may be allowed to cool further or may be immediately ejected from the mold by actuating the ejection cylinder, which pushes ejector pins through the mold, thus pushing the casting out to a waiting operator or automatic unloading device. The casting is removed from the area to a cooling line or quench bath. The operator checks for any flash that remains in or around the mold and removes it, after which the cycle is ready to begin again. Total cycle time for the low-pressure process is dependent on the number and size of castings being made as well as the mold cooling methods employed but typically ranges from 2 to 7 min for aluminum castings. Low-Pressure Furnace. In all cases, the molten metal supply to the low-pressure casting process is from below the machine/mold. This requires a furnace of sufficient size to produce many castings, yet small enough to fit under the casting machine. Aluminum furnace capacities can range from 450 to 1500 kg (990 to 3300 lb), depending on the style of furnace and size of casting machine. Furnaces can be gas fired or electric resistance crucible style, where the crucibles are mainly silicon carbide. In magnesium low-pressure casting, steel crucibles are commonly used. Another common low-pressure furnace is a refractory-lined furnace heated by electric radiant (glow bar) or electric resistance (immersion) heaters. Immersion-style heaters have almost twice the average efficiency of radiant glow bar heaters and cost less to operate over time. Crucible-style furnaces will typically support a single fill tube, while refractory-lined furnaces can accommodate 1 to 5 fill tubes. Figure 6 shows a refractory-lined immersion-style low-pressure casting furnace. Regardless of furnace style, each must be refilled periodically as metal is used up. A critical feature in maintaining a successful

Fig. 5

Sprue and sprue bushing area of a low-pressure casting mold. Castings must not solidify below the choke point, to avoid a die-locked condition.

Fig.

6

Refractory-lined immersion-heated pressure furnace. Courtesy of Equipment and Engineering, Inc., AuGres, MI

lowCMI

production cycle is the ability to refill the casting furnace with no or minimal interruption to the casting cycle. Refractory-lined furnaces are equipped with a fill door through which freshly treated and degassed molten metal can be added. A refractory-lined filling trough is typically used to aid in filling the furnace through the fill door (Fig. 7). The refill occurs after the pressure has been released during a casting cycle, while the casting is cooling in the mold. It must be completed before another pressure cycle may begin, or the furnace will not build pressure to make another casting. Crucible furnaces may be filled in a similar fashion, or, in some cases, the entire furnace is swapped out for another with a full supply of fresh metal. As the metal in the casting furnace is depleted, both the volume of space that must be pressurized and the height that the metal must travel to reach the mold increase. These two factors must be considered to assure that the casting mold is filled consistently every time. Through PLC programming, a compensation factor can be added to the filling profile to assure that

Low-Pressure Metal Casting / 703

Fig. 7

Fig. 8

Fill trough for replenishing molten metal in a low-pressure furnace. Courtesy of CMI Equipment and Engineering, Inc., Augres, MI

Low-pressure wheel mold with four moving side cores to create the external geometry of a wheel

the metal front reaches the mold at the same time. This compensation factor can be determined in various ways and can be based on calculated metal volume decrease per shot, weight of remaining metal in the furnace, or metal level in the furnace, among others. Direct feedback from a load cell or metal-level sensor can provide the best information on which to determine a compensation factor but is also the most costly. Pressurization of the casting furnace is typically achieved by using dry compressed air. The air must be substantially free of oil and moisture, because these elements will lead to

contamination of the metal bath through excessive oxidation and hydrogen pickup over time as the moisture is vaporized above the molten bath. Dry air is acceptable for most low-pressure casting applications; however, for some high-integrity components, an inert gas such as nitrogen or argon is substituted for dry air, either intermittently or completely, to avoid the problems described previously. Low-Pressure Tooling. The molds used in low-pressure casting can range from hard sand molds to lost foam flasks but are most typically constructed of tool steel or cast iron. In applications where only forced-air cooling is used, iron molds can be used very effectively and at a lower cost than steel. To take full advantage of the enhanced mechanical properties capable in the low-pressure process, water-cooled steel molds are used. Strategically placed internal water cooling channels can be added to steel molds to promote rapid directional solidification of the casting from the extremities back to the sprue. This technique results in a finer casting microstructure and increased mechanical properties as well as reduced castingto-casting cycle time. Low-pressure casting machines can be fitted with computercontrolled water (and air) cooling solenoids that allow customization of casting solidification to eliminate shrinkage defects. To achieve the maximum quality of cast components in low pressure, care must be exercised when designing low-pressure tooling. Since each casting is typically filled through a single entry point, and external risers are not normally used, the in- gate or sprue of the casting must be positioned so that good directional solidification back to the sprue is achieved. This may be enhanced through the use of water or air cooling, as described previously. In the absence of good directional solidification, castings may yield significant unwanted shrinkage

porosity. Successfully designed low-pressure castings have some of the best casting-toshot-weight yields of any process, due to the lack of risers and the single sprue feeding. Casting yields over 90% are typically achieved in aluminum castings. It is important to note that because the casting fills from the bottom up, the metal front must force the air out of the cavity through properly located vents. If good venting practices are not followed, any air trapped in the mold can accumulate in localized areas, resulting in unfilled features of the casting. Standard practice incorporates the use of vented ejector pins, because they are typically located in the upper mold and on the uppermost surfaces of the casting, where air is most likely to accumulate. Ejector pin vents are typically self-cleaning of any flash; however, if excessive soldering buildup occurs, the venting capability may be diminished. In situations where ejector pins are not feasible or where casting geometry may cause air to be trapped, several styles of in-mold vents can be used to exhaust the cavity. It is important to mask in-mold vents when applying mold coatings so that they are not coated over, rendering them useless. Vents that become plugged during casting must be cleaned or replaced to maintain their effectiveness. Proper vent cleaning/replacement should be exercised during every mold maintenance interval. As is the case for gravity permanent molds, low-pressure permanent molds for aluminum castings made from iron or steel must be coated with a refractory coating to prevent soldering of the aluminum to the iron-base mold. There are many commercially available permanent mold coatings with custom-tailored thermal properties ranging from highly insulating to chill coats. The proper application of mold coating is critical to the longevity and performance of the coating, and strict adherence to manufacturer’s recommended application techniques must be exercised. Cores. The low-pressure process can readily accommodate the use of almost any core technology. Hydraulically operated mechanical cores are easily integrated into the PLC control logic of the machine and automatically controlled. These cores can be used to create internal passages in castings or, as in the case of wheels, used to create external geometry of the casting. Figure 8 shows a typical low-pressure wheel mold installed in a low-pressure machine. Four side cores make up the external geometry of a wheel. The low-pressure process can also accommodate semipermanent molds, where cores can be expendable sand, shell, or plaster. The incoming metal velocity in low pressure is slow and does not typically displace or erode these types of cores. Automatic core- setting equipment can be easily integrated into the casting process. Air-set precision sand has been widely used to make cores; however, it is possible to create entire casting molds as well. With some

704 / Permanent Mold and Semipermanent Mold Processes adaptations to the low-pressure machine, these types of molds may be positioned over a lowpressure furnace and fill tube and filled in much the same manner as iron or steel molds. With this approach, the sand casting will benefit from smooth, controllable bottom-up filling, the trademark of the low-pressure casting process.

Counterpressure Casting The counterpressure casting process is an evolution/extension of the basic low-pressure die casting process. The equipment used in the counterpressure casting process is similar to a low-pressure die casting machine and is essentially a vertical press with a crucible furnace mounted under the press. A unique feature of the PCPC (Intermet Corp.) process is that the

horizontally parted die is surrounded by an airtight mold chamber. The basic pressure counterpressure casting (PCPC) machine is shown in Fig. 9. The PCPC process is capable of producing high-quality, cost- effective aluminum castings if proper attention is paid to critical aspects of the process. Aluminum castings that are to be used in safety-critical automotive applications or military applications are candidates for the PCPC process (Ref 4-7). Counterpressure casting has a long-established history of process development and practical implementation. Due to the high integrity of the castings this process can produce, counterpressure castings have replaced aluminum forgings and squeeze castings, with significant savings to the customer. Ductile iron parts have been replaced at a 40 to 50% weight savings, a good value to the customer. History of Counterpressure Casting. Counterpressure casting evolved out of research conducted at the Bulgarian Academy of Sciences in the 1950s. In 1961, Angel Tontcheve Balevsky and Ivan Dimov Nikolov were awarded a patent for a method for casting under pressure. This patent described the use of pressure acting inside a mold cavity and was termed counterpressure. The process was then named counterpressure casting. In 1964, the first vertical semiproduction counterpressure casting machine was constructed for casting aluminum alloys. To date (2008), counterpressure casting machines are producing millions of castings per year. Automotive safety-critical castings, such as wheels, front steering knuckles, control arms, and rear knuckles, are being produced. Power train components such as engine blocks and cylinder heads have also been produced. Military applications such as tank track wheels have also had counterpressure castings replace more expensive aluminum forgings. Pressure Differential (DP) and Mold Filling. In standard low-pressure casting, the height of the metal above the metal bath is calculated by: H ¼ p=g

Fig. 9

Basic pressure Components

counterpressure

Machine

where H is the height of the metal column above the furnace metal level, g is the specific mass of the alloy melt, and p is the pressure applied above the alloy melt. Since there are two pressures acting in counterpressure casting (Fig. 10), the height of the alloy melt is given by: H ¼ ðp2  p1 Þ=g

Fig. 10

Counterpressure casting method

where H is the height of the metal column above the furnace metal level, g is the specific mass of the alloy melt, p1 is the pressure applied to the cavity in the mold chamber, and p2 is the pressure applied above the alloy melt in the furnace. Since DP is defined at p2  p1, the formula can be simplified to resemble the calculation for low pressure by substituting DP for p2  p1.

Thus, the formula for calculating the height of the metal in the fill tubes over the metal bath is: H ¼
p=g

For example, if the density of the aluminum alloy is 2550 kg/m3 and the DP is 300 mbar, then: H ¼ 0:30bar=2550 kg=m3 H ¼ 30 kPa=2550 kg=m3 H ¼ 30; 000 Pa=2550 kg=m3

Since 1 kg = 9.81 N, then: 30; 000 Pa ¼ 3058:1 kg=m2 H ¼ 3058:1 kg=m2 =2550 kg=m3 H ¼ 1:20 m

By knowing the diameter of the fill tube and thus its area, this can be compared to the area of the mold in the x- y plane at any z height. The correct fill profile for the part being cast can be established as a result of using the conservation of mass flow principle. Counterpressure Casting Cycle Steps. The following describes the basic counterpressure casting cycle steps (Fig. 11): 1. The die closes to initiate the casting cycle. The mold chamber is set up to seal at the same time as the die halves. 2. The furnace is pressurized at the same rate as the mold chamber. 3. Once the furnace pressure and the air inside the mold chamber reach the pressure set point, the pressure holds for an equalization period. This period is approximately 5 s. After the equalization is over, the pressure in the furnace increases. The pressure in the chamber remains constant as it is bled off when the metal rises up the riser tubes and compresses the air displaced from the tubes and die cavities. This pressure difference is called the delta pressure. During this phase, the delta pressure is slowly increasing to permit the die to be filled. The delta pressure curve should resemble the fill profile of a typical low-pressure fill profile. The metalfilling characteristics are controlled by the delta pressure profile programmed into the casting machine. 4. Once the die is filled, the delta pressure is held constant for a period of time to stabilize. This is the delta P-maintain time programmed into the casting machine (usually 5 s). After the delta P-maintain time expires, the pressure in the chamber is released, and the delta pressure increases by the value of the system pressure setting. 5. The pressure is maintained on the furnace (feeding time) until the sprue has solidified. Once the sprue has solidified, the furnace pressure is released, and the casting cools until it has hardened enough to withstand the forces applied to it during demolding

Low-Pressure Metal Casting / 705

Fig. 11

Counterpressure casting process steps

and ejection from the upper die half. The die then opens, and the casting is ejected during the mold-open phase of the casting cycle. The casting can now be processed through the remaining manufacturing process, such as degating, deflashing, inspection, heat treatment, and packaging. Advantages of Counterpressure Casting. The advantages of the PCPC process are as follows:  High yield rate: This translates to lower

operating costs, reduced melt loss, and less material handling. Yields are similar to low pressure or vacuum riserless/pressure riserless casting and are typically approximately 85 to 90%. On larger castings, yields above 90% are possible.  Good fill control: Fill is controlled by programmed pressure curves, and precise fill dynamics can be achieved.  Good mechanical properties: Properties are typically better than sand and gravity

permanent mold and low pressure, but PCP is capable of matching the properties of the more expensive squeeze casting or semisolid/thixocasting processes. The net result of these advantages translates into very high- performance castings at relatively low costs. The good mechanical properties of PCPC castings are attributed to:  Filling characteristics: Since the casting

process is conducted with a working pressure greater than atmospheric pressure, the metal is less prone to turbulence during the filling process. This reduces the potential for oxide films to be generated that will reduce the mechanical properties of the casting.  Feeding characteristics: This includes the increased feeding pressure that the crucible furnace can generate and the ability to create a near-impulse increase in net feeding pressure when the pressure in the mold chamber

is released. By controlling how much pressure is applied to the casting during solidification, areas prone to solidification shrinkage can have improved internal integrity. This results in less microporosity from solidification shrinkage, and thus, mechanical properties are improved.  Mold/casting heat-transfer efficiency: Due to the increased pressure imparted on the casting from the furnace, the casting is kept in contact with the mold surface with greater force, and the heat from the casting can be extracted from the casting with greater efficiency. This results in smaller secondary dendritic arm spacing and thus better mechanical properties.  Melt cleanliness: Since the crucible furnaces are completely exchanged once the metal is consumed during the casting process, each new crucible of metal has the opportunity to be fluxed, alloyed, degassed, and thoroughly cleaned. This is a big advantage in metal quality compared to other low-pressure processes, because the casting machine

706 / Permanent Mold and Semipermanent Mold Processes furnaces are typically continually topped off, and the whole furnace is not cleaned unless the machine is taken out of production. Effective degassing of a fixed furnace in continuous operation is essentially impossible, a because gas is introduced each time it is topped off. This increased control on metal preparation contributes to improved mechanical properties.  Low levels of gas porosity: Since the solidification of the casting is conducted at pressures greater than atmospheric as a result of the counterpressure applied to the mold chamber, any dissolved hydrogen will stay in solution longer and will be less likely to evolve out and create microporosity in the casting. This results in less gas microporosity and better mechanical properties. Special considerations for PCP operations include preventative maintenance, die access, and die coating considerations. Preventative Maintenance. Since the control of the filling profile depends on two air systems working in harmony, any problems with sealing in either of the systems can result in uncontrolled mold filing, casting problems, machine damage, and potential operator injury. There are several rubber seals on counterpressure casting machines that must be replaced on a periodic basis. This must be done and the schedule rigidly adhered to. If this is not done, the casting machine operating equipment efficiency will suffer. This results in slightly increased repair and maintenance costs when compared to low- pressure casting. Die Access. Since there is a steel chamber surrounding the die, access to the mold is restricted. The mold cannot be seen during the mold-closed phases of the casting cycle. In order to detect if there is a die spill (metal that flows between the parting line of the die halves and onto the casting machine base), the operator must rely on a spill detection wire to signal the machine to depressurize when molten aluminum hits it. If this wire is set up incorrectly or not installed, the machine can completely fill up the mold chamber with molten aluminum, which will spray out of the mold chamber once the rubber seal is burned up. The upper die access is limited to the die cavity faces. Access to the outside of the upper die is only possible if the chamber is removed. This can result in longer die change times and make die repairs on the machine more laborintensive. The mold chamber also significantly reduces the effect of radiation heat transfer and convective heat transfer during the casting cycle. Thus, the PCPC casting process must rely on effective water or air cooling circuits in the die. Die Coating Considerations. Like gravity permanent mold and low-pressure permanent mold casting, the counterpressure casting process relies on die coatings. These are applied to the cavity surfaces to promote complete filling of the die cavity, to control solidification,

Fig. 12

Example of vacuum riserless/pressure riserless cast automotive chassis and suspension components

and to provide lubrication for demolding the casting from the die cavity. Due to the increased metal pressure exerted on the casting during the solidification phase of the casting cycle, metal will infiltrate into the pores of the die coating and increase abrasion of the coating. This demands that excellent die coating procedures be in place in a counterpressure casting operation. If the die coating practices are neglected, casting performance will suffer due to premature mold coating wear.

Vacuum Riserless/Pressure Riserless Casting The vacuum riserless/pressure riserless casting (VRC/PRC) process for casting aluminum is a modified permanent mold casting method that incorporates the use of vacuum and pressure to bottom fill multiple mold cavities from a hermetically sealed molten-metal source. The use of vacuum in the production of permanent mold castings dates back to World War II for the production of aircraft cylinder heads and was further developed by Alcoa throughout the 1940s to the 1960s with the addition of pressurized filling, expendable cores, and watercooled molds (Ref 8). In the late 1990s, the process was reengineered by CMI Equipment and Engineering for use as a high- volume casting process for automotive components, with manufacturing costs ranging between standard permanent mold and squeeze casting costs.

The VRC/PRC castings are readily heat treatable, and mechanical properties approach those of squeeze castings. Products such as safety-critical automotive subframes, steering knuckles, control arms, and brackets are produced with the VRC/PRC process (Fig. 12) A schematic of a typical VRC/PRC casting machine is shown in Fig. 13. The equipment consists of a hydraulically operated machine that houses the mold package in order to open and close the mold as well as eject the castings. Figure 14 shows a typical production machine. The machine is situated over a molten-metal casting furnace that can be an electrically heated silicon carbide crucible furnace or a refractory-lined electric immersion-heated furnace. Immersion-heated furnaces (Fig. 6) provide significant efficiency advantages over the crucible-style furnaces and have become the furnace of choice in new installations. A key feature of immersion furnaces is that the lower platen/mold combination can act as the furnace lid, providing a hermetically sealed chamber to which pressure may be applied to fill the mold cavities. Multiple fill tubes are connected to the lower mold or molds, through which metal is introduced. By design, the entire surface of the molten-metal bath is accessible, providing significant flexibility in positioning molds and fill tubes to maximize productivity. A “family” of individual molds, each with a single fill tube, can rest on a base plate and be filled simultaneously. In the same fashion, a large singlemold cavity incorporating multiple fill tubes

Low-Pressure Metal Casting / 707

Fig. 13

Schematic of vacuum riserless/pressure riserless casting machine

Fig. 14

Vacuum riserless/pressure riserless production casting machine. Courtesy of CMI Equipment and Engineering, Inc., AuGres, MI

may be used. If expendable cores are used, they are placed in the mold either manually or by an automated mechanism. When a casting cycle is initiated, the machine closes the mold, providing a critical airtight seal around the parting line of the mold. A controlled vacuum is applied through engineered passages (vacuum track) and vents in the upper mold half to evacuate air and other contaminating atmosphere from the cavity. As a result of the vacuum, molten metal begins to rise into the cavities. The amount of vacuum is dependent on the number and size of castings being filled but is limited to only what is necessary to fill the cavities. Pressure is then applied to the molten-metal bath to maintain the metal

height in the fill tube and cavity, providing feed metal until solidification proceeds to the fill tube. Both vacuum and pressure are PLC controlled and follow a preprogrammed profile to achieve shot-to-shot consistency. Another significant feature of the process is that the fill tubes protrude only a few inches below the metal surface. Close coupling of the fill tubes to the molten metal allows lower metal temperatures, minimizes oxide formation, and provides maximum feeding of the casting. As a result of the minimum fill tube depth, a unique method of replenishing the casting furnace is employed so that a nearly constant metal level in the furnace is maintained. Fresh metal is transferred to a holding furnace via a filtered launder system from a reverberatory melting furnace, where it is thoroughly degassed before being dosed to the casting furnace after every shot. The solidification of the castings in the VRC/ PRC process must proceed from the extremities back to the feed metal in the fill tubes to provide sound, shrink-free castings. To facilitate rapid directional solidification, strategically engineered and sequenced water cooling is used. Up to 150 electrically operated and PLC-programmable water solenoid valves are incorporated on the largest machines, providing optimum thermal control of the casting process. In addition to minimizing cycle time, the rapid solidification results in a finer microstructure, which improves the mechanical properties of the castings. Automotive subframes manufactured using the VRC/PRC process, A356-T6, exhibit tensile strength of 310 MPa (45 ksi), yield strength of 225 MPa (33 ksi), and elongation of 13%. The extensive thermal control of the mold also provides repeatable process temperatures throughout the casting cycle, minimizing process variation during production. After the castings have solidified, the machine opens the top mold. By design, the castings remain in the upper mold and are subsequently ejected onto an automated extraction

mechanism. The castings are removed to a quench station, and the machine begins another cycle. One operator is necessary to prep the mold between casting cycles by checking for and removing any flash or debris remaining from the previous shot, visually inspect castings, and touch up any areas where mold coating may be compromised. The VRC/PRC process is a hybrid of the lowpressure process and as such has similar casting yields of 85 to 95%. Production rates for this process are dependent on casting size, complexity, and weight but typically range between 100,000 and 600,000 castings per year. Casting weights produced by VRC/PRC typically range between 3 and 25 kg (7 and 55 lb); however, these should not be considered as the limits of the process. High-volume-production VRC/ PRC machines can be built to accommodate molds as large as 132 by 168 cm (52 by 66 in.). Smaller machines will accommodate 107 by 130 cm (42 by 51 in.) molds (Ref 9). In a typical installation, the casting furnace resides in a shallow pit below the VRC/PRC machine. The entire casting machine can be raised off the furnace and shuttled forward to gain access to the metal bath and mechanical components of the furnace for cleaning and service. The casting furnace is typically cleaned at least once per shift to avoid dross buildup at the air-metal interface and near the filling port. The VRC/PRC machines are equipped with a tiltout upper platen capable of carrying the weight of the entire mold. This feature aids in loading molds into the machine by allowing overhead crane access to the platen. The tooling can be secured to the upper platen and subsequently tilted up to the locked position. The upper platen is then lowered to secure the bottom (drag) mold to the fixed platen. Supporting mold straps that hold the mold together are removed, and the machine is ready to start production. Due to the large number of water cooling lines employed, a quick-disconnect manifold is standard for the upper and lower molds to speed the process of mold changes. This also helps avoid improperly connected water lines, which can cause significant problems when relying on proper directional solidification to produce high-integrity castings.

ACKNOWLEDGMENTS Gordon Peters thanks Thomas Prucha of AFS and George Roussev of CPC for their assistance in supplying supporting material on counter pressure casting. REFERENCES 1. J.L. Jorstad and W.M. Rasmussen, Aluminum Casting Technology, 2nd ed., The American Foundry Society, 1993, p 181 2. T.E. Prucha, Metal Mold Processes — Gravity and Low Pressure Technology, International Conference on Structural Aluminum

708 / Permanent Mold and Semipermanent Mold Processes Castings, Proceedings (Orlando, FL), The American Foundry Society, Nov 2003 3. J. Campbell, Castings, 2nd ed., Elsevier Butterworth-Heinemann, 2003 4. T. Prucha, Recent Advancements for the design and Manufacturing of Reliable Structural Aluminum Castings, AFS International Conference on Structural Castings, (Orlando, FL), 2003 5. Y. Arsov, E. Momchilov, K. Daskalov, G. Bachvarov, The Theoretical and

Technological Fundamentals of Gas CounterPressure Casting. Prof. Marin Drinov Academic Publishing House, Sofia, Bulgaria, 2007 6. Wurker, Lars, Aluminum Chassis Parts Produced by Counter Pressure Casting, Casting Plant and technology international, Vol. 1, 2006, p 38–49 7. G. Ruff, T. Prucha, J. Barry, D. Patterson, Pressure Counter Pressure Casting (PCPC) for Automotive Aluminium Structural Components, SAE 2001-01-0411

8. T.F. Arnold, “Vacuum Permanent Mold Casting,” Aluminum Company of America, Permanent Mold Castings Division 9. VRC/PRC Offers Squeeze Casting Properties at Permanent Mold Costs (Case History: Vacuum Riserless Casting and Pressure Riserless Casting), Mod. Cast., Vol 94 (No. 6), June 1, 2004, p 48(1)

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 709-711 DOI: 10.1361/asmhba0005265

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Low-Pressure Countergravity Casting Paul Mikkola and Gary Scholl, Metal Casting Technology, Inc.

THE METHOD of countergravity mold filling has been practiced for many years. The technique is the result of applying a differential pressure between the molten metal and the mold to be filled, much like a soda straw is used to drink a liquid. During the last 30 years, the technology has matured, with many variations currently in practice. The differential pressure countergravity processes are used to produce aluminum and nonferrous alloys, low-alloy steels, stainless steel, high-alloy steels, and superalloys in weights ranging from a few grams to approximately 20 kg (44 lb). Production volumes range from 50 castings per year to 180,000 per day. The process is widely used for investment castings, precision sand molds made of resin or coldbox bonded sands, and semipermanent molds. The markets commonly served by these processes include automotive, hand tools, sports, medical, military, and aerospace. The general principle of the process involves placing a gas-permeable mold in an isolating chamber such that the sprue (inlet runner) extends downward from the bottom of the chamber. For investment casting, the molds are a common tree design with numerous part cavities attached around a central runner. The process starts with the mold and melt surface at the same pressure (typically 101 kPa, or 1 atm). The extended sprue is then positioned below the melt surface, and a differential pressure is created between the mold cavity in the chamber and the top of the melt. Usually, this is accomplished by applying a controlled vacuum to the mold chamber, but it can also be accomplished by applying a controlled pressure to the melt surface. This differential pressure causes metal to flow up the sprue and into the mold cavities. After the mold is completely filled, the liquid metal is held in position with this differential pressure while solidification takes place. For investment casting, typically only the castings and their in-gates are allowed to solidify before the differential pressure is removed, causing the molten metal in the central runner to be returned to the furnace for use in the next casting cycle. The process offers many advantages. Because metal is taken from below the center of the melt surface, slag that collects on the

melt surface (or migrates to the furnace lining) is not washed into the castings. The metal filling the mold is exposed to less atmosphere, and fill rates can be closely controlled. This enables routine filling of thin sections at lower metal temperatures because the liquid front encounters less resistance. There are also fewer opportunities for reoxidation and splashing, which cause entrainment of inclusions in the casting. Less superheat is required to attain fillout since the metal does not need the intermediate step of transferring into pouring ladles. This saves energy, eliminates ladle costs, and improves production throughput. The ability to cast at lower metal temperatures generally results in smaller grain sizes. By returning liquid metal from the central sprue, both energy and metal yield are improved over gravity pouring. These processes normally provide 70 to 90% metal yields, with only a small in-gate that needs to be removed. The process is accomplished in a chamber that is independent

Fig. 1

of the mold geometry, and mold filling is accomplished by computer-controlled differential pressure. Therefore, the casting process is easier to automate, which reduces labor and process variation. In the investment casting process, the single largest advantage is in sprue and mold loading. Because the central runner is removed before it solidifies, parts can be spaced more closely on a sprue, with no need to allow space for cutoff saws. This advantage can enable 10 to 15% more castings per sprue, which improves mold costs and throughput. Several production implementations that use differential pressure countergravity mold filling methods are discussed.

Mold Filling Countergravity Low-Pressure Air (CLA) Process. The CLA process (Fig. 1) has been used to produce investment castings in a wide

Schematic diagram of the countergravity low-pressure air process steps. (a) Preheated investment shell mold is placed into a casting chamber. (b) Chamber with mold is positioned and lowered into a molten-metal source. Vacuum is applied to the chamber, and mold filling is complete. (c) After casting and in-gates have solidified, vacuum is released, and all of the central sprue flows back into the melt.

710 / Permanent Mold and Semipermanent Mold Processes

Fig. 2

Schematic showing the steps of the countergravity low-pressure vacuum process. (a) Metal is melted in a vacuum chamber that is then flooded with argon. (b) A preheated mold is introduced into a separate upper chamber that is evacuated and then flooded with an equal pressure of argon before opening the separation valve. (c) The melt is raised, causing the fill pipe to be submerged in the metal, and then a controlled vacuum is applied to the upper chamber, causing filling of the mold cavities. (d) After solidification of the in-gates, the mold chamber is again flooded with argon, causing liquid metal to be returned to the furnace. The melt is lowered, and the separation valve is closed.

Fig. 3

Schematic showing, steps of the countergravity low-pressure inert atmosphere process. (a) Metal is melted in a vacuum or inert atmosphere, the chamber is filled with argon at +102 kPa (+1 atmt), and a hot mold is introduced into an independent mold chamber. (b) The mold chamber is then positioned above the melting furnace, and the fill pipe is lowered into the melt chamber but above the bath. Vacuum is applied to the mold chamber, purging the mold cavity with argon. (c) The fill pipe is then lowered into the molten bath, and additional vacuum is applied to the mold chamber, causing liquid metal to flow into the mold. (d) After a dwell time to allow the gates to solidify, the vacuum is released, and the remaining molten metal is returned to the furnace crucible.

variety of air melt alloys for over 30 years. These parts are supplied to a wide variety of industries. Applications for the automotive industry include steering system components, transmissions parts, diesel precombustion chambers, fuel injection components, rocker arms, and others. Parts are supplied to the aircraft and aerospace industries, such as temperature probes, fuel pump impellers, missile wings, brake parts, pump housings, and structural parts. Other applications include golf club heads, machine components, wood router tool

bits, tin snip blades, small wrenches, lock components, gun parts, valves and fittings, and power tools. Countergravity Low-Pressure Vacuum (CLV) Process. The CLV process (Fig. 2) is similar to the CLA process except that the mold and melt are contained in chambers that support both hard vacuum and argon atmospheres, to protect reactive alloys. The two chambers are joined by a slide valve that enables the downward-extending fill tube to contact the melt. For highly reactive alloys such as MAR-M-509

and Rene´ 125, which are noted for hafnium and zirconium inclusions, the CLV process consistently delivers parts with low levels of oxide inclusions. Castings produced by this process are used in gas turbine engines. Thin-wall components (0.5 mm, or 0.02 in.) previously made as weldments can be produced, thus enabling design freedom for shaped castings that maximize heat transfer, ease assembly, and reduce thermal fatigue. These design improvements often enable higher operating temperatures. Important cost and quality improvements have been achieved in cast airfoils, turbine seals, conduits, fuel swirls, and clamps. Countergravity Low-Pressure Inert Atmosphere (CLI) Process. The CLI process (Fig. 3) combines attributes of the CLA and CLV processes. Here, the melt and mold chambers are independent, enabling integration of the CLI melt chamber with a conventional casting environment. This allows reactive alloys to be cast using essentially the same casting machine that is used to produce high-volume steel castings. This process is used for high-volume automotive turbocharger wheels and vanes as well as gas turbine engine components for temperature probes and struts. Countergravity Pressure Vacuum (CPV) Process. The CPV process (Fig. 4) is similar to other countergravity reactive alloy processes except that the mold and melt are contained in chambers that remain under hard vacuum. Argon atmosphere is applied locally to the melt box, providing differential pressure for casting. The hard vacuum surrounding the melt box and mold chamber reduces the tendency for any oxygen in the outside room atmosphere to come into contact with the molten metal. The lack of any gas atmosphere in the mold during casting enhances fillout and eliminates the variable of mold permeability on casting fill. For the highest reactive alloys, the CPV process has demonstrated the lowest levels of oxide inclusions. Castings produced by this process are used in gas turbine engines. Wall thicknesses down to 0.5 mm (0.020 in.) have been successfully cast at metal temperatures only 28  C (50  F) over the solidus temperature of the alloy, thus providing a very fine, uniform grain structure. The higher detail, finer grain size, and reduced oxide inclusion count help to enable product designs that will operate at higher temperatures, improving efficiency of the gas turbine engine. Important cost and quality improvements have been achieved in cast airfoils, turbine seals, afterburner seals, and combustor liners. Supported Shell (Mold) Technique. The supported shell technique (Fig. 5) is an enhancement of the previously mentioned processes and enables use of thinner investment casting shell molds and sand molds. The products produced are the same as indicated previously. Loose Sand Vacuum (LSVAC) Process. The LSVAC process (Fig. 6) is used to produce thin-wall automotive components. Exhaust

Low-Pressure Countergravity Casting / 711

Fig. 5

Fig. 4

Schematic showing the steps of the countergravity pressure vacuum process. (a) Metal is melted in a vacuum chamber, and a hot mold is introduced into an isolated casting chamber. The mold fill pipe protrudes from the bottom of the support plate. The mold chamber is then sealed to the support plate, and both mold and casting chambers are evacuated. (b) An interlock between the melt and the casting chambers is opened, the melt is moved into the casting chamber, and the fill pipe is submerged in the melt. (c) Argon pressure is applied to the casting chamber, causing metal to fill the mold. The mold chamber may or may not be pressurized to achieve the desired differential pressure.

Schematic showing supported shell techniques. (a) The mold is placed in an empty chamber so that the fill pipe protrudes from the bottom. (b) Sand or other particulate media completely fills the void between the shell mold and the chamber wall. (c) A porous vacuum head/separator is placed atop the chamber, and casting operations proceed as described for the countergravity low-pressure air process in Fig. 1.

Fig. 7

Fig. 6

Schematic showing the loose sand vacuum process. (a) Several molds, produced using any of the bonded sand technologies, are placed atop a sheet of aluminum foil and protrude slightly below an open bottom tube, with the mold fill runners facing downward. (b) The region between the casting tube, the molds, and the foil is then filled with a particulate support medium, and a porous vacuum head/separator is placed at the top of the tube. (c) Vacuum is applied to the chamber through the vacuum head, and the assembly is moved to the melt furnace. The bottom of the mold assembly is submerged into the melt, thus causing metal to flow into the individual molds. (d) After the metal has solidified, the mold tube assembly is moved to the shakeout area, and vacuum is discontinued.

manifolds are produced in stainless steel. The process has been demonstrated to produce hollow camshafts in chill iron and nodes for space frame construction in low-alloy steels. Countergravity Centrifugal Casting (C3) Process. The C3 process (Fig. 7) makes use of the supported shell technique to hold the mold

in the casting chamber. Centrifugal force not only retains liquid metal in the part cavities, but after the central runner has been removed, it creates a strong, radial pressure gradient that enhances feeding and floats lighter oxides and gases to the vacant center. Unlike castings that are connected to a central runner and solidify

Schematic showing the steps of the countergravity centrifugal casting process. (a) A mold is placed in a chamber using the supported shell technique. (b) The mold is filled using the countergravity low-pressure air or countergravity lowpressure inert atmosphere casting technique. (c) The mold chamber begins to spin, and atmospheric pressure returns to the chamber, allowing liquid metal in the central sprue to return to the furnace but retaining liquid metal in the parts and in-gates by centrifugal force. Spinning continues until the parts and in-gates have completely solidified.

at different metalostatic pressures based on their height on the runner, each part cast using the C3 process solidifies under exactly the same pressure conditions. This promotes an exceptionally high level of uniformity when compared to parts produced with other casting processes. Many parts originally cast using the CLA and CLI supported shell processes are now cast using the C3 process.

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 715-718 DOI: 10.1361/asmhba0005266

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

High-Pressure Die Casting William A. Butler, General Motors (Retired), Consultant for NADCA

HIGH-PRESSURE DIE CASTING is characterized by the use of a source of hydraulic energy that imparts high velocity to molten metal to provide rapid filling of a metal die. The die absorbs the stresses of injection, dissipates the heat contained in the metal, and facilitates the removal of the shaped part in preparation for the next cycle. The hydraulic energy is provided by a system that permits control of actuator position, velocity, and acceleration to optimize flow and force on the metal as it fills the cavity and solidifies.

Die Casting Alloys and Processes The variety in die casting systems results from trade-offs in metal fluid flow, elimination of gas from the cavity, reactivity between the molten metal and the hydraulic system, and heat loss during injection. The process varieties have many features in common with regard to die mechanical design, thermal control, and actuation. Four principal alloy families are commonly die cast: aluminum-, zinc-, magnesium-, and copper-base alloys. Lead, tin, and, to a lesser extent, ferrous alloys can also be die cast. Parts produced by the die casting process are shown in Fig. 1. The primary categories of the conventional die casting process are the hot chamber process and the cold chamber process. These main process variations have been enhanced in recent years to produce high-integrity die castings, which provide mechanical properties suitable for use in structural and safety applications previously not considered for conventional high-pressure die castings. Hot Chamber. The hot chamber process is the original process invented by H.H. Doehler. It continues to be used for lower-melting materials (zinc, lead, tin, and magnesium alloys). Hot chamber die casting places the hydraulic actuator in intimate contact with the molten metal. The process minimizes exposure of the molten alloy to turbulence, oxidizing air, and heat loss during the transfer of the hydraulic energy. However, the prolonged intimate contact between the molten metal and the system components presents severe materials problems

in the production process, and the process is limited in the size of the part that can be produced. Details are presented in the article “Hot Chamber Die Casting” in this Volume. The cold chamber process solves the materials problem of higher-temperature operation by separating the molten metal reservoir from the actuator for most of the process cycle. Cold chamber die casting requires independent metering of the metal and immediate injection into the die, exposing the hydraulic actuator to the molten metal for only a few seconds. This minimal exposure allows the casting of higher-melting-temperature alloys such as aluminum, copper, and even some ferrous alloys. The cold chamber process is used to make parts of all sizes, from small brackets and connectors to automotive housings and engine blocks. A further discussion is found in the article “Cold Chamber Die Casting” in this Volume. High-integrity die casting processes have been developed in recent years that expand the

Fig. 1

potential markets for high-pressure die castings. These processes enhance various aspects of the conventional die casting process to provide castings with very low levels of porosity that can be heat treated and/or welded to meet stringent product requirements. Examples of highintegrity die castings are shown in Fig. 2. There are three different types of high-integrity die casting processes: squeeze casting, semisolid casting, and high-vacuum die casting. Squeeze casting is a term commonly used to refer to a die casting process in which liquid alloy is cast without turbulence and gas entrapment and subsequently held at high pressure throughout the freezing cycle to yield high-quality heat treatable components. Direct squeeze casting originally was developed as a liquid forging process in which liquid metal was poured into the lower half of a vertically oriented die set and subsequently closed-die forged. This is really not a die casting process. There are also several indirect squeeze casting approaches involving

Typical high-pressure die castings. Courtesy of NADCA (Ref 1)

716 / High-Pressure Die Casting to a high-pressure casting process using extremely low vacuums, in the range of 10 kPa (100 mbar) or less, in order to assure heat treatable castings. High-vacuum die casting relies on the vacuum system to effectively eliminate cavity gases from the shot sleeve and die, thereby eliminating the possibility of gas entrapment. In most other respects, the process is much like conventional die casting and uses thin, trimmable gates coupled with high metal-injection velocities to fill thin-wall sections. High-vacuum die casting generally produces thin-wall parts in which ribs are often used to provide strength. A complete discussion of vacuum die casting is included in the article “Vacuum High-Pressure Die Casting” in this Volume.

Advantages of High-Pressure Die Casting

Fig. 2

Typical semisolid high-integrity die casting products. Courtesy of NADCA (Ref 1)

injection of metal into a cavity via massive gates, which allow adequate feeding of solidification shrinkage while the casting freezes. In commercial practice today (2008), there are vertical or horizontal injection systems with the parting line of the die oriented horizontally or vertically, respectively. Squeeze castings can be made from the full range of heat treatable (and non-heat-treatable) aluminum alloys used in the permanent mold processes. Only the hot-short 2xx, 5xx, and 7xx aluminum alloys, sometimes used in sand and plaster mold casting, are not suitable for squeeze casting. Hot shortness is a tendency of an alloy to crack during or immediately after solidification (Ref 2). Squeeze castings are not limited to the higher-silicon (higher-fluidity) alloys needed for die casting. These processes are discussed in more detail in the article “Squeeze Casting” in this Volume. Semisolid metal (SSM) casting differs from squeeze casting in that it uses a novel semisolid material. In all other respects, however, the basic rules remain. The same low turbulence during filling followed by application of high pressure are critical elements for good process control. The principal difference is that the higher-viscosity semisolid metal allows higher metal velocities to be implemented before the onset of turbulence, and, of course, the metal is already half-solidified at the time of casting. These attributes allow SSM casting to achieve remarkable thin-walled components and high production rates. The feed material for SSM casting shown in Fig. 3 is aluminum A357. Feed material is typically 50 to 60% solid, with a globular

microstructure comprised of solid spheroids suspended in the melted portion to produce the consistency shown. There are three types of SSM processes: thixocasting, rheocasting, and thixomolding. Feedstock for SSM was originally prepared in specially cast billets produced on continuous casting systems, equipped with electromagnetic stirrers. The billets were cut to length and then heated to the proper temperature by induction heating before being placed into the casting machine. Recently, several systems have been developed to prepare on-demand feedstock for SSM casting. These systems offer the potential of reducing the need for special billets and billet heating, thereby simplifying the process and reducing cost. A complete discussion of SSM casting methods is provided elsewhere in this Volume. Vacuum die casting uses a vacuum pump to evacuate the air and gasses from a die casting die cavity and metal delivery system before and/or during the injection of metal. The term vacuum die casting can mean any vacuum in the die cavity below atmospheric pressure. Many die casting companies use vacuum to aid their die casting process. Some manufacturers use vacuum to not only evacuate the air and gasses from the die cavity but also to suck liquid metal into the shot sleeve prior to injection into the cavity. Vacuum is used not only in conventional high-pressure die casting but also to some degree in the high-integrity processes of squeeze casting and SSM casting. There is another term, high vacuum die casting, which describes a process used to produce high-integrity castings, that generally refers

The high-pressure die casting production rate is much higher than for gravity or low-pressure casting. The ability to produce castings with close dimensional tolerances greatly reduces machining operations. Die castings have good surface finish, which is a prime requirement for plating. Die castings can be produced with much thinner wall thickness, reducing overall casting weight. In the die casting operation, more complex parts can be produced, thereby reducing the number of components in an assembly. Inserts can also be cast in place during high-pressure die casting.

Fig. 3

Semisolid metal (SSM) feedstock material. This SSM is A357.0 (UNS A33570), an Al-Si-Mg alloy. Courtesy of NADCA (Ref 1)

High-Pressure Die Casting / 717

Disadvantages of High-Pressure Die Casting The high tooling costs of high-pressure die casting make short production runs generally uneconomical. There are a limited number of alloys suitable for die casting, and internal porosity restricts the heat treatment or welding of the finished castings, except for those produced by the high-integrity processes described previously. Iron or steel alloys are normally not die castable. There are restrictions in die casting on the casting size and casting wall thickness, which eliminate the possibility of die casting some parts. Die casting machine and machine maintenance costs are higher than for other casting processes due to their complexity and high pressure requirements.

Product Design for the Process Product design and die design are intimately related. The high-speed nature of the die casting process allows the filling of thin-wall complex shapes at high rates (of the order of 100 parts per hour per cavity). This capability places additional demands on the casting designer, because traditional feeding of solidification shrinkage is almost impossible. The inability to feed in the traditional sense demands that machining stock be kept to a minimum; high-integrity surfaces should be preserved. A factor in cost is the parting-line topology. The parting line is the line on the casting generated by the separation between one die member and another. The simplest and lowest-cost die has a parting line in one plane. Casting design should be adjusted if possible to provide flat parting lines. Draft is required on the die casting walls perpendicular to the parting line or in the direction of die motion. An important characteristic of good design is uniform wall thickness, which is necessary for obtaining equal solidification times throughout the casting. Die castings have wall thicknesses of approximately 0.64 to 3.81 mm (0.025 to 0.150 in.), depending on casting shape and size. Bosses, ribs, and filleted corners always cause local increases in section size. In particular, bosses that must be machined require consideration of the entire product-manufacturing cycle. The machinist will find it easier to drill into a solid boss; cored bosses may require floating drill heads in order to align the drill with the cast tapered hole that preserves the high-integrity skin of the casting. Cores and slides provide side motions for undercuts. A core body is generally round and buried within the cover or ejector die. A slide body has a rectangular or trapezoidal shape and crosses the parting line of the die. As with the cover and ejector dies, the impression steel is often separate from the holder steel. Cores and slides are actuated by various methods, including hydraulic cylinders, rack and pinion, and angle

pins. Innovative die design permits radial die motion at a price of die expense. There are die casting processes that use complex-shaped disposable cores similar to those in gravity casting processes. Cores and slides provide the casting designer with tremendous flexibility at the expense of an increase in die complexity. A standard set of cores—fixed core pins for small holes that are screwed in, or bolted-in inserts— can be used to reduce die construction cost and to permit rapid replacement. In certain cases, a reentrant shape needs to be cast into the part where there is no space for core/slide mechanisms. In such a case, the die designer can use a loose piece. A loose piece is placed in the die before each shot is made. It is then ejected from the die with the casting and separated manually or by a fixture. Although it provides design flexibility, the load/unload sequence required for loose pieces slows the process, thus increasing cost. Similarly, the die casting process can allow the part designer great flexibility in local material properties by the use of cast-in inserts of other materials, such as steel, iron, brass, and ceramics. The bond between insert and casting is physical, not chemical, in nature. Therefore, the insert should be clean and preheated. The insert should be designed to prevent pullout or rotation under working loads; knurling, grooves, hexagons, or flats are commonly used for this purpose. Proper support of hollow inserts will prevent crushing of the insert under the high metal-injection pressure. The wall thickness of the casting surrounding an insert should be no less than 2.0 mm (0.080 in.) to prevent cracking by shrinkage, hot tearing, and excessive residual stresses. Trimming. The die-cast part is ejected from the die with a variety of appendages (gates, overflows, vents, flash, and robot grasping lugs) that must then be removed. This secondary process is called trimming. Although trimming can be done manually, the high production rates characteristic of die casting demand automation. Trim presses are used to remove the excess material. Castings are often trimmed immediately after the casting process, because their higher temperature reduces the strength of the metal. It is common for the die casting machine and trimming operation to be located in a cell arrangement, with a robot providing the material handling. Trimming conditions directly influence the design of the part and the die casting process, especially the gating and parting-line definition. Trimming is facilitated by flat parting lines. The relatively rough edge that results from trimming may be acceptable and is often left as is. In some cases, this rough edge is not acceptable and must be removed by machining or grinding. The direction of flash must be such that the edge is machinable. Dimensional variation is determined by die design, the accuracy of die construction, and process variation. The most accurate dies are those machined using computer numerical control methods. Close control of alloy composition,

temperature, casting time, and injection pressure will lead to more consistent casting dimensions. The minimum variation in dimensions is found for those features contained entirely within one die half. Therefore, machining locators should ideally be placed in the same die half. Tolerances are a function of casting size and projected area. Features across parting lines have added variation because of the accuracy of repeated die closing. Die temperature, machine hydraulic pressures, and die cleanliness are the principal factors to be controlled. Finally, further dimensional variation occurs if the feature is in a moving die member such as a slide or core. Detailed product specifications for both conventional and high-integrity die castings are available from the North American Die Casting Association (NADCA) (Ref 1). In summary, a cost-effective die casting design demands proper attention to the dimensional variation of the process. Inattention to dimensional factors will lead to an inability to provide consistent products within economic process conditions. The product designer and the die caster must therefore initiate a dialog early in the product cycle.

Metal Injection The distinguishing characteristic of the die casting process is the use of high-velocity injection. The short fill time (of the order of milliseconds) allows the liquid metal to move a great distance despite a high rate of heat loss. Proper process performance depends on the delivery of molten metal with high quality, as defined by temperature, composition, and cleanliness (gas content and suspended solids). The molten alloy is prepared from either primary ingot or secondary alloys. A melting furnace is used to provide the proper temperature and to allow time for chemistry adjustment and degassing. The alloy is often filtered during transfer to a holding furnace at the casting machine. Most die casting machines provide the ability to control the piston acceleration in a linear fashion. Parabolic velocity curves are also available on some controls. This phase of injection can be accomplished in several steps. The third phase of injection is activated as the cavity is close to being filled. This intensification phase draws on an accumulator of highpressure hydraulic fluid or multiplies pressure using conventional piston intensifiers. This increases the pressure on the metal to force the rapidly freezing alloy into incipient shrinkage cavities. The gate is the controlling entry point into the casting. The gate serves a fluid flow need, but it must later be removed from the casting by trimming. Therefore, the gate cross section should be the smallest in the gating system. The cross section is determined by the desired fill time and flow rate that the casting machine can provide.

718 / High-Pressure Die Casting

7000 Machine power 6000 Theoretical Fill rate

Metal pressure, psi

5000

Machine

__________

Hyd. cyl .size

__________

Hyd. press. used

__________

Dry shot speed

__________

Plunger tipsize

__________

4000 Die line(Gate area)

Operating Window 3000

Gate velocity limits 2000 Operating Point

1000

0 0 200 300 1400 2100

400 2800

500 3500

600 4200

700 4900

800 5600

900 6300

1000 7000

Q , in.3/s

Fig. 4

PQ2 diagram generated by process control software. The volumetric flow rate, Q, is represented with a nominal 0 to 1000 scale to clearly show its nonlinearity. The 0 to 7000 scale models a specific process. Courtesy of NADCA (Ref 1)

The selection of a fill time for the casting is based on experience and experimentation. The limited selection of plunger diameters for a given machine restricts the design. The cold chamber process links the volume of metal to the plunger diameter by filling the shot sleeve approximately one-half to two-thirds full. Flow Modeling. It has been recognized by the die casting industry that the ability of the casting machine to provide this metal volume flow, while keeping the dies closed during injection, must be considered. The tool that has been developed for this purpose is called the PQ2 diagram (Fig. 4). It can be shown that the pressure, P, on the metal and hydraulic system is proportional to the square of the injection velocity and therefore the volume flow rate, Q. The line with the negative slope is the machine characteristic line. The charac-

teristic line moves to the left and right with changes in hydraulic pressure, shot valve throttling, and plunger diameter. The line that starts at the origin of the graph is a measured relationship of pressure to flow rate for the particular casting and gate being cast. The effect of adjusting the gate area is that the angle of the line changes. In order to have a successful operation, the operating point should be inside the operating window. The flow characteristics are nonlinear, as illustrated by the Q axis, 0 to 1000 scale, on the PQ2 diagram (Fig. 4). The 0 to 7000 scale below this nominal scale is developed for the specific process being evaluated. Optimization of these various parameters for the casting and machine provides the process engineer with a powerful tool for process definition and debugging. A number of micro-

computer programs are available to base gate design on this hydraulic approach. The overflow is the final component in the fluid flow system. Although they add to the weight of remelt, overflows do serve a variety of purposes. They can act as a reservoir for metal to be removed from the cavity, and they can provide an off-casting location for ejection pins, robot holds, or instrumentation points. REFERENCES 1. North American Die Casting Association (NADCA), 241 Holbrook Dr., Wheeling, IL 60090, (847) 279–0001, www.diecasting.org 2. J.G. Kaufman and E.L. Rooy, Aluminum Alloy Castings, ASM International, 2004, p 295

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 719-723 DOI: 10.1361/asmhba0005267

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Hot Chamber Die Casting Frank E. Goodwin, International Lead Zinc Research Organization

DIE CASTING ALLOYS that do not attack die casting machine components used for injection or delivery of metal to the die during long periods of immersion can be hot chamber die cast. These casting metals are processed at relatively low temperatures and include zinc alloys with casting temperatures between 400 and 450  C (750 and 840  F). Also, lead and tin alloys, together with other systems having casting temperatures below 250  C (480  F), are routinely hot chamber die cast. The injection end of a hot chamber die casting machine is shown in Fig. 1.

Melting Process The holding furnace, which can also serve as a melting furnace if a central melting facility is not used, has traditionally been gas fired and constructed of cast iron. In recent years, great improvements in energy efficiency have been

realized by constructing these pots from ceramic material and using either gas or electric immersion heaters to supply heat to the bath. If a central melting facility is used, liquid casting alloy is delivered either by ladle or by using a launder system. If melting is carried out directly in the pot, it is recommended that ingots of no greater than 6.8 to 9.0 kg (15 to 20 lb) be used. Each is gradually lowered into the bath using a clockwork type of mechanism. The casting alloy is in the form of a long, thin rectangular ingot with a hole through which the hook of the lowering mechanism is inserted. This design of ingot is termed a margash shape (Fig. 2). Gradual lowering of the ingot into the bath rather than sudden additions of whole ingots greatly improves temperature uniformity of the bath and reduces dross and oxides. For zinc, the pot should be equipped with controls so that the temperature of the molten bath can be maintained within 6  C (10  F). The furnace capacity required depends on the size of the casting machine and the production rate.

Shot cylinder

Accumulator

Overflow

Pump

Overflow gate Die cavity

Control valve Ingate nozzle Gooseneck

Gooseneck plunger

Generally, a holding furnace should be able to hold at least four times the amount of metal required for 1 h of operation. If the furnace is also used for melting, its capacity is typically 450 to 9000 kg (1000 to 20,000 lb), although immersion tube furnaces can range up to 18 metric tons (40,000 lb). Total melting furnace capacity is usually five to seven times the amount of metal required per hour.

Injection Components In the molten metal pot, a gooseneck plunger and cylinder arrangement is located that allows introduction of casting metal to the cylinder when the plunger is retracted. These machine components are normally made from tool steel or stainless steel. Also, the sealing rings in the plunger are made of a similar material. This plunger-and-cylinder arrangement injects metal into the gooseneck, so called because of the shape of this channel, which is typically made of cast iron. The gooseneck connects the melting pot to the die casting machine. The gooseneck is shaped to allow for smooth flow of the rapidly injected casting metal and is also maintained at a temperature that avoids solidification of the casting metal in the gooseneck channel. The end of the gooseneck that delivers metal to the die casting machine has a semispherical end that allows it to be fit into the back of the cover or stationary half of the die casting die in a conventional machine. This spherical joint is tightly clamped to prevent leakage of casting alloy.

Hydraulic fluid reservoir Sprue pin

Runner

Die casting machine frame

Fig. 1

Holding furnace

Basic components of a typical hot chamber die casting system

Fig. 2

Two margash-form zinc alloy ingots for feeding die casting alloy to the holding furnace. Note the hole that accepts a hook that slowly lowers the metal into the furnace. Source: Courtesy of Allied Metal Company, Chicago, IL

720 / High-Pressure Die Casting The injection cycle for a conventional hot chamber die casting machine is shown in Fig. 3. To begin the cycle, the die is closed, and the gooseneck is filled with liquid casting alloy. The plunger then pushes the liquid alloy through the gooseneck and nozzle into the die cavity. The alloy is held under pressure until it solidifies. After a predetermined time, the die opens, and the cores, if any, retract. The casting stays in the ejector half of the die. The plunger returns, pulling metal back through the nozzle and gooseneck. The ejector pins push the casting out of the die. As the plunger uncovers the filling hole, molten metal flows through the inlet to refill the gooseneck. Casting Machines. In recent years, so-called four-slide machines have become increasingly popular. These feature a metal platen on which four quadrants of a die casting die are mounted to open and close (Fig. 4). Typical die dimensions are 160 by 160 mm (6 by 6 in.), with casting weights of 245 to 650 g (8.7 to 23 oz). Cycle times can reach 450 shots per hour. In a four-slide machine, the gooseneck is fitted to the back of the platen rather than any portion of the die. The portions of the die close together

Materials of construction for the gooseneck nozzle seat are chosen to resist high wear and have been made from nitrided alloy steel, hot worked tool steels such as H13, high-speed tool steel, and stainless steel. The gooseneck, plunger, and cylinder are always designed to allow easy replacement. Pressure. Injection energy is produced by either hydraulic or pneumatic means. Injection pressure used for zinc die casting alloys usually ranges from 10.3 to 20.6 MPa (1500 to 3000 psi). The lower pressures are used for simpler castings, while higher pressure is for more complex ones. The best practice is to use the lowest pressure that will produce acceptable castings; however, a minimum pressure of 10.3 MPa (1500 psi) is usually required for obtaining an acceptable combination of soundness, surface finish, and mechanical properties. The injection characteristics of the machine can be changed by increasing or decreasing the size of the cylinder or the plunger that fits into it, although some adjustment of plunger stroke length can be made using a machine control system; the amount of metal required for different dies is best adjusted to the die by using a proper-sized cylinder and plunger combination. Movable die half

Fixed die half Nozzle

Ejector pins

Gooseneck

v, F

Plunger Cavity Pot

to form a leakproof cavity into which the casting metal is injected.

Distinctions Between Hot and Cold Chamber Processes Beyond the injection section, a conventional hot chamber die casting machine closely resembles a cold chamber die casting machine. A major difference is the cycle time at which hot chamber die casting machines can typically run. This is due to two reasons. The use of the hot chamber injection system allows for much higher cycling of the submerged plunger in the cylinder, because metal is readily available to fill the injection cylinder after each cycle. In a cold chamber machine, metal must be either manually or automatically ladled into the cold chamber cylinder, which will always require more time. Second, casting metals processed in hot chamber machines have lower melting points than those processed in cold chamber machines; therefore, less heat extraction is needed to solidify the metal before ejection of the part and completion of the casting cycle. The design of the runner and gate channels in the die that transmit liquid casting alloy to the casting cavity also differs in some ways from cold chamber dies. As in the cold chamber process, hot chamber die casting dies can be constructed in singlecavity, multiple-cavity combination, or unit dies. However, dies that are designed to produce zinc alloy castings can seldom be used to produce castings of aluminum alloys or other metals that are cast at higher temperatures, because zinc alloys can be cast at thinner sections, smaller radii, and closer tolerances than aluminum,

Chamber Cross head Slide (a)

V

Die block

(b)

V

Core

Fig. 4

(c)

Fig. 3

Operating sequence for the hot chamber die casting process. (a) Die is closed, and hot chamber (i.e., gooseneck) is filled with molten metal. (b) Plunger pushes molten metal through gooseneck and nozzle and into the die cavity. Metal is held under pressure until it solidifies. (c) Plunger returns, pulling molten metal back through nozzle and gooseneck. Die opens and cores, if any, retracts. Casting stays in ejector die until ejector pins push casting out of ejector die. As plunger uncovers filling hole, molten metal flows through inlet to refill gooseneck.

Four-slide mechanism for production of complex shapes in a single casting. The multi-slide tool is made up of the die block, slide, crosshead and cover plate. Each die block has either a cavity and/or cores on its face, which together form the complete cavity and runner profile into which the molten metal is injected. These die blocks are mounted onto sliders, which fit precisely into a crosshead, ensuring repeatable opening and closing operations. A cover plate, bolted onto the top of the tool, holds all the components together.

Hot Chamber Die Casting / 721 magnesium, or copper alloys. However, a die design for casting the higher-melting-point alloys can be used for casting zinc alloys. The precision capabilities of zinc casting alloys vary with the casting size, and they are described in the most recent edition of NADCA Product Specification Standards for Die Castings (Ref 1). For zinc die castings, both a normal and precision tolerance level are given. These allow many components in zinc to be cast to near-net shape or even net shape condition, avoiding the need for further operations, such as machining, that would add cost to the part. On the opposite side of the gooseneck nozzle bushing, the orifice that allows metal to enter the die cavity faces the beginning of the feeding system in the moving or ejector side of the die, termed the sprue. The purpose of the sprue is to allow for uniform introduction of the injected liquid metal to the runner system that feeds the die cavities. The volume of this sprue is much smaller than the volume of the biscuit typically used in cold chamber die casting and can be a much smaller proportion of the untrimmed casting weight in comparison with a part made in a cold chamber die. The design of the gate and runner systems used in hot chamber die casting is otherwise similar to those in cold chamber die casting.

Gate and Runner Design An example of a typical runner, gate, and overflow configuration for a faucet fixture casting is shown in Fig. 5. This casting is both highly decorative and functional; the use of overflows allows hot metal to fill areas critical to appearance. Casting metal enters the sprue, where it is distributed to one or more runner systems that feed casting cavities. The main runner for a casting can be divided into branch runners, but the sum of the cross sections of all branch numbers cannot exceed the cross section

of the main runner. Generally, slightly tapered runner systems are used to accelerate the flow and prevent air entrainment. Each runner ends with a narrow gate in which the casting alloy is further accelerated as it fills the die cavity. General rules to follow in selecting a feed system configuration are:  Minimize runner length  Provide uniform flow across the die cavity,

such that air is forced out of the cavity ahead of the molten zinc into overflows and/or vents Feed from the side with the most critical surface requirements Flow across the short dimension of the cavity Avoid flow across large openings by center gating (Fig. 6) or provide runners across large openings Select runner and gate configurations that control flow direction through the cavity so that the metal tends to follow desired paths

   

The dark sections shown in Fig. 5 and 6 are trimmed off, leaving the final unshaded casting. The lines drawn on the casting are segment lines, which are a useful way to visualize the intended metal flow paths in the casting cavity.

Temperature Control The temperature of the die casting dies for zinc die casting is relatively low, ranging from 160 to 245  C (325 to 475  F); therefore, hot worked tool steels are not generally required for dies. However, for extremely long runs and high dimensional accuracy, hot worked tool steels such as H13 will provide optimal die life. The die hardness is also less critical for alloys of higher casting temperature. Steels prehardened by the manufacturer to any maximum hardness content with reasonable machinability can be used. The typical hardness is 29 to 34 HRC (Rockwell C hardness) or 280 to 320 HB Overflows

Tangent runner 1 Uniform runner 1 Second cavity Uniform Fan gate runner 3 (main runner)

2

1

9

10

3

(Brinell hardness). For the slides and cores, hardenable stainless steels such as 440B can be used in many cases. If significant wear occurs, such as in slides, hot worked tool steels such as H13 can be used. Because these tool steels respond to nitriding, they can be selectively hardened, and therefore, a component made from one of them can be treated for different properties in different areas. For example, the main portion of cores and slides can be nitrided for wear resistance, and the end can be masked to resist nitriding for better resistance to heat checking and spalling. Lubricating the slides and cores with molybdenum disulfide or colloidal graphite helps to ensure smooth action and minimize wear. With hot chamber die casting of zinc alloys, the surface life of the die casting die is normally much longer than that for casting aluminum, magnesium, or copper alloys. It is not unusual for die to last for one million shots, and the very small castings made on four-slide machines can reach one billion shots before extensive maintenance is required. Maximum die life depends largely on having well-designed dies, trained operators, and a rigidly enforced program of machine and die maintenance. Lower die temperatures, near 160  C (325  F), are normally used for thicker-section die castings, whereas the higher end of the range, 245  C (475  F), is used for thin-section castings. When hardware finish is required, temperatures at the high end of the range are generally required, regardless of the casting thickness. Some casting shapes require localized heating or cooling of the die above or below the established temperature. Metal overflows are often used to heat die areas surrounding the parameters of castings that have thin sections far from the main runner. This method of local heating helps to fill thin sections and to improve casting finish. Conversely, water channels are frequently placed behind the runner area immediately adjacent to the sprue to provide localized cooling and to prevent soldering of the molten metal to the die. Control of die temperature is one of the most important parameters affecting production rates and casting quality. A number of die design programs are available to allow the designer

4

5 6

Sprue base

Turning vane Uniform runner 2

7 8

Typical cross section through runner and ingate

Fig. 5

Tangent runner 2

Sketch of general runner, gate, cavity, and overflow configuration of a faucet fixture casting. The numbered sections indicate the flow pattern in the finished part. The dark runner area and the overflow areas are trimmed from the finished part.

Fig. 6

Center-gated casting. An alternative method of gating a large casting with a large center opening. Dark areas will be trimmed from finished part. Lines indicate flow pattern from center to perimeter.

722 / High-Pressure Die Casting to determine proper placement of cooling lines in the die and the coolant capacity required. Generally, dies are designed with excess cooling capacity, and then a temperature control system is used to modulate the flow of coolant in the die. Heating and cooling of the die can be accomplished using oil as a heat-transfer medium. Water cooling can also be used, but in this case, cartridge heaters or other means of preheating the die prior to beginning the casting campaign are required. Many die casters control die temperature by observing the casting being produced and then manually adjusting valves to regulate the flow of die cooling water through passages in the die until a satisfactory casting is produced. If the cooling passages are properly located, if the machine is operating consistently in a very repetitive manner, if the machine operator is skilled and is giving sufficient attention to the die temperature, and if the part does not require close temperature control, this method is satisfactory. However, as casting conditions become more critical, as they are with thin-wall die castings, it is virtually impossible to control the flow of cooling water manually so as to obtain satisfactory castings consistently. Automatic temperature control should be employed to continuously monitor die temperature and adjust the coolant flow rates to maintain the optimal die temperature for critical applications. Studies on some large castings indicate that scrap is reduced between 15 and 40% when temperature is automatically controlled (Ref 2–4). The reasons for the scrap reduction are:  No coolant flow occurs during start-up until

the die is up to temperature, reducing startup time and amount of scrap produced during start-up.  Die temperature variations are reduced during steady running.  Whenever an interruption occurs in the production process, the coolant is automatically shut off, reducing the number of scrap parts that must be made to bring the die back up to operating temperature.

Alternatively, a hot oil medium can be used instead of die coolant, which provides accurate control of die temperature. In either case, a basic die temperature controller of the type diagrammed in Fig. 7 is suggested. The basic components are a thermocouple, located approximately 6.4 mm (0.25 in.) below the die cavity surface, a temperature controller that determines whether the temperature as measured at the thermocouple location is above or below a setpoint temperature, a solenoid valve to regulate coolant flow, and a manual valve to control the flow rate. If the manual valve is not used, the temperature will tend to swing through large extremes, and a generally unsatisfactory performance will result. For complex castings, or where more control is desired, a multichannel temperature controller is desired. This monitors temperatures from thermocouples located at selected locations in the die and controls the flow of coolant to each selected part of the die. An example is the location of thermocouples on an automobile headlight bezel (halfsection view) (Fig. 8).

Ejection and Postprocessing Robotics. Operations such as die lubrication, casting removal, and automation of the die casting process using robotics are common to hot chamber and cold chamber machines. Trimming of the die casting after it is ejected from the machine is automatically performed for many zinc die castings because of their small size. This requires that the gate thickness be carefully controlled. Tumble degating of many zinc die castings is possible. The conduit connector (Fig. 9) is an example of a high-volume part designed for tumble degating. However, for larger castings, a trim press may be required. In finished trimming, the part must almost always be quenched in a water-soluble coolant or lubricant similar to the lubricants used for machining zinc castings. Failure to include soluble oil usually causes the trim die plates to solder, with consequent tearing of the trimmed parts.

Quenching several hundred pounds of shots per hour in a small tank requires either a heat exchanger for the quenching fluid or connection to a larger remote reservoir so that heat can be dissipated naturally. Trim presses that remove the sprue and runner system, together with overflows and flash, typically deliver these trimmed portions of the shot to a tank that allows for recycling of this scrap after cleaning. In-die degating can also be designed, allowing for simplification of the postprocessing area. Recycling. Clean scrap from trimming operations, together with scrap casting, can be recharged into the die casting circuit in unlimited portions, although the use of 50% maximum of scrap per charge is recommended. There are safeguards that should be employed to ensure that remelt does not disrupt the desired alloy chemistry. The ability to remelt zinc process scrap many times without losing properties is of significant advantage to the zinc die caster; however, caution should be taken to keep this material clean and free of unwanted substances. It should be stored separately from other metals if it is accumulated in batches. If conveyed back to a central melter, conveyers and the furnaces should be covered when maintenance work is being done overhead. Floors and tables should also be kept clean. If there is any doubt as to the purity of the scrap, it should not be used in any proportion until it has been analyzed. Scrap returned for recycling must be free of oil and moisture. A safety hazard is created when oil and moisture are present on the metal being charged into the furnace. Some zinc scrap may be electroplated. This material should be remelted separately and added back in small quantities or, better still, sold outright. The electrodeposited metals will separate and float to the top of the bath, where they can be skimmed off. Agitation will increase copper, nickel, and chromium levels and therefore should be avoided. Die castings that have been joined using lead- or tin-base solders should not be added to the scrap feed under any circumstances. The very low limits

A Thermocouple B

Die cavity

C

Controller

Fig. 9

Water main Solenoid valve

To drain

Fig. 7

Manually operated valve

Components of a commercial single-channel die temperature-control system

Fig. 8

Location of thermocouples in a die for a headlight bezel. A, B, and C indicate approximate positions of thermocouples to be mounted 1 6.4 mm (4 in.) below the die cavity surface.

Design of a casting for automatic degating. A two-cavity die is used to make electrical conduit connector castings. Use of thin gates and placement of the mass of the casting at the end of a runner allow the use of tumbling to automatically degate this high-volume part. Zinc casting metal enters through a central sprue and is divided to flow through the two runners in opposite directions. One of the castings is attached, while the other has been removed to show the location of the gate. Source: Courtesy of Bridgeport Fittings, Bridgeport, CT

Hot Chamber Die Casting / 723 of impurities in casting composition specifications make only very small quantities of contamination by lead, tin, or cadmium capable of bringing a large holding pot or melting furnace out of specification. This includes castings that may have pressed-in bronze bearings that, in many cases, contain lead or tin—introducing such items into the scrap feed should be avoided at all cost. Even when zinc alloys are highly contaminated, a zinc refinery can refine die-cast scrap, allowing for complete recycling. Worldwide, 400,000 metric tons of zinc alloy die castings are recycled, mainly from shredded automobile scrap (Ref 5). Fluxing of zinc alloys during melting and holding is generally not required and can change the percentage of alloy elements. However, 1.4 to 2.3 kg (3 to 5 lb) of a chloride or fluoride flux

should be added for each 450 kg (1000 lb) of metal when the melting stock partially or totally comprises trimmings, gates, and rejected castings. A few pounds of flux per ton of alloy will reduce the magnesium content, and greater flux additions can make the magnesium disappear completely. Therefore, it is necessary to control temperatures continuously, to flux properly, and to check the analysis, particularly for aluminum and magnesium contents. Adjustments should be made as required.

REFERENCES 1. NADCA Product Specification Standards for Die Castings, 6th ed., North American Die Casting Association, 2006

2. B.K. Dent and R. Fifer, Production Operation with the ILZRO-Battelle Multichannel Temperature Controller, Paper 5572, Transactions of the Seventh SDCE International Die Casting Congress, The Society of Die Casting Engineers, 1972 3. R. Reddi, “Temperature Control of Die Casting Dies,” Paper G-T75-125, Presented at the Eighth SDCE International Die Casting Congress, March 17–29, 1975 (Detroit, MI), The Society of Die Casting Engineers 4. Zinc, (No. 1), Zinc Institute, Inc., 1974, p 1 5. “Zinc Recycling—The General Picture,” International Zinc Association—Europe, 2005

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 724-726 DOI: 10.1361/asmhba0005268

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Cold Chamber Die Casting Robert J. McInerney II, BuhlerPrince

THE COLD CHAMBER die casting process is used with higher-melting-point alloys, such as aluminum and magnesium. Since the cold chamber is located outside of the furnace, as compared to hot chamber, it requires a means of moving the molten metal from the holding furnace to the cold chamber. The cold chamber is attached between the die casting machine front platen and the die. The transport of the molten metal is typically done with a ladle mechanism, either manually or automatically, when casting aluminum alloys. Casting cycle times can range from 10 s for a small machine up to 2 min for a large machine.

Machine Components The cold chamber high-pressure die casting machine performs several functions during the

Fig. 1

die casting process and is essentially two machines that work together. The two machines are called the clamp end and the shot end. The machine functions are controlled, directly or indirectly, and machine variables are monitored by the processing engineer. Figure 1 shows the main components of a horizontal cold chamber die casting machine. Items to the left of the front plate are considered part of the clamp end, and items to the right of the front plate are considered part of the shot end. The machine is powered by a high-performance hydraulic system, consisting of pumps, cylinders, valves, and oil/nitrogen accumulators. Oil/nitrogen accumulators are used because they store a large amount of potential energy. This energy can be released easily and accurately in terms of time and pressure by the control valves in the system. Clamp End. The functions of the clamp end of the machine include opening and closing the die,

Typical parts of a horizontal cold chamber die casting machine

developing clamping force, and providing power for sliding cores and for ejecting the casting from the die cavity. The machine must open and close the die on a straight line to keep the die parting line parallel. In most cases, the faces of the die are flat, and their plane determines the parting line. If this is not done properly, there will be incomplete closure and interference between the die and the casting. This interference can cause the casting to be bent and the tool to wear prematurely. The design of the machine is robust to provide the needed rigidity. To open and close the die straight:  The machine must be level.  The plate support rails and plate supports must

be in good condition for operation and leveling.

 The tie bars and locking mechanism must be

parallel with the base.

Cold Chamber Die Casting / 725 The shot end includes the cold chamber, plunger rod, and plunger tip (Fig. 2). The primary mechanical requirement of the shot end is to provide good alignment so that proper clearance is maintained between the plunger tip and cold chamber. Misalignment will result in a drag, which will affect plunger velocity. The cold chamber container is also called the shot sleeve. To the right of the plunger rod are the hydraulic system piston (shot cylinder), shot accumulator and intensifier accumulator, and control valves that provide velocity to the plunger. This velocity is changed at predetermined positions, providing the output force (pressure). The velocity changes are linear or parabolic. This output force allows the plunger to move forward and is increased after the cavity is full for the intensification phase.

Process Parameters

Fig. 2

Typical parts of the shot end of a horizontal die casting machine

Fig. 3

Shot trace from machine control panel. Process variables monitored are plunger position and velocity and cylinder rod and shot cylinder head pressure.

 The front plate must be perpendicular to the       

base. The four plates must not be bent. There must be caved-in areas where the die makes contact with the plate. The tee slots that hold the die to the plate must be in good condition (no cracks). The tie bars and load indicators must be in good working condition. The linkage and bearing housing can have only a minimal amount of wear. The traveling plate support must carry the traveling plate (no weight can be on the tie bars). Die carriers must be used when needed.

The machine must also provide the correct amount of clamping force to seal the die parting line, operate core slides and locks, and hold the die closed against the force of the molten metal. The required force to hold the die closed is determined by the projected area of the casting and the amount of pressure on the metal. Mechanical locks are often needed. More details on die components and fluid flow dynamics are found in the article “Die Casting Tooling” in this Volume. Shot End. The purpose of the shot end is to inject the molten metal (the shot) into the die. This injection must be done at a controlled velocity and pressure.

The shot profile, as seen on a machine control screen for a typical aluminum die casting process, is given in Fig. 3. The velocities in this process are called slow shot and fast shot. The slow shot is typically split into two different phases: a pour hole close speed and a speed used to control the wave of molten metal in the cold chamber as the cold chamber is filling. The shot must start forward with as smooth a motion as possible as soon as the ladle mechanism finishes the pour, to prevent metal from splashing out of the shot sleeve (cold chamber). The size of the die cavity, including gating and vents to be filled, determines the size of the shot. This pour hole close velocity (0.25 m/s, or 10 in./s) indicates the plunger forward motion until the pour hole is closed. The second portion of the slow shot phase is the critical slow shot. The plunger diameter and the percentage that the cold chamber is filled determine this critical slow shot velocity (0.4 m/s, or 15 in./s). Some die casting machines are capable of providing an engineered linear or parabolic acceleration rate to the critical slow shot velocity, reducing the possibility of air entrapment even more. Fast Shot. The final velocity phase provided by the casting machine is the fast shot velocity (2.8 m/s, or 110 in./s). This fast velocity is determined by the required time to fill the cavity and the plunger velocity needed to push the metal through the gate for atomized flow. Figure 3 shows a complete shot profile, including the shot velocity and the related shot cylinder piston and rod pressures. From the plunger distance traveled and its velocity, most of the shot cycle is consumed by the slow shot portion (2 s), and the fast shot takes approximately 0.1 s. The die must be filled quickly to minimize heat loss in the melt. The intensification phase lasts a few milliseconds. Intensification Phase. The final phase of the cold chamber casting shot process is the intensification phase. The aluminum alloy will

726 / High-Pressure Die Casting shrink approximately 7% as it solidifies, and as the casting shrinks, there will be tears, cracks, and voids in the casting. To help overcome this shrinkage, which would result in a porous, lowquality casting, there must be a means of pushing metal through the gate to fill the shrinkage. The intensification system on the casting machine is designed to increase the output force of the shot cylinder, which produces higher flow pressure to force metal through the gate and narrowing flow paths. This will help reduce or eliminate the shrink porosity.

Component Size. Sizing of the cold chamber and plunger tip is highly dependent on the casting that is being cast. For small castings, the die can be designed with multiple cavities to match the machine capabilities and to increase productivity. The process engineer will use the casting size, wall thickness, and final function to determine the volume of melt required and the corresponding correct cold chamber size. Knowing these variables will also allow the engineer to calculate the correct shot velocities and intensification phase pressure.

The size of the cold chamber/plunger tip diameter may change from 50 mm (2 in.) for a small casting, such as a power saw component, to 180 mm (7 in.) for a large die that produces automobile parts. The die casting process and tooling are also discussed in the articles “High-Pressure Die Casting” and “Die Casting Tooling” in this Volume.

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 727-731 DOI: 10.1361/asmhba0005269

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Squeeze Casting* Rathrindra DasGupta, National Science Foundation

Squeeze Casting Squeeze casting (Ref 1–4) is considered a high-integrity process and has emerged as an alternative to conventional casting techniques (gravity permanent mold, low-pressure permanent mold, sand casting) and forging. Squeeze casting builds on conventional (high-pressure, high-velocity) die-casting practices and, in recent years, has been widely used to manufacture automotive components (essentially aluminum) requiring high impact strength, high fatigue strength, and pressure tightness or wear resistance. Two types of squeeze-casting machines are commercially available: vertical squeeze cast (VSC) and horizontal-vertical squeeze cast (HVSC). Figure 1 shows the schematic of a typical HVSC machine that is more widely used. Major features of the HVSC machine include:  Horizontal die clamping  A vertical, high-pressure delivery system  A tilt-docking injection unit in which the

 Smaller dendrite arm spacing and a fibrous

 Use of localized squeeze pins to eliminate

eutectic silicon morphology resulting from the rapid solidification rate  Heat treatable. Unlike conventional (highpressure, high-velocity) die castings, squeeze castings can be subjected to solution treatment with minimal or no blistering.  Used for sections >2.5 mm (0.1 in.)

shrinkage porosity in areas otherwise difficult to feed Avoidance of varying wall thicknesses (thick-thin-thick) to aid in metal flow Casting layout in the die (Fig. 3). It is evident from the comparison that a more optimized flow pattern is achieved with the vertical gating and not the horizontal gating (vortex formation). Avoidance of sharp corners (to achieve nonturbulent flow) and undercuts Use of heat treated, premium-grade H13 steel or equivalent for the cavity/retainer Use of larger gates and runners (in comparison to conventional die casting) to assure planar cavity fill and to aid in directional solidification Flat area for the gate, since the gate must be sawed off and not trimmed, as in conventional die casting

Casting and Tooling Design for Squeeze Casting The casting/tooling design process (Ref 1, 2) for squeeze casting (HVSC) requires the following considerations:  Location of cooling channels or high-ther-

mal-conductivity-material inserts (Fig. 2) in the tool to extract heat from the tool quickly and to achieve directional solidification

sequence involves die closure, transfer of molten metal into a vertical chamber (shot sleeve), tilting of the shot unit to injection position, docking of the shot sleeve, and injection of molten metal into the die cavity

 

  



Cover half

Eject half Moving platen

Cavity Stationary platen

Molten metal

In the squeeze-casting process, degassed and filtered molten metal is ladled into a vertical shot sleeve and slowly forced into a preheated and lubricated die cavity at a speed slower than that used for conventional die casting. Solidification of metal then occurs under a continuous application of pressure. The typical advantages of squeeze casting include:

Shot cylinder

 Minimal air entrapment in the matrix result-

ing from reduced turbulence during the flow of metal into the die cavity  Reduction or absence of shrink porosity due to high cavity pressure (70 to 100 MPa, or 10 to 15 ksi) and rapid solidification rate (from the contact of the molten metal with the lubricated die surface)

Fig. 1

Schematic of a horizontal-vertical squeeze-cast machine

*Reprinted with permission from the North American Die Casting Association

728 / High-Pressure Die Casting

Factors Affecting Squeeze-Cast Products (Ref 1, 2) As in any casting process, monitoring and control of select process parameters influencing the integrity of squeeze castings is also necessary to produce defect-free (absence of porosity and/or inclusions) components. The following sections outline a few of the key process characteristics.

Cube insert

Metal Temperature and Metal Cleanliness  Metal temperature control (to less than +/ 5 After filling completes, heavy section is very hot.

Fig. 2

High-thermal-conductivity inserts

 C, or +/ 3  F) helps minimize hydrogen absorption by the melt (thus retarding the occurrence of gas porosity in castings), sludge formation, and/or oxidation of the molten metal (thereby hindering the occurrence and entrapment of hard spots in castings).  The cleanliness of the metal is typically achieved via continuous degassing (lance or porous plugs) of molten metal in the furnace and the use of ceramic filters. The overall cleanliness of the molten metal can be easily determined using either the reduced pressure test or the Prefil-Footprinter (ABB Group) (see the article “Measurement of Metal Quality” in this Volume).

Casting (Cavity) Pressure  Pressure levels of 70 to 100 MPa (10 to 15

ksi) are generally used to help reduce shrink porosity and to promote rapid solidification of the molten metal in the die. The magnitude of cavity pressure is dictated by the part geometry, and any increase in cavity pressure beyond the optimal pressure required for a given component does not necessarily contribute to significant improvements in mechanical properties. Oxides stay on the flat surface

Less metal pressure drop

Type of Lubricant  Care must be taken in selecting the type of

lubricant in squeeze-casting operations, since the reaction between the molten metal and the lubricant may lead to undesirable gases (e.g., hydrogen) being generated. Entrapment of these gases in castings can contribute to the formation of blisters during heat treatment.

Gate Position, Area, and Velocity  Gates are normally >3 mm (0.1 in.) thick to

Fig. 3

Casting layout in the die. Vertical gating (right) provides a more optimized flow of metal.

avoid premature freezing during component solidification. Gate placement is at the thickest part of the component to promote directional solidification.

Squeeze Casting / 729  Gate velocity is under 500 mm/s (20 in./s).

The slow gate velocity helps reduce turbulence and entrapment of gas in the metal.

Presolidified Layers (Cold Flakes) from the Shot Sleeve  Cold flakes form on the inner wall of the

shot sleeve when molten metal, in contact with the inner walls of the shot sleeve, solidifies prior to injection of the metal into the die cavity. Entrapment of cold flakes in castings contributes to a nonknitting condition and thus can affect product performance, including premature failure of the casting. Maintaining the appropriate sleeve temperature is therefore critical in preventing the occurrence of cold flakes.

melt cleanliness, and shot velocity to help reduce gas entrapment.

Squeeze-Cast Applications Squeeze casting has been successfully used in manufacturing a variety of products using traditional nonferrous casting alloys (particularly aluminum) (Ref 1–4). Examples of squeeze-cast applications include: Axle Carrier. Cast from a modified 383 aluminum alloy with a T6 temper and shown in Fig. 4, the part requires high strength (yield strength >280 MPa, or 40 ksi; tensile strength >350 MPa, or 50 ksi) and high stiffness.

Axle Cover. This part requires pressure tightness. It is shown in Fig. 5 cast from a modified 383 aluminum alloy with F temper. Front Steering Knuckle. The mechanical property requirements for these knuckles are: yield strength, >207 MPa (30 ksi); tensile strength, >276 MPa (40 ksi); and elongation, >6%. An example cast in A356.2-T6 aluminum alloy is shown in Fig. 6. Suspension Links. The mechanical property requirements for the links are: yield strength, >207 MPa (30 ksi); tensile strength, >276 MPa (40 ksi); and elongation, >7%. Various

Heat Treatment  Too high a solution-treatment tempera-

 

 

ture may cause incipient melting of the casting; on the other hand, too low a temperature may cause inadequate dissolution of Mg2Si and other phases. Too long a time at temperature may coarsen the eutectic silicon particles, thereby contributing to reduced ductility. Care must be taken in selecting the appropriate quench medium (air, water, or polymer) as well as the quench medium temperature. Although the use of hot water as a quenchant remains popular, forced-air cooling or quenching in a polymer bath following solution treatment are techniques often employed to minimize distortion of very large and complex components. The quench delay is typically less than 30 s. Too high an aging temperature or too long a time at temperature must be avoided to prevent overaging (softening) of castings.

Fig. 4

Axle carrier cast from modified 383-T6 aluminum alloy

Fig. 5

Axle cover cast from modified 383-F aluminum alloy

Porosity  A critical feature for the proper use of

squeeze casting is the application of process simulation and modeling prior to casting development trials to predict areas that may be difficult to feed, that is, determining areas prone to shrink porosity. Optimization of cooling line configurations and cooling sequences is one way to reduce or eliminate porosity in these locations. The other option involves the use of separate squeeze pins (localized squeeze), as described earlier.  Gas entrapment in squeeze castings can lead to the occurrence of blisters during heat treatment. Thus, attention must be paid to proper vent design/location, use of appropriate lube-and-dilution ratio,

730 / High-Pressure Die Casting examples cast in A356.2-T6 aluminum alloy are shown in Fig. 7.

Commonly Used Alloys, Properties, and Microstructures (Ref 1–5) Table 1 compares the tensile, hardness, and impact properties of select squeeze-cast aluminum alloys with those obtained from conventional casting processes (e.g., gravity permanent mold, or GPM, and conventional die casting). Key observations from Table 1 are: 

Fig. 6

For a given heat treatment process and aluminum alloy (A3256.2, 357, or 206), squeeze casting exhibits higher ductility than GPM; however, both processes reveal comparable strengths. The higher ductility of squeeze-cast components is attributed essentially to the lack/absence of porosity in the matrix. In addition, squeeze castings yield a more refined microstructure (more rounded eutectic silicon particles) than GPM. Typical microstructures of squeeze-cast A356.2-T6 and 390-T6 alloys are shown in Fig. 8 and 9, respectively.  Reduced porosity and refinement of microstructure are factors also responsible for squeeze-cast modified 383-F (F = as-cast) exhibiting higher tensile properties than conventional die-cast (high-pressure die cast, or HPDC) 383-F aluminum alloy.  In comparing the impact properties, it is evident that the 390-T6 aluminum alloy exhibits the lowest impact strength. The lack of ductility in this alloy (<1%) is a major factor contributing to its poor impact strength. On the other hand, the highest impact strength, associated with squeezecast 206-T4 aluminum alloy, can be attributed to both its high strength and ductility.

Front steering knuckle cast from A356.2-T6

Compression Camber

Camber Tension

Fig. 7

Suspension links cast from A356.2-T6

Table 1

Various properties of squeeze-cast aluminum alloys Yield strength(a)

Alloy

A356.2-T6 A356.2-T6 357-T6 357-T6 383-F(d) 383-F(d) 383-T4(d) 383-T6(d) 390-F 390-T6 206-T4 206-T4

Process

Squeeze GPM(c) Squeeze GPM(c) HPDC(e) Squeeze Squeeze Squeeze HPDC(e) Squeeze Squeeze GPM(c)

MPa

221–234 207–228 241–262 248–262 152–172 145–159 234–255 296–317 172–193 N/A 248–269 248–283

ksi

32–34 30–33 35–38 36–38 22–25 21–23 34–37 43–46 25–28 N/A 36–39 36–41

Tensile strength(a) MPa

296–310 283–303 324–338 331–345 241–262 269–290 359–386 379–421 193–228 352–396 400–421 407–441

ksi

43–45 41–44 47–49 48–50 35–38 39–42 52–56 55–61 28–33 51–57 58–61 59–64

Impact strength(b) Elongation(a),% Hardness, (HRB)

10–14 3–5 8–10 5–7 1–2 2.75–3.5 5–7 3–5 0.5–1.0 <1 21–26 14–18

48–63 45–58 52–68 50–65 50–60 50–60 55–70 73–84 50–60 80–90 63–75 60–75

J ft
lb

14–18 (10–13) 8–10 (6–7) N/A N/A N/A N/A 7–11 (5–8) 5–8 (4–6) <1 (<1) <1 (<1) 47–56 (35–41) 35–48 (26–35)

(a) All tensile data based on samples machined from actual castings. (b) Cross section of impact specimens is 10  3.3 mm (0.4  0.1 in.); all specimens machined from actual castings. (c) GPM, gravity permanent mold. (d) Modified chemistry. (e) HPDC, high-pressure die casting

In addition to tensile properties and impact strength, fatigue strength is often required of aluminum alloys used in manufacturing certain automotive products, including brake and suspension components. The mean fatigue strength for a commonly used squeeze-cast aluminum alloy (e.g., A356.2-T6) based on the “staircase” method at 107 cycles is 106 MPa (15 ksi). In comparison, the mean fatigue strength of squeeze-cast modified 383-T4 aluminum alloy was found to be 113 MPa (16 ksi). The high fatigue strength of the modified 383 aluminum alloy may be due to its higher silicon level (versus A356.2), which contributes to increased strength and stiffness by reducing the level of porosity (fluidity increases with increasing silicon) and pore size. Table 2 compares wear properties of select aluminum alloys and processes. High wear resistance (associated with reduced material volume loss) is also a characteristic necessary of certain automotive products, including brake cylinders, calipers, and steering gear housings.

Squeeze Casting / 731 ideal material for lightweight constructions. Work recently performed with AZ91D magnesium alloy reveals the following:  Wall thickness does not significantly affect

tensile properties.

 The Mg17 Al12 phase (in as-cast microstruc-

ture) does not form a continuous network.

 Squeeze-cast AZ91D-T4 magnesium alloy

(a)

Fig. 8

yields higher elongation than that achieved from squeeze-cast 380-F castings.  Squeeze-cast AZ91D-T6 magnesium alloy yields properties that are comparable to those obtained from 380-F conventional die castings.

(b) Squeeze-cast A356.2-T6 microstructures. (a) Original Magnification: 200. (b) Original Magnification: 500

REFERENCES Table 2 Wear properties comparison Alloy

380-F A356.2-T6 A356.2-T6 357-T6 357-T6 383-F(b) 383-T6(b) 390-T6

Process

Conventional die cast GPM(a) Squeeze GPM(a) Squeeze Squeeze Squeeze Squeeze

3

Volume loss, mm

0.462–16 0.418–0.665 0.140–0.375 0.320–0.566 0.140–0.363 0.137–0.241 0.123–0.189 0.050–0.100

(a) GPM, gravity permanent mold. (b) Modified chemistry

Fig. 9

Typical squeeze-cast 390-T6 microstructure reveals primary silicon (arrows) and eutectic silicon particles. Original magnification: 200

It is evident from the data shown that squeezecast aluminum alloys exhibit improved wear resistance compared to GPM or HPDC. Reduced porosity in the matrix and refined microstructures (presence of more rounded

eutectic silicon versus platelike silicon particles) are factors responsible for the enhanced wear resistance. Although squeeze casting has been widely used with aluminum alloys in recent years, the compatibility of this process with magnesium alloys is now being explored. Low density, high specific strength, and excellent castability of magnesium qualify magnesium alloys as an

1. R. DasGupta and Y. Xia, Squeeze Casting: Principles and Applications, Die Cast. Eng., NADCA, Jan 2004, p 54–58 2. R. DasGupta, Y. Xia, and B. Szymanowski, High Volume Squeeze Cast Applications, Proceedings of the Second International Light Metals Technology Conference, June 8–10, 2005 (St. Wolfgang, Austria), p 269– 274 3. S. Corbit and R. DasGupta, “Squeeze Cast Automotive Applications and Squeeze Cast Aluminum Alloy Properties,” Paper 1999– 01–0343, SAE International Congress 4. R. DasGupta, C. Barnes, P. Radcliffe, and P. Dodd, “Squeeze Casting of Aluminum Alloy Safety Critical Components for Automotive Applications,” Paper 33, 67th World Foundry Congress, June 5–7, 2006 (Harrogate, U.K.) 5. P. Burton, Z. Brown, and R. DasGupta, “Microstructure and Mechanical Properties of Squeeze Cast Magnesium Alloy AZ91D,” Paper 2005–01–0330, SAE International Congress and Expo

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 732-733 DOI: 10.1361/asmhba0005276

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Vacuum High-Pressure Die Casting William A. Butler, General Motors (Retired), Consultant to NADCA

VACUUM HIGH-PRESSURE DIE CASTING uses a vacuum pump to evacuate the air and gases from the die casting die cavity and metal delivery system before and during the injection of molten metal. The term vacuum die casting can mean any reduced pressure in the die cavity below atmospheric pressure. The method is widely used not only in conventional high-pressure die casting but also to some degree in the high-integrity, high-pressure die casting processes. Effective vacuum die casting removes air and gases from the die casting die, improves the fill of the part, reduces dissolved gases in the molten metal, enhances leak tightness, allows T6 heat treating of aluminum and welding of the parts, addresses surface finish and coating issues, and can increase the burst strength of rotating components. Vacuum can also increase the operating window for the parameters of the die casting machine. Vacuum riserless casting (VCR) is a variation of low-pressure die casting that uses vacuum alone to draw metal into the die. The VCR process, discussed elsewhere in this Volume, is limited by the size and cost of dies but has the advantages of high volume and improved surface and internal quality, like vacuum high-pressure casting.

Vacuum Die Casting An alternative to displacement venting is the extraction of the air ahead of the metal with a vacuum system using an accumulator. Once the die is closed and the shot hole sealed, the vacuum valve is opened, and the air in the cavity is drawn into the accumulator. The vacuum valve is either closed after a preset time or by metal impingement, which shuts the valve to prevent metal from filling the vacuum system. The accumulator is then pumped down for the next cycle. The vacuum pump should have the ability to evacuate the air and gases quickly. Open flow rates are 3 to 6 m3/min (100 to 200 ft3/min, or 3000 to 6000 in.3/s). Mechanical pumps powered with 4 to 8 kW (5 to 10 hp) motors are used to evacuate a single die casting die. Typically, the vacuum

High-Vacuum Die Casting High-vacuum die casting refers to a highpressure die casting process that uses extremely low-pressure vacuums, 10 kPa (100 mbar) or

Vacuum die casting of aluminum

Conventional Die Casting

Vacuum valve

Air venting is an important aspect of conventional high-pressure die casting. Back pressure in the die cavity directly affects fluid flow, which is controlled by differential pressure. The molten metal injected into the die must displace the gas initially within the die cavity or absorb it as compressed bubbles. The common belief that die castings cannot be heat treated is an indication that castings often contain entrained high-pressure gas pockets. Effective venting eliminates entrained gas, resulting in optimal metal properties and even heat treatable castings. The most common method of gas displacement is the use of thin channels, called vents, that are open to the outside of the die. The total area of vents should be approximately 20% of the gate area. The machined channels are no more than 0.18 to 0.38 mm (0.007 to 0.015 in.)

pump and accumulator are connected to the die casting die via a combination of hard and flexible piping. In some variations of the process, the vacuum is also used to suck the molten metal through a tube into the shot sleeve, and the shot hole is closed mechanically or by the advancing piston (Fig. 1). The theory behind vacuum high-pressure die casting is simple. If atmosphere in the shot system and die cavity is essentially eliminated, then there can be little or no air entrapped in the molten metal during the turbulent cavity fill. To that end, some vacuum systems (or the way in which they are employed) achieve only partial vacuum (40 to 80%) but still provide excellent positive venting and can greatly improve the fill and integrity of resulting parts. However, these partial vacuums may be unable to make parts that can consistently be solution heat treated without any blistering.

thick to ensure that the metal will freeze off before the edge of the impression block is reached. The turbulence of flow makes complete removal of cavity gases difficult by displacement venting. Vent placement is a matter of experience, with final design often done by trial and error on the casting machine.

Vacuum system valve opens, evacuating air from die cavity and shot sleeve prior to fill by plunger

Shot sleve

Fig. 1

Plunger

A vacuum high-pressure die casting system for casting aluminum. Source: Adapted from schematic courtesy of NADCA

Vacuum High-Pressure Die Casting / 733

Fig. 2

The light-colored components of this automobile front suspension system are made of high-vacuum die-cast components. Source: Courtesy of NADCA

less, in order to assure the production of heat treatable and weldable castings. This is less than 10% of normal atmospheric pressure. This process is more common in Europe than in North America and is particularly associated with the new manganese-bearing aluminum alloys, such as alloy 365.0 (UNS A03650), that offer good combinations of strength and ductility. The process has proven its effectiveness for the production of castings with extremely low gas and oxide content. The process permits the production of highly ductile heat treatable structural and chassis components, thin-walled weldable body parts, and parts with surface conditions suitable for coating. Some parts of the suspension system of an automobile (Fig. 2) were produced with this high-vacuum die casting process. Process Steps. In this process, the liquid metal is taken from an electrically heated

furnace and dispensed into the injection sleeve through a suction pipe by the differential pressure of the vacuum in the die and the injection sleeve. The metal volume is set in the injection sleeve by the length of the charging time. The die is then filled in the conventional way by the mechanical plunger. As a result of the metal filling process in a vacuum, the vacuum can act on the die significantly longer than in the conventional vacuum-assisted die casting process. This results in a lower absolute pressure in the die and the injection sleeve. Also, the metal is charged not only by vacuum above the metal surface but also from positive pressure below the metal surface, which reduces the turbulence and the gas entrained during the charging process. The goal of high-vacuum die casting is to virtually eliminate the atmosphere in the shot

system and die cavity and thus make parts that are sound, heat treatable, and have predictable and reliable mechanical properties. To accomplish this, the shot and die cavity system must be well sealed to avoid aspirating ambient air while under vacuum. Also, a tortuous path is required between the die cavity and the vacuum source to avoid molten metal invading the vacuum system when terminal pressure is achieved at the end of each shot. Special nonvolatile die lubricants and practices are required that do not create undesirable cavity gases during the casting cycle. The ejection part of the cycle is the same as for conventional die casting. Applications. Driven by the need to conserve fuel, the trend toward lighter-weight construction materials continues unabated in the international automotive and other transportation markets. Aluminum and magnesium alloys are increasing their share of the automotive market. It is normally considered that die-cast parts have a great deal of design flexibility, but high ductility is not usually a characteristic of die-cast parts. High-vacuum die casting permits the production of sophisticated shapes while producing parts that can be heat treated to obtain outstanding ductility characteristics. As a result of the remarkable structural properties achieved by this process, high-ductility and weldable car body panels can be manufactured with high process reliability. Another advantage of the process is that almost all aluminum alloys are castable, which opens up a wide spectrum of potential new applications for this technology. Another emerging market for this technology is the field of safety components based on the high energy absorption of the cast parts, which is approximately 50% higher than that of sheet metal.

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 747-758 DOI: 10.1361/asmhba0005292

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Automation in High-Pressure Die Casting Yeou-Li Chu, Ryobi Die Casting

HIGH-PRESSURE DIE CASTING (HPDC) is a fast (or perhaps the fastest) method for net shape manufacturing of parts from nonferrous alloys. It is ideal for the economical production of high-volume metal parts with moderately accurate dimensions. The automotive industry provides the largest market for die cast components (Ref 1). The rapid cycle time of die casting provides economical production with the capability for a high degree of automation. The process is adaptable to the use of metal-saving designs and can also replace an assembly of separate parts with a single casting. Usually, a minimum of machining or finishing is needed to turn a raw part into a finished component. One basic process limitation is that part geometry must allow removal from the die cavity. Die casting is generally limited to metals with low melting points. Aluminum is the most commonly used, followed by zinc, magnesium, copper, tin, and lead. Zinc, tin, and lead alloys are considered to be low-melting-point alloys, while aluminum and magnesium alloys are considered to be alloys with moderate melting points. Copper is considered to be a high-melting-point alloy, and there are some copper-base alloys for die casting. Suitable high-temperature die materials are critical for economical die casting of copper alloys. This article briefly reviews the automation technologies for the different stages or steps of the die casting process. The different levels of automation can be semiautomatic and fully automatic. Some manual aspects of the operations (together with automation options) are also mentioned. Many die casters also combine the finishing steps into the die casting production cell (Fig. 1). Detailed finishing steps may involve finish trimming, detailed deflashing, shot blast cleaning, and quality checks. Automation of the postcasting process is also discussed.

immersed in molten metal in a furnace attached to the machine. These machines are quick in operation and are normally used for lowermelting-point alloys, such as zinc, lead, and tin alloys, that will not erode the metal chamber, cylinders, or plungers. Some small-sized magnesium parts are also cast using the hot chamber machine (Ref 2). In the cold chamber machine, however, the construction of the injection mechanism is different. The plunger and cylinder are not submerged in molten metal. The molten metal is poured into a cold chamber, and the plunger rod is pushed forward hydraulically to fill the die cavity. The cold chamber machine is used for alloys such as aluminum and copper that have a high melting point or the capability to erode the plunger. Due to the high injection power, larger-sized magnesium parts are also produced by a cold chamber machine (Ref 2). Regardless of the type of machine used, all die casting machines have three main sections: the shot end, the clamping section, and the

Die Casting Cycle Two basic types of die casting machines are used for production: the hot chamber and the cold chamber. The hot chamber machine consists of an injection mechanism that is

Fig. 1

Die casting production cell

motor power section. They all have very similar cycle production steps. From a die casting production point of view, the detailed die casting production can have the following steps:        

Liquid metal pouring Injection Solidification Die open Part extraction Die lubrication Insert loading (if needed) Die close

Depending on the die caster’s practice, the die casting cycle can be started from any one of these steps. All of the steps listed can have different levels of automation. Again, depending on the requirement, the die caster may also perform additional operations, such as:  Finish trimming  Detailed deflashing

748 / High-Pressure Die Casting  Shot blast operation  Casting quality check (leak test, x-ray, ultra-

sonic inspection) Operators may perform all, none, or one or two of these steps in the finishing cell. High-Pressure Die Casting Automation. Automation in the HPDC application is an integrated technology concerned with the application of both complex hardware (robots, computers, and electronic controllers) and software (simulation, statistical process control, and management information software) to control various steps of production. Depending on the part complexity and the part finishing requirements, the die casting production involves three major steps: liquid metal preparation, die casting, and finishing. The robot, accompanied by computer software, can be a critical element for advanced HPDC development. The secret to HPDC success lies in the integration of all those automated entities to form a good production system so that it can achieve good-quality casting at a low cost and still be flexible. However, not all die casting production systems can integrate 100% automation. The extent of the automation depends on product volume, complexity, production cycle time, and product life. Additionally, the floor space and available capital must be considered. For instance, for small-part production with automatic ladle pouring and automatic spray cooling, the cycle will be increased due to all the equipment mechanical movement. Each die caster has different issues. Integrated automation systems can tie together the liquid metal pouring, injection shot control, die cooling system, casting extraction robot, and die spray application. In the finishing operation, the finishing cell can either tie with the casting process system or have an individual finishing operation. Independent finishing is more flexible, with the robot assisting in an integrated trimming, deburring, and inspection. The success of integrated automation depends on management and workforce involvement. Optimizing pieces of the automation equipment is easy, but it does not mean that the whole automation system is optimized. In conclusion, when considering using automation in the HPDC process, the following questions should be asked:    

Will Will Will Will

it boost production efficiency? it reduce scrap? it make the job simple? the cost-savings justify the investment?

If the answer is yes to these questions, then automation should be considered. Planning a good integrated die casting production system requires the gathering of process information and merging this information into the decisionmaking process. This can be done through the networking and awareness of different social, political, and business attitudes within a given production setting. Once the decision has been made to move forward, planning should follow some simple rules:

 Plan to use the least amount of equipment

and space to achieve the process.  Plan for flexibility to grow and handle different parts.  Reduce the number of process steps, if possible.  Plan to streamline the part flow. Of course, different HPDC applications will dictate different levels of sophistication and cost. A few of the bigger issues in HPDC production are:  Production









efficiency improvement: In today’s (2008) HPDC production, it is not uncommon to pour more than 23 kg (50 lb) in a large die casting production, such as engine blocks or transmission casings. This intense physical labor can tire a human body in a short amount of time. However, a robot can handle loads of over 45 kg (100 lb) without tiring or needing breaks. Process consistency and repeatability improvement: The die casting process is very dynamic. More than 20 factors will affect the process simultaneously. For a human to monitor and control all of these factors is beyond the normal operator’s capability. Computer monitoring and control systems can monitor more than 20 channels of information and feed this information back to the machine to perform automatic adjustments. Cost reduction: The effective use of automatic equipment can generate enough savings in parts of the production methods to offset the initial investment. According to the industry, the payback for an automatic spray application can be in a 6 month to 1 year period. The cycle-time reduction, lubricant uniformity, and reduction and consistency of the quality will be improved (Ref 3). Product quality improvement: Porosity defect is the biggest issue in die casting operation. Porosity comes from entrapped air and shrinkage. However, with human eyes, there is no way to detect this type of defect. Fortunately, with x-rays and automatic defect-recognition software, an in-line mass-production quality check can be done. Safety improvement: Die casting injuries are ranked very high among the manufacturing industry. If an operator misses a minor detail, he can be burned from hot liquid metal. The movement of a machine sleeve or die can cause a serious injury to the operator, if safety procedures are not observed or are forgotten. Automated equipment can replace an HPDC employee working in dangerous and/or hazardous conditions, such as the automatic pouring device.

of the liquid metal. Automation for this area has evolved from the earlier manual hand-pour ladle to the robotic-pour ladle system and the automatic dosing pour system. The consistency in using automated systems to replace hand pouring is a big advantage, especially for productions where a large amount of pouring is necessary. A number of automatic pouring systems are available in the HPDC application; basically, they can be divided into two different categories: robotic ladling and dosing-type feeding systems. Robotic Ladling System. This may be the most popular automatic pouring system in the die casting operation so far. The current robotic ladling can have simple two axis chain motion control system to complicated six axis servo control motion and multiple function system. The advantages of this furnace are:  Operation is easy.  Pouring size can be adjusted easily.  Troubleshooting and maintenance are easy.

The die casting industry uses an automatic pouring ladle with robot-like movements that is controlled by a programmable controller. The cycle begins at the holding furnace. The ladle dips precise amounts of molten aluminum from a dipwell, carries it to a position directly over the shot sleeve, pours it quickly into the sleeve, and then injects the liquid metal into the mold (Fig. 2). A simple adjustment on the controller determines the amount of liquid metal dipped by the ladle. A sensor detects the level of the liquid-metal surface to ensure accurate shot repeatability throughout the metal-level height range. For the ladling type of pouring, the holding furnace must have an open top for the ladle to access. The temperature variation and the energy usage are higher. Additionally, for the open-top holding furnace, oxide and inclusion generation is greater. Many die casting machine suppliers and independent automation companies provide this type of pouring system. Dosing Feeding System. A dosing type of furnace is basically a closed chamber with a pressure-tight seal-assisted furnace. This type of furnace can have different feeding methods. The advantages of this type of furnace are:  Temperature consistency: The liquid metal

Automatic Pouring

is stored in a closed chamber, so there is less temperature variation.  Energy consumption is lower: Compared with the traditional open-top holding furnace, the energy consumption can be 40 to 50% less.  Metal quality is better: Since it is a fully closed chamber, there is less formation of oxidation and inclusions than in the opentop traditional furnace.  Metal quantity control is better: According to all of the equipment suppliers, the dosing furnace can maintain +/ 2% or better accuracy than the set-up requirement.

The pouring control system for any HPDC operation is critical for achieving maximum yield per run and also maintaining good quality

There are three common types of dosing feeding furnaces on the market: pressure pushing, vacuum suction, and motor pumping.

Automation in High-Pressure Die Casting / 749

Fig. 2

Fig. 3

Two axis motion robotic ladle pouring system

A suction type of dosing, such as the Vacural concept, is shown in Fig. 3. The metal is metered from a holding furnace, located between the injection unit and the fixed platen. The liquid metal is sucked to the shot sleeve through vacuum pressure. The vacuum operation not only provides sleeve feeding but also assists the injection of liquid metal inside the cavity. The vacuum reduces the air inside the sleeve and cavity; therefore, a high-quality, less porous part can be produced. A pressure pushing type of dosing furnace is shown in Fig. 4. The concept is similar to a low-pressure casting process approach. The pressure is applied to the whole furnace chamber liquid surface. According to the pressure regulation, it controls the filling amount. When the preset filling quantity is reached, the dosing procedure is terminated by reducing the pressure. This type of dosing furnace can even install porous plugs in the bottom of the furnace and conduct the degassing operation (see the article “Transfer and Treatment of Molten Metal—An Introduction” in this Volume). A third type of automatic dosing holding furnace is the feeding pump system (Fig. 5). A feeding pump is widely used to transfer the liquid metal; it is fast and precise, especially for a large melting service. The liquid metal feeds into a small control chamber through vacuum or other processes. The liquid metal is then pumped out from this small chamber to a feeding channel (i.e., launder) and charged into a shot sleeve. This type of operation can achieve very accurate pouring, on the order of +/ 1%. The dosing-type furnaces are very popular in the European die casting industry, since the greatest advantage of this type of furnace is the energy savings. With a closed-chamber controlled environment, the energy loss is fairly

Vacuum suction tube dosing pouring system

small. Generally speaking, in European countries, energy is more expensive than in the United States. However, due to energy costs, some U.S. die casters have also started to use this type of furnace. Another reason for the die caster to consider this furnace is the metal-quality advantage. The die casting part has been advanced from simple housing-and-cover-type castings toward structured or strength-needed products. In order to meet these demands, metal quality is very important. Fewer inclusions and oxides, which improves metal quality, also cause the die caster to consider using this type of furnace.

Injection Process The most important step for die casting is liquid metal injection, which is known as a shot. The injection phase has the most influence on casting quality once the die and machine are fixed. Therefore, it is very important to have a consistent and repeatable shot control system. Automation can reduce human error and process variability. During the past 20 years, die casting process monitoring systems and closed-loop shot control systems have been developed to measure and control the injection phase and other process variables. For die casting to become a fully automated process, the die casting machine and its control system must be able to sense, measure, and control the variables that occur in both the machinery and the process. In past years, many die casting machine suppliers and individual monitoring and control system development companies have dedicated efforts in this area (Ref 4–9). The most important variables during injection are the velocity and position of the shot cylinder, the hydraulic pressure of the shot system, and

the temperatures of the die and molten metal. Encoders are the best type of velocity sensor for measuring this high-speed process. Specifically, those encoders are directly embedded in the hydraulic cylinder itself. The advantages of this approach are that no new moving parts are added, and the direct sensing from the rod that is being controlled is the most accurate. For pressure monitoring, hydraulic pressure transducers are easily installed. The result is a perfectly coordinated profile of velocity versus position and pressure versus position. Due to the high-speed nature of this event, a special-purpose and dedicated highspeed data acquisition board is employed to gather the information. Once the event has been captured, the data are moved to a computer, where the various process parameters can be further calculated and stored in a database. Die Casting Shot Process Monitoring System. By measuring the plunger stroke and the hydraulic pressure of the injection units, it is possible to plot the injection profile, more commonly known as a shot profile. Based on the shot profile, further parameters can be derived through computer calculation. Examples of important process parameters include biscuit size, fill time, filling velocity, intensification pressure, and cycle time. Injection profiles have been plotted in two basic ways. A time-based injection profile was more commonly employed in the early days of die casting shot monitoring. The time-based injection profile plotted a time scale on the x-axis. On the y-axes were plotted a position versus time curve, velocity versus time curve, and pressure versus time curve(s). This sampling technique provided adequate data during the slow-shot phase and during intensification but was a poor method of gathering velocity

750 / High-Pressure Die Casting

Fig. 4

Air-pressure-type dosing pouring system

Fig. 5

Pumping-type dosing pouring system

data during the fast-shot phase, where the casting is actually being made. The more intuitive injection profile plots velocity and pressures versus position during the filling phase—a so-called position-based injection profile. This position-based plot provides an easily understood graph of the filling phase. The result from a position-based injection profile could easily employ an overlay technique to display shot-to-shot variation with great clarity. A further refinement to the position-based injection profile—a dynamic switchover at the point of impact to a time-sampling mode—took advantage of the greater resolution of the timesampling technique when the plunger is moving very slowly (as it does during intensification). A second and connected graph that plots plunger displacement and pressure of the

injection cylinder versus time provides the best visual image of the intensification phase. This hybrid plot has become the standard method for plotting a shot profile, which is the term for an entire die casting injection event (Ref 8, 10). In-Line Automatic Sorting and Monitoring. It is not uncommon in die casting production to pay a large sorting cost to identify defective castings. This can happen after shipment of the parts to the customer, when value-added operations, such as machining, plating, or painting, have already been done. The die casting process monitoring system can be successfully deployed to ensure most of the product quality issues prior to any further processing of the casting. Measuring process parameters and comparing those parameters to predetermined ranges provides a method to accept or reject

castings as they are made. In addition, if process limits are exceeded, signals can be generated to automatically sort the casting, discard it, or set it aside for further evaluation. Using a computer and specialized software, shot profile data can be used to calculate and derive other measurements that are of interest to the process engineer, including gate velocity of the metal, fill time, metal pressure, biscuit length, intensification pressure, intensification rise time, and intensification squeeze distance, to name just a few. This calculated or derived information can also provide the first-line automatic quality check. In addition, the shot profile and the process parameter data can be used to diagnose problems with the process. These, in turn, will guide an experienced process engineer to consider the factors that could negatively influence performance (Ref 11). Some common process anomalies that can be detected easily by using the process monitoring system include insufficient biscuit length (could indicate metal pour problems or flashing of the die), plunger tip drag, intensification (too slowly, too late, or not at all), and poor fill (may indicate a cold die, cold metal, or insufficient filling velocity). These are just some examples. There are many such parameters, and high and low limits can be easily set to provide signals that summon attention or to sort parts automatically. Off-Line Automatic Statistical Process Control Analysis and Monitoring. The process parameter data are easily stored in a computer database. This information can be used for statistical evaluation, which can be done on-line in real-time and also stored for further analysis. Trend charting of process parameters can be a useful tool for quickly spotting deterioration in important variables. Statistical analyses, such as X-bar/R charting, can be used to measure machine capability. This is fundamental information that the process engineer uses to determine if a machine has the capability to run a particular part. X-bar charting can also be used to verify that a process is statistically repeatable and if important process variables can be maintained within a processing window that is required to make a particular casting. Statistical analysis can also be used to determine the upper and lower limits for the critical process parameters. Finally, statistical information can be useful for establishing routine die casting machine maintenance intervals. Continuous Process Monitoring and Control. Process monitoring provides the foundation for process control. It is far more important that once critical parameters are identified and understood, they must be controlled. To control the process, monitoring must be done continuously and have the capability to automatically set signals to indicate off-standard performance. Only in this way will quality demands be met. The technology to implement process control is readily commercially available, and due to advancements in computer technology, the cost of this equipment is very

Automation in High-Pressure Die Casting / 751 low, especially when considered against the cost of the equipment in the die casting cell and the costs associated with producing scrap (Ref 12). Real-Time, Programmable, Dynamically Adjustable Shot Control Systems. The traditional die casting machine is capable of a twostep or three-step shot profile. In the modern commercial die casting environment, simple two-step and three-step capability may not be enough to make complex part geometries. The traditional die casting shot control does not have acceleration-control capability. The traditional shot system has two cartridge valves and a hand wheel to adjust for flow rate (and therefore speed control); the sequencing of these valves is limit-switch controlled. Modern die casting machine shot control systems feature programmable, all-digital velocity control with integrated process-monitoring systems. Special two-way servo-piloted throttling valves have been designed and integrated with the shot control electronics. The throttling valve can be on either the meter-in or meter-out side of the shot cylinder. It is even possible to retrofit such controls onto many existing machine designs. These electronically controlled shot systems are capable of running in fully closed loop mode. The velocity profile is programmed in engineering units. The control then attempts to execute the programmed event. It generates an electronic command to the servo pilot valve, which modulates the throttling valve. The throttling valve has a linear variable differential transformer that reports valve gate position back to the controller, which compares the programmed event to the actual. Any difference results in a corrective signal from the controller to the valve. This is the inner control loop (Ref 11, 12). The control, once again, has an event that it tries to execute—the programmed event. The control generates an electronic command to the throttling valve in order to execute the velocity profile. As the event occurs, the electronic control monitors the actual position and velocity of the shot cylinder to ensure that it is on path. Any difference between the programmed event and the actual event results in a corrective output signal. This represents the outer control loop. The shot control system compares in realtime the programmed position of the valve versus the actual position of the valve and the programmed position/velocity of the shot cylinder versus the actual position/velocity. It dynamically adjusts for both (continuously) as the event occurs. Total Integrated Die Casting Information, Monitoring, and Control System. The modern die casting manufacturing cell consists of a number of auxiliary programmable devices that provide integrated sequential tasks in the cyclic operations of the die casting machine. These include automatic ladles, automatic extractors, and automatic sprayers. Robotics are commonly used for these applications.

Typically, each device has its own programming terminal. The current state-of-the-art is to have a computer-integrated system select a recipe from each device programmable controller. Communication links between the programmable logic controller, other auxiliary programmable devices, and the computer-based information system can be used to facilitate diagnostics and machine and equipment maintenance. A networking or integrated die casting information and control system structure can function as follows (Ref 12):  Level 1: The system begins with the digital

position/velocity and pressure sensors to provide an accurate, reliable, and responsive signal of the shot-end motion. The system provides for almost any other sensor input, so that accurate and continuous monitoring of other important casting parameters, including temperature, stress, and flow rate, can be achieved.  Level 2: Data are acquired with a high-speed microcomputer-based system integrated with a personal computer that displays the results to the operator or process engineer in easily understood formats. Signal outputs provided by this data-acquisition and monitoring system can be used to implement a first-line casting quality check or a closed-loop control for temperature or cooling water flow rate.  Level 3: The system provides funneling information from machines on the floor to a host supervisory workstation computer. Multiple monitors located at various locations throughout the plant can also receive this information and perform numerous tasks, including shot profile monitoring, detailed off-line analysis, statistical process control analysis, scrap reporting, and downtime reporting.  Level 4: Plant operations information is funneled into a plant- or corporate-wide computer system, permitting data archiving on a corporate-wide computer system. Future analysis and report generation is implemented. Ethernet can provide more direct and longdistance data transfer and review. The computer monitoring and control system may have different structures, but the essential concepts are still the same. A die casting facility can start with the sensor and a stand-alone computer system, on the shop floor, that are capable of monitoring the machines on an individual basis, then build to an integrated plant- or even world-wide system.

Solidification and Heat Removal The die casting process is basically a heattransfer process. The hot liquid metal is injected into a metal cavity die. Through a heat-transfer process, the heat of the liquid metal is removed through the die itself and through the internal cooling line inside the

die. When the part solidifies to have enough strength, it is ejected from the metal die (Ref 10, 13). The part will continuously cool in the air or through a quench medium, such as water, for further operation. The die surface is then cooled further by spray water or lubricant. A number of heat flow paths are available for the die casting process (see the article “Die Casting Tooling” in this Volume). The thermal behavior or the temperature of a die plays an important role in die casting quality. Many die casting defects are related to the thermal management of the die. For instance, a hot die surface in conjunction with a heavy casting wall thickness may create a shrinkage defect. A cold die surface in conjunction with a thin wall may create cold shut or lack-of-fill issues. Automation in the solidification step can have two aspects: internal heat management (solidification stage) and external heat management (die lubrication stage). This section focuses first on internal heat management automation. External heat management, that is, spray-related cooling, is discussed later. Internal heat management automation does not use heavy equipment, such as a robot, but is a more computer-aided engineering application. Die Internal Heat-Removal Management. Die internal heat removal involves two issues:  Design of the die internal cooling line  Control of the die cooling line

Each of these two areas has different degrees of automation. In terms of design, computer-aided tools have advanced the design process for the die internal cooling line. In earlier days, cooling line design was based on experience and trial and error. Quite often, the design had more cooling lines than was necessary. If a die runs too cold, the operator can always close some redundant cooling lines. Die casters also do not want to receive a big penalty for running a hot die, because it is very time-consuming and costly to add cooling lines after a die is completed. In the 1980s the computer thermal and flow management software for die casting was introduced on the market. The ability to accurately simulate the effect of cooling lines in the die provided a valuable analytical capability. Cooling line location and size are critical but can be difficult to design. With the computer simulation program, one can quickly add, remove, or move cooling lines and perform a new simulation (Ref 14). With computer simulation results, trial-and-error efforts can be reduced dramatically. For die internal heat management, the die caster can use the thermal profile of both die and part to further define the die open time, spray focus, cooling line flow rate, and so on. Computer-aided engineering improves not only die cooling design work but also process optimization (Ref 15). Control of the internal cooling line is the second important part of heat-removal management. Even with the sound engineering design of a cooling line, the result may not be optimal

752 / High-Pressure Die Casting without good practice or operation. In addition to casting quality, die life is another issue that the die caster must consider. The durability of the die is determined by the quality of the die steel (including heat treatment), the cooling line, and the operating conditions. A good cooling line design has been introduced; it is a more advanced engineering concept. The internal cooling line control can be divided into three different levels: manual, semiautomatic, and fully automatic. Manual-mode operation is purely dependent on the die casting operator’s experience and habit. Different operators may have different understandings, so the setup and operation are different. It is quite common for die casting operators to forget to turn on the cooling line, and a shrinkage defect or a part sticking in the die may occur. When the operator finds the defect and rushes to turn on the cooling line to compensate the hot die, it may create a die gross-cracking (thermal shock) issue. Therefore, the manual die cooling line control must be very carefully and routinely checked. In semiautomatic control mode, the cooling operation is based on a preset programmable logic controller device to control the cooling water on/off and some other operations, such as air purge. One of the largest Japanese die casting companies developed a jet-cool system for intricate die areas, such as small-diameter core pins or small inserts (Fig. 6). The jet-cool system has the ability to program the cooling time and cooling medium, such as air or water. According to the preset cooling time and air purge time, the system performs an automatic operation of cooling water discharge and air purge in response to the start signal from the die casting machine. After discharge of the cooling water, the water remaining in the cooling tubes and core pins is removed by air (air purge) to prevent overcooling. Depending on the cooling requirement, this system can provide various water and air cooling combinations with different pressure levels

Fig. 6

Schematic of a semiautomatic cooling control jet-cool system

and time intervals. An example profile of a semiautomatic setup is shown in Fig. 7. For manual die cooling control, when the cooling line is turned on, it will always stay on until the operator shuts the cooling line off. Compared with the manual die cooling control, the semiautomatic die cooling control has been advanced from a simple on/off to a scheduled on/off and can even select different cooling media. This is more appropriate for the die casting operation. Undercooled and overcooled die situations can be reduced through this semiautomatic programming cooling device application. A simple automatic cooling line control is a thermocouple plus an on/off solenoid valve controller. The electrical signal from the die thermocouple is sent to an indicating controller, which indicates the temperature and has an adjustable set-point. When the die temperature reaches the setting, the controller turns on a solenoid valve to allow cooling water to flow through the die. Since the on/off setup may repeat, the manual cooling control has a die gross-cracking issue; therefore, a modified automatic cooling control concept is available. The modified automatic cooling line control adds a parallel bypass line. This bypass line always has a small amount of water flowing through it. Since water flow always finds the easiest route, once the solenoid valve in the main cooling line is open, the water will flow through this main line at a high flow rate. This approach compensates the drawback of the on/off control (Ref 10). The fundamental purpose of die temperature control is to ensure that the production is in a consistent state. Most die casters understand that if the die casting runs in a consistent state, the production is smooth and the efficiency is better. In semiautomatic and simple automatic cooling thermal management, the die caster can set up the desired cooling program. However, in the die casting operation, many factors are not that easy to control. For instance, the ambient temperature and the moisture level on a hot summer day and a cold winter day are

Fig. 7

Example of a semiautomatic cooling system setup

different. No one can exactly predict or assess the impact. If the die caster can do something to compensate those uncontrollable factors, it is always helpful. In the advanced, fully automatic control die cooling operation mode, the cooling system is based on the real production die temperature and adjusts the cooling flow rate to ensure the die cooling is at an expected level. The fully automatic die cooling control system has been available for a few years. Due to cost or some other issues, the fully automatic die cooling control system has not been very popular. One of the largest die casting die construction companies in North America has invented a fully automatic die cooling control system (Ref 13). Some research institutes also published a closed-loop study about the fully automatic die cooling control system (Ref 16). The system consists of temperature sensors, data acquisition, input/output interface, pump, and programming. The input is from real die temperature-sensor thermocouples. There is also a predetermined die temperature setup inside the computer. The setup and real die temperature difference will trigger a proportional-integral-derivative control program and output the signal to adjust the cooling line water flow. The predetermined setup temperature normally can be based on either experience or computer simulation. For a large transmission die, it is quite typical to have 80 to 100 cooling lines. In reality, it is impossible for every cooling line to have one of the closed-loop control system setups. For limited closed-loop temperature-control devices, the issue is which die temperature will be used to represent the real die temperature and compare with the setup temperature to trigger the output control. While die casters can install many thermocouples, it is not feasible due to space limitations and maintenance difficulties. This issue is the same as in the simple automatic cooling control concept—which thermocouple data will be used to represent the true die temperature?

Automation in High-Pressure Die Casting / 753 Temperature zone control is a more feasible approach. In practice, different areas of the die have different temperatures during operation. For instance, the biscuit area is always hotter due to heavy casting mass. The zone idea is that the die can be divided into different sectors. Each sector or zone has a unique temperature requirement and temperature-response characteristic. Zoned temperature control allows control of the solidification temperature and time in various parts of the cavity (Ref 10, 13, 17). For control purposes, the zone is defined as a region whose temperature is affected by a group of waterlines. This group of cooling lines can be considered as a single cooling unit, and their flow can be regulated by a single valve with temperature feedback from a single thermocouple. Each zone is considered to be independent (Fig. 8). For example, one entire slide or an insert in the die can be considered as a single zone. From operation experience or computer calculations, die casters can bundle and treat a few temperature profile areas as a single zone. This approach significantly reduces the total number of control areas and installations for the fully automatic cooling control system (Ref 13).

Fig. 8

Automatic die cooling control unit and zone concept. TC, thermocouple

Fig. 9

Robot extracting a high-pressure die casting engine block. Courtesy of the Buhler Group

Casting Extraction and Insert Loading One of the most mature automation technologies is the use of robots to handle material movement (Ref 18, 19). Robots for extracting or inserting loads are very common in HPDC applications, especially when die casting heavy or large components (Fig. 9). Casting extraction from an HPDC operation can be automated using either a traditional extractor or a robot. There are two immediate benefits:  Eliminating the need for the operator to reach

between the dies. This removes the operator from a potentially dangerous and ergonomically challenging situation. Reaching between the dies poses the unlikely risk that the dies will close while the operator is reaching between them, and also the more prevalent risk of the operator overreaching, causing back and shoulder injuries, and slipping and falling due to the die casting environment.  Keeping the production cycle time consistent. Due to the production demand, especially large castings cannot be cooled to a low enough temperature for the operator to handle. The normal temperature for a robot gripper to handle is in the 260 to 315  C (500 to 600  F) range. In this temperature range, the castings would be too hot for humans to handle efficiently and safely. With a robotic operation extraction, the cycle time has no interruption, and the possibility of human injury from the casting temperature and weight is removed.

Some key design elements to consider when automating the extraction process include:  Extra processes in the die casting cell: Will

machine? The robot gripper is much easier to design if it grabs the biscuit. The gripper may not need to be changed, even if the cell produces different parts.

the casting need to be quenched (air or water) and trimmed within the die casting cell? If so, a robot is probably a good solution due to it flexibility.  Gripper design: Can the casting be gripped by the biscuit, or must it be gripped by the part to be extracted cleanly from the die cast

Insert Loading. The loading of inserts into an HPDC die can usually be done robotically (Fig. 10). The main benefit is the same as mentioned previously: It eliminates the need for the operator to reach between the dies. Normally, the inserts need to have a tight tolerance when

754 / High-Pressure Die Casting grips from the side of the biscuit, whereas a three-jaw gripper grips from the back of the biscuit (on center).  Will the runner system be removed in the casting cell using a saw or break-off device? If so, then gripping by the features on the casting is generally the desired approach.

Automatic Spray

Fig. 10

Robot inserting cast iron component for high-pressure die casting engine block production

placed into the die. A large die casting production is beyond the operator’s reaching capability. Before the die closes, the temperature may still be in the 150 to 230  C (300 to 450  F) range. This will prevent the operators from performing an efficient operation. If multiple inserts need to be loaded, the cycle time becomes a big variable. Furthermore, when a robotic insert-loading process is added to a die casting machine that is already otherwise fully automated (ladle, die spray, and extraction), consistent cycle times result. This allows for more accurate production planning. Some key design elements to consider when automating the insert-loading process include:

get the insert(s) and load it into the die? Typically, the die spray process happens immediately after the casting is extracted by the robot; this gives the robot time to set the casting down on a conveyor and pick up the inserts to be ready to load them, without increasing cycle time.  Will the casting be cooled and trimmed in the casting cell? If so, then a separate robot for insert loading may be required.  How many inserts are to be loaded in the die? It may not be possible to combine extraction and insert-loading grippers on the same robot end-of-arm tool due to space constraints.

 Are there “lead-in” features in the die/insert

Gripper Design. Depending on the requirement, the gripper design will affect the robot or extractor operation flexibility. Some of the following should be considered by the die caster:

that allow the insert to be easily placed in the die? If the inserts are difficult to load manually, then it may not be possible robotically.  Are the manufacturing tolerances of the insert controlled closely enough to ensure that loading into the die is consistent?  How will the inserts be presented to the robot? This can be done in several ways.  Do the inserts have to be preheated before putting into the die? Will the robot load/ unload this heater, or will the inserts already be heated when presented to the robot? Combination Insert Loading and Extraction. In many cases, the same robot can be used to place the inserts into the die as well as to extract the casting. In addition to the previously listed considerations, combining these processes into one robot also raises the following design considerations:  Is there enough cycle time to allow the robot

to extract the casting and set it down, then

 Will the casting be quenched (using air or

water) in the die casting cell? Will the casting be loaded into a trim press after quenching? If so, it is usually better to grip the casting by the biscuit, using either a twojaw or a three-jaw gripper.  Gripping by the biscuit allows the orientation of the gripping position to be changed during the exchange in a cooling rack or even a quench tank. This potential to reorient the gripping position allows more flexibility in the cell layout and, in some cases, is the only thing that makes loading a trim press possible.  The choice to use a two-jaw or three-jaw gripper depends on the clearances in the casting machine and the downstream quench and trim processes. How much room is there for the gripper? Generally, a two-jaw gripper

Spray cooling of the die is not only done for the thermal benefits. The spray contains a lubricant that is applied on the die surface to help the flow of aluminum in the cavity and to prevent soldering of the alloy with the die. When the casting solidifies, the element contents of the lubricant will help to release the casting from the die. Therefore, the main functions of die spray can be summarized as release of the casting, lubrication of moving parts, protection of the die surface, and external thermal management of the die (Ref 20). The principle of all types of spraying is essentially the same. The differences lie in the equipment used to obtain the desired result. Past studies have shown that 80% of the die release used for cooling runs off the die and is wasted (Ref 21). For a normal die casting operation, the die lubrication step may occupy approximately 30% of the whole cycle time (Ref 3). How to boost the productivity and reduce the cycle time is always one of the top issues in the die casting industry. For a large powertrain component production, such as a transmission case, the entire die casting production cell will be over three million dollars; therefore, the cycle time reduction to boost higher production is extremely important. For a small, multiple-slides die, a uniform and consistent spray is important to ensure good casting quality. Furthermore, due to the size of the die, a successful manual spray is not feasible. During the spray, the noise and mist lubricant are also not healthy for humans. There is no doubt that automatic die lubricant equipment provides effective process improvements in terms of:    

Reduced spray cycle Reduced die lubricant use Improved casting quality Environmental friendliness

Lubricant application has advanced, with many options compared to the traditional manual spray gun. Approaches range from simple reciprocating spray to six-axis robots, with many variations of spray head design and spray application available with multiple nozzles or tubetype manifolds and multiaxis robots. Some of the sprayer suppliers even provide the extractor-and-sprayer combination version. However, this combination version can only be used in a small and simple part production. As the parts become more complex and larger, with the robot to handle both the casting and spraying

Automation in High-Pressure Die Casting / 755 the casting may get in the way of allowing a proper spray operation (Ref 21). All automatic sprayers are equipped with programmable controller electric panels, and many start-stop, spray, or blow cycles can be inserted into the program to ensure proper lubricant coverage of the most intricate dies (Ref 3). There are many options in automating spray application and the integration of technologies. In addition to a robotic sprayer or reciprocator, three other key developments have also been applied in this area:  Spray head design  Spray nozzle technologies  Spray strategy

Spray Head Design. Based on various concepts, the spray head design can be handled differently. One concept is to have fewer spray nozzle assembly heads and, depending on the robot movement, to spray different die areas. With this concept, the die spray cycle time will be longer. The space for the spray head assembly is smaller, and movement is more flexible. The spray head movement can be reprogrammed to fit different products. This concept is applied to small-die production. Another concept is to use multiple nozzles to match the casting configuration. By using an actual casting or die drawing the spray head can be designed quickly with a computer-based program (Ref 22). The final design matches the die contour and covers more areas with multiple spray nozzles. It can be more efficient and precise to spray the target areas. The robot movement for this concept is limited due to the size of the spray assembly. Since the multiple nozzles can work simultaneously and cover large areas, the spray time is decreased. This type of concept is for a dedicated product, especially large brackets or panel production. In reality, one die casting machine may be used to produce various parts. A flexible modular design for a spray head is very practical and handy. The spray nozzle heads can be assembled or disassembled like a fitted building block, and the nozzle itself can be reused (Fig. 11). This modular spray head design concept, in conjunction with the programming capability, enables the die caster to quickly set up and adjust for new production. Spray Nozzle Technologies Development. When a fine droplet of spray hits a hot surface, it is easily vaporized. From a physics point of view, it absorbs more heat from the liquid phase and changes to the vapor phase. In other words, if the spray lubricant drains off of the die surface, the spray is not efficient. In the conventional spray technique, it is hard to control the size, distribution, and timing of spray droplets. From years of experience, the die caster recognized the importance of the spray. It is why die casters now emphasize the need for a good atomized spray (Fig. 12). Newly developed nozzles have a direct air path and less turn; therefore, the velocity of the flow stream

is faster, stronger, and more effective in removing the heat. Depending on the application purpose, there are several types of spray nozzles, such as the external mixing spray nozzle, internal mixing spray nozzle, needle-type shutoff nozzle, and spool-type shutoff nozzle. The most common and advanced type is the external mixing spray nozzle. This type (Fig. 12) directs air into the liquid after it has left the nozzle, so this nozzle has excellent spray atomization (Ref 13). Spray Strategy. The spray timing directly impacts the cycle time, and the spray volume directly impacts the economics. The die casting design prefers to have uniform wall thickness. However, in reality, it is hard to achieve. The complicated die casting will face issues of deep pocket versus shallow pocket or heavy wall versus thin wall. The heat content in various die areas is different. It is challenging to reduce the temperature variation in different die surface areas. The traditional approach is to adjust the spray head direction and target the hot spot. The spray lubricant volume is normally driven by the hot spot temperature. A smart spray strategy is to provide enough lubricant to where it is necessary by not only controlling the spray direction but also the spray timing and spray volume. To achieve this, independent zone control ideas are needed. Zone control capability allows a setup for different spray volume and spray time for various areas. The computer program can direct separate nozzles to turn on or shut off. Normally, there are two to eight different spray circuits to be used, depending on the size and complexity of the part. The relationship between quality and operational reliability strongly impacts die casting economics. Problems attributed to die lubricant and spray obviously affect both of them. When

Fig. 12

part extraction and lubricant spraying are performed manually, more variation occurs in the cycling of die temperatures. This was demonstrated with an infrared temperature profile of a die surface from a die casting cell where part extraction and lubricant spraying were performed manually (Ref 23). In contrast, the temperature profile of a fully automated die casting cycle had minimal variation in cycle time and die temperatures (Ref 23). For defect casting troubleshooting purposes, it will be very hard for a die caster to evaluate if the problem comes from a manual lubrication approach, since the manual spray drives both the cycle time and the die surface temperature difference.

Finishing Automation A die casting die construction is composed of various portions: the biscuit, runner, gate, cavity, and overflow. The useable portion is known

Fig. 11

Robotic sprayer with good atomized spray head operation

Spray head with modular design and quickdisconnect valve plate

756 / High-Pressure Die Casting as the cavity; all other portions need to be removed and remelted. Additionally, when the die ages and becomes old, the die surface will begin to have small cracks (i.e., heat checking). This will affect the casting appearance. When a die does not close tightly, flash in the die parting line area will generate. Most of the biscuit, gate system, and flash can either be removed through a rough trim operation or through an operator manual knockout. The fine flash or heavy gating stub normally needs to be handled separately. A small heat-checking problem could need a shot blast or surface polish. Robots have been used for die casting extraction, insert loading, molten metal delivery, and spray for many years. Extending robotic technology to serve in the finishing area is relatively new. This is where robots will be proven to be indispensable. During finish die casting operations, robot automation should be reviewed to determine what parts of the processes will be integrated (Ref 24). There are many examples in the postcasting finishing cell to use the robotic integration concept. For instance, the robot can be used in conjunction with the trim die to remove any flash perpendicular to the line of die draw or punch draw. The robot can be used to work with a deburring cutter or sanding wheel for parting line flash or contours that cannot be reached by the trim die. No matter what kind of finishing operation, the robot is the key. Depending on the requirements, the current automatic finishing cell can be designed to provide the following robotic functions:  Trimming by a robot with a trim press  Deflashing and deburring by a robot with

cutting or sanding tools

 Removing shot blast residue with a robot  Robot integrated with x-ray or ultrasonic

equipment for quality checking The trimming or shot blast residual removal is a simple loading/unloading operation. It is similar to part extraction from the die, which was introduced earlier. This section discusses more about deburring and quality checking. Automatic Flash Removal. After a robot has been programmed with a path to remove gating and parting line flash, it will be so precise that several operations performed manually can be eliminated. Although the robot can follow the programmed path well, due to the die life and different cooling situations, the casting may have dimension or flash differences. This could create trouble for the robot cutting application. An off-line programming package has been developed to compensate for the casting profile variation. By using a casting model, the program generates points to indicate the exact casting profile. This will assist the robot in making adjustments (Ref 25). With the ability to remove flash or gate stub to a consistently close tolerance, robotic automation often reduces machining requirements. The robotic flash removal also has a big impact on the

operator. The working environment to remove the fine flash or gate stub is normally noisy and dusty. Without care, it may easily hurt the operator. For a large, heavy casting flash removal, this is a challenge. Some of the component assembly work must be done by hand. If the casting flash removal is not consistent or good, it will become more difficult for the following process operator. The robotic application eliminates repetitive-motion injuries to the hands and arms caused by the constant use of hand tools. Robotic application in finishing also optimizes the abrasive life of belts and bonded wheels. Due to the advanced controls and tight tolerance of the robot, it performs repetitive grinding with consistent pressure against the belt. This provides a high-quality product with the least amount of wear on the tools (Ref 24). The automatic finishing cell layout has improved in recent years. The earlier layout had more robots involved. There was a robot to hold the part and put it in the trim press or stationary cutting fixture. Another robot with cutting tool bits or a sanding wheel followed the programmed tool path to perform the finishing job. When changing the different types of parts, the fixture needed to be changed. This type of setup required more robots. The cycle time was also longer due to loading and unloading of the part from the fixture. The newer design layout has a robot grabbing the part and moving toward both the trimming and fixed cutting or deburring stations. This type of layout has less investment, less space, is more flexible, and is faster. For instance, a newer-designed robot gripper can hold four different types of parts with no need to change the gripper. The trimming and cutting can be programmed simultaneously. The robot utilization is more efficient. Automatic Finishing Integrates with Vision System. Vision cameras check the part shape and key features (Fig. 13). Advances in productivity have indicated there is a wide range of applications that may use the vision system during the finishing of parts (Ref 24). In the engineering world, the vision system can provide the following functions (Ref 27):      

Identification of shapes Measurement of distances and ranges Gaging of sizes and dimensions Determining orientation of parts Quantifying motion Detecting surface shading

Robots for handling parts are also integrated with vision systems for saving labor and space (Ref 26). In the old way of handling the finished work, each product line may have a separate setup. With the new automatic setup, the vision system can play a job-dispatching function. All of the castings can be set in the same areas. The robot can access different programs to perform different finishing requirements. Less labor, less equipment, and less space are obvious benefits.

Fig. 13

Vision camera view with measuring coordinates of some key features. Source: Ref 26

The drawbacks for the vision-integrated automatic finishing cell are its reliability and stability. The vision sensor is a sensitive piece of equipment. For instance, background light may affect the vision sensor function. If any components or the setup are not correct, the whole finishing line may shut down. Protecting the vision sensor from outside interference may need to be considered. Some of the vision functions that have been used in conjunction with robots in the die casting finishing field in recent years include:  Detecting the casting and its type  Identifying the coordination of the part  Identifying the die number or special fea-

tures of the casting Detecting the Casting and Its Type. This is the first step in identifying the shape of the casting. When discussing the automatic finishing, the first step is to identify the part for finishing. The vision system compares an image with one that is stored in the system to identify what type of casting it is. After the system recognizes the part, the system can automatically dispatch the part to the right operation channel or conveyor for the next operation. Identifying the Coordination of the Part. One advantage of the robot vision system is its capability to adjust the gripping coordination to fit the part location, whether the part is placed on the conveyer manually or by automation. By doing this, the robot determines the orientation of the part. For example, if the part is placed on the working table manually, the part will not be placed in the exact position every time. If the part is transferred from a conveyer, the vibration caused by movement of the conveyer can change the position of the part as it moves down the line. With the vision system, the robot head can automatically adjust its position and grab the part precisely to conduct the following finishing procedure. Identifying the Die Number or Special Features of Casting. Quite often, one part is produced from different dies. During the model year change of the product, the same die may have different engineering changes. Various

Automation in High-Pressure Die Casting / 757 issues will make automatic finishing more difficult. Fortunately, the vision system can help to identify detailed features and direct the part to the next operation. For instance, a deburring program is heavily part-dependent from the die number identification; the robot can access the right deburring program to conduct the cutting job.

Die Casting Internal Quality Check Automation Porosity is the most common die casting internal defect. It involves shrinkage and two types of air porosity. Shrinkage porosity is due to heavy wall casting thickness or a bad cooling situation. Most of the liquid metal during phase change, that is, from liquid to solid, will have volumetric shrinkage. Normally, fast cooling or uniform wall cooling for casting will not result in significant shrinkage. With a heavy wall and a slow cooling situation, shrinkage porosity will easily form. The air-entrapped porosity is due to quick cavity filling of the liquid metal and entrapped gas. Porosity is an internal condition, and there is no way to detect this by visual inspection, unless there is very severe shrinkage porosity close to the casting surface. Testing for porosity requires either machining or nondestructive tools, each with its own merits. Nondestructive testing is faster without scrapping the part. Destructive evaluation provides more direct results and can also check part distortion issues; however, the casting will need to be remelted. There are many nondestructive quality-check tools in die casting production, such as x-ray, ultrasonic, liquid penetration, eddy current, and magnetic particle testing, to help identify the casting defects. All of these tools require skilled people to operate and interpret the results. Each of the methods has its own unique niche; for instance, liquid penetration is very good for a casting surface crack check. The most common way to check the inside casting porosity is by x-ray. Although ultrasonic has also been used to check the internal porosity of the casting, the ultrasonic approach is limited to relatively simple and small-sized casting; otherwise, the echo pattern may not be easy to interpret (Ref 28). In-line x-ray radiographic inspection of mass-produced castings has three key elements: loading/unloading, x-ray, and image processing software. Depending on the volume, size, and number of locations that need to be checked in a casting, manual, semiautomatic, or fully automatic x-ray modes can be applied. The manual mode includes loading/unloading, operating the x-ray, and interpreting the results—all implemented by a human. The semiautomatic mode can have automatic loading/unloading, and x-ray operation, but a human interprets the x-ray result. In the fully automatic mode, no human interrupts the

operation. The interpretation of the x-ray result is conducted by the preset computer program. An unqualified casting will automatically dump to the scrap bin. Fully automatic x-ray inspection systems have been operational since the mid- to late 1990s. Early x-ray images lacked clarity and perceivable contrast. Both the computer hardware and x-ray image processing made significant advancements in the early 1990s. Currently, most commercial x-ray image analysis in the casting field can be fully digitalized and used in a 16-bit dynamic range. The 16bit image processing has significantly more image information to reduce the false defect rate. The image is sharp, and the inspection result is more reliable. Additionally, x-ray detectors have progressed from conventional image intensifiers to advanced amorphous silicon detectors, making the resolution of x-ray images even better (Ref 29). Automated Loading and Unloading for Radiography. There are a few concepts for loading and unloading selection during the x-ray operation. The part can be set in a fixed turntable between the x-ray probe and x-ray detector to take the image. The turntable can have a program to drive a three-axis movement. This concept is for a column-type casting. The turntable can be replaced by a robot. The robot moves the casting toward the x-ray probe and x-ray detector. The robot stops at the appropriate location, and the x-ray image is taken. This concept is for a part with multiple features in different directions. For a special tailor-made system, the x-ray probe and detector are mounted on a fixed robotic head. The robotic head is moved toward the parts. This application is for a long but consistent-shaped part, such as long tubing. The cycle time is very fast. No matter the type of setup, the robot plays an important role. Automatic Defect Recognition System. The challenge in x-ray checking is the same as in the automatic injection shot control system. It needs to be reliable and repeatable. Even with a sharp image generated from x-ray image processing, different operators may have different interpretations. If a computer can replace the human in explaining the x-ray check result, it will be more consistent. Additionally, the human eye will become tired from watching the computer screen. Digital images have been widely used in the x-ray inspection system. A digital image can be stored and archived for further use. Additionally, digital images can be enhanced for more accurate inspection results with tools such as integration to reduce noise or by introducing complex mathematical algorithms. Automatic defect recognition (ADR) is a sophisticated imageenhancement software that automatically makes a decision based on the gray-scale levels for each region of the x-ray image, thus eliminating operator subjectivity. Basically, all ADR systems use the same procedure to recognize defects automatically. Starting with the

current digital x-ray image to be analyzed (original image) and assuming that an ideal image is known, a different image of both is computed (Ref 30). As the image is displayed, the image data are also sent to a computer for evaluation. A pass/fail decision is made by the system, and the robot is signaled to handle the part as either accepted or rejected. Castings may be routed to the appropriate chute or conveyor. Accepted castings continue through the manufacturing process, while rejected castings are eventually melted down and recast (Ref 31). The ADR systems for nondestructive testing have been used for the last 10 years due to their reliability and efficiency. Currently, many safety-related castings that need 100% inspection go through this process. This porosity-checking approach is far more efficient and reproducible than that done by humans. The limitation for ADR inspection is based on the x-ray device, such as the power of the x-ray tube and the types of detection sensors. However, the typical detection limitations are 3% for material thickness, >0.5 mm2 (0.0008 in.2) for defect size, and <2% for false detection rate. In order to reduce the false detection rate, deflashing or accurate trimming prior to x-ray are important (Ref 29). ACKNOWLEDGMENTS The author would like to thank the following individuals and companies for their assistance:    

Jack Branden, Visi-Trak Mark Riekert, Rimrock Jeff Walters, BuhlerPrince Frank Wiencek, Yxlon International

REFERENCES 1. “The ADCI Die Casting Data Access File,” American Die Casting Institute, Inc., NADCA 2. “Introduction to Die Casting,” American Die Casting Institute, Inc., NADCA 3. W. Levine, Automation—It Pays for Itself, NADCA Trans., 1985 4. J. Vann and Y. Shen, “Computer Integrated Information and Control Systems for Improved Quality and Profit,” 15th International Die Casting Congress and Exposition, NADCA, 1989 5. D.C. Wilson, Working with the Machine Training Book, 1999 6. “Central Management System for Die Casting Machines Specification,” UBE Machinery 7. N. Iwamoto and H. Tsuboi, “Trend of Computer Controlled Die Casting,” 13th International Die Casting Congress and Exposition, SDCE, 1985 8. J.T. Branden and M. Brown, “A Method for Developing Uniform Cavity Pressure, Extending Die Life by Integrating SoftShot Technology onto Automotive Transfer Case,” Cast Expo, NADCA, 2005

758 / High-Pressure Die Casting 9. “PMSI Programmable Multi-Stage Intensification,” BuhlerPrince, Inc. 10. E.A. Herman, Die Casting Process Control, Chap. 7, NADCA Training Book, 2003 11. J. Vann, “Real Time, Dynamic Shot Control for Improved Quality and Productivity,” 17th International Die Casting Congress and Exposition, NADCA, 1993 12. K. Young and K. Brissing, “The SC Machine: Real-Time Power and Control for Advanced Application,” 17th International Die Casting Congress and Exposition, NADCA, 1993 13. W. Bishenden and R. Bhola, Increased Production Rates with Automatic Die Temperature Control and Effective Die Spray, NADCA Trans., 2003 14. M. Osborne, C. Mobley, and A. Miller, Modeling of Die Casting Processes Using Magmasoft, NADCA Trans., 1999 15. R. Rosch, H. Becker, S. Kluge, and L. Kallien, Simulation Aided Design and Process Development for Mg Die Casting, Die Casting Eng., July/Aug 1998

16. H. Hu, F. Chen, and X. Chen, A Computerized Data Acquisition and Control System for Die Thermal Management, NADCA Trans., 2003 17. W. Bishenden and R. Bhola, Die Temperature Control, NADCA Trans., 1999 18. E. Wrench, Error Proofing Your Process— Man or Machine? Die Casting Eng., March 2008 19. D. Murry, “Automated Die Cast Insert Loading Using a Robot,” International Die Casting Congress and Exposition, 1983 20. Acheson Deltacast liquid powder presentation, NADCA, 1999 21. D. Kraklau, Automatic Die Release Spraying, Die Casting Eng., March 2008 22. P. Robbins, The Efficient Application of Die Lubricant, Die Casting Eng., July 2003 23. A. Koch, Die Lube Selection: An Opportunity for Cost Saving, Die Casting Eng., Nov 2004 24. D. LaRussa, Robotic Automation: The Key to Reliability, Versatility and Profitability, Die Casting Eng., July 2001

25. Software Advances Robotic Casting Finishing, Foundry Manage. Technol., Jan 2006 26. C. Cooper, Quality Control Through Automatic Visioning, Die Casting Eng., Sept/ Oct 1999 27. J. Meyer, Machine Vision and Robotic Inspection Systems, Nondestructive Evaluation and Quality Control, Vol 17, ASM Handbook, ASM International, 1989 28. S. Palanisamy, Automated Ultrasonic Classification of Defects in Al Die Castings, NADCA Trans., 2003 29. F. Wiencek, Automatic defect recognition PXV 5000 presentation, Yxlon International, 2007 30. F. Herold, “A New Analysis and Classification Method for Automatic Defect Recognition in X-Ray Images of Castings,” Eighth ECNDT, June 2002 (Barcelona, Spain), Spanish Society for NDT 31. R. Pontefract, Automatic X-Ray Inspection of Structural Castings, Die Casting Eng., Sept/Oct 1999

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 761-763 DOI: 10.1361/asmhba0005272

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Semisolid Casting—Introduction and Fundamentals Q.Y. Pan, SPX Corporation Diran Apelian, Worcester Polytechnic Institute John Jorstad, J.L.J. Technologies, Inc.

SEMISOLID METAL (SSM) PROCESSING, also known as semisolid metal casting, semisolid forming, or semisolid metal forging, is a special die casting process wherein a partially solidified metal slurry (typically, 50% liquid/50% solid instead of fully liquid metal) is injected into a die cavity to form a die-cast type of component. The origin of SSM can be traced to experiments conducted by David Spencer at Massachusetts Institute of Technology (MIT) in 1971 as part of his doctoral thesis under the supervision of Professor Merton Flemings (Ref 1, 2). Spencer discovered that when the dendritic structure of a partially solidified Sn-15wt%Pb alloy was fragmented by shear in a Couette viscometer, a globular structure resulted. The apparent viscosity of the globular structure was dramatically lower than that of the dendritic, and the slurry thus formed had fluidity approximating that of machine oil. That semisolid material at rest held its shape like a solid; however, when a shear stress was applied, it became sufficiently fluid to be injected into a die casting die — a property known as thixotropy. The key finding of Spencer’s work — that forming a globular structure in a partially solidified alloy reduces the viscosity of the slurry under shear and makes it castable — ultimately led to SSM processing as it is known today (2008). Since 1971, further development work has been carried out at MIT, Worcester Polytechnic Institute, as well as numerous research laboratories and companies around the globe whose interest includes advanced casting techniques. An international conference is held every two years, dedicated to addressing recent developments and applications of semisolid processing, as well as the scientific issues being pursued by researchers around the world. The tenth such conference is scheduled to be held in Aachen, Germany, during 2008. After more than three decades of continuous development and refinement, SSM processing is a commercially viable technology for the production of high-integrity die-cast-type components. The key to SSM processing is to generate a semisolid metal slurry that contains a globular

primary phase (surrounded by the enriched liquid phase) and exhibits thixotropic behavior. That is, the viscosity of the slurry decreases continuously under shear deformation, whereas the viscosity value can be recovered once the shear action ceases. The appropriate semisolid slurry does not have a dendritic primary phase. Instead, it has a globular, or at least largely globular structure that allows it to exhibit thixotropic behavior. Figure 1 shows the fraction solid versus temperature curve of A356, a popular alloy for SSM processing. Note that upon reaching the aluminum-silicon eutectic temperature, the alloy has formed 40 to 60% solid primary alpha-aluminum phase, an ideal scenario for SSM processing (Fig. 2). Semisolid processing has now been in commercial use for more than 30 years. However, the thixocasting route, requiring production of a billet precursor, has been deemed too expensive to enjoy widespread applications. Thus, the recent driving force to reduce semisolid processing costs has shifted research and development efforts toward rheocasting process variations. Thixomolding (Thixomat, Inc.), specific to the semisolid processing of magnesium, remains another

Fig. 1

A356 alloy fraction solid versus temperature

commercially viable SSM processing route. The semisolid processing route that proves least cumbersome, least complicated, and most manageable for any specific facility and product will prevail.

Advantages of SSM Processing Reduced Enthalpy: Longer Tool Life. Because the slurry is approximately 50% solid when cast, much of the sensible heat and heat of fusion has already been released before injection into a die, and the thermal load on tooling is thus greatly reduced. Figure 3 illustrates enthalpy versus fraction solid for A356 alloy. It can be seen that liquid squeeze casting from a typical melt casting temperature through solidification and a typical casting ejection temperature releases 900 J/g (390 Btu/lb) to the die, whereas semisolid casting at 0.5 fraction solid (where enthalpy is only 300 J/g, or 130 Btu/lb) releases 500 J/g (215 Btu/lb) through casting ejection, which results in 45% less energy to the tool compared to squeeze casting. That reduced energy load, together with a smaller temperature difference between cast material and the tool, provides semisolid casting a vastly superior tool life.

762 / Semisolid Casting Reduced Cycle Time. The reduced energy release also affects solidification time, typically reducing it by nearly 50%. Reduced Solidification Shrinkage. Injecting a slurry with 50% fraction solid into a die cavity means that 50% solidification shrinkage associated with the primary phase has already taken place, thus reducing the tendency for shrinkage pores and the need for shrinkage feeding. Improved Casting Integrity and Properties. Reducing solidification shrinkage and porosity ultimately improves mechanical properties. Also, the eutectic morphology typical of semisolid processed material favors high ductility, even in less-pure alloys such as secondary material made from scrap recycling.

SSM Processing Routes There are three major semisolid processing routes: thixocasting, rheocasting, and Thixomolding, and several variations within those. The following is a brief description of the concept for the processing routes. More details about processing routes and various processing methods are given in other sections and articles in this Volume. Thixocasting (Billet Feedstock). In the thixocasting route, one starts from a nondendritic solid precursor material that is specially prepared by a manufacturer, using continuous casting methods. Upon reheating this material into the mushy (two-phase) zone, a thixotropic slurry is formed, which is the feedstock for the semisolid casting operation (Fig. 4). Thixocasting was the first commercialized SSM process. It consists of two separate stages: the production of billet feedstock having the appropriate globular structure, and reheating of billets to the semisolid temperature range, followed by the die casting operation. Billet Feed Stock for Thixocasting. There are three major processing methods to make thixocasting feedstock:  Magnetohydrodynamic (MHD) stirring pro-

Fig. 2

Typical semisolid A356 alloy structure consisting of globular alpha-aluminum surrounded by eutectic

cess: This process combines electromagnetic stirring during freezing with multistrand, direct-chill (DC) casting techniques to produce bars of up to 152 mm (6 in.) diameter having a cellular or rosettelike primary phase for shipment to casting facilities, where they are cut into slugs that are reheated to the semisolid region to develop

Fig. 3

Enthalpy versus fraction solid for A356 alloy

Fig. 4

Schematic of the thixocasting semisolid metal (SSM) route. EM, electromagnetic

the fully globular structure for forming into SSM parts.  Chemical grain refining: This alternative for the production of thixocasting feedstock merely reheats billets that were chemically grain refined and fast cooled during DC casting to generate the small rosette grains that quickly become globular during reheating for SSM processing.  Strain induced melt activation (SIMA) process: This process involves cold/hot deformation of a starting material to induce sufficient strain, followed by reheating the cold/hot-worked material to the semisolid region to develop an extremely fine recrystalized globular structure for SSM forming. The SIMA process is applicable to various alloy systems, including aluminum, magnesium, copper, and ferrous alloys, but is only applicable to SSM feedstock with relatively small diameters. Rheocasting (Slurry). In the rheocasting route (also known as slurry-on-demand), the liquid state is the beginning point, and a thixotropic slurry is formed directly from the melt via special thermal treatment/management of the solidifying system; the slurry is subsequently injected into the die cavity (Fig. 5). Rheocasting is favored over thixocasting in that there is no associated cost premium for producing billet precursor, and trim and scrap can be recycled within the processing facility without first being turned into billet. In the early days of SSM development, it was thought that the melt had to cool into the twophase region and the dendrites had to fragment (i.e., melt agitation via mechanical or MHD stirring) to produce the globular slurry required for semisolid processing. However, recent research and development at the Advanced Casting Research Center of the Metals Processing Institute of Worcester Polytechnic Institute (WPI), sponsored by the Department of Energy (Ref 3, 4), has led to the discovery that dendrites did not need to break off to produce the semisolid structure. Instead, if the temperature of the melt was such that many nuclei were produced (copious nucleation), and if the nuclei did not grow past a certain point (i.e., suppression of dendritic growth) nor melt back into the bulk liquid, then a slurry with an ideal semisolid structure could be produced directly from the melt. This concept is the genesis of various processes and methodologies to generate semisolid slurries from the liquid state. The concept relies on controlled nucleation and growth, as opposed to the previous theory in which a dendritic structure is modified into a globular structure via shear forces. Numerous Rheocasting Process Variations. Recently, the driving force to reduce semisolid processing costs has led to the development of several rheocasting process variations. These include UBE’s New Rheocasting (NRC) process, MIT’s Semisolid Rheocasting (SSR) process, THT’s Subliquidus Casting (SLC) process, Alcan’s Swirl

Semisolid Casting—Introduction and Fundamentals / 763

Fig. 5

Schematic of the rheocasting semisolid metal (SSM) route compared to thixocasting. MHD, magnetohydrodynamic; GR, grain refining

Enthalpy Equilibration Device (SEED), Brunel’s Rheo-Diecasting process, Mercury-Marine’s Slurry-on-Demand process, and WPI’s Continuous Rheoconversion Process (CRP). These are covered in some detail in the article “Rheocasting” in this Volume. Thixomolding (Magnesium Pellets). The Thixomolding process was developed by Dow Chemical and subsequently purchased by Thixomat Corp.; today (2008), it is the commercial process for semisolid forming of magnesium alloys and the most-used of all semisolid processing routes. The method is analogous to the injection molding of polymers (Fig. 6). In this process, solid chips or pellets of conventionally solidified magnesium alloys are fed into a heated injection system containing a reciprocating screw. Upon heating, the metal chips are converted by the shear action of the screw into a thixotropic, low-solid-content slurry (solid fraction less than 0.3), which is fed into the shot accumulator by the rotating screw. Once the accumulation chamber is filled, the slurry is injected in the mold. A major advantage of the process is that it effectively combines both slurry making and slurry injecting into a onestep process, leading to high productivity and energy savings. In addition, the process avoids the safety problems usually associated with melting, handling, and die casting molten magnesium. Details are provided in the article “Thixomolding” in this Volume. REFERENCES

Fig. 6

Schematic of Thixomolding machine

1. D.B. Spencer, Ph.D. thesis, MIT, Cambridge, MA, 1971 2. D.B. Spencer, R. Mehrabian, and M.C. Flemings, Metall. Trans., Vol 3, 1972, p 1925–1932 3. Report DE-FC07-98ID13618, Department of Energy 4. M.C. Flemings, Semi-Solid Processing — The Rheocasting Story, Test Tube to Factory Floor: Implementing Technical Innovations, Proceedings of the Spring Symposium, May 22, 2002, Metal Processing Institute, Worcester Polytechnic Institute, Worcester, MA, 2003

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 764-772 DOI: 10.1361/asmhba0005273

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Thixocasting Justin Heimsch, Madison Kipp Corporation

SEMISOLID METAL CASTING had its genesis in Merton Flemings’ lab in 1971 with the rheological experiments done by David Spencer as part of his doctoral thesis. The discovery that shearing a semimolten metal produced potentially castable rheologies ignited tremendous interest in what would be an entirely new area of casting technology. It took almost a quarter century of investigation before semisolid metal casting matured into a fully commercialized process (Ref 1). At the heart of all semisolid metal casting processes is the need for material comprised of a globular solid phase surrounded by a lower-melting-point liquid phase (Fig. 1). When such a material is sheared, the interconnections between the solid globules break down, allowing the material to flow as a viscous liquid. There are two distinct process routes to attain a globular structure suitable for semisolid casting. The term thixocasting was adopted to describe processes that require specially processed billet material that is reheated to the semimolten state just before injection (Ref 2). Alternatively, processes that create semisolid structures directly from the liquid state are referred to as rheocasting (Ref 2). The goal of this article is to provide a general

overview of the thixocasting process and to discuss the concepts that are important to the practical application of this technology. The thixocasting process involves two casting processes. The first casting process is required to make the feedstock that must be reheated to achieve the structures necessary for casting. The second casting process combines billet sawing, reheating, and the actual injecting of material into the mold. The thixocasting and rheocasting processes are closely related, and both have several distinctions that set them apart from more conventional highpressure die casting processes. A brief summary of the distinctions includes:
Exceptional part quality—extremely low










levels of shrinkage or gas porosity, excellent leak tightness and weldability Excellent mechanical properties—T5 heat treated thixocast parts are able to achieve properties typically found in T6 heat treated permanent mold castings. Fast cycle times—casting rates in excess of 100 shots per hour possible Long die life—limited thermal shock to the tooling, less heat checking Near-net shape—low draft angles possible, reduced machining stock compared to competitive casting processes High-pressure process requiring additional machine capabilities Lower material utilization compared to competitive processes Higher material costs associated with precursor material production

With the general understanding of the distinctions that separate thixocasting from other more conventional processes, each step of the process is now examined in more detail.

Bar-Making Processes

Micrograph showing typical a-phase morphology found in reheated magnetohydrodynamic material. Specimen has been etched to darken the eutectic phase.

Fig. 1

During the 1990s, companies began production of semisolid metal castings in large numbers, using a variety of thixocasting processes. Central to these processes was the use of a precast bar stock that had a modified grain structure, which allowed for reheating to a

castable semisolid state. Historically, there were four distinct production methods used to produce suitable material for reheating: mechanical stirring, strain-induced melt activation, grain-refined direct-chill casting, and magnetohydrodynamic casting. These processes have been used to produce a wide variety of materials, ranging from simple binary alloys to exotic high-temperature materials for thixocasting production and laboratory study. When discussed in the context of tons of alloy cast, the aluminum-silicon family of casting alloys is the clear leader. As such, much of the following discussion focuses on the aluminum-silicon casting alloys methods of production. The basic principles applied to this family of alloys crosses over to many others. Today (2008), the predominant method of thixocast bar production uses magnetohydrodynamic (MHD) stirring to form the structure necessary for good castability after reheating. Central to the MHD process, and most of the alternative processes previously employed, is the use of sophisticated foundry techniques originally designed to achieve aerospace-quality cast and wrought materials. Exceptional metal quality is the cornerstone of the thixocasting process and the factor that further distinguishes it from lower-cost casting alternatives. However, the additional cost associated with the production of billet stock has relegated thixocasting to a niche market focused on highintegrity, high-strength, heat treated components often used as cast iron or aluminum forging replacements (Ref 3). Before examining the four historically significant processes in detail, it is useful to summarize the steps normally taken to produce a bar stock suitable for thixocasting. A brief summary of these steps follows (Ref 4): 1. Select high-grade primary alloy base material. 2. Melt material and perform initial chemical analysis. 3. Perform metal treatment and stirring. 4. Analyze precasting metal quality. 5. Cast on vertical or horizontal bar-casting machines, using continuous metal treatment via flow-through degassing and filter boxes before entering the billet molds.

Thixocasting / 765 6. Analyze product quality: material density and conductivity. 7. Prepare for shipment. What should be apparent in the previous summary is that special attention is taken to maintain metal quality throughout the casting process. The use of large heats of tightly controlled alloy during the bar-casting process allows as many as 10,000 billets to be produced from a single heat and helps to ensure high levels of consistency of the billet produced (Ref 4). The bar-casting process is often highly automated, which helps to ensure microstructural consistency during the casting of a single heat, as well as from heat to heat. The following variables are typically monitored and controlled throughout bar stock casting (Ref 5):










Metal chemistry Stirring power (MHD processes) Cast metal temperature Cast metal flow rate Cooling water temperature and flow rate Lubricant flow rate Degassing practices Filtration techniques Mold thermal conductivity

These variables control microstructural features such as a-phase size and morphology, as well as the level of alloy segregation seen from the exterior to the interior of the billet. For instance, higher cooling rates and increased metal agitation result in smaller a-phase particles. The size of the a-particles becomes important during the reheating phase of the thixocasting process. With the general barmaking overview complete, the four historically significant bar production methods are now examined. Mechanical Stirring. The earliest forms of semisolid precursor material production used mechanical agitation to control the formation of extensive dendrites found in normally solidified alloys. The fluid velocity gradients set up during stirring shears off the dendrite arms, resulting in a microstructure characterized by small rosettes of the primary a-phase. During reheating, the a-phase rosettes coarsen slightly and become more spherical or globular. This globular structure is necessary to achieve a castable rheology after reheating. While attempts were made to commercialize mechanical stirring, the process never gained wide acceptance and today (2008) is no longer in use. This failure was due, in part, to the increased tendency for oxide and atmospheric entrainment into the metal during mechanical stirring and to the cost associated with maintaining the stirring mechanism, which is required to function in an extremely damaging thermal and chemical environment (Ref 1). The strain-induced melt activation (SIMA) process involves cold working of standard extruded or rolled bars, followed by a post-coldworking reheating cycle. As this cold-worked

material approaches the solidus temperature during the reheating process, numerous small, equiaxed grains are formed from the strained matrix. Once melting begins, these small grains coarsen quickly due to high diffusion rates at the liquid-solid interface and form the globular structure necessary for thixocasting. When compared to grain-refined or MHD bar, the scope of market penetration of SIMA bar has been limited. This limited market penetration is due to the high cost of cold working larger-diameter bars sufficiently to ensure recrystallization throughout the entire cross section during the reheating process. For this reason, the scope of SIMA usage is often limited to academic investigations or to the commercial casting of smaller components made of wrought or ferrous alloys. Typically, SIMA bars are no larger than 50 mm (2 in.) in diameter. Grain-Refined Billet. The semisolid casting boom of the mid-to-late 1990s centered around two primary billet production processes: grainrefined direct-chill casting and MHD casting. The use of specialized grain refinement techniques, coupled with high cooling rates during the casting process, produced bars with nondendritic or semidendritic a-phase structures capable of being thixocast after reheating. The average a-phase particle size for grain-refined raw material is between 75 and 125 mm (3000 to 5000 min.). Figure 2 illustrates an average billet cross section before reheating. The relatively small a-phase rosettes ripen during reheating, forming the globular solid-phase structures necessary for semisolid metal casting. Typically, grain-refined bars are no larger than 90 mm (3.5 in.) in diameter. This size limitation stems from difficulty achieving cooling rates fast enough to ensure the correct a-phase morphology in the center of the bar. Figures 3 to 5 illustrate how a-phase particle size and morphology can change throughout the cross section of a larger-diameter grain-

refined billet. Figure 6 illustrates the a-phase morphology of a grain-refined material after reheating. Compared to mechanically stirred billet production processes, the grain-refined billet-casting process has several distinct advantages. The grain-refined casting process typically incorporates inline degassing and filtering techniques, as well as launder systems to ensure quiescent filling of the casting molds. The use of launders eliminates the turbulence normally associated with mechanical stirring and improves metal quality by reducing the tendency for oxide generation and atmospheric entrapment within the solidified bar. When compared to MHD-stirred billet, grain-refined billet typically exhibits reduced melt loss during the reheating process. This reduced melt loss can be attributed, in part, to the larger size

Micrograph showing a-phase morphology of a typical grain-refined raw material before reheating. Specimen has been etched to darken the eutectic phase.

Micrograph showing a-phase morphology midway between the center and exterior of a grain-refined billet. Specimen has been etched to darken the eutectic phase.

Fig. 2

Micrograph showing a-phase morphology near the faster-cooling exterior of a grain-refined billet. Specimen has been etched to darken the eutectic phase.

Fig. 3

Fig. 4

766 / Semisolid Casting The morphology of the high-melting-point aphase within the raw billet stock is the factor most critical to the castability of the material. The aphase structures found in commercially available MHD billets typically have average rosette diameters in the range of 25 to 75 mm (1000 to 3000 min.), with some variation throughout the billet cross sections. If the scale of the a-phase is too large, or the features are excessively dendritic throughout the bulk material, the reheating process will lead to insufficient spheroidization of the a-phase. Billets with excessive dendritic interconnection do not exhibit adequate fluidity under an applied shear to be useful even after longer periods of reheating.

Billet Sawing and Volume Control Micrograph illustrateing a-phase morphology found at the center of a larger-diameter grainrefined billet. Specimen has been etched to darken the eutectic phase.

Fig. 5

Micrograph showing a-phase morphology of a grain-refined material after reheating and casting. The eutectic phase is lighter in this micrograph because it has not been etched.

Fig. 6

and increased dendrite arm branching of the aphase rosette. During reheating, the coarser dendrite arms trap eutectic liquid within the aphase particle as it ripens into a pseudospherical globule. This entrapped eutectic liquid does not affect the castability of the reheated billet, but additional time at temperature is often necessary to achieve the required a-phase ripening. The increase in heating time should be weighed against the 5 to 10% improvement in material utilization possible when compared to MHD billet. This consideration is more important than ever, since the raw material cost associated with billet production has increased drastically in recent years as worldwide demand for clean primary aluminum scrap has expanded. MHD-Stirred Billet. The most predominant billet-casting process is MHD casting. This process combines conventional direct-chill casting

Micrograph showing a-phase morphology typical of magnetohydrodynamic material prior to reheating. Specimen has been etched to darken the eutectic phase.

Fig. 7

technology with induction stirring of the molten metal near the liquid/solid interface. The ability to control the rate of shear and heat transfer within the solidifying metal allows for a finergrained a-phase, improved homogeneity of the a-phase size distribution within the cast bar, as well as reduced edge-to-center segregation of alloy constituents such as silicon. An average MHD bar microstructure before reheating can be seen in Fig. 7. Typically, the a-phase rosettes range in size from approximately 25 to 75 mm (1000 to 3000 min.). When compared to grain-refined billet, MHD billet often requires less energy during reheating to achieve castable morphologies. This reduction in energy is related to the improved spherical character of the a-phase found in MHD raw material. The melt losses associated with MHD billet can be somewhat higher than those encountered in the reheating of grain-refined billet. The actual melt losses for both material types is a function of the reheating equipment and recipe used; vertical reheating processes tend to have more melt loss. Suitable Billet. Fortunately for the thixocasters of the world, foundry issues are not ones they need to worry about, since the most crucial phase of billet preparation is taken care of by companies with specialized foundries accustomed to producing high-quality materials using directchill cast or MHD-stirred bar-casting processes. Central to the thixocasting process is the use of fine-grained, nondendritic cast billets having exceptional metal quality. Since the typical semisolid component end user is looking for high-strength, leaktight, or weldable castings, particular emphasis is placed on controlling oxides, nonmetallic inclusions, and dissolved gas. The likelihood of having serious billetrelated issues is extremely small, but the following discussion emphasizes the most important factors influencing the performance of the raw billet stock when reheated and cast.

With the casting of the semisolid precursor strands, the next phase of the thixocasting process is billet sawing. There are two possible process routes to choose from. The first route is to have the billet cut by the bar vendor or another outside sawing facility. This option may be the best choice for smaller-scale casters wanting to avoid the additional cost of bar-cutting equipment. The second option is to purchase the equipment necessary to cut the longer strands to the correct cast length. This is the most common approach used by today’s (2008) thixocasters. When selecting a cutting method, several factors should be considered. These factors include the ability to automate the process, the quality and repeatability of the cut, and the total cost of ownership of the capital equipment necessary. In practice, thixocast billet cutting is usually performed using automated circular saws. Carbide-bladed circular saws are capable of precision cutting of nonferrous bar stock at extremely high production rates. The rigid blade design of a circular saw improves cut perpendicularity and surface finish. The perpendicularity of the cut to the long axis of the billet is important when vertical induction reheating equipment is used. Good cut perpendicularity reduces the tendency for billet tipping during the final stages of the reheating process. The largest billet diameter to be cut will determine the minimum blade size. Typically, the workpiece diameter should be no more than 1=3 the overall blade diameter. Typical billet diameters range from approximately 50 to 152 mm (2 to 6 in.), depending on the bar production method, with the most common sizes falling within the 75 to 115 mm (3 to 4.5 in.) range. Thus, saw blades are often in excess of 325 mm (13 in.) in diameter. These large-diameter blades are coupled to 20 to 40 kW, frequency-driven motors in order to maintain high production rates. The use of light oil blade lubricants improves cutting speeds as well as blade life and does not have adverse downstream effects. With proper process optimization, blade life can be as high as 100,000 cuts before resharpening is required, and blades

Thixocasting / 767 can be resharpened several times before they need to be retired. Another critical saw-related factor to consider is the blade kerf, or the material removed during each cutting cycle. While optimizing the kerf is important to maximize the profitability of any cutting operation, it is even more important for cutting semisolid billet. The chips generated during semisolid bar cutting can lose 50% or more of their value. Because of this loss, extra efforts should be focused on minimizing blade kerf. A variety of available options improve the automation and performance of these circular saws. High-end automated saws commonly feature inclined bar magazines and roller conveyers with push carriages, which allow semicontinuous operation. Hydraulically dampened length gages with coarse and fine adjustments allow for rapid setup and improve cut-to-cut repeatability. Cut tolerances between þ0.1 and  0.2 mm (0.004 and 0.008 in.) are possible at production rates as high as 300 to 600 cuts per hour or more. Billet process control is very easy, assuming a good-quality saw is used and the vendor has adequate control of its process. The goal of the billet-cutting process is to deliver a consistent metal volume to the casting process. The billet diameter, density, and cut length are the variables that affect the delivered volume. The billet diameter and cut length are easy to measure and control. The billet molds and saw blades may wear over time, but the time frame for the changes is long relative to the process cycle, so shot-to-shot volume fluctuations will be insignificant. For example, a casting using a 102 mm (4 in.) diameter billet with an overall cut length variation of  þ0.2 mm (0.008 in.) will have a volume variation of approximately 3.3 cm3 (0.20 in.3). Assuming the casting machine uses a plunger diameter of 108 mm (4.25 in.) for this billet size, the overall injection stroke variation due to this volume difference would be less that 1 mm (0.04 in.). This level of process variation is insignificant to part quality. The volumetric change associated with billet diameter variation increases at a rate proportional to the square of the diameter. Fortunately, bar-to-bar diameter variations are typically held to a minimum, and bar diameter drift associated with mold wear is quite slow. This slow wear rate allows adequate time to retarget the sawing process, diminishing the impact that slight diameter changes would have on the casting process. Consider a 254 mm (10 in.) long billet having a 102 mm (4 in.) nominal diameter. Assuming this billet is to be injected using a 108 mm (4.25 in.) diameter plunger, a 1 mm (0.04 in.) increase in the nominal diameter would decrease the injection stroke of the casting machine by approximately 4.5 mm (0.18 in.). This 1 mm (0.04 in.) variation in billet diameter would be considered quite large if it were a bar-to-bar variation, but the resulting injection variation would still fall within normal die-casting process variation limits.

Density variation should be managed by the bar vendor during the course of normal process audits. Ultrasonic examination of individual cast strands is one method used to ensure that excessive internal porosity does not exist in outgoing thixocasting bar stock. To date, the author has processed more than 25 million kg (28,000 tons) of thixocasting billet and has yet to encounter billet density issues that negatively impacted the cast product quality. The two simplest methods for controlling the billet sawing process are to weigh billets or to measure their length. The billet-weighing method is the most straightforward method to gain volumetric information regarding the billet being prepared. Unfortunately, this method increases the difficulty in setting up the saw if tight weight controls are in place, particularly when attempting to control the weight of the last cut piece of each longer strand. This piece is subject to additional length variation due to the bar vendor’s sawing process tolerance and can impact the final weight of the piece. While length measurement tells less about the volume actually being delivered to the casting process, it is easy to measure and improves the ease of saw setup. High-quality sawing equipment is capable of maintaining cut length variations of  þ0.2 mm (0.008 in.) or less. As has been previously mentioned, the thixocasting process can easily accommodate the variation typically associated with bar production and sawing. The last step of the sawing process is to package the billet in the appropriate container for the reheating process. This packaging could involve simple metal baskets or more complex loading magazines for fully automated reheating cells using robots for both loading and unloading of the heating stations. Billet reheating is discussed in the next section.

Reheating and Billet Handling After cutting, the billet must be heated to a temperature above the alloy solidus temperature for a period sufficient to ripen the small a-phase rosettes into semispherical globules. The time required to achieve adequate spheroidization is dependent on the initial microstructure of the material and the reheating process employed. Over the years, a wide range of billet-heating methods have been investigated. Currently, induction heating is the most commonly used method. This dominance is due to the ability of the induction heating process to achieve high production rates within the confines of the casting cell, while giving the process engineer high levels of heating control. When examining the induction heating systems that are currently used for thixocasting, two distinct design philosophies become apparent. One style of heating system couples a single power supply to a single induction coil. This approach gives the process engineer nearly unlimited control of the reheating process and

has the potential to improve cycle time and material utilization. An additional benefit of multipower supply systems is the ability to create dwell cycles to reduce material losses during short periods of machine downtime. The biggest drawback to the one-supply-toone-coil design is upfront capital cost. In order to achieve rapid production rates, these multipower supply systems may require as many as 12 or more coil-power supply pairs within the casting cell to achieve subminute shot-to-shot cycle times. This demand increases the capital outlay for the system dramatically and can put the total price tag of a single high-speed induction cell into millions of dollars. In addition, the increased complexity of the heating equipment can lead to additional maintenance costs throughout the life of the system. The competing induction system design compromises some of the process control to enhance the affordability and simplicity of the final heating system. The key feature of this style of induction heating system is the use of a single power supply connected in series to as many as 18 induction coils. A carousel system is often used to index billets from coil to coil during the reheating process. After a preset heat time, the induction coils rise to allow the index table to move the billets to the next coil in the heating sequence. Where the multipower supply systems allow for nearly unlimited computer-controlled process variations, a single power supply system allows for programming of three basic parameters: heat-on time, heat-off time, and power input. Further process refinement is achieved by varying the geometry of the coils arranged in series. It is possible to increase or decrease the power induced within the exterior of the billet at a particular heating station by increasing or decreasing the number of coil turns or the coil diameter (Ref 6). Refining the reheating process using three or more different coil geometries within an 18-coil system is not uncommon. Another difference between the single power supply system and the multipower supply and coil system is the orientation and movement of the billet during heating. Typically, single power supply systems have vertically oriented coils that are raised and lowered via a hydraulic lift between each heating cycle. The necessity for billet indexing, via a lower carousel table, limits the length-to-diameter ratio of the billet to be heated. At larger length-to-diameter ratios, the billets become increasingly unstable and can topple during the later stages of reheating. A good target length-to-diameter ratio (2 to 1) minimizes the negative process implications of excessively large billet diameters while still maintaining adequate stability. Multipower supply systems often—not always—employ horizontal coils that allow for larger length-to-diameter ratios. Because the billet is supported by a ceramic-lined process boat, billet stability is not a critical factor. Horizontally reheated billets may exhibit less eutectic runoff after reheating. Allowing this

768 / Semisolid Casting eutectic liquid to be placed within the shot sleeve for casting may increase process yield at the expense of overall product quality. The metallic boats used in horizontal heating systems are well suited to robotic automation of the billet loading and unloading process. In these systems, the robot grasps locators on the metallic boat so it can be repeatedly picked up and delivered to the die-casting machine. Maintaining boat cleanliness is the major process problem for this application, but the robot allows for integration of automated boat-cleaning cycles into the casting process. In vertical heating systems, the billet may tilt slightly as it nears the final stages of the reheating process. Unfortunately, the axis of tilting is not necessarily repeatable, so simple robot extraction of reheated billets may be plagued by material waste. The use of machine vision systems may be helpful in this regard, but manual billet loading has also been applied successfully for smaller shot weights. The materials selection and gripper design are critical to robust robotic loading, because grippers will repeatedly encounter hot semimolten alloy. Gripper cleanliness is essential to prevent presolidified material from entering the casting.

Rheological Tests The rheology of semisolid materials has been researched extensively, and a variety of models have been proposed to describe the complex fluid properties. It is widely accepted that semisolid flow is non-Newtonian; beyond this, agreement breaks down over the specifics of semisolid rheology. This disagreement stems from the time-dependent nature of a semisolid material response to shear. Experiments measuring shear viscosity under steady-state conditions generally find semisolid materials to be shear thinning. When rapid transient conditions are used, the semisolid material exhibits shear thickening behavior. In production, mold filling occurs too rapidly to allow for steady-state conditions to form, making the rapid transient condition intuitively more reasonable. Fitting this concept within the mathematical framework necessary to more fully illuminate the rheology of semisolid fluids as they are used in realworld application is no trivial matter. Several computational fluid-dynamics programs have been developed that incorporate the constitutive models commonly associated with semisolid fluid flow to simulate mold filling for industrial application. The most common models associated with semisolid rheology are power law, Bingham, Herschel-Bulkley, and an internal variable model developed at Massachusetts Institute of Technology during the early 1990s (Ref 7–9). The internal variable model adds an additional agglomeration term to the conventional non-Newtonian power law model from which the Bingham and HerschelBulkley are developed. This agglomeration term incorporates the breakup and rewelding

of the a-phase as a result of time-dependent shear and temperature variation into the calculation of semisolid viscosity (Ref 7, 8). The specific mathematics related to the rheology of semisolid materials is well documented and beyond the intended scope of this section. The remainder of this section focuses on tests used to develop and maintain production processes. A variety of rheological tests have been used to help in the development and maintenance of reheating recipes. The most basic of these tests involves the use of a knife in manual cutting of the reheated slug. The knife test is the most frequently used test because it provides immediate information regarding overall billet consistency, and it needs no expensive secondary equipment. In experienced hands, the knife tests will readily expose reheating issues, such as discontinuous edge-to-center consistencies, allowing for on-thefly improvements to the reheating process. The major drawback of the knife test is its qualitative nature, which requires the tester to understand semisolid reheating principles to fully harness the power of the test. Internal and external billet temperature measurements have been used to indirectly gain insight into the rheological state of reheated billets. A typical system may use a thermocouple probe or a noncontact pyrometer at an intermediate reheating stage where the billet is still fully solid, as well as at least one thermocouple probe near the final stages of reheating to gather information on internal billet temperatures. This quantitative information can be entered into a decision matrix for the machine operator so that the reheating process can be maintained. With the appropriate reheating equipment, this

Fig. 8

information could be entered directly into the system programmable logic controller, creating a closed-loop reheating process. The major issue with contact measurement systems is material buildup on the final stage probes. Excessive material buildup can skew temperature readings as well as lead to contamination of the cast material should the buildup break away from the probe, lodging in the billet about to be cast. Several direct rheological tests have been developed that measure the insertion force of a probe pushed into a billet near the final stages of reheating (Ref 1). The quantitative nature of the force measurement data allows for improved process monitoring and control, but the high-temperature metal contact of the insertion device makes probe cleanliness a potential problem. One offline billet rheology tester used a potentiometer in conjunction with a metal inert gas welder wire-feed system to ensure the billet-probing device was free of debris before each test. Another offline, direct-measurement technique involved squashing a large portion of a billet cross section to gain insight into the rheology of the reheated billet (Fig. 8). More advanced techniques continue to be developed that allow for noncontact determination of the billet rheology. Several of these techniques use closed-loop inductive feedback circuits (Ref 10, 11) as well as acoustic resonance (Ref 11). The electrical conductivity drop that occurs during billet reheating leads to changes in the penetration depth of the electromagnetic field induced by the coils of the billet heater; the inductive feedback circuits measure this change, allowing for continuous measurement and control of the power input into the

Example of billets that underwent one method of offline rheological testing. Courtesy of S.C. Bergsma, Northwest Aluminum Co.

Thixocasting / 769 billet. The billet acoustical signature undergoes a similar distinct change as the billet transitions from the solid to liquid state. After empirical calibration, this acoustical signature allows for a verification of the billet relative solid-to-liquid ratio prior to casting.

Basics of Semisolid Tooling Design Without properly designed tooling, the best raw material and reheat equipment will not guarantee a successful casting process. Unfortunately, much of the wisdom gained from decades of conventional high-pressure die casting and squeeze casting has no immediate applicability to semisolid metal casting; a number of things make the most common forms of thixotropic casting unique. The extremely low iron content of commonly thixocast aluminum-silicon alloys increases the prevalence of solder formation during die filling. For this reason, die surface modifications or special die lubricants are typically required to improve casting performance and tooling longevity. However, a wide array of surface-modification techniques and coatings are available. The most common techniques involve forms of physical vapor deposition, chemical vapor deposition, ion-and plasma-enhanced gaseous diffusion processes, or salt bath diffusion processes. Typically, the most successful coatings for semisolid casting are based on higher-temperature diffusion or chemical-reaction-based processes such as chemical vapor deposition, ferritic-nitrocarburizing, or salt bath diffusion. Physical vapor deposition coatings often lack the tenacity to withstand the extreme level of erosive wear common to semisolid casting. The reheating processes commonly used in commercial thixocasting environments produce reheated billet exhibiting heavy exterior oxide layers. This oxide layer must be managed during die filling to ensure excellent part quality, particularly for structural and safety components. Several oxide-stripping mechanisms have been developed to minimize the opportunity for oxide entrapment within the cast part. During injection, the basic oxide-stripping concept attempts to turn the billet inside out so that the oxide-laden material ends up in the biscuit. Implementation of the oxide-stripping mechanism sometimes leads to increased die complexity and extra expense during die construction. Another unique aspect of thixocasting is the low level of thermal energy available within the cast material at the time of metal injection. This limited heat content makes proper runner and gating design essential for ensuring optimum part quality. While this reduced heat content diminishes the thermal shock to the casting die and allows for decreased solidification times, it also makes part filling and consolidation more difficult. Runner concepts should use only one inlet gate per part, whenever possible, to help reduce the number of separate flow fronts that must rejoin during cavity

filling. Locating overflows where two or more flow fronts intersect can help to reduce incomplete welding defects caused by the low heat content of the cast material. A properly designed gate geometry and location will also help to reduce cast defects associated with the low heat content of the thixocast material. The gate must have sufficient cross-sectional area to allow for rapid cavity filling while maintaining laminar flow. Typically, inlet gates should be at least as thick as the mean wall thickness of the casting. Gating into the heavy part features allows for longer intensification times throughout the part and can improve overall part soundness. Interestingly, the viscous nature of a thixocast material allows for laminar flow at gate velocities as high as 5 m/s (16 ft/s) (Ref 12). By comparison, conventional casting processes must maintain gate velocities an order of magnitude slower to achieve laminar flow. Moreover, when developing gating schemes, the gate location should ensure that material will flow from the thickest section of the casting toward the thinnest section. This approach maintains laminar flow during filling and improves part consolidation. Often, overall part quality can be improved by gating directly into critical features—maximizing shrinkage feeding during intensification. If this is not possible, select a gate location that minimizes the flow distance to the critical features. The following examines how the gating concepts discussed previously may be applied to a real-world part. The cross section shown in Fig. 9 illustrates a segment of runner and the inlet gate into a thixocastable part. Notice how the inlet gate thickness is very similar to the average casting wall thickness, allowing for rapid filling and good intensification. While the part wall thickness does not strictly obey the optimum thick-to-thin filling criterion, it comes close to achieving this goal by minimizing abrupt cross-sectional changes that may negatively affect part filling and intensification. The inlet gate location minimizes the number of recombining metal fronts and allows for a relatively direct, unobstructed intensification path to the slightly heavier sections within the part. The geometry of this example part is quite simple, but application of these concepts to more complex parts will increase process robustness. The venting requirements of the thixocasting process are quite different from those developed for high-pressure die casting. Conventional die casting uses gate velocities between 30 to 50 m/s (100 to 165 ft/s) to help disperse the excess unvented air throughout the part, as well as to improve part surface finish. The viscous nature of a semisolid fluid front inhibits the entrapment of cavity air within the flowing material. Pockets of entrapped air generate backpressures that inhibit part consolidation and the rejoining of fluid fronts in the final areas of the cavity to fill. The use of conventional vent area determination equations, such

Fig. 9

Sketch illustrateing some key design concepts used in thixocasting. Notice the thick gate and runner cross sections feeding the part. Also, the part wall thickness conforms fairly well to the thick-to-thin filling ideal, while the gate has been located to minimize the potential for recombining metal fronts.

as the one shown subsequently, results in inadequate venting of the thixocasting tool: Vent area ¼

Volumetric flow rateðm3 =sÞ 200 m=s

Practical semisolid casting experience has shown that typical semisolid vent cross-sectional areas exceed the standard recommendation by four or more times. In application, thixocasting vents are as thick as 0.0635 cm (0.025 in.) at their thinnest and run from the cavity to atmosphere through the holder block. The high solid fraction of the cast material makes it possible to use large vent cross sections without metal escaping from the casting die. By comparison, high-pressure die-casting vents are typically no thicker than 0.018 cm (0.007 in.). This vent restriction allows air escape but keeps the molten metal within the die.

770 / Semisolid Casting The more sophisticated venting tools used in conventional die casting can also be used in thixocasting. Power vents, also know as chill or waffle vents, can be used to increase the venting area while limiting the potential for metal escape from the tool. Similarly, vacuum venting can be employed to help reduce the backpressure experienced in the last areas to fill. No venting technology is without drawbacks, so it is important to select the best method for a given application. Power vents require periodic cleaning and add additional cost to the die. Vacuum-assisted venting requires additional maintenance, since the vacuum system ingests all the effluent of the casting process after the die closes. This maintenance drives additional machine downtime. Even with their drawbacks, the application of these methods may enhance overall part quality and machine utilization. The life of thixocasting dies is generally significantly longer than conventional high-pressure casting dies. With conventional die-cast tooling, heat checking and gross cracking are the two predominant failure modes. Both heat checking and gross cracking are the result of the cyclic high and low temperature extremes experienced by the mold during cavity filling and die lubrication. Over time, metal injected at high temperatures can temper the mold surface, reducing its resistance to crack formation and propagation. This tempering, combined with the thermal shock of spraying the casting die with water-based die lubricants, contributes to the deterioration of the casting mold. By comparison, the thermal energy transmitted to the mold surface during injection of a thixocast material is approximately 60% less than that of the same alloy conventionally cast. A typical aluminumsilicon thixotropic material is cast when it is between 40 to 60% solid; therefore, it has a drastically reduced percentage of its latent heat of fusion, as well as a lower superheat, available to contribute to the die surface. As a result, thixocasting dies typically require additional internal heating (via hot oil passages or electrical resistance heater cartridges) to obtain surface temperatures similar to those seen in conventional high-pressure casting processes. The predominant mode of die deterioration for a thixocasting die is parting-line erosion near the final places to fill. This erosion is the result of high material viscosity, extremely short intensification rise times, and high intensification pressures. Features experiencing longer durations of fluid impingement, such as cores directly across from the gate, may also experience erosive wear. This form of erosive wear can be managed by proper selection of core materials and surface modifications. The shot life of a typical thixocasting die will exceed that of a similar conventional die-cast tool by 150 to 200% or more. While not typical, semisolid casting dies have occasionally exceeded the 1 million shot mark, a level unheard of in conventional die-cast tooling for significant structural components.

In summary, the runner and gating design should address the following:
Limit the number of gate inlets to minimize

rejoining metal fronts.


Select inlet gate locations for thick-to-thin

flow—flow should progress from the thickest sections to the thinnest sections.
Gate thickness should be close to nominal part wall thickness to ensure rapid laminar filling and adequate shrinkage feeding.
Overflow locations should coincide with rejoining metal fronts.
Increase the vent areas to ensure proper filling.

Casting Process and Parameters The machine and process requirements for successful semisolid casting are different from those found in high-pressure die casting. Figure 10 illustrates the differences between a conventional high-pressure die-casting injection profile and the thixocasting injection profile used to produce the same part. The ability of the thixotropic billet to maintain its shape until it is sheared allows for rapid acceleration of the metal down the shot sleeve, until it begins entering the runner system. With the metal entering the runner, the injection velocity is reduced so that laminar flow can be maintained during runner filling. Since the runner crosssectional area is larger than the gate area of the part, the injection velocity is again reduced just before metal enters the part, to maintain laminar flow during part filling. While developing an injection velocity profile, remember that

Fig. 10

laminar flow can be maintained at gate velocities as high as 5 m/s (16 ft/s) when casting thixotropic materials in the 40 to 60% solid range. By comparison, conventional die-casting injection profiles use low velocities to reduce air entrapment that the initial plunger displacement causes during sleeve filling. After the shot sleeve is completely filled, the injection velocity is increased until the part-filling velocity is achieved. The relative aggressiveness of the transition from the slow to fast injection velocities is dictated by the runner design and the part being produced. What is important to learn from Fig. 10 is that the injection profile used for thixocasting is nearly a mirror opposite of that used for conventional high-pressure die casting. Another set of casting parameters that are significantly different in thixocasting are the intensification pressure and rise time. Figure 11 shows a typical high-pressure die-casting and a thixocasting intensification profile. The pressures applied during the thixocasting process are much more aggressive than those typically seen in standard high-pressure die casting. The low heat content and viscous nature of the thixocast material make it possible, and actually require, the application of pressures in this manner in order to achieve the highest-quality castings. Intensified metal pressures in the range of 1000 to 1500 bar (14,500 to 22,000 psi) or more are not uncommon for the thixocasting process. Using similar pressure profiles in highpressure die casting would typically lead to dangerous levels of metal ejection from the mold. While no casting is truly porosity-free, thixocast parts created using these high pressures typically have porosity levels low enough to avoid detection using conventional x-ray

Contrast between high-pressure die-casting (HPDC) injection velocity profile and thixocasting injection velocity profile

Thixocasting / 771

Fig. 11

Contrast between high-pressure die-casting intensification profile and thixocasting intensification profile

Table 1 Casting machine requirement comparison between thixocast and highpressure die-cast parts General information Required shot weight, kg (lb) Casting projected area, cm2 (in.2) Required metal pressure, bar (psi) Maximum hydraulic system pressure, bar (psi) Locking force required (opening force), kN (tonf) Semisolid metal specific data Required shot weight, kg (lb) Melt loss, % Required billet weight, kg (lb) Length-to-diameter ratio Billet diameter, mm (in.) Tip diameter, mm (in.) Tip area, cm2 (in.2) Required injection force, kN (tonf)

3.175 (7.0) 581 (90.0) 1,000 (14,504) 172.5 (2,502) 5,806 (653)

3.18 (7.0) 15 3.74 (8.25) 2.0 to 1 95.3 (3.75) 104.8 (4.125) 86.2 (13.36) 862 (97)

High-pressure die-casting specific data Tip diameter (based on 50% full sleeve), 76.2 (3.0) mm (in.) Tip area, cm2 (in.2) 45.6 (7.07) Required injection force, kN (tonf) 456 (51)

analysis techniques. Thixocast parts have been used in a variety of applications requiring low leakage rates, such as master brake cylinders and automotive fuel rails. In general, thixocast aluminum-silicon alloy parts exhibit the same structural integrity as permanent mold castings. Compared to permanent mold castings, the thixocast component will be nearer net shape and will often require less expensive T5 heat treatments to achieve properties similar to T6 heat treated permanent mold parts of the same alloy. Casting Machine Requirements. The methodology used to determine the appropriate machine size for a thixocasting project is slightly different from that used for high-pressure die casting. For thixocasting, the most important die-casting machine criterion is the ability to achieve high levels of intensified metal pressure. Another beneficial machine

characteristic for semisolid casting is closedloop injection profile control. The injection profile of the machine can be tailored to maintain laminar flow throughout the filling process. The machine selection methodology is now discussed in detail. The ability of a die-casting machine to deliver the intensification pressure is driven, in part, by the selection of the plunger diameter. Before the plunger diameter can be selected, the total shot weight and the length-to-diameter ratio of the unheated billet must be understood. The reheated billet target percent solid, the billet reheating and slug delivery method, and the available die-casting machine injection stroke factor into the selection of an appropriate billet diameter. Typically, billet length-to-diameter ratios range from 2 to 2.5:1 when calculated using unheated billets. This ratio helps to maintain slug stability during reheating with vertical orientation induction systems and allows the billet to readily fit within the stroke constraints of the die-casting machine. It is important to remember to factor in the melt loss that occurs during the induction reheating when selecting the overall billet weight required. With these factors in mind, the engineer can calculate the billet diameter based on the total shot weight. With the billet diameter determined, the plunger diameter can be selected. Typically, the plunger diameter is approximately 6.5 to 10 mm (0.26 to 0.40 in.) larger than the unheated billet diameter; this extra space improves billet loading into the shot sleeve. This size differential is particularly important when using vertical induction systems, because the billet undergoes a slight increase in diameter near its base during reheating. Unfortunately, this increased tip size reduces the ability of the die-casting machine to achieve higher intensified metal pressures. The following examines the die-casting machine shot end capabilities necessary to cast

a 3.2 kg (7.1 lb) shot having 580 cm2 (90 in.2) of projected area using 1000 bar (14,500 psi) of intensified metal pressure. The use of 1000 bar (14,500 psi) in this example should be considered the lower limit for determining the machine capability. The use of higher intensification pressures is not uncommon when casting products with geometries that are not specifically tailored to semisolid casting or when exceptional leak tightness is required. Table 1 compares the die-casting injection and locking force necessary to produce the hypothetical part using high-pressure die-casting and thixocasting methods. The hydraulic system pressure used in Table 1 was chosen because it is indicative of the maximum recommended system pressure found on the vast majority of die-casting machines worldwide. The injection force requirements for the two processes are the critical details to be gleaned from this table. It can be seen that the thixocast process requires significantly more injection force to achieve the target metal pressure, even though the opening force exerted by the two processes is identical. Therefore, the selection of a die-casting machine for a thixocasting project will focus more on shot end capability than on locking force. If machines are selected based on their locking force, additional multiplication cylinders must be integrated to upgrade the shot end performance. Machines selected to meet the injection force criterion without the use of additional multiplication technology will be significantly larger due to the additional locking force capabilities inherent in their design. In either case, additional capital considerations are required when compared to the standard high-pressure die casting of the same part. Since the fluidity of a thixotropic material is strongly dependent on the sheer it experiences, full control of the injection profile is a useful tool for optimizing a thixocasting process. Die-casting machines with closed-loop shot end control allow a variety of injection velocities and positions to be programmed. While this level of control is not necessarily as important as the ability to reach high levels of intensified pressure, the ability to maintain laminar flow fronts during runner and cavity filling will reduce scrap rates. This accuracy is particularly important since the cost of thixocasting scrap is so high relative to conventional die casting. Several die-casting machine manufacturers have developed equipment specifically tailored to thixocasting. These machines have tremendous injection force and good shot velocity control. Typically, these thixocasting machines incorporate larger injection cylinders with additional hydraulic multiplication systems, as well as thicker platens and larger-diameter tie bars to accommodate the high injection forces. Due to the extra material costs and engineering necessary to achieve the higher injection forces and velocity control, these specialized thixocasting machines are more expensive than conventional die-casting machines.

772 / Semisolid Casting REFERENCES 1. J. Jorstad, Practical Considerations of Processing, Science and Technology of SemiSolid Metal Processing, A. de Figueredo, Ed., North American Die Casting Association, 2001, p 4–1 to 4–33 2. M.C. Flemings, Structure Control in Cast Metals, Proceedings of the First Army Materials Technology Conference: Solidification Technology, Oct 1972 (Portsmouth, NH), J.J. Burke, M.C. Flemings, and A.E. Gorum, Ed., Metal and Ceramics Information Center, 1972, p 10–14 3. J.L. Jorstad, M. Thieman, and R. Kamm, SLC, The Newest and Most Economical Approach to Semi-Solid Metal (SSM) Casting, Proceedings of the Seventh S2P: Advanced Semi-Solid Processing of Alloys and Composites, Sept 2002 (Tsukuba, Japan), National Institute of Advanced Industrial Science and Technology, Japan Society for Technology of Plasticity, 2002, p 701–706 4. A. Kraly, SAG, personal correspondence 5. M.P. Kenney, J.A. Courtois, R.D. Evans, G.M. Farrior, C.P. Kyonka, A.A. Koch,

6. 7.

8.

9.

and K.P. Young, Semisolid Metal Casting and Forging, Casting, Vol 15, Metals Handbook, 9th ed., ASM International, 1998, p 327–338 C.A. Tudbury, Basics of Induction Heating, Vol 1, 1960, p 1–11 S. Brown, P. Kumar, and C.L. Martin, Exploiting and Characterizing the Fundamental Rheology of Semi-Solid Materials, Proceedings of the Second International Conference on the Processing of SemiSolid Alloys and Composites, June 1992 (Cambridge, MA), TMS, 1992, p 183–192 S. Brown, P. Kumar, and C.L. Martin, Flow Behavior of Semi-Solid Alloy Slurries, Proceedings of the Second International Conference on the Processing of Semi-Solid Alloys and Composites, June 1992 (Cambridge, MA), TMS, 1992, p 248–262 A. Alexandrou, Rheological and Numerical Simulations of Flow of Semi-Solid Suspensions, Science and Technology of SemiSolid Metal Processing, A. de Figueredo, Ed., North American Die Casting Association, 2001, p 5–1 to 5–28

10. R. Cremer, A. Winkelmann, and G. Hirt, Sensor Controlled Induction Heating of Aluminum Alloys for Semi-Solid Forming, Fourth International Conference on SemiSolid Processing of Alloys and Composites, June 1996 (Sheffield, England), Dept. of Engineering Materials, University of Sheffield, 1996, p 159–164 11. A. Scho¨nbohm, H.M. Ritt, H. Rake, and D. Abel, Temperature Estimation and Control of Inductively Heated Aluminum Billet, Proceedings of the Seventh S2P: Advanced Semi-Solid Processing of Alloys and Composites, Sept 2002 (Tsukuba, Japan), National Institute of Advanced Industrial Science and Technology, Japan Society for Technology of Plasticity, 2002, p 299–304 12. S.P. Midson, L.E. Thornhill, and K.P. Young, Influence of Key Process Parameters on the Quality of Semi-Solid Metal Cast Aluminum Components, Fifth International Conference on Semi-Solid Processing of Alloys and Composites, June 1998 (Golden, CO), Colorado School of Mines, 1998, p 181–188

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 773-776 DOI: 10.1361/asmhba0005274

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Rheocasting Q.Y. Pan, SPX Corporation Diran Apelian, Worcester Polytechnic Institute John Jorstad, J.L.J. Technologies, Inc.

RHEOCASTING is one of the three methods of semisolid metal (SSM) processing (see Fig. 5 in the preceding article, “Semisolid Casting: Introduction and Fundamentals,” in this Volume). In the rheocasting route, one starts from the liquid state, and the thixotropic slurry is formed directly via thermal management of cooling and solidification. A slurry is formed having the appropriate globular structure and fed directly into a die cavity. Rheocasting is currently the favored SSM route for processing aluminum because there is no cost premium for a billet precursor, and whatever trim or scrap generated in the process can be recycled into new product without first being turned into billet. It was once thought that producing appropriate semisolid feed stock could only be accomplished by cooling liquid alloy into the two-phase region and fragmenting dendrites as they formed to produce slurry having a globular microstructure. However, more recent research has led to the understanding that properly managed thermal conditions in the melt can accomplish the same structure. If copious nucleation of alpha grains is somehow provided and if those nuclei are not allowed to grow beyond a certain point or to melt back into the bulk liquid, then an ideal semisolid globular structure develops directly from the melt. This concept is the genesis of various processes and methodologies to generate semisolid slurries from the liquid state. The concept relies on controlled nucleation and growth, as opposed to the previous (thixocasting) theory in which a dendritic structure was modified to globular via

Fig. 1

shear forces. The recent emphasis on reducing the cost of semisolid processing has provided the driving force to develop several rheocasting variations. These include UBE’s New Rheocasting (NRC) (Ref 1), Massachusetts Institute of Technology’s (MIT’s) Semisolid Rheocasting (SSR) (Ref 2), THT’s Subliquidus Casting (SLC) (Ref 3), Alcan’s Swirl Enthalpy Equilibration Device (SEED) (Ref 4), Brunel’s RheoDiecasting (Ref 5), Mercury-Marine’s Slurryon-Demand process, and Worcester Polytechnic Institute’s Continuous Rheoconversion Process (CRP)(Ref 6). The following is a brief overview of those process variations that have been commercialized or are under final development.

Methods New Rheocasting (NRC) process was patented by UBE Industries, Ltd., and has been commercialized in Europe and the United States. The process relies on thermal treatment of alloy melt to generate SSM slurries. As illustrated in Fig. 1, the process involves several steps: 1) pouring low superheat melt to a specially prepared and coated crucible, 2) overcooling the melt and then reheating the melt to an appropriate semisolid processing temperature, and 3) transferring the slurry to a shot sleeve of a vertical casting machine and casting into a final product. Semisolid Rheocasting (SSR) is a process developed by MIT and licensed to IndraPrince, which was acquired by Buhler in 2006. In this

Schematic of UBE’s New Rheocasting (NRC) process. Source: Ref 1

process, a stirrer that also provides the cooling action is inserted into an alloy melt held a few degrees above the liquidus temperature. After some seconds of stirring, the melt temperature decreases to a value that corresponds to a fraction solid of only a few percent, and the stirrer is withdrawn, as illustrated in Fig. 2. The combined stirring/cooling action results in copious nucleation of the primary phase and distribution of nuclei in the alloy melt. A globular structure of an appropriate fraction solid can be achieved during subsequent cooling. The Subliquidus Casting (SLC) process was developed by THT Presses, Inc., based on their vertical die casting machines. Figure 3 illustrates the process concept. In the SLC process, a well-grain-refined melt is introduced into the shot sleeve at a temperature only a few degrees above liquidus temperature. The primary phase initially forms as tiny rosette grains that then spheroidize and ripen into a globular slurry, which has a fraction solid of approximately 0.5 as it cools to near the eutectic temperature. The slurry is then injected through a gate plate into the die. The gate plate concept provides an opportunity for positioning gates to minimize both flow and solidification shrinkage feeding distances. The large shot diameter, short shot stroke, and gate plate feature make the THT equipment well suited to a slurry form of semisolid processing. The Swirled Enthalpy Equilibration Device (SEED) was developed by Alcan. As illustrated in Fig. 4, the process involves two steps: 1) Transfer molten metal to a vessel, where the

774 / Semisolid Casting

Fig. 2

Schematic of Massachusetts Institute of Technology’s Semisolid Rheocasting (SSR) process. Source: Ref 2

injected into the die cavity. Figure 6 shows the process during production. Recent advances include a method of producing and delivering high-liquid-fraction slurries of alloys that do not easily form a semisolid mixture, such as near-eutectic aluminum-silicon alloys. The Continuous Rheoconversion Process (CRP) is a novel process developed at Metal Processing Institute/Worcester Polytechnic Institute. The process is based on a passive liquid- mixing technique in which the nucleation and growth of the primary phase are controlled using a specially designed reactor. The reactor provides heat extraction, copious nucleation, and forced convection during the initial stage of solidification, thus leading to the formation of globular structures. Experimental results with various commercial aluminum and magnesium alloys indicated that the CRP is highly effective for the manufacture of high-quality semisolid feed stock (Ref 7, 8). The process has recently been scaled up for commercial applications. Figure 7 shows the original concept of the process, and Fig. 8 shows typical settings of optimized/scaled-up CRP reactors mounted in a horizontal die casting machine and a vertical die casting machine, respectively, to generate SSM slurries. The advantages of the process include process simplicity, low cost, a wide process window, the feasibility of recycling scrap metal within the process flow stream, and so on. In conclusion, although thixocasting has been commercialized for many years, the driving force to reduce process cost has recently shifted research and development efforts toward slurryon-demand processes. There are various rheocasting-based SSM processing methods, each different and yet all of them yielding a globular (nondendritic) microstructure suitable for semisolid processing. The process that proves easiest to adapt, is least cumbersome, least complicated, and most manageable for any given facility and product will prevail.

REFERENCES

Fig. 3

Schematic of THT Presses’ Subliquidus Casting (SLC) process. Source: Ref 3

melt is cooled to a solid fraction between 0.3 and 0.4. During the cooling process, the vessel is swirled to ensure that nuclei of the primary phase generated on the walls of the vessel are distributed uniformly in the melt. 2) Drain away the excess liquid to form a compact, self-supporting slug for subsequent forming operations. The Rheo-Diecasting process was developed by Brunel University. As shown in Fig. 5, a twin-screw extruder is employed to provide a shearing and mixing action of molten metal. Slurries with fine, globular structure are produced and directed to the shot sleeve of a die casting machine. For relatively large components (>4 kg, or 9 lb), a slurry accumulator is needed.

A blade-type stirrer is fitted in the accumulator to prevent particle agglomeration and to maintain slurry uniformity. The Slurry-on-Demand process is a rheocasting process originally developed by AEMP and subsequently acquired by Mercury Marine. In this process, molten alloy is ladled from a furnace into a slurry-producing unit, where the melt is simultaneously stirred via a magnetic field and cooled in a controlled manner to form a semisolid billet with a desired solid fraction (typically approximately 50%). Once the billet is made, it is ejected from the slurry-producing unit and transferred immediately into the cold chamber of a die casting machine, where it is

1. Method and Apparatus for Shaping Semisolid Materials, European patent EP 0 745 694 A1, UBE Industries Ltd., 1996, p 117 2. J.A. Yurko, R.A. Martinez, and M.C. Flemings, “SSR: The Spheroidal Growth Route to Semi-Solid Forming,” Eighth International Conference on Semi-Solid Processing of Alloys and Composites (Limassol, Cyprus), 2004 3. J.L. Jorstad, “Fundamental Requirements for Slurry Generation in the Sub-Liquidus Casting Process and the Economics of SLC Processing,” Eighth International Conference on Semi-Solid Processing of Alloys and Composites, (Limassol, Cyprus), 2004 4. D. Doutre, J. Langlais, and S. Roy, “The SEED Process for Semi-Solid Forming,” Eighth International Conference on SemiSolid Processing of Alloys and Composites (Limassol, Cyprus), 2004

Rheocasting / 775

Fig. 6

Fig. 4

Mercury Marine’s Slurry-on-Demand process in action. Courtesy of Mercury Marine

Schematic of Alcan’s Swirled Enthalpy Equilibration Device (SEED) process. HPDC, high-pressure die casting. Source: Ref 4

Fig. 7

Fig. 5

Schematic of Brunel University’s Rheo-Diecasting process. Source: Ref 5

Schematic of the original Continuous Rheoconversion Process concept

5. Z. Fan, S. Ji, G. Liu, and E. Zhang, “Development of the Rheo-Diecasting Process for Mg-Alloys and Their Components,” Eighth International Conference on Semi-Solid Processing of Alloys and Composites, (Limassol, Cyprus), 2004

776 / Semisolid Casting 6. Q.Y. Pan, M. Findon, and D. Apelian, “The Continuous Rheoconversion Process (CRP): A Novel SSM Approach,” Eighth International Conference on Semi-Solid Processing of Alloys and Composites, (Limassol, Cyprus), 2004 7. M. Findon, “Semi-Solid Slurry Formation via Liquid Metal Mixing,” M.Sc. thesis, Worcester Polytechnic Institute, 2003 8. Q.Y. Pan, P. Hogan, and D. Apelian, “Optimization of Commercial Alloys for Semi-Solid Processing,” 110th AFS/NADCA Congress, 2006

Fig. 8

Typical settings of scaled-up Continuous Rheoconversion Process reactors mounted in a horizontal die casting machine (left) and a vertical die casting machine (right) to generate semisolid metal slurries

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 777-782 DOI: 10.1361/asmhba0005275

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Thixomolding* Robert D. Carnahan and Raymond F. Decker, Thixomat, Inc.

THIXOMOLDING is a method of molding thixotropic semisolid magnesium alloy pellets in a machine that resembles an injection molding machine in physical appearance and operation (see Fig. 6 in the article “Semisolid Casting: Introduction and Fundamentals” in this Volume). From the pioneering research at Massachusetts Institute of Technology by Fleming, Mehrabian, and Joly (Ref 1, 2), several methods of semisolid processing of metals have been developed with continuing research and development (Ref 3). In contrast to billet and slurry casting techniques, Thixomolding is a semisolid injection molding process (Ref 4) that has the virtue of permitting the production of parts over a wide range of solid fractions, fs, nominally of the order 0.05 to 0.60. The thixotropic characteristics are developed in situ as parts are being produced. Starting granules are subjected to mechanical shear as they are heated into the region between liquidus and solidus and accumulated for injection molding in a self-contained and one-step unitized machine. Obviating any external handling or melting, the alloy is protected from high-temperature oxidation by an argon environment that further allows 100% recycling of sprues, gates, runners, and scrap without secondary refining.

Process Description Since the invention of the thixomolding process by Dow Chemical Company (Ref 5, 6), the process and equipment have been improved, refined, and commercialized by Thixomat, Inc., its machine-building licensees, and its molder licensees. The process obviates external melting furnaces and other elements of the die-casting practice, as well as the requirement for a specially prepared semisolid metal processing (SSP) billet used in many other SSP processes. Constituting the ability to operate a foundry-free environment, Thixomolding addresses the contemporary issues of energy and materials conservation, recyclability, waste disposal and management, safety, and environment in costeffective ways. *Thixomat, Inc.

In contrast to many other SSP processes that require a relatively costly precast billet as a starting material, the process uses feedstock granules in particle sizes of 2 to 5 mm (0.08 to 0.2 in.) as chipped from as-cast ingots. The particulate feedstock is conveyed from a storage container to a volumetric feeder that automatically controls the level of material in the barrel feed throat. Argon gas, introduced into the feed throat, displaces ambient air in the system and protects the magnesium alloy from high-temperature oxidation during processing. Air is prevented from entering the barrel through the injection nozzle by means of a solid plug of magnesium formed at the conclusion of each shot. The plug is ejected upon shot initiation and is entrapped in the sprue post, wherein it is conveniently recycled. Vacuum-assisted molding is simplified by the plug, which also serves as a seal on the inlet side of the die. The desired barrel temperature profile, preselected and programmed into the programmable logic controller (PLC) system, permits heating the alloy to the desired temperature between the liquidus and solidus. The rotation of the screw provides mechanical mixing and shear forces that result in the formation of the classic thixotropic mixture of primary aprim solid particles dispersed in liquid. The screw action prevents local solids agglomeration in the accumulation zone and nozzle. The shear action of the screw ensures that any solid aprim phase that should precipitate in the machine takes on a spherical shape (Ref 7). The micrographs in Fig. 1 demonstrate the wide range of microstructures and solid-to-liquid ratios that are produced in this process by control of temperature. The white aprim solid-phase particles in Fig. 1(a–c) are approximately 30 to 55 mm (1200 to 2200 min.) in diameter at 5 to 18% solids and increase in size to approximately 38 to 70 mm (1500 to 2800 min.) with 25 to 35% solids (Ref 8, 9). The dark regions in the figures had been liquid in the machine and had solidified in the mold to constitute the proeutectic apro phase of 3 to 5 mm (120 to 200 min.) grain size surrounded by a divorced eutectic of aeut and b-phase (Fig. 1d,e). In these examples of AZ91D, the process has been controlled

at temperatures ranging from 560 to 590
C (1040 to 1094
F). Researchers (Ref 10, 11) have established experimentally that the thixotropic characteristics of AZ91D, AM60, and AM50 obtain an extended range of solid fractions in studies of the shear rate dependence of apparent viscosity for these alloys (Fig. 2– 4). The authors confirmed that magnesium alloys exhibit a pseudoplastic rheological behavior analogous to that of the tin-lead alloy system under isothermal conditions. Upon accumulating an appropriate thixotropic shot charge, the rotation—at approximately 150 rpm—and retraction of the screw are automatically interrupted, and the screw is accelerated forward under PLC control, injecting the SSP charge into the die cavity. Fast fill times of 0.02 to 0.05 s are critical to filling out the die with fast-freezing magnesium alloy. Therefore, screw shot velocity must be 1.8 m/s (5.9 ft/s) or more for optimum mechanical properties (Tables 1, 2). The screw thrust applies a pressure of approximately 70 to 105 MPa (10 to 15 ksi) to pack the die. The screw then resumes its previous action, preparing and accumulating the charge for the next shot as the die clamp is being opened and the molded part is being removed. Production cycle times depend on the size of shot injected. Cycles range from 10 s for 100 ton machines to 20 s for 280 ton machines to 30 s for 650 ton machines and up to 60 s for 1000 ton machines. More than 400 commercial machines are fabricated and supplied by Japan Steel Works (JSW) and Husky Injection Systems with dieclamping capacities ranging from 75 to 1600 metric tons, capable of molding parts as small as 15 g (0.03 lb) and as large as 9 kg (20 1b), with projected areas larger than 3750 cm2 (580 in.2). Commercial parts have been molded as thin as 0.3 mm (0.012 in.) and as thick as 25 mm (1 in.). Individual shot sizes, that is, volume of thixotropic charge, are controlled by the distance of retraction of the screw. The temperature profile, established by means of individually controlled electrical resistance band heaters, is very stable due to the relatively large thermal mass of the

778 / Semisolid Casting

Fig. 1

Fig. 2

Steady-state apparent viscosity, Z, versus g for AZ91D (various isothermals)

Fig. 3

Steady-state apparent viscosity, Z, versus g for AM50 (various isothermals)

Fig. 4

Steady-state apparent viscosity, Z, versus g for AM60 (various isothermals)

Microstructures of semisolid Thixomolded AZ91D at solid aprim levels of (a) 4.4%, (b) 28%, and (c) >60%. (d) Distribution of aprim, apro, aeut, and beut. (e) Transmission electron micrograph of aeut and b

barrel and screw compared to the small quantity—approximately five shot equivalents—of alloy in process and its short residence time in the barrel. Measured shot-to-shot variation is less than  þ2
C (4
F) (Ref 13). The small inventory of metal being processed also leads to unusual system flexibility in startup and shutdown modes as well as rapid exchange of alloy systems. Again, as in injection molding of plastics, with a rapid changeover system and preheated dies, the machine can be quickly purged and a new alloy introduced for production of a different part within minutes. A cold startup requires less than 1 h to power up and produce parts, while shutdown merely requires turning the power down to an idle mode or off. As an example, during a

commercial production run, only the initial four shots were scrapped preceding a campaign of 1500 parts. All 1500 met customer specifications and were accepted.

Emerging Developments At the turn of this century, the Thixomolding process was in a relative state of commercial infancy. Having been introduced in the mid1990s, it is replacing die-cast magnesium and injection-molded plastic in myriad products in the marketplace. Many newly designed systems can be produced only by the Thixomolding process. Recent process developments have seen

the introduction of direct injection, hot sprues, hot runners, new magnesium alloys by the blending of solid feedstocks, recycling, and

Thixomolding / 779 Table 1 Effect of Thixomolding shot velocity of screw on mechanical properties of AZ91D Shot velocity

Ultimate tensile strength

Shot velocity Alloy

m/s

ft/s

MPa

ksi

Elongation, %

1.2 1.5 1.72

3.9 4.9 5.64

245 255 275

36 37 40

1.5 2.0 6.0

Source: Ref 12

Fig. 5

Table 2 Effect of Thixomolding shot velocity of screw and aluminum content on tensile properties

AM60 AM60 AZ71 AZ71 AZ80 AZ80 AZ91D AZ91D

Yield strength

m/s

ft/s

MPa

ksi

MPa

ksi

Elongation, %

6 6 7 7 8 8 9 9

1.8 2.2 1.8 2.2 1.8 2.2 1.8 2.2

5.9 7.2 5.9 7.2 5.9 7.2 5.9 7.2

121 121 ... 131 144 144 152 158

17.5 17.5 ... 19.0 21.0 21.0 22.0 22.9

241 251 ... 252 248 257 247 259

35.0 36.4 ... 36.5 36.0 37.3 35.8 37.6

12.5 15.9 9.2 10.7 7.2 8.6 5.5 6.6

Japanese Steel Works Thixomolded personal computer housings with and without hot runners. Source: Ref 14

joining by friction welding. The process is further enhanced by advances in flow modeling and tool design. Hot Sprues and Hot Runners. As widely practiced in tool design and manufacture in the plastic injection molding industry, bulky sprues, gates, and runner systems have been eliminated by JSW and Husky Injection Systems. This technology has been reduced to practice in recent commercial magnesium injection molding. For the most part, this has been necessitated by the need to develop methodologies such as center gating to control shrinkage, to optimize dimensional control, and to improve cosmetics (e.g., to avoid flow line boundaries). Due to the similarities between Thixomolding and plastic injection molding, namely the molding of a

Ultimate tensile strength

Aluminum, %

pseudoplastic semisolid slurry, success has been achieved in adapting both hot sprue and hot runner systems (Fig. 5). The hot metal path into the die cavity remains filled with hot slurry between shots by sealing the nozzle closer to the part cavity, thus obviating the need to freeze off a relatively large sprue. This increases metal yields to 80% or more and speeds cycle times. The economic advantages of both are evident. Hot runner systems, while having been demonstrated on multicavity tools for cell phone parts and on large automotive parts, continue to be the subject of advanced manufacturing development. Recently, under DOE USCAR sponsorship, a large ford “shot gun” automotive component has been molded on a 1000 ton Thixomolder using direct injection, that is, no sprue and

runner, achieving a 96% total yield (parts out versus chips in). Alloy Design by Blending and Thixoblending. The injection molding of plastics has benefited greatly from the relative ease of modifying properties with additives. Additives are used for improving mechanical properties, thermal conductivity, and ultraviolet stabilization, as well as for economics. Modification is typically accomplished by dry blending and compounding. In contrast to metal foundries, modifying and casting alloys with the objective of improving mechanical and thermal properties, for example, strength, fatigue, and creep resistance, is a time-consuming undertaking. Thixomolding, due to its relative process simplicity versus foundry practice, is amenable to dry blending of primary master alloy feedstock with a suitable additive either prior to or during introduction to the feed hopper. This step, termed Thixoblending (Thixomat, Inc.), has permitted the blending of AM60 and AZ91D to produce two other alloys: AZ71 and AZ81. As seen in Fig. 6 and 7 and Table 3, the yield strengths have been tailored inversely from 121 to 158 MPa (17.5 to 22.9 ksi) and the elongations from 6 to 16% among the four alloys. The same technique has been practiced to add calcium-containing alloy at various levels to the base alloy. In addition to the advantages of blending existing commercial alloys to achieve new compositions, Thixoblending offers the agility to rapidly change compositions while simultaneously performing tool cavity changeover, also reducing inventory of alloy stocks. Other molded alloys of commercial interest include those shown in Table 4. Effect of Scrap Recycling on Mechanical Properties. Table 5 summarizes results from two independent studies of the mechanical properties of recycled AZ91D. In both studies, the source of material was sprues, gates, runners, and scrapped Thixomolded parts. The scrap was reground using techniques similar to that for virgin feedstock. No remelting or refining steps were used. The data clearly demonstrate the ease and utility of recycling Thixomolding scrap (Ref 15, 16). The preservation of tensile properties—especially the lack of degradation in percent elongation—clearly indicates the absence of additional oxide particles and other inclusions, which can

780 / Semisolid Casting Table 4 Compositions of magnesium alloys that have been Thixomolded Nominal composition (magnesium balance)

Alloy

AZ91D AZ81 (Thixoblended) AZ71 (Thixoblended) AM60 AM50 AM30 AE42 AJ62 AJ52 AM-Lite MRI 153 AS41 JSW AC ZAC8506 ZAC8506/AZ91 (Thixoblended) Melram

9Al-0.7Zn 8Al-0.5Zn 7Al-0.3Zn 6Al 5Al 3Al 4.3Al-0.3Zn-2.9RE 6Al-2Sr 5Al-2Sr 12Zn-4Al-0.2Ca-0.3Mn 8.6Al-0.6Zn-0.9Ca-0.3Mn 4Al-2Si 6Al-1–5Ca 5Al-8Zn-0.6Ca 6–8Zn-3–5Al-0.2–0.4Ca 6Zn-1.2Cu-0.8Mn-12 vol% SiC

Table 5 Tensile properties of Thixomolded recycled AZ91D

Fig. 6

Strength of Thixoblends as a function of blend composition and aprim solid fraction

Table 3 Mechanical properties of Thixoblended alloys at 4 to 18% solids Yield strength

Elongation of Thixoblends as a function of blend composition and aprim solid fraction

dramatically affect ductility in both virgin alloys and directly recycled scrap.

Mechanical Properties and Microstructure The mechanical properties depend on porosity; volume fraction solid, faprim ; grain size of apro; and fs and size of eutectic intermetallic phases (e.g., b-Mg17Al12). Porosity derives from gases and shrink. Partially filled shot sleeves and accumulation zones can generate gas porosity in parts, as will turbulent flow in air-containing molds. High superheat above the liquidus and moisture in the feedstock encourage H2 pickup.

Ultimate tensile strength

Recycled, %

MPa

ksi

MPa

ksi

100(a) 100(b) 50(a) 50(b) 10(a) 10(b) 100(c)

166 167 170 167 165 162 ...

24.1 24.2 24.7 24.2 23.9 23.5 ...

260 254 262 263 259 256 296

37.7 36.8 38.0 38.1 37.6 37.1 42.9

Elongation, %

7.0 6.1 6.0 6.5 6.3 7.0 10.3

(a) and (b) Solid fractions faprim of <0.05 and 0.1, respectively. Source: Ref 15. (C) Carried out with faprim <0:02. Source: Ref 16

Ultimate tensile strength

Alloy

MPa

ksi

MPa

ksi

Elongation, %

AM60 base AZ71 blend AZ81 blend AZ91 base

121 131 144 158

17.5 19.0 21.0 22.9

252 257 255 256

36.5 37.3 37.0 37.1

16 12 8 6

Source: Ref 9

Fig. 7

Yield stress

Several process factors in Thixomolding reduce gas porosity compared to die casting. The accumulation zone is tightly packed by the screw action to avoid trapped gas. The more uniform fill front in the part expels air through the vents. Vacuum can be applied in approximately 1 s to further reduce gas porosity. The 80
C (144
F) lower melt temperature of Thixomolding minimizes H2 pickup. Feedstock moisture content is maintained below 0.2%, as can be afforded by heating of the hopper if one is operating in a humid environment. As for shrinkage porosity, the semisolid content and 80
C (144
F) lower metal temperature diminish the thermal cycle of the mold and thus the gross part shrinkage. Shrinkage porosity can be minimized by a combination of design, increasing the relative solid fraction of aprim, and controlling the location of residual liquid metal in the part. Reduction of porosity below 2% is very beneficial to the strength and ductility of Thixomolded AZ91D (Table 6). In the case of AS41B containing 11% aprim, an increase of Thixomolding shot velocity from 1.3 to 1.8 m/s (4.3 to 5.9 ft/s) reduced porosity

Table 6 Effect of bulk porosity on mechanical properties of Thixomolded AZ91D at 11.3 to 15.6% aprim Yield strength Porosity, % MPa

2.70 1.96 0.6

ksi

Ultimate tensile strength MPa

ksi

166 24.1 229 33.2 156 22.6 254 36.8 180 26.1 300 43.5

Elongation, % Reference

4.0 5.7 10.0

17 18 12

from 1.1 to 0.6% (Ref 19). Studies of die-cast AM60 (Ref 20) established the following quantitative correlation between tensile elongation, e, and total area fraction of all defects, fd: e ¼ eo ½1  fd n

(Eq 1)

where eo and n are empirical constants that depend on test temperature, casting process parameters, and alloy chemistry. Elongation correlated better with reduced gas and microshrink porosity on the fracture path than with bulk porosity. Significant reductions in porosity relative to die-cast counterparts (e.g., 50%) are typical from studies. The absence of an external melting step and processing within a protective argon atmosphere obviate the need for fluxes, preclude the formation of dross and oxides,

Thixomolding / 781 Table 7

Tensile properties of thixomolded magnesium alloys

Range of properties over range of percent aprim of <5 to 25 at shot velocities of 1.72 m/s (5.6 ft/s) or more Ultimate strength Alloy

AZ91D AS41A AM60 AM50 AJ52 MRI 153 Melram (metal-matrix composite) ASTM International specifications AZ91D AM60 AM50

ksi

MPa

ksi

Elongation, %

245–299 235 240–278 269 232 ... 218

35.5–43.4 34.1 34.8–40.3 39.0 33.6 ... 31.6

145–169 ... 120–150 140 144 ... 147

21.0–24.5 ... 17.4–21.8 20.3 20.9 ... 21.3

5–10 8 12–19 20 6 ... 4.5

230 210 220

33.4 30.5 31.9

150 125 130

21.8 18.1 18.9

3 10 8

and consequently minimize the presence of planar defects such as cold shuts and folded oxidized surfaces. With proper tool design, there is an absence of continuous porosity that permits the production of gastight hermetic enclosures, eliminating the necessity for a costly secondary resin-impregnation step common in aluminum die castings. The tensile properties are shown in Table 7 for Thixomolded AZ91D, AM60, AM50, and AS41, the most commonly used magnesium alloys. It is noted that AZ91D meets or exceeds the ASTM International die-casting specification at solid fractions of faprim ¼ 0:25 or lower. At higher solid fractions, heat treating is required to achieve maximum strength levels, a requirement discussed earlier for SSP aluminum alloys formed using the billet/preform process. Alloys AM50 and AM60 are likewise seen to exceed the ASTM International diecasting specifications when Thixomolded. The data tabulated were sourced from both independent investigations and round-robin studies of common batches of standardized test samples (Ref 13, 15–18, 21–26). Influence of aprim and apro. Using the selfcontained features of Thixomolding, the pseudoplastic thixotropic state is generated in situ, affording the unique opportunity to select the fraction of solids, faprim , varying from <0.05 to >0.6, by choosing the correct temperature profile for processing. Thixomolding generates solid aprim particles in the machine, which then are injected into the part. These particles of approximately 50 mm (2000 min.) diameter are lean in aluminum content, reporting to the part at approximately 2.4% Al in AM60 and 3.7% Al in AZ91 (Ref 9). Then, apro precipitates as the metal cools in the sprue, runners, and part, freezing at aluminum content above 6% in AM60 and above 9% in AZ91D, and with grain size of 3 to 5 mm (120 to 200 min.). Thus, the microstructure represents a composite of two a-phases of different grain size and solid-solution hardening. Therefore, the strength was related to the rule of mixtures of the two a-phases (Ref 25): salloy ¼ faprim saprim þ fapro sapro

Yield strength

MPa

(Eq 2)

where f is volume fraction, and s is strength. A term was added (Ref 9) representing the

Hall-Petch rule for grain size to provide for the mixture of a grain sizes and strength of each a-phase: saprim or apro ¼ so þ mBn Cn þ kd1=2

(Eq 3)

where so, m, B, C, n, and k are constants for each phase, and d is the grain size of each phase. Indeed, as predicted by this law of mixing, yield and tensile strengths with 0.044 to 0.185 faprim were higher than in the range of 0.28 to 0:352 faprim (Fig. 7). Therefore, in commercial practice, 15 to 20% solids is often chosen for parts with thicker sections and 5% for very thin sections. In fact, ASTM International specifications for AM60 and AZ91D are exceeded with up to faprim of 0.25 (Table 7). As the solid fraction is increased further, the apparent viscosity of the SSP slurry increases, tending to increase the uniform flow front behavior as the slurry enters the die. When Thixomolding with 30 to 60% solids with heavy sections, one can solution heat treat to homogenize the a and then age to recover some strength. Also, the increase in strength with shot velocity was attributed to refinement of apro grain size from 5 mm (200 min.) at 1.27 m/s (4.17 ft/s) to 3 mm (120 min.) at 1.72 m/s (5.64 ft/s) (Ref 12). Superior as-molded ductility was also found in the 0.044 to 0:185 f aprim interval. Effect of Eutectic Phases. Intermetallic eutectic phases form in the very final stages of solidification, as is the case for b-(Mg17Al12) with fb = 0.02 in AM60 and fb = 0.12 in AZ91D. The morphology is generated in a divorced eutectic reaction that tends toward elongated forms enveloping the apro particles. These hard and brittle b-particles are prone to crack according to the generic equation: scrack ¼ ðEg Þ1=2 =2r

(Eq 4)

where scrack is the cracking stress, Eg is the surface energy of b, and r is the size of the b-particles. As seen in Table 2, elongation decreases as b increases when the aluminum content is raised from 6 to 9%. Refinement of the b size is one of the factors in the increase of elongation with shot velocity. Since fracture initiates

in the apro/b regions and is fostered by more aluminum and more b, the shunting of additional aluminum to these regions as solids are increased could be another cause of the effects seen.

REFERENCES 1. R. Mehrabian and M. Fleming, Trans. AFS, Vol 80, 1972, p 175 2. P. Joly and R. Mehrabian, J. Mater. Sci., Vol 11, 1976, p 1393 3. S2P 2006, Ninth Int. Conf. on Semi-Solid Processing of Alloys and Composites, Sept 11–13, 2006 (Busan, Korea), Pusan National University 4. D. Ghosh, R. Carnahan, R. Decker, C. Van Schilt, P. Frederick, and N. Bradley, Proceedings of the International Conference on Magnesium Alloys and Their Applications, Dtsch. Ges. Materialkd., April 1992 (Garmisch-Partenkirchen, Germany) 5. R. Busk, U.S.Patents 4,694,881 and 4,694,882 6. N. Bradley, R. Wieland, W. Schafer, and A. Niemi, U.S. Patent 5,040,589 7. F. Czerwinski, Scr. Mater., Vol 48, 2003, p 327 8. F. Czerwinski, Metall. Mater. Trans. A, Vol 33, Sept 2002, p 2963 9. N. Nandy, R. Messing, J. Jones, T. Pollock, M. Walukas, and R. Decker, Metall. Mater. Trans. A, Dec 2006 10. D. Ghosh, R. Fan, and C. Van Schilt, in Third International Conference on Processing of SSM (Tokyo, Japan), 1994, p 85 11. R. Fan, D. Ghosh, and C. Van Schilt, Light Met., 1998 Me´taux Le´gers, Met. Soc. CIM, 1998, p 251 12. T. Tsukeda, K. Takeya, K. Saito, and H. Kubo, J. Jpn. Inst. Light Met., Vol 49 (No. 7), 1999, p 287 13. K. Kitamura, Sokeizai, Vol 36 (No. 5), 1995, p 1 14. T. Yamaguchi, Al, Serial 403, Light Metals Correspondence Al Co., Ltd., p 14 15. R. Carnahan, R. Kilbert, and L. Pasternak, 51st World Mg Congress (Berlin, Germany), IMA, 1994 16. T. Tsukeda and K. Saito, Keikinzoku, Vol 47 (No. 5), 1997, p 298 17. R. Carnahan, R. Hathaway, R. Kilbert, L. Pasternak, and P. Rohatgi, Light Met.’93, Me´taux Le´gers, Met. Soc. CIM, 1993 18. D. Ghosh, K. Kang, C. Bach, J. Roemer, and C. Van Schilt, Light. Met.’95, Me´taux Le´gers, Met. Soc. CIM, 1995 19. T. Tsukeda, K. Saito and H. Kubo, J. Jpn. Inst. Light Met., Vol 49 (No. 9), 1999, p 421 20. A. Gohkale, “Structural Cast Mg Development Project,” USCAR, U.S. DoE, 2006 21. R. Carnahan, R. Decker, R. Vining, E. Eldener, R. Kilbert, and D. Brinkley, Die Cast. Eng., May/June, 1996 22. K. Saito, 53rd World Mg Congress (Ube City, Japan), IMA, 1996

782 / Semisolid Casting 23. Nikkei Mech., Vol 1998 (No. 3), March 1998, p 522 24. S. LeBeau, Y. Yamamoto, and K. Sakamoto, SAE Technical Paper Series, 980087, Feb 23–26, 1998 (Detroit, MI) 25. R. Carnahan, Third International Conference on SSP of Alloys and Composites, June 13–15, 1994 (Tokyo, Japan) 26. R. Kilbert, P. Frederick, N. Bradley, R. Newman, and R. Carnahan, 16th Int. NADCA Congress and Exposition Trans., May 1991 (Detroit, MI)

 R. Carnahan, Magnesium and Magnesium 

  

SELECTED REFERENCES  R. Beals, S. LeBeau, O. Roberto, and

P. Shaskov, SAE Technical Paper Series, 2003-01-0181, March 3–6, 2003 (Detroit, MI)  J. Brown et al., Magnesium Properties and Applications for Automobiles, SAE, March 1993, p 27

 

Alloys, ASM Specialty Handbook, ASM International, 1999, p 90 P. Gregg, Beschichtung von Magnesium— Dekorative Anwendungen fur High–End Fahrzeuge, EFM, 14th Magnesium Automotive and User Seminar, Sept 28–29, 2006 (Aalen University of Applied Sciences, Germany) J. Hillis and S. Shook, SAE Technical Paper Series, 890205, 1989 T. Larek, Precis. Met., Aug 1998, p 27 S. LeBeau and R. Decker, Fifth International Conference on SSP of Alloys and Composites (Boulder, CO) Colorado School of Mines, 1998 S. Lebeau and M. Walukas, Plastics in Portable and Wireless Electronics, Society of Plastic Engineers, Jan 25, 1999 (Mesa, AZ) S. LeBeau, M. Walukas, R. Decker, P. LaBelle, A. Moore, and J. Jones, SAE

      

Technical Paper Series 2004-01-0137 (Detroit, MI), 2004 R. Macary, Arlington Plating Co., Palatine, IL, private communication, Dec 2006 A. Moore, C. Torbet, A. Shyam, J. Jones, D. Walukas, and R. Decker, Magnesium Technology 2004, TMS, p 263 K. Nakajima and K. Higuchi, in Proceedings of the 52nd Congress, IMA, 1995, p 15 I. Nakatsugawa, Japan Magnesium Association Seminars, 1998 Norsk Hydro data S. Shirazi et al., in Conference Proceedings —Magnesium on the Move, May 12–15, 1992 (Chicago II), p 50 T. Tsukeda et al., JSW Annu. Tech. Report, No. 54, 1998, p 665

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 785-811 DOI: 10.1361/asmhba0005322

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Introduction to Cast Irons* George M. Goodrich, American Foundry Society

CAST IRONS ARE A CLASS of alloys within the larger ferrous family of metals. The term cast iron designates a group of materials that have some similar attributes and components but exhibit other features that affect resultant properties, manufacturing characteristics, and ultimate use. They date to 600 B.C. in China, with the original use being cookware and architectural applications. The Industrial Revolution brought a major increase in both the production and application of cast irons due to the invention of the cupola for melting in the mid-17th century. The fluidity and quality of iron castings due to the addition of silicon allowed them to rapidly outpace cast steel as an engineering material. Steels and cast irons are composed principally of iron alloyed with carbon and containing some silicon. Steels contain less than 2.0% C (more commonly less than 1.0% C), and cast irons contain more than 2.0% C. During solidification, the maximum amount of carbon that can be dissolved in a single phase (austenite) of an iron-carbon alloy is approximately 2.0%. As a result, cast irons are heterogeneous alloys that contain more than one constituent in their microstructure due to that excess carbon. Variations in that additional constituent are observed in the various types of cast irons. Cast irons contain an appreciable amount of silicon (typically 1.0 to 4.0%) and thus are Fe-C-Si alloys. Compared to steels, cast irons have lower melting temperatures, higher fluidity, and are less reactive with mold materials. The high carbon content results in lower density and improved castability, and silicon provides strength in the microstructure. However, the cast iron alloys are not as easily forged, mechanically worked, or as weldable as steels. There are five generic types of cast irons based on the form of the excess carbon present: white, malleable, gray, ductile, and compacted graphite. Unlike steel, the various types of cast iron cannot be designated by their chemical composition, since the range of element analysis for the types overlaps (Table 1). However, during solidification, the excess carbon present in the iron forms graphite and/or iron carbide

constituents that result in unique characteristics, such as the fracture appearance and graphite morphology for which these irons are named. Alloyed cast irons are produced from these generic types by the addition of chromium, nickel, molybdenum, silicon, aluminum, and/ or copper. Minor amounts of various elements are also important in controlling the casting microstructure and resultant properties. For example, inoculants are used to control graphite type and size in gray and ductile irons; bismuth or tellurium is used in malleable iron production to avoid graphite formation during solidification; small amounts of magnesium are added to molten iron to form spheroidal graphite (ductile iron) or compacted graphite. The microstructure of irons depends on process factors rather than just chemical composition (Fig. 1). Accordingly, large and small castings produced to the same mechanical property requirement will require different compositions, and the same casting produced in different foundries may require different compositions. The cast irons, then, are designated and purchased on the basis of their mechanical properties, except where service requirements (such as corrosion resistance or high temperature) necessitate that certain chemical elements be present.

Classification of Cast Irons Cast irons are classified by their microstructures, primarily by the form of the excess carbon (the majority of the carbon in the alloy).

This excess carbon may be present in the form of a compound, such as Fe3C (iron carbide or cementite), or as graphite. Each of the various types of cast iron may be moderately alloyed to obtain certain properties, and some may be more heavily alloyed (high-alloy irons). White irons have excess carbon present entirely as iron carbide, Fe3C, also called cementite (Fig. 2a). This compound is hard and brittle, exhibits little ductility (white crystalline fracture seen in Fig. 2b), high compressive strength and wear resistance, and will retain these characteristics even when heated to red heat for a limited time. Iron carbide forms during solidification of the iron. The amount of iron carbide increases with lower carbon and/or silicon contents when the solidification cooling rate is increased. With higher carbon and/or silicon contents, or slower cooling rates, solidification results in gray graphite formation. Mixed structures containing iron carbide and graphite are mottled and are generally undesirable in white iron castings. Cooling Rate. Since the formation of iron carbide in a given composition of cast iron depends on the solidification cooling rate, cast irons are sensitive to section size. Localized rapid solidification of cast iron can promote the formation of white iron in selected areas at the casting surface, referred to as chill. This may be accomplished intentionally by using higher-thermal-conductivity mold material (chills) in that area to extract heat from the iron very rapidly. The thickness of this chill layer depends on the molding material, the composition of the iron, and foundry practices. Examples of desired chill formations are on the

Table 1 Range of compositions for common cast irons Composition, % Type of iron

Gray Compacted graphite Ductile White Malleable

C

Si

Mn

P

S

2.5–4.0 2.5–4.0 3.0–4.0 1.8–3.6 2.2–2.9

1.0–3.0 1.0–3.0 1.8–2.8 0.5–1.9 0.9–1.9

0.2–1.0 0.2–1.0 0.1–1.0 0.25–0.8 0.15–1.2

0.002–1.0 0.01–0.1 0.01–0.1 0.06–0.2 0.02–0.2

0.02–0.25 0.01–0.03 0.01–0.03 0.06–0.2 0.02–0.2

*This article is adapted with permission from the Iron Casting Engineering Handbook edited by George M. Goodrich (ISBN 0-87433-260-5), published by the American Foundry Society (www.afsinc.org), 2006.

786 / Cast Irons

Fig. 1

Basic microstructures and processing for obtaining common commercial cast irons

periphery of a cam, the tread of a wheel, or the edge of a knife blade. Hard, white iron regions may also develop in otherwise soft iron castings where the casting geometry (thin sections, fins, sharp corners) results in rapid heat transfer (Fig. 3) or where pouring conditions result in more rapid localized cooling rates (low pouring temperatures, slow pouring rates). If not desirable, these hard areas, or chilled edges, can be converted to iron and graphite using suitable heat treatments. Castability. Because of the lower carbon and silicon content compared to other cast irons, white irons solidify at higher temperature and have reduced castability. Carbon present as iron carbide, rather than graphite, results in increased density of the solid iron. Since graphite expands when formed during solidification, graphitic irons exhibit less solidification shrinkage, while white irons require the judicious use

and placement of risers to achieve effective feed metal transfer. Malleable iron was invented by Seth Boyden circa 1820, while experimenting with the heat treatment of white iron castings. All malleable iron is the result of iron cast as white iron and then subjected to an extended solid-state heat treatment at approximately 900
C (1650
F). This heat treatment causes the iron carbide (Fe3C) to dissociate and form irregularly shaped nodules of graphite, referred to as temper carbon. This temper carbon forms only at elevated temperatures as iron carbide dissociates into austenite and graphite. It is therefore necessary that castings solidify entirely as white iron, thus limiting the casting section size to that which is practical for malleable iron production. The thickness of acceptable castings that can be cast white and still respond to annealing is determined by the iron composition

(typically approximately 2.5% C, 1.5% Si) and by the use of limited amounts of bismuth and tellurium added to the molten iron. The temper carbon of malleable iron is distributed within a matrix structure that can be varied so that a wide range of mechanical properties can be obtained. Slow cooling through the critical temperature range (roughly where the color changes from red to black) results in a matrix of ferrite (soft, ductile, tough, lower strength); moderate cooling rates (or higher manganese contents) result in the formation of pearlite (harder, stronger); more rapid cooling rates (air or oil quenched) result in the formation of martensite (hard, high strength, low ductility), which can be tempered (high strength and toughness). Standard or ferritic malleable iron, also called blackheart (Fig. 4a) (ASTM A 47, Ref 1), possesses a matrix of ferrite, while pearlitic malleable iron (ASTM A 220, Ref 2) matrices

Introduction to Cast Irons / 787

Fig. 4

Malleable iron. (a) Sand-cast and annealed blackheart iron. Structure is a mixture of ferrite, pearlite, and irregular graphite. Ferrite content increases at the center of the casting, while pearlite occurs mainly on the edge. (b) Sand-cast and annealed whiteheart iron. This surface region shows the transition from a completely pearlitic zone to a zone where cementite surrounds the pearlite colonies. Contributed to ASM Online Micrograph Center by L.E. Samuels

Fig. 2

White iron. (a) Microstructure of as-cast, sandcast white iron (3.6C-0.41Si-0.46Mn-0.98Cr0.15P-0.024S). Carbon equivalent: 3.7%. White area is iron carbide (cementite). Gray areas are solidified as austenite and were transformed to pearlite during solidstate cooling. Etched with 2% nital. Original magnification: 100. Contributed to ASM Online Micrograph Center by L.E. Samuels. (b) Historically, the white fracture surface of the hard, brittle white iron gave it its name

Fig. 5

Gray, or flake or lamellar graphite, cast iron (3.6C-2.0Si-0.2P). Carbon equivalent: 4.31%. As-cast transverse section of sand casting. Original magnification: 100. Contributed to ASM Online Micrograph Center by L.E. Samuels

Fig. 3

The effect of section size and chill rate is evident in the wedge-shaped test casting. At the tip (left), rapid cooling results in white iron. A mottled section is adjacent, and a gray fracture surface occurs on the right where cooling was slower. Courtesy of C.R. Loper, Jr., University of Wisconsin

are typically tempered martensite or, less commonly, pearlite. The decarburized iron made chiefly in Europe is called whiteheart (Fig. 4b). Gray iron is also called flake or lamellar graphite cast iron due to the shape of the graphite present in these alloys. The generic term, cast iron, is often improperly used to mean gray

cast iron. While gray iron has been produced for over 3000 years, modern gray cast irons require precise control of factors that affect both the solidification and the matrix structures of these alloys. Over a range of appropriate composition and cooling rates, the solidification of molten iron will occur in the form of a eutectic cell consisting of interconnected graphite flakes and austenite. The graphite crystal is hexagonal in form, with flakes growing as a result of the edgewise extension of the hexagonal plate. The austenite transforms on cooling to pearlite (Fig. 5), a combination of ferrite and cementite in lamellar plates. Fracture of gray iron occurs as the fracture path follows the graphite flakes, resulting in the characteristic gray color of the fracture and giving rise to the name of this family of cast iron alloys. As with other cast irons, the properties of gray irons are determined by the microstructure. The size, shape, distribution and amount of the

graphite flakes, and the type and hardness of the matrix determine properties in gray iron. These characteristics are determined by both the composition of the iron (primarily carbon and/or silicon) and by the cooling rates (during both solidification and solid state). Slower casting cooling rates and a higher carbon and silicon content result in more and larger graphite flakes, a softer matrix, and lower strength. Fast cooling rates through the eutectoid transformation temperature and/or the addition of some alloying elements result in a harder matrix and higher strength. Properties are dependent primarily on the amount and morphology of the graphite flake, while compressive properties (e.g., hardness) depend on the matrix structure. The presence of graphite flakes provides gray iron with unique mechanical properties (e.g., excellent machinability at hardness values that produce superior wear resistance, excellent damping characteristics attributed to the flake graphite morphology as well as a nonlinear stress-strain relationship at low stresses, resistance to galling), physical properties (thermal conductivity), and casting characteristics (superior castability, reduced shrinkage, etc.). The specific shape of flake graphite varies with composition and casting conditions that affect nucleation and growth during solidification. ASTM A 247 (Ref 3) classifies the graphite by distribution, size, and type. As noted previously, these characteristics derive from foundry processing as well as composition, so that it is preferable to specify desired properties for gray iron rather than the factors that affect those properties. Furthermore, since section size affects casting cooling rate, it is desirable to specify the section and location in which the desired mechanical properties are to be monitored and measured. Gray cast irons are typically specified by their service requirements.

788 / Cast Irons Ductile iron was invented by Keith Millis and Al Gagnebin at the International Nickel Company during the 1940s and resulted from the addition of a small amount of magnesium to a low-sulfur-content molten iron. At approximately the same time, Henton Morrough at the British Cast Iron Research Foundation found that a small amount of cerium would spheroidize graphite in cast irons. Ductile iron is also referred to as nodular or spheroidal graphite iron because the magnesium treatment results in the graphite assuming a spheroidal shape during solidification (Fig. 6). The magnesium acts to deoxidize and desulfurize the melt and to control the spheroidal growth of the graphite. Excess magnesium will promote increased amounts of dross (oxides and sulfides) as well as carbides. The base iron composition must be carefully controlled to prevent the introduction of certain minor elements that interfere with spheroid formation. Process controls have been developed that make the processing of ductile iron very reliable. Typically, ductile irons are produced at high carbon or silicon contents that provide processing advantages. Because of the spheroidal form of the graphite, the properties of ductile iron are chiefly dependent on the matrix structure rather than carbon and/or silicon content. The various grades of ductile iron, then, are related to the specific matrix structure, rather than to carbon and/or silicon content, which is affected by cooling rate, minor uses of alloying elements (primarily to promote pearlite formation), and the number of spheroids formed. A wide range of mechanical properties can be obtained in ductile irons due to the various matrices possible. Ductile irons, however, can also be more highly alloyed to produce irons having unusual properties for special applications. Compacted Graphite Iron. During the period of ductile iron development, it was realized that, when an insufficient amount of

magnesium or cerium was added to molten iron, a vermicular or compacted graphite shape would form during solidification. It was later realized that the co-addition of a small amount of titanium permitted more reproducible production of this type of cast iron. In more recent years, technologies have been developed to commercially produce cast irons having a compacted or vermicular graphite structure. This morphology is often considered to be intermediate between the flake graphite of gray irons and the spheroidal graphite of ductile irons (Fig. 7). The interconnected, stubby graphite flakes along with a small amount of graphite spheroids result in both properties and foundry processing characteristics that are intermediate between gray and ductile irons. Compacted graphite irons exhibit some of the castability of gray iron, have higher strength and ductility than gray iron with good thermal conductivity, wear characteristics, and machinability. High-alloy irons are base irons with more or higher percentages of alloying elements. Most of these alloys are included in ASTM specifications A 436, A 439, A 518, A 532, and A 571 (Ref 4–8). The unique properties of these irons (corrosion resistance, elevated- or low-temperature service, hardness, wear resistance, nonmagnetic properties) are derived from characteristics of the composition. The form and amount of the excess carbon phase (carbide or graphite), as well as the nature of the matrix, may be strongly affected by this alloying. High-alloy irons are usually produced by foundries that specialize in their production, because special foundry procedures are required to accommodate the 3 to 30% alloying content. These irons are typically specified both by their mechanical properties and by composition (primarily alloy content, secondarily by carbon and/or silicon content) and are applied to a wide range of applications resulting in classification according to their service: corrosive service, elevated-temperature service, wear, and abrasion resistance.

Composition of Cast Irons While the metallurgy of cast irons has similarities to the metallurgy of steels, there are significant differences. Essentially, most common steels are leanly alloyed and can be considered as binary iron-carbon alloys, where the alloying elements principally influence the kinetics of transformations. Under these conditions, the binary iron-carbon phase diagram can be used to interpret microstructures developed through slow cooling or near-equilibrium transformations (Fig. 8). This diagram shows the phases present in various iron-carbon alloys at various temperatures. Cast irons, on the other hand, contain considerable silicon, making them ternary Fe-C-Si alloys so that a more complex phase diagram applies. Figure 9 represents the metastable equilibrium at 2.0% Si on the Fe-Fe3C-Si diagram. Compared to Fig. 8, the eutectic composition (4.3 versus 3.6% C), the maximum solubility of carbon in austenite (2.0 versus 1.67% C), and the eutectoid composition (0.80 versus 0.60% C) are lowered by the presence of 2.0% Si. Eutectic solidification and eutectoid transformation (when the matrix forms from austenite) occur at higher temperatures and over a range of temperatures when the silicon content is increased. Another factor, not apparent in these diagrams, is that, while the metallurgy of steel can be described in terms of the Fe-Fe3C metastable system shown in Fig. 8, the metallurgy of cast iron involves both Fe-Fe3C and iron-graphite (stable) systems. The microstructure of cast irons is therefore more complex and more susceptible to processing conditions than is the microstructure of steel. Carbon Equivalent. The influence of both carbon and silicon on ferrous alloys is shown in Fig. 10. The carbon content of the eutectic composition indicated by the upper dashed line is reduced by the presence of silicon in the following manner: %C þ 1=3%Si ¼ 4:3

(Eq 1)

Alloy compositions that satisfy this relationship are eutectic alloys. If the alloy composition plots above this line, the alloy is hypereutectic; if it is below this line, it is hypoeutectic. Because of this relationship of carbon and silicon to the eutectic, it is often convenient to describe cast iron compositions in terms of carbon equivalent (CE), where: CE ¼ %C þ 1=3%Si

(Eq 2)

This relationship can be modified by including factors for other elements, such as phosphorus:

Fig. 7

Fig. 6 Microstructure of spheroidal graphite in cast ductile iron. Graphite (dark) is surrounded by ferrite (white) in a pearlite matrix. Original magnification: 250. Courtesy of Bruce Boardman

Compacted graphite gray iron (3.7C-2.3Si0.82Ni-0.21Mn-0.03P). Carbon equivalent: 4.48%. The compacted graphite structure improves mechanical properties of flake gray iron. Original magnification: 100. Contributed to ASM Online Micrograph Center by George Vander Voort

CE ¼ %C þ

%Si þ %P 3

(Eq 3)

This CE was given for some of the micrographs (Fig. 2a, 5). Another means of illustrating this effect (more commonly used in

Introduction to Cast Irons / 789 treatment, and is not usually included in the chemical analysis of cast irons. Sulfur. All commonly produced ferrous alloys contain some sulfur resulting from the fuels used to process these alloys, from additions made in the preparation of these alloys, or from the raw materials employed in their production. The acceptability of sulfur in the iron depends on the type of iron being produced. For example, the production of ductile iron typically requires that the base iron contain <0.02% S prior to magnesium treatment, since sulfur in the melt will react with the magnesium, reducing the amount available for spheroidization. On the other hand, higher sulfur content is desirable in malleable and gray irons to assist in the nucleation of graphite, and, in producing those irons, sulfur may be added to the melt in the form of pyrite or by the use of higher-sulfur-content additives. Manganese. In all ferrous alloys, it is essential that the effect of sulfur be considered relative to the manganese content. In the absence of manganese, sulfur will combine with iron to form FeS, which segregates to intercellular areas or grain boundaries during solidification. Whenever sufficient manganese is present, it is assumed that MnS will form and be distributed more or less uniformly throughout the structure. However, the formation of MnS effectively removes a portion of the manganese from the alloy so that it is no longer available to serve as an alloying element. Due to complete solid solubility between iron and manganese in the formation of the sulfide, it is likely that both iron and manganese are present in the resulting compound, although the iron concentration will decrease as the manganese content increases. The actual compound, then, is (MnxFey)S. The stoichiometric relationship is then: %S  1:7 ¼ %Mn

Fig. 8

Iron-carbon phase diagram, where solid curves represent the metastable system Fe-Fe3C and dashed curves represent the stable system iron-graphite

Europe) is to define a saturation ratio (Sc), where: Sc ¼ %C=½4:3  1=3ð%Si þ %PÞ

(Eq 4)

This relationship describes how near to the eutectic composition (Sc = 1.0) an alloy is. A given CE or Sc can be obtained with any combination of carbon, silicon, and phosphorus content satisfying the relation. While alloys having the same CE or Sc may possess similar solidification temperature ranges and require similar foundry characteristics, they will not provide identical properties. For example, carbon is more than twice as effective in preventing solidification shrinkage than the CE equation indicates, and silicon is more effective in preventing chill in thin sections. Nevertheless, this relationship is commonly used in discussions concerning cast irons. The temperature at which austenite starts to form during the solidification of hypoeutectic cast iron compositions is noted by the line

separating the “liquid + austenite” and the “liquid” regions of the phase diagrams presented in Fig. 8 and 9. The temperature at which austenite begins to form from the melt correlates to iron composition as follows: %CEL ¼ %C þ 1=4%Si þ 1=2%P

(Eq 5)

Note that this definition of carbon equivalent (CEL) relates to the austenite liquidus temperature and differs from that used to describe the position of an alloy chemical composition with respect to the eutectic composition. Combined carbon content is a term used to describe the difference between the total carbon content of a cast iron and the amount of graphitic carbon. Combined carbon (Fe3C) is not a dependable indicator of either the matrix structure or properties of iron because of the variability in the combined carbon content of pearlite with silicon content. Also, the combined carbon content will be strongly affected by the cooling rate, either as cast or during heat

(Eq 6)

The balance of sulfur and manganese that will generally result in optimum balance between tensile strength and hardness in gray iron has been found to be: %S  1:7 þ ð0:30 to 0:50%Þ ¼ %Mn

(Eq 7)

Manganese contents toward the lower end of this range are reported to maintain higher tensile strength and hardness, while values toward the higher end of this range decrease both tensile strength and hardness. However, the use of manganese to control the amount of pearlite in the matrix is not recommended because it detrimentally affects the ratio of strength to hardness, that is, achieving a given tensile strength only at a higher-than-typical hardness. Additions of elements such as copper and tin have been found to be more effective as pearlite promoters. Phosphorus. Another element typically present in all ferrous alloys is phosphorus, where a small amount will be retained in solution in ferrite, increasing its hardness and strength. Greater amounts will result in the formation of another microstructural constituent, steadite,

790 / Cast Irons

Fig. 10

Approximate composition ranges of carbon and silicon for groups of alloys. The upper dashed line (%C + %Si/3 = 4.3) shows the eutectic composition. The lower dashed line (%C + %Si/6 = 2.0) illustrates the decreasing solubility of carbon in austenite

Fig. 9

Section through Fe-Fe3C-Si ternary phase diagram at 2.0% Si

which segregates to the last area to solidify. There are small amounts of phosphide eutectic (steadite) in Fig. 5. This characteristic can be observed in gray irons by examining the structure after etching. The reagent plates copper onto the polished sample surface, except where phosphorus is concentrated, thereby delineating the cell boundaries of the eutectic in which phosphorus segregates. This treatment is most effective in gray irons containing >0.05% P.

This network of steadite may reduce the machinability and impact strength of gray iron but may increase wear resistance in some applications. In some situations, phosphorus may also be increased to 0.06 to 0.11% to minimize the burn-on or bonding of molding sand to the gray iron casting surface. Phosphorus in the ferritic grades of malleable and ductile irons will increase their impact transition temperature—the temperature below

which an impact fracture tends to be brittle rather than ductile. This effect is much less evident when significant amounts of pearlite are present in the matrix. Phosphorus has a significant effect on the fluidity of the iron, along with other composition variables and temperature. Minor Elements. It is not uncommon for a number of other elements to be present in cast irons, either intentionally or incidentally, and those elements have an important effect on the microstructure and properties of the iron. Depending on the element and the type of iron being considered, these effects could be advantageous or detrimental. Small amounts of these elements may affect nucleation during eutectic solidification, others may result in a degradation of the graphite shape, some may embrittle the matrix structure or influence the diffusion of carbon within the matrix, and others may segregate strongly within the cell boundaries. Among these elements are aluminum, bismuth, calcium, lead, tellurium, titanium, tin, and nitrogen. Foundries must practice careful scrap selection to avoid building up adverse concentrations of these elements in the irons they are producing. Alloying Elements. When an element is present in amounts typically greater than 0.10%, it is referred to as an alloying element. (Two exceptions to this are nitrogen and boron, which may be used as alloying elements in amounts of approximately 0.001%.) Alloying elements are usually employed to modify or enhance the properties of the base iron by influencing the matrix structure. However, some of these elements may also affect the solidification of the cast iron. A summary of these effects is presented in Table 2.

Introduction to Cast Irons / 791 Table 2

Structural effects of some elements in cast irons

Element

Al Sb Bl B (<0.015%) B (>0.015%) Cr Cu Mn Mo Ni Si Te Sn Ti (<0.25%) V

Effect during solidification

Effect during eutectoid reaction

Strong graphitlzer Little effect in amounts used Carbide promoter but not a carbide former Strong graphitizer Carbide stabilizer Strong carbide former. Forms complex carbides, which are very stable Mild graphitizer Mild carbide former Mild carbide former Graphitizer Strong graphitizer Very strong carbide promoter but not a carbide former Little effect in amount used Graphitizer Strong carbide former

Silicon is present in all cast irons, although greater amounts than normal may be employed to promote a ferritic matrix. High silicon contents result in an increase in the impact transition temperature so that the silicon content may be restricted, for example, <2.75%. Higher silicon contents may be used to develop excellent oxidation and corrosion-resistance properties. Nickel is also a graphitizer and decreases carbide stability, while it increases pearlite fineness and the stability of pearlite, thereby increasing strength. Nickel is considered to have no undesirable effects in cast irons. Nickel is also an austenite stabilizer. Copper is used to promote pearlite formation, believed to occur by acting as a barrier to carbon diffusion from austenite to graphite. Another strong promoter of pearlite is tin, which is more effective than copper. Antimony acts in a similar manner but is more effective than either copper or tin. Antimony has also been demonstrated to improve nodule shape in ductile irons. Chromium promotes carbide formation during both solidification and the eutectoid reaction. By increasing the tendency for the formation of white iron and the retention of a higher combined carbon, chromium improves wear resistance. High levels of chromium also retard oxidation and growth at elevated temperatures and provide improved corrosion resistance. Manganese and molybdenum are used to enhance matrix properties. Manganese will also participate in carbide formation but not to the extent of chromium. Molybdenum actually retards pearlite formation, thereby promoting ferrite, but refines pearlite and promotes the formation of bainite.

Solidification of Cast Irons The structure of the cast iron is determined during solidification and depends on the composition of the iron, the rate of cooling, and the use of nucleating agents that initiate

Promotes ferrite and graphite formation Strong pearlite stabilizer Very mild pearlite stabilizer Promotes graphite formation Strong pearlite retainer Strong pearlite formation Promotes pearlite formation Pearlite former Promotes ferrite and bainite formation Mild pearlite promoter Promotes ferrite and graphite formation Very mild pearlite stabilizer Strong pearlite promoter and retainer Promotes graphite formation Strong pearlite former

formation of the graphite phase. The formation of graphite is promoted by the use of higher carbon and/or silicon contents, slower solidification cooling rates, and the presence of other elements. On the other hand, lower carbon and/or silicon contents, the presence of certain elements, and more rapid solidification promote carbide formation. The production of white irons requires compositions that have a sufficiently low carbon and/or silicon content, suitable carbide-forming alloying additions, or the addition of carbidepromoting elements that inhibit the formation of graphite. An important factor in controlling the solidification microstructure is the presence of nuclei in the liquid metal. These nuclei are capable of inducing the formation of graphite in gray irons, ductile irons, and compacted graphite irons. Once nucleated, the shape of the graphite, however, is a growth-controlled process distinct from that of nucleation. Thermal Analysis. A convenient means of estimating the chemical composition of molten cast irons is derived from the correlation established where the liquidus temperature is observed to correlate to carbon, silicon, and phosphorus through the relationship of Eq 5. During solidification, heat is evolved by the formation of austenite. The lower the CE of the iron, the higher the temperature at which austenite begins to solidify. The amount of latent heat evolved can be documented by measuring the temperature change in a standard sample (Fig. 11), thereby obtaining an accurate measure of the CE of the melt. If a sufficient amount of a carbide promoter (such as tellurium) is placed in the sample cup, an austenite-carbide eutectic will result. Correlation of the temperature of the austenite liquidus with the white iron eutectic temperature can be used to estimate the carbon content of the iron; by difference, an estimate of the silicon can be obtained as well. This technique enables accurate and rapid estimation of the CE value and carbon and silicon contents on the melt floor. Recognizing the fact that the evolution of latent heat during solidification will depend on the type and amount of phases formed during

Fig.

11

Sample cup for thermal analysis measurements of composition in cast irons. The resin-bonded sand cup has a chromel-alumet thermocouple. Courtesy of C.R. Loper, Jr., University of Wisconsin

solidification, the solidification of a suitable sample of iron may also be used to estimate the size, shape, and distribution of graphite and/or carbide formed during eutectic solidification. It should be recognized, however, that metallurgical characteristics of the iron may be affected by factors other than chemistry. These factors are often referred to as foundry processing variables and may include charge materials, melting techniques and procedures, melt handling and treatment, time-temperature history of the melt, and so on. Lack of control of these processing variables will affect the accuracy of thermal analysis estimations of carbon and silicon. Chill Test. Cast irons are sensitive to variations in solidification cooling rate. Rapid cooling during solidification promotes a carbidic structure (or white fracture), whereas slower solidification cooling rates promote a graphitic structure (or gray fracture). The tendency of a given iron to solidify white or gray can be estimated from the fracture of a properly selected test casting. Two types of these test castings are used: the chill wedge and the chill plate. The chill wedge consists of a wedge-shaped casting (Fig. 3) where the high surface-areato-volume ratio at the tip of the wedge induces rapid solidification and chill formation. The chill plate consists of a plate-shaped casting poured with one edge of the casting against a clean copper, steel, or cast iron plate, to induce rapid solidification. After cooling to a black heat, these chill castings are fractured. The extent of clear chill formation, or total chill (including mottle), is measured and may be used as an indicator of the metallurgical character of the melt. Since the propensity of iron to solidify white is reduced by increased carbon and/or silicon, by the presence of graphitizing elements, and by the effectiveness of inoculation (as well as by foundry processing variables), this test serves as a measure of the metallurgical quality of a given melt and an indicator of the structure that will develop in a given casting section. When the chill test is used in combination with thermal analysis, tighter structural control can be realized. Inoculation. Solidification of a particular phase must be initiated by the presence of a

792 / Cast Irons foreign substrate upon which that phase can nucleate and grow. In cast irons, that substrate serves to initiate the formation of graphite, either as the proeutectic graphite formed in hypereutectic compositions or as the graphite phase in eutectic or hypoeutectic compositions. These foreign substrates may be present in charge materials and additions made during the melting process or may be intentionally added to the melt as inoculants. Controlled amounts of these inoculants, in the range of 0.10 to 0.50%, are typically used to minimize the formation of eutectic iron carbides and to produce the desired form of graphite. Inoculation promotes the formation of type A flake graphite (see the section “Gray Iron Graphite” in this article) in gray irons (most commonly with an increase in the number of eutectic cells) and increases the nodule count in ductile and compacted graphite irons. The nucleating effect of inoculants fades with time, and therefore, inoculation is not considered as alloying. On the other hand, additions may be made to molten iron that are intended to suppress the formation of graphite during solidification. It is common in the production of malleable iron to add minute quantities of bismuth or tellurium to extend the section thickness within which white irons can be produced, by causing solidification totally as white iron without mottle or free graphite present.

Graphite Morphologies The forms of graphite in cast irons have been classified into seven basic types in ASTM A 247 (Fig. 12). See ASTM A 247 (Ref 3) for ordering plates to be used for evaluation. This figure is for concept only. Types I and II graphite are characteristic of ductile irons, with type I being the usual desirable form, although type II has little adverse effect on mechanical properties. Types IV, V, and VI represent degraded, unacceptable graphite in ductile irons. Type IV is meant to illustrate compacted or vermicular graphite, which is attributed principally to a low effective magnesium level in the iron and/or the presence of excessive titanium. Type IV is also the characteristic graphite structure found in compacted graphite irons. Type V represents deleterious graphite due to the presence of an excessive amount of certain elements detrimentally affecting spheroidal graphite growth during solidification. Type VI is termed exploded graphite, a morphology attributed primarily to high carbon equivalents and the resultant flotation of graphite nodules during the solidification of thick-section iron castings. Type III graphite is temper carbon distributed in nodules resulting from the graphitization heat treatment producing malleable iron. Type VII represents the flake graphite forms observed in gray irons. Gray Iron Graphite. A subclassification of type VII graphite includes the six types of flake graphite illustrated in Fig. 13. This

Fig. 12

The seven types of graphite as established by ASTM A 247 (Ref 3). Types I and II are simply called that. Type III is also called temper carbon; type IV is vermicular graphite; type V is crablike graphite; type VI is exploded graphite; and type VII is gray iron (flake graphite)

classification is an attempt to summarize the variety of flake graphite forms observed in gray irons. However, using them to characterize or specify flake graphite structures is less than fully objective, since it is recognized that these graphite types are not discrete but are rather representative of discrete steps in the continuous change in morphology developed during solidification. Type A graphite is uniformly distributed and randomly oriented and is formed during eutectic solidification in cells of austenite and flake graphite. These cells form at the highest temperatures during eutectic solidification. Type A graphite is generally considered the preferred graphite type for gray iron. When nucleation of the eutectic is less effective, so that few nuclei are activated, the near-eutectic composition liquid undercools at a greater rate until the austenite-graphite eutectic is nucleated, forming

a finer graphite flake at the center of the cell. Heat of fusion evolved during subsequent solidification causes the temperature to increase so that the eutectic cell is coarsened. The result is a larger cell size and rosette distribution with random orientation of type B graphite. Hypoeutectic iron compositions start solidification with the formation of proeutectic austenite dendrites so that eutectic solidification occurs between the dendrite arms. When the iron is sufficiently undercooled, the dendritic austenite is well developed so that eutectic solidification occurs interdendritically in a cellular manner but with finer, randomly oriented graphite flakes (type D). Type E graphite is similar, except that the carbon equivalent is even lower and the cooling rate faster, so that a greater fraction of solidification occurs as austenite dendrites. This restricts the interdendritic areas even more, so that the graphite flakes are

Introduction to Cast Irons / 793

Fig. 13

The six types of flake graphite as established in ASTM A 247 (Ref 3) as a subclassification of type VII in Fig. 12

forced to assume a preferred orientation as they align parallel to the dendrite arms. Both types D and E graphite develop in irons containing sufficient silicon to inhibit carbide eutectic formation. The result is that portions of the microstructure are devoid of flake graphite. This type of iron is typical of that produced in permanent mold castings. When the matrix structure is fully pearlitic, the graphite-free regions of the microstructure act to reinforce and strengthen the iron. As a result, in pearlitic gray irons of the same hardness, those containing type D graphite will exhibit higher strength than those containing type A graphite. However,

obtaining a fully pearlitic structure with type D graphite is difficult due to the short diffusion distance between graphite flakes. Graphite that forms as the proeutectic phase in hypereutectic irons is classified as type C. In this case, graphite forms in the liquid iron independently of austenite and grows unencumbered by other solid phases. Therefore, it appears as straight plates, with some branching, that grow in size according to the solidification cooling rate. These graphite flakes are entrapped in the structure as solidification progresses. However, due to the buoyancy of graphite in the melt, this proeutectic graphite will have a

tendency to float to the surface, concentrating in the upper portions of a casting section. Graphite that collects at the upper casting surface, or that forms on the surface due to the high carbon concentration in the melt, is termed kish graphite. Type F is Widmansta¨tten graphite associated with high lead (>0.0005%) and sometimes high phosphorus (>0.20%). ASTM A 247 also establishes a standard for evaluating the size of graphite flakes. Since all graphite within a eutectic cell is interconnected, there is a direct relationship between eutectic cell size and graphite flake length. Malleable Iron Graphite. All graphite present in malleable iron is the result of solid-state graphitization of the white iron eutectic. The shape of the temper carbon nodules is generally unaffected by normal process considerations. However, the number of these nodules (nodule count) is a function of a number of variables, including but not limited to composition, the introduction of boron to the melt, the specific heat treatment cycle, and so on. An increased nodule count will affect certain mechanical properties, but its greatest effect is in reducing the time required for first-stage graphitization. Ductile Iron Graphite. Acceptable graphite forms in ductile irons are types I and II nodules, and only approximately 3% vermicular graphite can be tolerated in meeting ASTM minimum property specifications. However, the nodule count in ductile irons increases with effective inoculation and increased solidification cooling rate. The amount of ferrite in ductile iron depends on composition, cooling rate, and nodule count. The thickness of the ferrite ring around the nodule is significantly affected by composition and cooling rate, so that a higher nodule count results in the formation of more ferrite with resultant lower hardness and strength. Higher nodule counts are also associated with improved graphite nodularity, reduced carbide formation, and more effective use of alloying elements such as, perhaps, bismuth. By far, the majority of ductile irons are produced from hypereutectic compositions where the first phase to form is graphite. As with type C flake graphite, these spheroids will become buoyant in the melt and, as they increase in size, will float upward. The degree of this graphite flotation will depend on the iron composition as well as the solidification time of the casting section. However, a higher concentration of carbon, as graphite spheroids, will be obtained at the upper portion of the casting section. As these spheroids rise to the surface, they encounter minor but significant temperature and concentration profiles that cause the well-formed spheroid to degenerate into the exploded form, type VI (Fig. 12). The formation of exploded graphite is further promoted by excess cerium. One graphite type that has not been incorporated into ASTM A 247 is referred to as chunky graphite, appearing in micrographs as a cluster of short, closely spaced, wormlike flakes. This deleterious graphite form develops

794 / Cast Irons near the thermal center of larger casting sections in magnesium-treated ductile irons containing an excessive amount of rare earths. It is promoted by higher carbon equivalents and higher silicon contents. Formation of chunky graphite can be inhibited by reducing the rare earth addition or by adding elements that will chemically react with the rare earths, for example, antimony, and will be reduced by lowering the carbon equivalent of the iron, particularly the silicon content. Compacted Graphite Iron. Ideally, compacted graphite irons will contain no spheroidal graphite. However, the process controls necessary to achieve this structure are not commercially feasible, and most specifications limit allowable nodularity. Nodularity results from excessive nodulizers used in melt treatment but is also attributed to more rapid solidification cooling rates and to inoculation that induces the formation of spheroidal graphite. Commercially acceptable compacted graphite irons may be specified to contain a wide range of nodularity, depending on specific applications. It is therefore essential that, when comparing various compacted graphite irons, the nodularity requirements of a specific grade be known, that is, maximum of 20% nodularity or 40 to 60% nodularity.

Matrix Structure Development The maximum amount of carbon is in solution in austenite when solidification is complete. Cooling to lower temperatures causes some of that dissolved carbon to precipitate out onto the graphite or carbide formed during solidification (Fig. 9). However, unless the cooling rate following solidification is extremely slow, the austenite will always remain supersaturated with carbon. On the other hand, when a cast iron is reheated to a temperature where the matrix will transform to austenite, carbon enters the austenite from the graphite or carbide until the austenite is saturated with carbon at a given temperature. Thus, the carbon content of austenite cooled after solidification can be expected to be greater than that developed during heat treatment. The matrix structure develops as a result of the transformation of the austenite on cooling. The structure or structures that form depend on the composition of the austenite; the shape, size, and distribution of the excess carbon (carbide or graphite); and the cooling rate through the critical temperature region. The four principal structures that develop during this continuous cooling are (in order) ferrite, pearlite, bainite, and martensite. Other matrix structures such as ausferrite may be formed during specialized heat treatments. Increased cooling rates cause the transformation of austenite to occur at increasingly lower temperatures, thus inhibiting the formation of ferrite. The chemical composition of the austenite influences the transformation, because

certain elements promote the formation of ferrite while others promote the formation of pearlite (Table 2). The role of the excess carbon phase (carbide or graphite) is more complex. The carbon content of ferrite is exceptionally low so that when ferrite forms, carbon must be rejected from the austenite. Graphite present in the structure or nucleation sites for graphite are locations where this carbon can be deposited. The ease with which this deposition takes place depends on the diffusion of carbon through austenite and the accessibility of the carbon to the preexisting graphite. When the excess carbon phase is carbide, as in white cast iron or chilled regions, graphite is not easily formed, so that carbon is retained in the austenite to lower temperatures, where it transforms to pearlite: gðausteniteÞ ! aðferriteÞ þ Fe3 C

(Eq 8)

where the ferrite and Fe3C are arrayed in lamellar form, termed pearlite. The spacing between the higher-carbon-content Fe3C and the lowercarbon-content ferrite depends on the time (cooling rate) permitted for carbon diffusion to occur and the composition of the austenite. The typical structure of white iron consists of iron carbides (formed principally during solidification) in a matrix of pearlite (Fig. 2a). Only when the cooling rate is very slow will the austenite decompose so that the preexisting carbide grows and a ferrite matrix develops. In gray irons, matrix development also depends on the distance over which carbon must diffuse to reach a graphite flake (spacing between flakes, typically related to flake size) and the crystallographic structure of the graphite flake. Carbon is more readily deposited on the edges of graphite flakes, resulting in preferred sites for ferrite formation. Smaller graphite flakes will therefore promote ferrite formation during solid-state transformation, since a greater number of graphite flake edge sites are available. For example, ferrite will form preferentially in the interdendritic structure of type D graphite. Certain elements (tin, antimony, copper) can be added to the melt to inhibit this ferrite formation by interfering with carbon diffusion or segregating preferentially on the surface of the graphite, serving to block the addition of carbon atoms. Ductile irons contain spheroidal graphite in which the hexagonal graphite basal planes lie roughly parallel to the spheroid surface. As a result, the edges of this crystallographic structure are readily available to accept carbon atoms diffusing from the austenite. Ferrite thus forms preferentially around the graphite spheroid, developing the bull’s-eye structure of Fig. 6. The thickness of the ferrite ring will depend on the composition of the austenite and the rate of cooling during ferrite formation. Increasing the nodule count of a given iron will then result in an increase in the total amount of ferrite present in the matrix. (This specimen is 35% ferrite.) Again, the use of certain elements will inhibit the ferrite formation.

Due to the irregular crystallographic arrangement of graphite found in compacted graphite irons, ferrite formation develops readily, and measures to inhibit ferrite formation are more difficult to implement. Just as the vermicular graphite in compacted graphite irons is somewhat intermediate between flake graphite and spheroidal graphite, the crystallographic nature of that graphite influences ferrite formation, just as it does in gray and ductile irons.

Physical Properties of Cast Irons Physical properties, such as density, thermal expansion, thermal conductivity, specific heat, electrical conductivity, magnetic properties, and acoustic properties, are of considerable significance in selection of materials for specific functions. These properties are presented for the various cast irons, and the influence that microstructure and chemical composition have on each property is discussed. Density. The true density (r) is the ratio between the mass and the volume occupied by the pore-free material. The apparent density is the ratio between the mass and the apparent volume (including pores). In the following discussion, the term density in the sense of true density, r, is used. Solid density is r without a subscript. Liquid density is r‘. The density of pure liquid iron is 7030 kg/m3 (440 lb/ft3). A wide range of values is found in literature (Ref 9) for the density of molten cast iron, extending from 6650 to 7270 kg/m3 (415 to 455 lb/ft3). More recent research (Ref 10) produced this empirical relationship:



r‘ ¼ 6993  73:2 ð%CÞ





 ð1:292  0:0874 ð%CÞÞ ðT  1550Þ (Eq 9)

where r‘ is the density of liquid iron in kg/m3, T is the temperature in
C, and %C is the carbon content in wt%. For liquid ductile iron, the following dependency of density on temperature was calculated, based on data from Grutter and Marrincek (Ref 11):



r‘ ¼ 8870  1:48 ðT Þ

(Eq 10)

Other experimental data (Ref 12) summarized in Fig. 14 indicate that, while the density of liquid ductile iron is smaller than that of gray iron, because of the higher carbon content, the density of solid ductile iron is higher than that of gray iron, due to the smaller volume occupied by graphite. It is also seen that the density decreases with higher temperature. However, upon the solid-to-liquid transition (melting), the density increases suddenly for both types of cast irons. These data confirm the older observations on solid iron floating in its melt. The increase in density is because the dissolution of graphite during melting produces a carbon-rich iron solution, which is denser than the graphite-iron solid mixture. The density of solid irons depends on the density of the microconstituents and the

Introduction to Cast Irons / 795 Table 4 Density of cast irons in solid state at room temperature Density Type of iron

Microstructure

Gray iron class 20 (high C) Gray iron class 30 (high C) Gray iron class 40 (med. C) Gray iron class 50 (low C) Gray iron class 60 (low C) Lamellar graphite Lamellar graphite, medium Si Lamellar graphite, high Si Lamellar graphite, high Al Lamellar graphite (Nicrosilal) Lamellar graphite (Ni-Resist)

Pearlitic-ferritic Pearlitic-ferritic Pearlitic Pearlitic Pearlitic Aclcular (bainitic) Ferritic Ferritic Ferritic Austenitic Austenitic Austenitic Austenitic Austenitic Ferritic Ferritic-pearlitic Pearlitic Austenitic Austenitic Austenitic ... Ausferrite Ferritic Pearlitic Ferritic Pearlitic Martensitic ...

Lamellar graphite Ductile iron

Fig. 14

Density as a function of temperature for gray and ductile iron

Table 3 Density of metallographic phases of irons at room temperature Density Metallographic constituent

kg/m3

lb/in.3

Ferrite Pearlite Cementite Graphite Phosphide eutectic Martensite Austenite MnS

7870 7840 7660 2250 7920 7690 7840 4400

0.284 0.281 0.277 0.081 0.264 0.283 0.276 0.159

Source: Ref 9, 13, 14

temperature. The densities of the major metallographic phases present in cast iron at room temperature are listed in Table 3. The composition affects the density as well as the relative amounts of the constituents. Silicon and aluminum lower the density of ferrite. The theoretical equation is:





ra ¼ 7870  56:2 ð%CÞ  55:4 ð%SiÞ





 51:7 ð%AlÞ þ 10:3 ð%NiÞ

(Eq 11)

The density of nonhomogeneous phases can be calculated with the rule of mixtures. For example, for pearlite:

 



rPe ¼ 0:865 ra þ 0:135 rFe3 C



¼ ð0:865 7870 þ 0:135 7660Þ ¼ 7840 kg=m

3

(Eq 12)

This equation is valid, assuming that the carbon content of pearlite is 0.8%, as given in the ironcarbon binary phase diagram. The density of pearlite can change from one alloy to the other, depending on how much carbon was present in the austenite from which the pearlite was formed. Table 4 is a summary of density values for various cast irons in solid state. It should be noted that the available literature data are inconsistent, in many cases, because the samples used for measuring may have had various amounts of porosity. However, some general trends are obvious. Because of the low density of graphite, the density of cast iron decreases

Ductile iron (Ni-Resist)

Ductile iron, high silicon Austempered ductile iron Maleable iron Compacted graphite White iron, unalloyed White iron (Ni-Hard) White iron, high chromium

Composition, wt%

kg/m3

lb/in.3

... ... ... ... ... 1.57 Mo, 1.87 Ni 4–6Si 14–17Si 28–30Al 20Ni, 4Si, 3Cr 15Ni, 6Cu, 2Cr 20Ni, 3Cr 35Ni 13Ni, 7Mn ... ... ... 18–32Ni 34–38Ni 34–37Ni, 5Si ... ... ... ... ... ... 3.3–4.7Ni, 1.4–2.6Cr 20–25Cr

6800–7000 7000–7100 7100–7300 7200–7300 7300–7400 7320 6810–7080 6980–7030 5300 7200–7400 7300 7400 7400 7200 6780–7060 6920–7200 7200 7410–7450 7680 6960 7100 7250–7350 7200–7270 7300–7500 7000–7200 7600–7800 7600–7800 7200–7750

0.246–0.253 0.253–0.257 0.257–0.264 0.260–0.264 0.264–0.268 0.265 0.246–0.256 0.252–0.254 0.192 0.260–0.267 0.264 0.267 0.267 0.260 0.245–0.255 0.250–0.260 0.260 0.268–0.270 0.278 0.252 0.257 0.262–0.266 0.260–0.263 0.264–0.271 0.253–0.261 0.275–0.282 0.275–0.282 0.261–0.280

Source: Ref 9, 15–17

as the amount of graphite increases. Since the tensile strength also decreases with increasing graphite content, the lower-strength irons of all types generally exhibit a lower density. Completely annealed ferritic malleable iron also has a lower density than do the pearlitic and martensitic matrix irons. The density of malleable iron, as shown in Table 4, is higher than that of gray and ductile irons because of its lower graphite content. White irons with essentially all of the carbon combined as iron carbide have densities of approximately 7700 kg/m3 (480 lb/ft3), which is significantly higher than the graphite-containing irons. For unalloyed irons, based on the analysis of some literature data, the density in solid state as a function of composition can be calculated with the following empirical relationship (Ref 18):





r ¼ 8110  233 ð%CÞ  91 ð%SiÞ



 71 ð%PÞ

(Eq 13)

where r is the solid density in kg/m3. This equation cannot be extrapolated to steel. Note that the decrease in density, because of silicon, is steeper than that calculated from the data on pure elements (Eq 12). Alloying elements can significantly affect density, depending on the amount. Aluminum and silicon additions will decrease density, while nickel will increase density in alloys such as Ni-Resist iron. Based on the rule of mixtures, the following relationship can be used to calculate the theoretical density of cast iron:



r ¼ ra ð1  fFe3 C  fGr  fPe  fpor Þ







þ rPe fPe þ rFe3 C fFe3 C þ rGr fGr

where f is the weight fraction of the various constituents, fpor is the porosity fraction, and ra and rPe are calculated with Eq 11 and 12, respectively. The use of this equation, assuming no porosity, gives a density of 7668 kg/m3 (479 lb/ft3) for a white iron having 2.3% C and 1.2% Si. This is in the range of experimental values reported in the literature and indicates that white iron has very little porosity, if any. The same iron after graphitization (malleable iron) will have a density of 7522 kg/m3 (470 lb/ft3). This is too high. However, when a porosity of 4% is assumed, the equation returns the value of 7215 kg/m3 (450 lb/ft3), which is in range for ferritic malleable iron. Similar calculations for gray iron match the experimental results, assuming porosity levels from 7% for low grades to 1% for high grades.

Thermal Properties Specific Heat and Enthalpy. When a solid material is heated, its temperature increases, which means that energy is absorbed. Heat capacity represents the amount of energy required to produce a unit temperature rise. It is expressed as: C ¼ ðdQ=dT Þ

(Eq 15)

where dQ is the energy required to produce a dT temperature change. The heat capacity, C, is expressed in J mol1 K1 and is also called the molar specific heat. At constant pressure, the change in energy can simply be expressed as the change in enthalpy. Specific ðEq 14Þ heat, c, represents the heat capacity per unit





796 / Cast Irons









mass in J kg1 K1 (J kg1 C1), or Btu lb1 F1. Since it is temperature dependent, the temperature at which c is evaluated must be specified. A mean specific heat can be used within a temperature interval.





A summary of literature data on the specific heat of pure iron and of metallographic constituents present in cast iron is given in Table 5. It is seen that the specific heat increases with temperature. Plotting mean specific heat data

Table 5 Specific heat of pure iron and some metallographic constituents present in cast iron Temperature


Material

Pure iron a-iron a-iron a-iron

C






J kg1

F

173 50–100 50–700 50–950 650–700 900–950 ? ? 1727

279 120–212 120–1290 120–1740 1200–1290 1650–1740

Fe3C

100 300 500 700 900

Graphite

20 138 642 896 100 300 700 900 1100 1250

g-iron g-iron d-iron Liquid

C


Specific heat

1



Btu lb1

F


1



Reference

3141

452 469 615 695 829 946 674 603 825

0.108 0.112 0.147 0.166 0.198 0.226 0.161 0.144 0.197

19 20 20 20 20 20 21 21 19

212 570 930 1290 1650

620 624 636 670 716

0.148 0.149 0.152 0.160 0.171

22

70 280 1188 1645 212 570 1290 1650 2010 2280

712 1063 1863 1901 846 1034 1369 1516 1616 1671

0.170 0.254 0.445 0.454 0.202 0.247 0.327 0.362 0.386 0.399

23

22

Table 6 Specific heat of several cast irons from 20–100
C (68–212
F) Type of iron

Gray iron Ductile iron Austenitic Austenitic Austenitic Malleable iron

Fig. 15

Composition, wt%



J kg1

... Unalloyed 20–22 Ni 30 Ni 35 Ni ...



Specific heat


C1

460–544 544–711 486 465 419 511



Btu lb1




1

F

0.11–0.13 0.13–0.17 0.116 0.111 0.100 0.122

Mean specific heat curves for iron and steel. Source: Ref 20

Table 7 Mean specific heats of gray and malleable iron Specific heat, J/kg


C

0–200 0–300 0–400 0–500 0–600 0–700 0–900 0–1100 0–1200 0–1300




 C (Btu/lb  F)





Gray iron, phosphorus content

Temperature range

over a range of 10
C (18
F) against temperature (Fig. 15), it is seen that, for pure iron and steel (0.48C-1.98Si-0.9Mn-0.047S-0.44P0.16Ni-0.64Cu), a jump is recorded at the Curie and a ! g transformation (Ref 20). The specific heat of nonhomogeneous structures can be calculated through the rule of mixtures. For example, for pearlite, the value of 489 J kg1
C1 (0.117 Btu lb1
F1) is obtained. Thus, graphitization should decrease the specific heat. Both the enthalpy and the specific heat of iron increase with the carbon content. In principle, for unalloyed irons, composition does not influence the specific heat significantly. However, in high amounts, nickel decreases the specific heat, as shown for the austenitic ductile irons in Table 6. Also, high phosphorus decreases the specific heat (Table 7). Graphite shape affects specific heat. Indeed, ductile iron has higher specific heat than does gray iron (Table 6). The specific heat of cast iron increases slightly with temperature up to the a ! g transformation temperature (Table 7) and then records a sudden increase in the transformation range, as shown in Fig. 16. It decreases abruptly at the end of transformation and then continues to increase up to the melting temperature. It is also seen that the specific heat of gray iron increases faster with temperature than that of white iron. At the melting temperature, the specific heat of cast iron increases again suddenly. This is illustrated in Fig. 17 for an iron with the composition 4.22C-1.48Si-0.73Mn-0.12P-0.032S. Latent Heat of Fusion and Solid-State Transformation. The latent heat of transformation is the change in the heat content of a unit mass, a mole of substance, occurring during phase transformation. It is expressed as energy per mass, kJ/kg (Btu/lb), or sometimes as energy per mole (J/mol). A summary of some literature data for the latent heat of cast iron of various compositions is presented in Table 8. Some significant differences are seen, in particular for the older data. In general, the latent heat of fusion is expected to decrease as the amount of carbon increases. The literature data (Ref 25, 26) for the latent heat of the a ! g transformation ranges from 15.09 to 15.21 kJ/kg (6.49 to 6.54 Btu/lb). Thermal Expansion. The thermal expansion and contraction of iron alloys is an important parameter for design of the casting cavity as well as a consideration for engineering applications of the components subject to temperature extremes or temperature changes. The effect is expressed as a linear coefficient:

F

0.147% P(a)

0.54% P(b)

0.88% P(c)

Malleable Iron(d)

32–390 32–570 32–750 32–930 32–1110 32–1290 32–1650 32–2010 32–2190 32–2370

561(0.134) 494(0.118) 507(0.121) 515(0.123) 536(0.128) 603(0.144) 678(0.162) 670(0.160) 871(0.208) 850(0.203)

377(0.090) 435(0.104) 465(0.111) 481(0.115) 502(0.120) 553(0.132) 653(0.156) 649(0.155) 850(0.203) 829(0.198)

289(0.069) 377(0.090) 423(0.101) 448(0.107) 469(0.112) 511(0.122) 632(0.151) 628(0.150) 833(0.200) 812(0.194)

523(0.125) 536(0.128) 557(0.133) 582(0.139) 611(0.146) 666(0.159) ... ... ... ...

(a) 3.71% C, 1.50% Si, 0.63% Mn, 0.069% S. (b) 3.72% C, 1.41% Si, 0.88% Mn, 0.078% S. (c) 3.61% C, 2.02% Si, 0.80% Mn, 0.080% S. (d) Lower temperature in temperature range column is 20
C (68
F). Source: Ref 13, 24



l ¼ ðl=lo Þ=T





(Eq 16)

where Dl is the change in length, lo is the gage length, and DT is the change in temperature. Units of al, also called the coefficient of thermal expansion (CTE), are usually expressed as 106/
C (106/
F). A length unit is not needed, although one finds al expressed as (mm/m/
C (min./in./
F). The CTE changes with temperature (generally increasing with temperature), so an

Introduction to Cast Irons / 797

Fig. 17

Mean specific heat for cast iron (4.22C1.48Si-0.73Mn-0.12P-0.032S) in vicinity of its melting temperature. Source: Ref 27

The SI units for thermal conductivity are
C. W/m Electrical conductivity, s, is a measure of the ability of a material to transfer electricity when a voltage potential is applied. In engineering applications, its reciprocal, the resistivity, r = 1/s, is convenient. Conductivity and resistivity can be expressed per volume or per weight. Volume resistivity is calculated as:



r ¼ ðA=LÞR

(Eq 18)

where R is the electrical resistance of the material through which the current is passing, A is the cross-sectional area perpendicular to the direction of current flow, and L is the length of the conductor. The SI units are O m. Since free electrons are responsible for electrical and thermal conduction of metals, they are related through the Wiedemann-Franz law:



L0 ¼ ðk=sT Þ

Fig. 16

Specific heat as a function of temperature for (a) gray iron and (b) white iron. Source: Ref 25, 26

(Eq 19)

where, theoretically, L0 is a constant independent of temperature and type of metal, k is the thermal conductivity, s is electrical conductivity, and T is absolute temperature. The theoretical value of L0 is 2.44 108 O W/K2. However, the experimental value for iron is 2.66 108. This relationship does not apply to graphitic cast irons. Once established for metal, the relationship can be used to estimate the difficult-to-measure thermal conductivity in terms of the much easier-to-measure resistance, r:



Table 8

Latent heat of fusion for cast iron Latent heat

Type of iron

Composition(a), wt%

kJ/kg

Btu/lb

Reference

Pure iron Gray iron Gray iron White iron White iron Gray iron Gray iron Gray iron Ductile iron White iron

0C 2.5–3.5 C; 1.5–2.5 Si 4.22 C, 1.48 Si, 0.73 Mn 4.31 C, 1.11 Si, 0.53 Mn 4.35 C 3.4 CE 3.9 CE 4.3 CE ... 4.35 C

274 209–230 197 195 247 264(b) 257(b) 246(b) 258(b) 247

118 90–99 85 84 106 113 110 106 111 106

21 9 22 27 28 29 29 29 29 30

(a) CE, carbon equivalent. (b) Data obtained from cooling curve analysis

average CTE is given for an applicable interval. The temperature range should be specified. For a comparison of alloys, Table 9 lists CTEs for the range 20 to 200
C (68 to 390
F). These are linear coefficients; the rate of volumetric expansion for isotropic material is taken as av = 3al. The amount of carbon in iron will decrease al, and in the form of graphite, al is anisotropic.

Conductive Properties Thermal conductivity, k, and electrical conductivity, s, are related. The ability of a material to transfer heat is expressed by: Q ¼ kðdT=dxÞ

(Eq 17)

where Q is the heat flux, k is the thermal conductivity, and dT/dx is the temperature gradient.

k ¼ L0 T =r





(Eq 20)

Thermal conductivity plays an important role in components subjected to thermal stress. The higher the thermal conductivity, the lower the thermal gradients throughout the casting, and therefore, the lower the thermal stresses. The microstructure of cast iron, especially graphite morphology, greatly influences thermal conductivity. Graphite exhibits the highest thermal conductivity of all the phases. The conductivity of graphite parallel to the basal plane is approximately four times higher than that perpendicular to its basal plane. Because of this, and because the path of heat transport in gray iron may be mostly through graphite, gray iron has higher thermal conductivity than does ductile iron.

798 / Cast Irons Table 9 Linear coefficient of thermal expansion of various cast irons





Temperature range: 20 to 200 C (68 to 390 F) Coefficient of linear expansion (CTE) Iron

106/
C

Composition or microstructure

Gray iron Gray iron, medium Si Gray iron, high Si Ductile iron Ductile iron Ductile iron Ductile iron Ductile iron Ductile iron White iron White iron

3.06–3.74% C 2.25% C-5.8%Si 0.41% C-14.7%Si Ferritic 60–45–10 Ferritic 60–45–10 Ferritic 100–70–03 Heat-resistant 30% Ni Pearlitic 80–60–03 Ni-Resist Martensitic Low-carbon abrasion resistant

11.1–12.2(a) 11.6(a) 13.7(a) 11.7–11.8 11.7–11.9 10.8 12.6–14.4 11.8–12.6 12.4–18.7 11.3–11.9(b) 12(b)

106/
F

Reference

6.17–6.78 6.44 7.61 6.50–6.56 6.50–6.61 6.00 7.00–8.00 6.56–7.00 6.89–10.4 6.28–6.11 6.67

31 9 9 15 32 22 32 15 32 16 32

(a) 0–300
C (32–570
F), CTE decreases with increasing %C. (b) 10–260
C (50–500
F)

Table 10

Thermal conductivity of iron, steel, and cast iron

Composition, wt% or microstructure Alloy No.

1 2 3 4 5 6 7 8

Material

Iron Steel

72 51 48 45 32

50 43 41 37 ...

19 33 33 33 34

59.5 58.6 55.7 49.0

46.9 46.1 46.9 40.2

9 9 9 9

53.2 44.0 41.4

43.5 40.2 39.3

9 30 30

41.4

40.2

30

46.9

42.7

30

Ferritic Ferritic Pearlitic-ferritic Pearlitic-ferritic Pearlitic Pearlitic Hardened and tempered

36–44 36.5 29–31 35.5 29–31 31.4 33.5

31–36 36.0 27–29 35.05 27–29 30.95 33.05

17 35 17 35 17 35 35

... ... ... ...

36 37 30 30

White Iron Gray iron High carbon High carbon High carbon Pearlitic

4.10 3.73 3.58 3.18

2.44 1.66 1.52 2.08

Ferritic (annealed) ... ... Cylinder head UTS = 306 MPa (44 ksi) 3.24 2.00 Cylinder head UTS = 342 MPa (50 ksi) 3.33 2.06

12

Cylinder head UTS = 364 MPa (53 ksi) 3.21 2.03

13

Cylinder head UTS = 263 MPa (38 ksi) 3.70 1.89

15

80–55–06

16

100–70–03, 120–90–02



400
C (750
F) Reference

Si

... ... 0.23 . . . 0.4 ... 1.2 ... 2.41 0.12

Ductile Iron 60–40–18, 65–45–12



100
C (212
F)

C

9 10 11

14

Thermal conductivity (W m1
C1) at temperature

Other

... ... ... ... ... 0.092 Mn, 0.068 S, 0.052 P 1.12 Mn, 0.059 S, 0.087 P 0.53 Mn, 0.101 S, 0.06 P 0.6 Mn, 0.112 S, 0.097 P, 0.14 Cr Same as above 0.70 Mn, 0.076 S, 0.10 P 0.71 Mn, 0.041 S, 0.14 P, 0.33 Cr, 0.32 Mo 0.68 Mn, 0.043 S, 0.13P, 0.32 Cr, 0.30 Ni, 0.33 Mo 0.69 Mn, 0.055 S, 0.12 P, 0.29 Cr, 0.33 Mo

17

100–70–03, 120–90–02

18 19 20 21

Compacted graphite iron Ferritic Ferritic Ferritic Pearlitic

3.61 2.54 0.05 Mn 3.41 2.05 0.4 Mn, 0.7 Ni, 0.28 Ti 3.70 1.70 0.3 Mn ... ... ...

38.5 41.9 49.4 29.3

22 23

Malleable iron Blackheart ferritic Blackheart ferritic

2.51 1.01 0.56 Mn, 0.183 S, 0.047 P 2.51 1.06 1.16 Mn, 0.159 S, 0.0.45 P

49 44

46 41

38 38

24 25

Alloy iron NI-Resist D2 (ductile) Nicrosilal (flake graphite)

... 1.81 ... ... 2.75 1.7 2.7 1.3 ... ...

13.4 30 38–42 38–42 37 21 33 16.5 14.2 13–15

... 26 ... ... 34 23 30 19.1(a) 18.8(b) ...

17 39 16 16 37 9 40 41 16 16

27 28 29 30 31 32 33

Ni-Resist (flake graphite) Sllal High chromium High aluminum Ni-Hard 1, Ni-Hard 2 Ni-Hard 3, Ni-Hard 4

... 6.42 4 ... 6.49 1.5 0.96 0.5 ... ...

UTS, ultimate tensile strength. (a) At 500
C (930
F). At 450
C (750
F)

Austenitic Austenitic Austenitic Austenitic

18–22 Ni 18.7 Ni, 2.0 Cr 20 Ni, 3 Cr 15–30 Ni, 2–5 Cr ...

30 Cr 0.58 Mn, 7 Al, 0.95 Cr 0.5 Mn, 28–30 Al White 3.3–5 Ni, 1.5–2.4 Cr White 4–6 Ni, 1.4–9 Cr

In turn, ductile iron will be better than steel from this standpoint. However, the improvement produced by spheroidal graphite over the steel matrix is relatively small. Not unexpectedly, as the amount of graphite increases, thermal conductivity is also improved. The higher conductivity of high-carbon irons makes them particularly suitable for applications involving thermal shock, such as ingot molds and brakes. The thermal conductivity of ferrite is reduced by dissolved alloying elements. The thermal conductivity of selected ferrous materials is given in Table 10. It is seen that carbon decreases the thermal conductivity of steel (alloys 1–3), where it acts as an alloying element (interstitial solid solution), but increases that of cast iron when much of the carbon is present as flake graphite (alloys 5–8). Increasing the carbon content in these irons results in higher conductivity. Ferritization increases surface energy, l, since it increases the amount of graphite (compare alloys 8 and 9). White iron (alloy 4), where the carbon is tied as cementite, has significantly lower thermal conductivity than both steel and gray iron. Compacted graphite iron has intermediate conductivity between gray and ductile iron. Alloying decreases the conductivity. White iron alloyed with chromium (alloy 29), or aluminum (alloy 31), or chromium and nickel (alloys 32, 33) has the lowest thermal conductivity. Thermal conductivity decreases linearly with increased temperature as described by: lT ¼ l0  aðT  100Þ

(Eq 21)

where l0 is the conductivity at 100 C (212
F) (in W/m
C), and T is the temperature. For a temperature range of 100 to 450
C (212 to 840
F), the coefficient a is 0.015 to 0.019 for gray iron, 0.008 for austenitic iron (Ref 9), and 0.01 for blackheart malleable iron. Data on compacted graphite (CG) iron (Ref 42) demonstrate that the thermal conductivity of CG iron is approximately 25% less than that of class 35 gray iron (minimum tensile strength of 250 MPa, or 36 ksi) at room temperature and 15 to 20% less at 400
C (750
F). Compositional changes within the practical production range of 3.5 to 3.8% C result in thermal conductivity variation of a maximum 10%. The data summarized in Fig. 18 show that an increase in nodularity from 10 to 30% causes a further 10% decrease in the thermal conductivity of CG iron relative to that of flake graphite iron. In addition, from Fig. 18 it is seen that while the thermal conductivity of gray iron decreases with temperature, that of CG increases. This is contrary to the data in Table 10. However, the presence of even relatively small amounts of flake graphite in a predominantly CG microstructure (1 and 3% nodularity) also causes conductivity to decrease with temperature. Electrical resistivity increases with both temperature and solutal elements. The resistivity of iron increases when other elements are added, both because elements dissolved in






Introduction to Cast Irons / 799 Temperature, °F

55

32

212

392

Temperature, °F

572

32

932

752

392

572

752

932

55

50

50 Theramal conductivity, W/mK

Gray, 3.25% C Theramal conductivity, W/mK

212

45

40

35

(−3%) (3%) (30%) (70%)

30

(100%)

Gray, 3.25% C

45

40

(−1%) (10%) −(75%) (95%)

35

30

25 0

200

0

500

Temperature,°C

(a)

Fig. 18

400

300

4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

300

400

500

Temperature,°C

(b)

Electrical resistivity of copper-and iron-base alloys Composition (wt%), type, or microstructure

1 2 3

200

Variation of thermal conductivity of gray and compacted graphite (CG) iron as a function of temperature and nodularity. CG iron 3.7–3.8 C. (a) CG iron 95–100% pearlite. (b) CG iron 70–80% pearlite. Source: Ref 42

Table 11 Alloy No.

100

Material

C

Electrolytic copper Iron Gray iron

... ... 3.8

Gray iron

3.01 3.05 3.05 3.52 3.52 3.52 2.95 3.00 2.99

Gray iron

Gray iron

Ductile iron Austenitic gray iron Nomag Ni-Resist 1 Ni-Resist 1b Ni-Resist 2 Ni-Resist 2b Nicrosllal Ni-Resist 4

<3 <3 <3 <3 <3 <2.5 <2.5

Austenitic ductile iron Ni-Resist D-2 <3 Ni-Resist D-4 <2.6 High-sillcon gray iron 2 2.76

High-aluminum iron

2.4 1 1.5 1.25

Si

Other

... ... 2.5



Electrical resistivity, mV m

Reference

... ... Coarse lamellar graphite Medium lamellar graphite Fine lamellar graphite 1.43 0.53 Mn 1.92 0.72 Mn, 0.03 S, 0.045 P 3.08 0.73 Mn, 0.02 S, 0.072 P 1.63 0.6 Mn, 0.04 S, 0.05 P 1.63 Same as above + 0.86 Ni 1.63 Same as above + 1.81 Ni 0.90 0.6 Mn, 0.07 S, 0.04 P, 0.06 Al 0.91 Same as above but 1.18 Al 0.93 Same as above but 3.34 Al 60–40–18, 65–45–12, 80–55–06 100–70–03

0.0159 0.098 1.036 0.914 0.774 0.537 0.713 0.904 0.864 0.991 1.032 0.441 0.635 0.937 0.50 0.54

9 9 35

1.5–3 1–2.8 1–2.8 1–2.8 1–2.8 4.5–5.5 5–6

1.2 1.4 1.1 1.7 1.2 1.6 1.6

43 43 43 43 43 43 43

1.02 1.1 1.5 1.88 1.31 2.18 0.5–0.63 2.4 1.5

43 43 44 45 45 45 44 44 44

0.8 0.85 0.295

46 46 9

13 15 15 20 20 20 30

Ni, Ni, Ni, Ni, Ni, Ni, Ni,

7 6 6 2 3 5 5

Mn Cu, 2 Cr Cu, 3Cr Cr Cr Si, 3 Cr Si, 5 Cr

1.5–3 20 Ni, 2 Cr 5–6 30 Ni, 5 Si, 5 Cr 4–6 Silal 4.76 0.05 Mn, as-cast Same as above, annealed 7.38 0.1 Mn, as-cast 16 0.5–3 21–28 Al <0.5 28–32 Al

Martensitic white iron Ni-Hard 1 3.3 0.4 Ni-Hard 3 1.3 0.5 Malleable iron 1.26 0.83

3.3–4.8 Ni, 1.5–2.6 Cr 4–4.75 Ni, 1.4–1.8 Cr Blackheart ferritic

9 9 9 9 9 9 9 9 9 17 17

ferrite increase its resistivity and because any other metallographic constituent has a resistivity higher than ferrite. Carbon strongly increases resistivity, mainly because of formation of graphite that has significantly higher resistivity than iron. Coarse graphite increases the resistivity more than fine graphite (alloy 3 in Table 11). Silicon is also a strong contributor to higher resistivity (alloys 4 to 6 in Table 11), since it increases the amount of graphite and also increases the resistivity of ferrite (solid-solution effect; its ionic radius is much smaller than that of iron). Phosphorus has little effect up to 0.2%. Above this value, its graphitizing effect and formation of phosphide eutectic will increase resistivity. The effect of other alloying elements on resistivity is presented in Fig. 19. Manganese has little effect as long as it is balanced by sulfur. Nickel seems to have little effect, too. However, according to other data, nickel can increase resistivity significantly (see alloys 7 to 9 in Table 11). The differences in the literature may be explained as follows. If nickel additions only affect the solid solution, the effect on resistivity is minimal, since it has an ionic radius very close to that of iron. However, nickel has a graphitizing influence. If nickel additions result in a higher amount of graphite, then a significant increase in resistivity may be expected. Aluminum increases resistivity considerably, both because it increases the resistivity of ferrite (ionic radius much smaller than iron) and because it is a graphite promoter. Some typical

800 / Cast Irons Table 12 Comparison of magnetic units Property

Magnetic field strength

Magnetic induction

Magnetic induction, intrinsic Absolute direct current (dc) permeability

Fig. 19

Influence of aluminum, manganese, and nickel on the electrical resistivity of gray iron at room temperature. Source: Ref 47

data for gray iron are given in Table 11 (alloys 10 to 12). The high resistivity of aluminum irons makes them suitable as power resistors. Ferritic ductile iron has the lowest electrical resistivity of all cast irons (alloy 13 in Table 11). For the same chemical composition, both the matrix and graphite shape will affect resistivity. The resistivity increases with silicon, but at the same level of silicon, a pearlitic iron will have higher resistivity, because pearlite has higher resistivity than ferrite (alloys 13 and 14). Finer pearlite appears to have higher resistivity than coarser pearlite. Austenitic irons have higher resistivity than pearlitic grades, because the alloyed austenite has higher resistivity than the pearlite (alloys 15 to 23). Annealing decreases resistivity when pearlite is transformed into ferrite, as seen by comparing alloys 25 and 26. Malleable irons have very low resistivity (alloy 33) because of low carbon and silicon contents and because graphite is present in a compact form.

Magnetic Properties Definitions of selected magnetic properties follow, and a comparison of magnetic units in preferred (SI) and customary (cgs-emu) forms is listed in Table 12 (Ref 48):  Magnetic field strength (H) is the change in  

 

space created by a change in current density. Amp(turn)/m Magnetic induction (B), or flux density, is the strength of the induced field that permeates space near a magnet or magnetic coil. Tesla (T) Magnetic permeability is m = B/H. Although the formula is simple, there are a number of magnetic permeability measurements. ASTM A 340 (Ref 49) provides 21 alternating and direct current permeabilities. m = m0 mr, where m0 is the permeability of free space, and mr is the relative permeability. Magnetic susceptibility (k) is closely related. k = mr 1 Remanent magnetism, or remanence, is the magnetic induction that remains in a

Permeability of free space Magnetic constant Permeability, normal dc Permeability, relative Permeability, intrinsic Susceptibility Susceptibility, mass Density

Symbol

Si

H (units) (Equals) (Equals) (Equals) B (units) (Equals) (Equals) B1 (formula) mabs (formula) (Units) (Equals) m0, mv (assigned values) Gm(assigned values) m (formula) mr (formula) mr (units) mi (formula) k (formula) (Units) w (formula) (Units) d (equals)

amp (turn)/m 1 amp (turn)/m (1000/4p) amp (turn)/m 79.59 amp (turn)/m tesla (T) 1T 104T BGmH DB/DH T/(amp/m) = H/m H/m 4p 107H/m 4p 107H/m B/H mabs/Gm Unitless mabs  Gm = Bi/H Bi/GmH = mr  1 Unitless k/d (cm3/g) kg/m3

Customary (egs-emu)

oersted (Oe) 4 p 103 Oe 1 Oe 1 Oe gauss (G) 10,000 G 1G BGmH DB/DH G/Oe ð1=4Þ 107 G/Oe 1 1 B/H mabs/Gm = mabs Unitless mabs  Gm = mabs  1 Bi/4pGmH = (mr  1)/4p Unitless k/d cm3/g 103 g/cm3

Source: Ref 48

magnetic circuit after the removal of an applied magnetic force.  Coercive force (Hc) is the opposing magnetic force that must be applied to remove the remanence. It is a measure of the strength with which the material retains its magnetism.  Hysteresis loss is the difference between the B/H ratio for increasing and decreasing values of magnetic intensity; it is an irreversible phenomenon due to energy dissipation (heating). The energy lost per unit volume per cycle is the magnetic core loss. Cast irons are employed in applications where their magnetic properties are of significance. The advantages of cast irons are their ability to be cast into intricate shapes, less effect of stress on magnetic properties, and a lower effect of temperature on the loss of magnetism. Magnets are described as hard or soft from the historical use of tool steels as magnets. Cobalt steels, cobalt rare earth alloys, Fe-NiCu, and Ni-Fe-Al alloys are used for hard permanent magnets. As with all other physical properties, the magnetic properties of iron-base alloys vary with both composition and structure. However, the influence of composition is indirect, that of structure being decisive. Some characteristic magnetic properties of cast iron alloys are given in Table 13. Typical magnetization curves for several iron-base materials are shown in Fig. 20. It is seen that pure iron has the highest magnetic induction (B). Carbon decreases B significantly. Note that the magnetic field strength is logarithmic. A more compact graphite morphology, such as that in ductile and malleable iron, reduces B less than flake graphite. In addition, spheroidal graphite will produce higher permeability and lower hysteresis loss when compared to flake graphite irons (alloys 3 and 7 in Table 13). The transition from coarse

type A graphite to fine type D graphite, accompanied by the total disappearance of pearlite, increases the induction, remanent magnetism, and the coercive force (Ref 50). Iron carbides reduce magnetic induction, permeability, and remanent magnetism but increase the coercive force and hysteresis loss. The magnetic induction is significantly affected by the matrix; ferrite is characterized by higher B than pearlite or martensite. This can be seen from the complete magnetization (B-H) curves of the gray irons (Fig. 21). It is also seen that the ferritic iron has a considerably steeper permeability curve than the pearlitic one. By comparing alloys 2 and 3 in Table 13, it is noticed that pearlite has lower permeability and higher hysteresis loss than ferrite. As the ferritic grain size increases, the hysteresis loss decreases. Austenite is nonmagnetic, but the presence of martensite can make austenitic irons weakly magnetic. Increasing the temperature results in a significant increase in permeability up to the Curie point. The Curie point is the temperature at which the ferromagnetic-to-paramagnetic transition occurs in materials. For pure iron, the Curie temperature is 768
C (1414
F) and decreases with increasing silicon. At 5% Si, it is 730
C (1350
F) (Ref 51). For silicon-free iron carbide (i.e., up to 4.8% Si), the Curie point is 205 to 220
C (400 to 430
F) (Ref 52). An austenitic matrix results in nonmagnetic properties. An austenitic matrix can be produced through the use of nickel and/or manganese. The drive for the use of manganese is mainly to replace nickel and thus lower the cost. Several alloy groups have been developed. On the basis of the main alloying elements, they can be classified as follows (Table 13):  Nickel irons (alloys 11 to 24) have good

mechanical, high-temperature and corrosion resistance.

properties,

Introduction to Cast Irons / 801 Table 13 Alloy No.

1 2 3 4 5 6 7 8 9 10

Magnetic properties of selected cast irons Material

Gray iron Pearlitic Ferritic Ductile iron 60–40–18 80–55–06 100–70–03 Ferritic Ferritic Pearlitic Blackheart malleable

Composition, %

3.1 C. 2.2 Si 3.34 C, 0.83 Si, 0.33 Mn Same as above ... ... ... 3.64 C, 1.41 Si, 0.3 Mn 2.9 C. 2.61 Si, 0.72 Mn, 2.2 Ni Same as above 1.26 C, 0.83 Si, 0.26 Mn

20 21 22 23 24

Austenitic gray iron Nomag ... Ni-Resist 1 ... Ni-Resist 1b Ni-Resist 2 Ni-Resist 2b Nicrosllal Ni-Resist 4 Austenitic ductile iron Nodumag Ni-Resist D-2 Ni-Resist D-2B Ni-Resist D-2C Ni-Resist D-4

13 20 20 22 30

25 26 27 28 29

Austenitic iron ... ... ... ... ...

5–10 7–12 7–16 7–15 <1.5

11 12 13 14 15 16 17 18 19

Remanent magnetism, T

Coercive force, A/m

Hysteresis loss, J/m3, at

mmax, mH/m

H for mmax, A/m

... 0.521

557 637

2696 2645

315 400

... 1114

0.618

199

938

1703

239

0.560 0.580 0.620 0.606

159 461 875 119

550 2050 2550 450

1432 517 314 2509

... ... ... 239

0.464

199

700

1445

318

0.540 0.619

915 119

3039 494

364 2664

1194 175 m at 0.01 T

Composition, % 13 Ni, 7 Mn 11–12 Ni, 4–7 Mn 15 Ni, 6 Cu, 2 Cr 15.5 Ni, 5.17 Cu, 1.81 Cr 15 Ni, 6 Cu, 3 Cr 20 Ni, 2 Cr 20 Ni, 3 Cr 20 Ni, 4 Si, 3 Cr 30 Ni, 5 Si, 5 Cr

Ni, Ni, Ni, Ni Ni,

7 Mn 2 Cr 3 Cr 5 Si, 5 Cr

Ni, 5–10 Mn, <5 Cu Mn, <4 Cu Mn, 1.5–3.5 Cu, 3.5–4.5 Si, 2–4.5 Al Mn, 4–5 Si, 1–5 Al Mn, <3 Cu, <5 Si, 20–34 Al

1.4–3.8 1.33 1.4–3.8 2.64 1.4–3.8 1.4–3.8 1.4–3.8 1.4–3.8 ...

1.27 ... 1.29 ... 2.26 1.31 1.63 1.76 >2.5 m at 0.03 T

... ... ... ... ...

1.27 1.28–1.31 1.31–1.36 2.43 1.26

1.4–3.8 1.4–3.8 1.4–3.8 1.4–3.8 1.265

... ... ... ... ...

Source: Ref 9, 16, 17

Fig. 22

Ultrasonic velocity versus (a) visually assessed nodularity in ferritic and pearlitic iron and (b) tensile and yield strength of ductile iron castings. Source: Ref 53

particular, in thin sections. As stated before, martensite increases magnetism. Chromium tends to promote carbides that reduce permeability.

Acoustic Properties

Fig. 20

Standard magnetization curves for selected iron-base alloys. Source: Ref 13

 Nickel-Manganese-Copper irons (alloy 25)

Fig. 21

are less expensive but have poorer corrosion and high-temperature properties.  Manganese irons (alloys 26 to 28) are the most economical but also have the lowest mechanical properties and corrosion resistance.

Ref 50

Increased carbon content may be necessary in some grades to avoid carbides. Higher

Complete magnetization (B-H) curves for ferritic and pearlitic gray irons. Source:

carbon will improve machinability but decrease mechanical properties. Changing graphite shape from lamellar to spheroidal remarkably improves the mechanical properties. The alloying element content must be selected so as to prevent martensite formation, in

Because ultrasonic velocity and resonant frequency (sonic testing) values are directly related to the microstructure, their measurement provides reliable methods for verifying the structure and properties of castings. They are both related to the modulus of elasticity. In cast iron, the change from flake graphite to nodular graphite is related to an increase in both modulus of elasticity and strength; therefore, ultrasonic velocity or resonant frequency measurement can be employed as a guide to nodularity, strength, and other related properties (Fig. 22). Measurement of ultrasonic velocity in cast iron (3.59 to 3.8 C, 1.92 to 2.29 Si, 0.21 to 0.31 Mn) produced by in-mold treatment (Ref 54) shows that the range of ultrasonic velocity for flake graphite iron was 4270 to 4620 m/s (14,000 to 15,000 ft/s), while for compacted graphite irons it was 5260 to 5690 m/s (17,000 to 19,000 ft/s).

802 / Cast Irons Since the sound velocity depends on graphite morphology and matrix structure, it can be used as an indicator of mechanical properties. Both tensile and yield strength of ductile iron increase as the sound velocity increases (Fig. 22b). Acoustic velocity and resonant frequency testing are routine methods of inspection and structure guarantee for foundaries (Ref 55). Typical correlations between resonant frequency and some tensile and yield strengths for compact graphitic iron are given in Fig. 23. More details on ultrasonic inspection can be found in Nondestructive Testing and Quality Control, Volume 17 of ASM Handbook (Ref 56).

Heat Treatment Iron castings afford the designer a desirable selection of mechanical and physical properties at low cost. The range of properties is achieved through a working knowledge of these materials, in terms of their response to casting and to various heat treatments. Heat treating is used to:     

Relieve internal elastic stresses Improve machinability Increase toughness and ductility Increase strength and wear resistance Reach a necessary step in the production process, that is, austempering gray or ductile irons, producing malleable iron castings, or annealing irons cast in permanent molds

It is obvious that most iron-casting producers would like to produce their castings to specification in the as-cast condition. However, there are situations in which the attainment of specified microstructural and mechanical properties requires heat treatment in addition to the as-cast

Fig. 23

Tensile and yield strength of spheroidal graphite iron test bars versus resonant frequency. Source: Ref 55

thermal cycle. It is estimated that a maximum of 5% of the iron castings require heat treatment, a number that has been steadily declining since the early 1980s. This decline has been due, in part, to the ever-tightening controls of chemical composition and processing. Heat treatment includes:  Stress relieving: use of special heating/cool-

   

ing cycles to relieve residual stresses caused by temperature differences in various parts of the casting Annealing: slow cooling of the austenite matrix to achieve maximum softness and machinability Normalizing: cooling the casting in air to obtain higher hardness and strength Through hardening: heating, quenching, and tempering to provide the highest possible hardness and strength Surface hardening: flame, induction, or laser heating the surface of a casting to increase its wear resistance

Caution should be exercised in specifying which heat treatment is to be used. For example, normalizing is neither stress relieving nor a normal procedure for iron castings. Some types of iron castings are routinely heat treated as a regular part of their production process. For example, malleable iron castings are subjected to a malleablizing treatment; the highest ductility grades of ductile iron are often annealed; permanent mold cast gray irons are annealed or normalized; and pearlitic malleable iron and high-strength grades of ductile iron may be either air or liquid quenched and tempered to obtain their specified strength and hardnesses. For more details of the metallurgy of heat treatment of cast irons, see Heat Treating, Volume 4 of ASM Handbook (Ref 57). Stress Relieving. Castings can retain residual elastic stresses, especially when they have a complex shape. In most castings where section size is relatively uniform throughout, these stresses are low in value and of no consequence. However, the presence of significant

Fig. 24

residual stresses can result in casting distortion or warpage, which is most likely to become evident during machining. If some of the stressed metal is removed by machining, the casting will change in shape to rebalance the remaining residual stresses. Chucking or cutting forces in the machining and grinding operation can also cause distortion. However, the large majority of iron castings in commercial production do not require a stress-relieving heat treatment, either because they do not contain significant residual stresses or because these stresses are removed during other types of heat treatment. Annealing is a process that softens cast iron by slow cooling the austenitic matrix through its critical temperature range, with the purpose of creating the equilibrium structure of ferrite and graphite. The annealing of iron includes one aspect not present in steel: The carbon content of the matrix (the combined carbon) can be reduced to zero by slow cooling so that only ferrite and graphite remain in the microstructure. The presence of significant quantities of silicon in cast iron promotes precipitation of the carbon on existing graphite. Several types of annealing can be employed to reduce the hardness of iron. These can also reduce the strength and increase the ductility and impact properties of ductile irons. Annealing can relieve residual stresses in castings if the slow cooling is continued to a low enough temperature. Annealing is mandatory for some grades of ductile iron. Figure 24 presents a schematic representation of three types of annealing (high, medium, and low or subcritical) for iron castings and compares them to stress relieving and normalizing. Annealing in its various forms provides a controlled means of achieving maximum softness and machinability or a partial softening effect with the retention of a higher strength. Recommended practices for annealing of gray and ductile iron are summarized in Tables 14 and 15.

Schematic representations of heating and cooling cycles used to stress relieve, anneal, and normalize iron castings. Source: Ref 58

Introduction to Cast Irons / 803 Table 14

Recommended practices for annealing gray iron castings Temperature(a)

Type of anneal

Purpose







C

F

Time

Cooling rate(b)

Furnace cool (55
C/h, or 100
F/h) to 315
C (600
F). Cool in still air from 315
C (600
F) to room temperature Furnace cool to 315
C (600
F). Cool in still air from 315
C (600
F) to room temperature

700–760

1300–1400

45 min/in. of cross section

815–900 Medium-temperature For conversion of pearlite full anneal to ferrite in irons unresponsive to lowtemperature annealing. For elimination of minor amounts of well-dispersed carbides in unalloyed irons 900–950 High-temperature Elimination of massive (graphitizing) carbides in mottled or full anneal chilled irons and conversion of pearlite to ferrite for maximum machinability 870–950 Normalizing anneal Elimination of massive carbides with the retention of pearlite for strength and hardness

1500–1650

1 h/in. of cross section

1650–1750

1–3 h plus 1 h/in. of cross section(c)

Furnace cool to 315
C (600
F). Cool in air from 315
C (600
F) to room temperature

1600–1750

1–3 h plus 1 h/in. of cross section(c)

Air cool from annealing temperature to below 480
C (900
F). May require subsequent stress relief

Low temperature (ferritizing)

For conversion of pearlite to ferrite in unalloyed irons for maximum machinability

(a) Preferred temperature of castings is dependent on silicon, manganese, and alloy contents. (b) Slow cooling from 540 to 315
C (1000 to 600
F) is to minimize residual stresses. (c) Time for graphitization will vary from 30 min to 1 h in unalloyed irons to several hours in irons containing carbide stabilizers. Source: Ref 59

Table 15

Recommended practices for annealing ductile iron castings Temperature(a)

Type of anneal

Low temperature (ferritizing) Full (for lower-siliconcontent iron)

High temperature (graphitizing)

Two-stage graphitizing and ferritizing

Normalizing and tempering stress relief

Purpose




C




F

Time

In absence of carbides. To obtain grades 60-45-12, 60-40-18 In absence of carbides. To obtain grade 60-40-18 with max. low-temperature impact strength In presence of carbides. To obtain grades 60-45-12, 60-40-18

720–730

1325–1350

1 h/in. of cross section

870–900

1600–1650

Only to equalize at control temperature

900–925

1650–1700

2 h minimum

In presence of carbides. To obtain grades 60-45-12, 60-40-18 where rapid cooling is practical In presence of carbides. To obtain grades 100-70-03, 80-55-08

870–900

1600–1650

1 h/in. of cross section

900–925

1650–1700

2 h minimum

Cooling cycle(b)

Furnace cool (55
C/h, or 100
F/h) to 345
C (650
F). Air cool Furnace cool (55
C/h, or 100
F/h) to 345
C (650
F). Air cool Furnace cool at 95
C/h (200
F/h) to 700
C (1300
F). Hold 2 h at 700
C (1300
F). Furnace cool at 55
C/h (100
F/h) to 345
C (650
F). Air cool Fast cool to 675–700
C (1250–1300
F). Reheat to 730
C (1350
F). 2 h/in. of cross section. Air cool Air quench with fans. Temper at 540–650
C (1000–1200
F). Furnace cool to 345
C (650
F) (55
C/h, or 100
F/h). Air cool

(a) Temperature of castings. (b) Slow cooling from 540 to 315
C (1000 to 600
F) is to minimize residual stresses. Source: Ref 59

Normalizing is the cooling of iron and steel, in air, from a temperature above its critical range. The term came from the wrought steel industry where air cooling is the normal procedure. It is sometimes improperly used to mean stress relieving. A normalizing heat treatment is typically applied to iron castings to obtain a higher hardness and strength than is obtained in the as-cast or annealed condition. The cooling rate of the casting, free of sand and in air, is usually more

rapid than it is in the mold. Normalizing typically produces a fine pearlitic matrix, which combines good wear resistance with reasonable machinability, and an excellent response to induction or flame hardening, provided the cooling rate is fast enough and the hardenability is sufficient for the section thickness. Iron castings may be normalized:  As a separate heat treatment, air cooling can

increase the hardness of castings that were

partially cooled in the sand mold and are too low in hardness.  Following a high-temperature heat treatment to remove excess carbides (free cementite), air cooling provides a microstructure of pearlite rather than the softer ferrite that would result from slow furnace cooling.  After a casting has solidified but before it has cooled to the critical temperature range, it can be removed from the mold and be freed of sand, so as to cool in air more rapidly than if left in the mold. This results in a higher hardness, but complex castings may then require stress relief. Normalizing Procedures. The heating rate for normalizing is not significant, except that large differences in temperature in a complex casting below 540
C (1000
F) can cause distortion or possibly cracking. The holding time at temperature in the heat treating furnace does not need to be extended beyond that necessary to obtain a uniform temperature distribution, because the carbon diffusion rate in iron is rapid at normalizing temperatures. These are typically approximately 50
C (100
F) above the critical temperature range of 810 to 870
C (1500 to 1600
F) for high-strength gray irons, 840 to 900
C (1550 to 1650
F) for lower-strength gray irons, and 870 to 900
C (1600 to 1650
F) for ductile irons. The determining factor is the silicon content, since an increased silicon content increases the critical temperature range. The cooling rate can be increased from that of still air using large fans. Heavier castings and those with thicker metal may require faster cooling. Uniform cooling is also desirable, so the castings must be separated during the air-cooling period. Some large, complex castings may not be suitable for normalizing. Alloyed iron castings and those with higher manganese content have higher hardenability in the pearlitic transformation range and can be satisfactorily normalized with slower cooling or in larger sections than unalloyed types. Complex castings with large differences in metal sections and cored castings with internal as well as external walls will have widely varying cooling rates in different parts of the casting. This not only can result in appreciable differences in hardness from one portion to another, but the casting may also be highly stressed from the differential cooling.

Through Hardening Heat treat hardening of iron castings by through heating, quenching, and tempering provides the highest possible hardness and thus the highest strength. A wide range in hardness is available by selecting the appropriate tempering temperature for use after hardening. Higher hardness provides improved wear resistance and strength, but of course, the ability of parts to deform without breaking decreases as hardness is increased. A quenched and tempered

804 / Cast Irons microstructure is more machinable than is a pearlitic structure of the same hardness. Irons with high silicon content either do not respond well to quenching or are prone to cracking. Heating. For proper hardening, iron must be heated to above its critical temperature range so that its structure becomes austenitic, in which carbon is soluble. The amount of carbon contained in saturated austenite will increase with the austenitizing temperature and decrease with increasing silicon content in the iron. The matrix microstructure or combined carbon of the iron before heating is not significant in furnace heating, because there is normally sufficient time for the solution of carbon into the austenite. With a very short holding time at temperature, such as in flame or induction hardening, the structure prior to heating is the significant factor that determines the carbon content of the austenite on quenching (Ref 60). The diffusion of carbon into austenite from adjacent graphite is rapid, but the solution of sufficient carbon to provide full hardness can require a few seconds to a few minutes at temperature, depending on the graphite spacing, the austenitizing temperature, and the presence of alloys. Castings to be hardened are austenitized in a furnace at a temperature 28 to 55
C (50 to 100
F) above the upper critical temperature for 2 min to 1 h per inch of section thickness , depending on their size and complexity. Furnace time is more dependent on obtaining a uniform temperature in the casting than it is on metallurgical considerations. Heating should be gradual through the lower temperature range so that thermal stresses and the possibility of cracking may be minimized. Heating rates that minimize temperature gradients are particularly important in the austenitization of complex and cored castings that would be subject to warping or distortion on rapid heating. Above 550
C (1000
F), heating to the austenitizing temperature may be conducted as rapidly as desired. Care should be exercised to avoid overheating, because of the adverse effect on properties, oxidation, and distortion. The usual differences in temperature between different locations in the furnace should be minimized. Quenching. After proper austenitization, the castings are quenched. The purpose of quenching is to suppress the diffusional transformations on cooling (too fast for ferrite, pearlite, or ausferrite) and thereby produce the desired hardest microstructure, martensite. In this step, the rate of cooling is very important, since a critical cooling rate exists for each composition or grade of iron essential for the formation of martensite on quenching. If the castings are cooled too slowly, the desired austenite-tomartensite transformation will not occur, and transformation will be to one or more of the softer constituents. Cooling rates, such as attained by water quenching, that are substantially in excess of the critical cooling rate—that cooling rate necessary to obtain the beginning of pearlite, as characterized by the distance measurement

to the first pearlite (Jdp) of Lee and Voigt (Ref 61)—can cause excessive warping, distortion, or even cracking. The seriousness of this problem becomes apparent when it is recognized that the transformation of austenite to martensite causes an appreciable expansion in volume. This expansion must occur, if martensite is to be formed, precisely at the time when adjacent untransformed portions of the casting are undergoing normal thermal contraction. The opposing forces of volume expansion because of martensite formation, and thermal contraction due to cooling, may set up severe internal stresses that can result in warping, distortion, and cracking. Thus, it is important to keep the cooling rate during quenching as slow as possible and still exceed the critical cooling rate. Hot quenching, as discussed in the next section, is a special technique that minimizes this problem. High carbon and silicon contents of iron can result in quench cracking and in retained austenite after quenching. The hardness of properly quenched martensite in various irons may be different, as measured by the conventional Rockwell or Brinell hardness test. Microhardness tests will indicate a Rockwell C hardness equivalent in the 60s, but conventional hardness test values are lower because of the soft graphite in the microstructure. Fully hardened gray irons can range from 48 to 55 HRC, depending on the amount and size of graphite flakes that are present. Fully hardened ductile iron should be in the 53 to 58 HRC range. The wear resistance of irons at

these hardnesses has been demonstrated to exceed fully hardened high-carbon steel. Tempering is a heat treatment that follows quenching in which the casting is raised to a temperature below the critical temperature. This process reduces the hardness of quenched steel or iron by allowing the martensite to transform to a + Fe3C. In the case of normalized irons, the tempering process softens the pearlitic structure by coarsening the existing iron carbides. It is advisable to temper-quenchharden iron castings immediately, to reduce residual stresses, reduce the probability of cracking, and reduce the amount of retained austenite. Where a maximum tempered hardness is desired, a tempering temperature of 150 to 200
C (300 to 400
F) is used. Tempering involves carbon diffusion; therefore, it is both time and temperature dependent, although temperature is much more important than time. The uniformity of tempering is improved if the time at temperature is at least twice the time necessary for the castings to reach the tempering temperature. Suitable tempering temperatures for a desired final hardness are greatly influenced by the composition of the iron and the nodule count or flake structure. Alloyed irons and those with higher hardenability require higher tempering temperatures than do unalloyed irons to obtain the same final hardness. The influence of alloys is demonstrated with gray irons in Fig. 25. A fully hard, martensitic structure will temper at a lower temperature than intermediate transformation

Composition, wt% Iron

A B C D E

Fig. 25

Quenching Temperature

Total C

Si

Mn

Ni

Cr

Mo

3.37 3.13 3.37 3.22 3.21

1.43 1.02 1.47 1.73 2.24

0.58 0.79 0.52 0.75 0.67

... 3.89 3.46 ... 0.02

... ... 1.56 0.03 0.50

... ... ... 0.47 0.52




C

847 788 815 ... ...




F

1550 1450 1500 ... ...

Influence of tempering temperature on Brinell hardness of five oil-quenched unalloyed and alloyed gray irons for 1 h tempering. Composition of irons given in table. Source: Ref 62

Introduction to Cast Irons / 805

Fig. 27

Austempering process for cast irons

Table 16

Summary of filler metals for iron castings

Filler metal

AWS A5.15 classification(a)

Iron base

RCI and ECI RCI-A RCI-B

Copper base

RBCuZn-A RCuZn-B RCuZn-C RBCuZn-D ECuSn-A ECuSn-C ECuA1-A2

Mild steel

ESt

Nickel base

ENi-Cl ENiFe-Cl ENiCu-A ENiCu-B

Composition, %

Fig. 26

Influence of quench and temper heat treatment on mechanical properties of an unalloyed, high-strength gray iron. Impact tests were unnotched. Source: Ref 63

structures (ausferrite). In comparison to normalized castings with a microstructure of fine pearlite, quench-hardened castings will temper to the same hardness with a lower tempering temperature than for normalized castings. Obtainable Properties versus Tempering Temperature. Tempering after hardening reduces the hardness and changes the mechanical properties of the iron. However, the relationship between hardness and tensile properties is not the same for quenched and tempered iron with a microstructure of tempered martensite as it is for as-cast or normalized irons with a microstructure containing pearlite. As shown in Fig. 25, the presence of alloying elements retards tempering. The effects of tempering on the properties of a high-strength unalloyed gray iron that was hardened are shown in Fig. 26 (Ref 63). Austempering is a special heat treatment that, when applied to irons and steels, produces parts that are stronger and tougher than conventionally heat treated, as-cast, or as-formed ferrous materials. Recently, much attention has been paid to this process and its application to ductile iron. The process creates castings that offer advantages over forgings, weldments, and assemblies. The austempering of cast irons is shown in Fig. 27 and consists of the following segments:  A-B: heating the part to the austenitic range

of 800 to 950
C (1475 to 1750
F)  B-C: holding the parts for a time sufficient to saturate the austenite with the equilibrium carbon level  C-D: cooling (quenching) the parts rapidly enough to avoid the formation of ferrite and pearlite to a temperature above the martensite start temperature (Ms); the austempering temperature

Carbon

Silicon

3.25–3.50 2.75–3.00 3.25–3.50 2.00–2.50 3.25–4.00 3.25–3.75

Manganese

Iron

Nickel

Copper

Tin

0.60–0.75 0.50–0.70 0.10–0.40

bal bal bal

Trace 1.20–1.60 0.50

... ... ...

... ... ...

... 0.25–1.25 0.25–1.25 ... ... ... 1.5

... ... ... ... ... ... ...

... 0.04–0.15 0.04–0.15 0.04–0.25 ... ... 0.10

... 0.10–0.50 0.01–0.50 ... ... ... ...

... 57.0–61.0 0.25–1.00 0.20–0.80 56.0–60.0 0.50–1.10 1.00 56.0–60.0 0.75–1.10 9.0–11.0 46.0–50.0 ... ... bal 4.80–5.8 ... bal 7.0–9.0 ... bal ...

0.15

0.03

0.30–0.60

...

...

...

...

2.00 2.00 0.35–0.55 0.35–0.55

4.00 4.00 0.75 0.75

1.00 1.00 2.25 2.25

8.00 bal 3.0–6.0 3.0–6.0

85.0 min 45.0–60.0 50.0–60.0 60.0–70.0

2.50 2.50 35.0–45.0 25.0–35.0

... ... ... ...

(a) See text for a description of classifications.

 D-E: austempering the part in the range of

240 to 400
C (460 to 750
F) for a time sufficient to produce a stable structure of acicular ferrite and carbon-enriched austenite. This structure, called ausferrite, is the structure that produces the desired, high-performance properties.  E-F: cooling the part to room temperature In the higher-silicon gray and ductile irons, the resultant structure is ausferrite. In the lower-silicon malleable irons, the structure is either a mixture of ausferrite and bainite or just bainite. Higher austempering temperatures result in coarser structures that exhibit good ductility and dynamic properties. Lower austempering temperatures produce finer structures that have higher tensile and yield strengths and superior wear resistance.

Welding of Cast Irons Welding of iron castings can enhance their function. Welding and other similar heatrelated joining techniques are used on iron castings for joining of simple parts to fabricate a complex assembly, improve wear or corrosion resistance, repair casting imperfections or

fractures, and to build up surfaces worn or corroded by service. See the article “Repair Welding of Castings” in this Volume for the repair of parts degraded in service. Cast irons require the use of more controlled techniques and parameters for welding or joining, as compared to more commonly joined materials such as steel, aluminum, and nickel alloys. In particular, the relatively high carbon content and low ductility of cast irons require precautions that are normally unnecessary for steels. The heating and cooling involved in these joining processes can directly cause cracking or degradation of material properties of the cast iron. The precautions may include preheating, filler metal selection, joint design, and postweld heat treatment. Some of the filler metals for cast irons are included in Table 16 and are specified in AWS specification A5.15. They are grouped as (Ref 64):  RCI—intended for oxyethylene welding  ECI—covered electrode with cast iron core  Copper group—intended for braze welding

and surfacing

 ESt—a covered electrode with a steel core.

The deposit is not machinable.

 ENi—a covered electrode with a nickel-base

core that is machinable

806 / Cast Irons Factors Affecting Weld Quality. The quality of a weld process is judged not only on the likelihood of cracking during welding but also the useful service life and soundness of the joint. The welding processes expose the weld and adjacent metal to a heating and cooling cycle that can significantly affect the structure and properties of the weld joint. Those portions of the weld altered by this heating and cooling process are commonly referred to as the fusion zone, the partial fusion zone, and the heataffected zone (HAZ). The zones in a typical fusion weld deposited on a flat surface are shown in the crosssectional sketch in Fig. 28. In the HAZ, the iron is not melted but subjected to a heat treatment. The resulting effect will depend on the microstructure and composition of the HAZ, which in turn depends on the maximum temperature reached and the cooling rate. The fusion zone consists of the base material that was melted during the welding operation and the part of the filler metal. Sufficient mixing of the melted base material and the filler metal usually takes place so that the resulting fusion zone is of relatively uniform composition. The amount of base material included in the fusion zone is related to the penetration of the fusion zone into the base material. Therefore, welding operations that have shallow penetration will form a fusion zone composed

Fig. 28

Fig. 29

Cross section of the zones in a typical fusion weld deposited on a flat surface

Ductile iron weld metal, upper left, has smaller graphite nodules than is typical in the ductile iron casting. Original magnification: 100

primarily of the filler metal, while a deeply penetrating process will have a weld bead nearer the composition of the base material. Spheroidal graphite can be obtained in ductile iron welds with the proper technique. The alloying element that induces the spheroidal graphite formation is magnesium. This not only oxidizes readily but vaporizes at molten iron temperatures; therefore, high metal temperature and oxidation must be minimized. Ductile iron or nickel-iron welding rods for ductile iron castings usually contain magnesium, and some also include cerium. When graphite spheres are formed in the weld deposit, the solidification rate is generally higher than that for the base casting, so a finer nodule structure is obtained (Fig. 29). In the zone of partial fusion, only a portion of the base material is melted. Melting usually starts at the interface of the graphite and iron in the structure. Because of the rapid cooling that typically occurs with arc welding, carbides are formed in these molten areas. Stresses. The differential heating and cooling that occur in welding produce thermal stresses in the casting. The base material surrounding the weld is heated and expands rapidly. The built-in mechanical constraint of the casting may result in the upsetting or distortion of the hot, expanding metal in the weld zone. On cooling, this section will not return to its original dimensions, which results in high internal stresses, warping, or even fracture. The solidification shrinkage of the weld metal may contribute to these stresses. When upsetting of the weld zone does not occur during heating, stresses can develop away from the weld zone, which may also lead to warping or fracture. Often, the welding process can be modified to prevent these problems by preheating the entire casting, by using a lower welding heat, or by stress relieving. Porosity can occur in cast iron welds because of entrapped gas. These gases are produced by the atmosphere surrounding the molten metals, by gas or moisture in the flux, by moisture or oil on the welding surface, or by a reaction of the weld metal with rust or slag. Care in welding preparation and procedure control will minimize porosity formation. Porosity can also be reduced if sufficient time is allowed for the entrapped gases to be released from the weld metal prior to solidification. Weld Preparation. The success of welding depends on the preparation and cleaning prior to welding. Any contaminants, such as slag, paint, oxide, grease, sand, and oil, should be removed. The grooves and cavities should be shaped to allow ease of access and manipulation of the welding torch or electrode. The casting should be positioned to allow for welding to be done in the most advantageous position. Generally, a flat weld is desired. Repairs of cracks or casting imperfections require thorough removal of the undesirable area. The weld must be deposited on sound base metal. Thermal removal of defects can

cause growth of the defect or crack. Drilling a hole at each end of a crack reduces the stress concentration at the crack tip and minimizes the probability of growth. Flame or arc gouging methods result in the formation of undesirable structures in the surface of the casting. These structures are similar to those resulting from fusion welding, so these techniques are not recommended. If these methods are used for economic reasons, then the skin and scale should be thoroughly removed by grinding or sanding. Preheating is effective in preventing weld defects because it reduces the temperature differential throughout the casting and reduces the rate of cooling after welding. Exposing parts to be repaired to temperatures of 370 to 480
C (700 to 900
F) should remove most oils and petroleum-based contamination. The decreased temperature difference results in slower cooling of the fusion zone during solidification, thereby reducing the tendency for carbide formation; the slower cooling of the entire weld at lower temperatures minimizes the possibility of martensite formation. Stresses resulting from the expansion of the welded part are also reduced because of the decreased temperature difference. As a result, preheating reduces cold cracking, residual stresses, and distortion. Postwelding Heat Treatment. Heating the welded area or the entire casting after welding may relieve the residual stresses introduced during welding. Gray irons have almost no ductility, so they should be gradually cooled from the welding temperature to ensure against cracks from residual stresses. In some cases, this gradual cooling may be accomplished by covering the casting with insulation or sand, after welding, or by controlling the cooling with a flame. However, it is desirable to put the castings back into the preheating furnace, as soon as they are welded, and then allow them to cool uniformly to room temperature. Where the casting geometry permits, preheating may provide sufficient heat to eliminate the necessity for post-heat treating. Where accuracy of dimensions in machining is important, postweld stress relieving is advisable. The minimum treatment should be heating to approximately 600
C (1100
F), followed by uniform cooling. This stress relief should be employed immediately after welding.

Machining and Grinding Most iron castings require one or more surfaces to be machined or ground when converting the raw casting into a finished component for a product. The machining of iron castings is readily adapted to computer numerical control (CNC) machines. Systems that include a dozen or more CNC machines of different types and associated robotics allow castings to be completely machined by many operations without being rehandled.

Introduction to Cast Irons / 807 Whether manually operated or machined by CNC, the individual machining operations are subject to essentially the same limitations. This section considers how the machining and grinding processes, commonly used in finishing iron castings, can be efficiently performed. Machinability Rating. The relative ease with which iron castings can be machined is an important advantage. The annealed grades can be cut at very high speeds without generating burrs. Coated carbide inserts and ceramic inserts can be used at rates as high as 900 m/min (3000 sfm) with adequate tool life. The harder grades of iron still machine at reasonable rates of speed, yielding economical machining. White irons and the alloyed gray and ductile irons that have an acicular or martensitic matrix are usually abrasively machined. Machinability is a commonly recognized attribute but not a specific property of a material. All machinability ratings are empirical and depend on the conditions that were selected to establish the rating. The machinability rating of materials can change appreciably with different standard conditions of testing. For this reason, depending on a machinability rating number without additional information can be a risky venture. There are three factors on which a machinability rating can be established: tool life, surface finish and accuracy, and power requirement and tool forces. The relationship between cutting speed and tool life is most commonly used as the criterion for machinability, because these factors directly influence machine tool productivity and product cost. Early machinability studies used tool wear as a criterion, with the limit on wear being efficient regrinding of the dull tool. The almost universal application of indexable inserts has transferred the criterion of tool wear to surface finish and dimensional accuracy. Tool life criterion is still important because the machining process must be interrupted to change cutting edges. Research on the machining of iron castings, to establish machinability data under controlled conditions, has kept pace with the developments in cutting tool materials. However, much of this work is privately funded and not published, giving the impression that little or no current machining data have been produced. In practice, the machinability of cast iron is an important economic factor affecting production rates and unit cost. Effect of Microstructure on the Machinability of Cast Iron. The machinability of iron relates directly to its microstructure, whether the evaluation is by tool life, surface finish, or power. The shape of the graphite in the iron determines the type of iron (gray, ductile, malleable, or compacted graphite). The presence of graphite provides the free-machining characteristic of iron, and the shape and amount of graphite establishes the potential surface finish obtainable with a specific machining operation. The microstructure of the metal around the graphite also determines the relative machinability

and the resulting range of cost-effective cutting speeds and feeds. An understanding of the influence of microstructure on machinability can provide insight into more efficient machining processes and the correct solution to problems, without being misled by traditional opinions. Small variations in microstructure can make a difference in machining, which could be very significant in production. Hardness is an indicator of machinability, but by itself it is not a totally dependable measure. Microstructural constituents and hardness effects are demonstrated by the test results shown in Fig. 30. The following sections relate the influence of the various matrix constituents on tool life. Ferrite is the essentially carbon-free constituent that constitutes the entire matrix of fully annealed irons. With the exception of graphite, ferrite has the lowest hardness of any constituent in iron. However, it is not as soft as the ferrite of low-carbon steel because the ferrite in cast iron contains silicon. The moderate hardening effect of the dissolved silicon gives ferrite a clean cutting property. Silicon contents in the usual range of 1.0 to 3.0% have very little effect on tool life, but special irons with higher

Fig. 30

silicon contents have reduced machinability. High-silicon irons, with 14% Si, are considered extremely difficult to machine. Pearlite is the common constituent in irons of medium strength and hardness. It is composed of alternate plates of soft ferrite and hard iron carbide. The thickness of the alternate plates can vary from coarse to fine. The finer structure is stronger and harder, and it machines at lower cutting speeds. The pearlite in iron provides the best combination of machinability and wear resistance. Martensite is the hard constituent formed by quenching during heat treating of cast irons. In its hard, untempered condition, martensite is very difficult to machine; but when tempered to a lower hardness, it can be machined very economically. The relative machinability of tempered martensite depends on the amount (time at temperature) of tempering. It is somewhat more machinable than is pearlite of the same hardness. When martensite is tempered to a relatively low hardness, it forms a structure of spheroidized carbides in ferrite, which is similar to ferrite in machinability. Bainite and other acicular structures usually occur in alloyed irons or are obtained by a hot

Relationship between the matrix microstructure of iron and tool life in machining is shown by these data on milling

808 / Cast Irons quenching heat treatment. They are generally of intermediate hardness and are somewhat difficult to machine. The acicular structures are generally machined at slower cutting speeds than tempered martensite of the same hardness. Austenite is the principal constituent in the high-nickel-content gray and ductile irons that are nonmagnetic. Ni-Resist is the most common name of this group of irons. Austenite is a relatively soft constituent and comparable to ferrite in machinability. However, some types of the austenitic irons contain sufficient levels of chromium to produce chromium carbides in the microstructure. The presence of chromium carbides reduces the machinability much more than the increased hardness would suggest. Carbides are extremely hard constituents, whether they are simple iron carbides or complex carbides containing alloying elements. When located in the thin plates in pearlite, carbides can be readily sheared. When larger carbides form in the iron as a separate constituent, they are very detrimental to tool life (Fig. 31). Carbides can be present at the edge of a casting because of very rapid solidification of the iron in this region. Steadite is a hard constituent that is formed by phosphorus in the iron. With a phosphorus content of less than 0.20%, there is no significant effect on tool life. At 0.40% p, a detrimental effect on tapping and higher phosphorus contents significantly reduce tool life. Manganese sulfide is a small, nonmetallic inclusion that has no measurable influence on the properties of iron. Mixed-matrix structures result in a tool life that is intermediate to those of the individual components, but the combined effect is likely to be somewhat less than a prorated amount. For example, a matrix structure with 50% ferrite and 50% pearlite will probably produce a tool life that is somewhat lower than the

Fig. 31

Free carbides in tool structure can cause a large reduction in tool life but affect only a relatively small increase in hardness.

average to be expected, with an entirely ferritic or pearlitic structure and an entirely pearlitic matrix of equivalent fineness.

Tool Life Of the three characteristics that are commonly measured in evaluating machinability, tool life is usually the most significant. For production operations, tool life is usually expressed as the number of pieces machined per tool change. In machinability testing, tool life is generally defined as the cutting time in minutes to produce a given wear land for a specified set of machining conditions. This cutting time can be converted to cubic inches of metal removed, knowing the machining parameters. Surface Condition. The casting skin or surface metal may not have the same machining properties as the interior metal because of conditions in the casting process, in heat treating, or in cleaning of the casting. The tool life can be appreciably shortened when making the first cut through the skin, if the surface metal is not free of abrasive materials such as sand or black oxide. Sand from the mold can adhere to the casting or actually be fused into the surface metal. Iron that is heat treated in an oxidizing atmosphere can have a surface layer containing oxides and silicates, which are very abrasive to cutting tools. With modern foundry techniques and highenergy cleaning machines, iron castings can be consistently produced with clean surfaces. However, small or complex internal passages may require special cleaning techniques, such as sand blasting with a probe or molten salt cleaning. There may be cases where accepting reduced tool life or lower cutting speed on the first cut would be more economical than are special cleaning techniques. However, with the exclusion of an unusual casting design or requirement, quality castings should not present a skin that is difficult to machine, particularly when a normal depth of cut is used on the first pass. In many modern production methods, absolutely clean casting surfaces are essential. The threading of pipe fittings is an example. It is standard practice to have the tap or die run directly onto the cast surface without a preliminary cut and without excessive tool wear. Automation or continuous automatic machining of iron castings is becoming common practice. The success of this process requires consistently clean and accurate castings. Production tool life is equally as dependent on clean and accurate castings as it is on the specified metallurgical properties of the metal. Hard Spots. Iron castings having a complex shape or large differences in wall thickness are not likely to have the same hardness and machinability throughout because of inherent differences in the cooling rate. Some castings are heat treated to minimize these differences, but with most castings, the difference in machinability from one section to another is too small to be significant. However, a local condition can be present in a casting that seriously affects its machinability.

A hard region in an iron casting can be formed as the result of rapid solidification. This harderthan-normal region may occur in a thin or a protruding section of the casting, at a sharp corner, or adjacent to a joint in the mold that formed a fin on the casting. These geometric conditions can cause abnormally rapid solidification and the formation of hard carbides. The occurrence of a chilled spot in a casting can usually be corrected by a change in the pattern or foundry practice. Hard spots can be removed from castings by heat treatment, either annealing or normalizing, unless the iron contains alloys such as chromium. Some welding methods can also produce a hard zone, as can some abusive grinding conditions. These effects can also be removed by heat treatment, but they are sometimes accompanied by cracking. Occasionally, the cutting edge of a dull tool will catastrophically fail and damage the last surface to be machined. It is possible to misinterpret the damaged surface as the cause of tool failure rather than a result.

Grinding Processes Honing is a low-velocity process using abrasive grains bonded onto sticks. These sticks are, in turn, fastened onto holders that move the abrasive sticks across the surface to be honed. Internal and external cylindrical surfaces and flat surfaces can be abrasively machined in this manner. Honing can produce a finish on iron that is excellent for holding an oil film. Cylinder linings for internal combustion engines are honed to a prescribed finish and scratch pattern to provide a proper seating of the piston rings. The honing tool for internal cylindrical surfaces has a number of abrasive sticks mounted around the periphery of the tool and parallel to its axis. As the tool is rotated and reciprocated, the abrasive sticks are forced out, radially, by a positive wedging action, generating an equal amount of force in all directions. Cylindrical bores from 3 to 1000 mm (1/8 to 40 in.) in diameter and up to 12 m (40 ft) long are being honed. Modern honing techniques remove stock rapidly and accurately. True cylinders are formed by the honing process as a result of the rotational and reciprocal motion of the honing sticks on the surface. Stock removals of up to 2.5 mm (0.100 in.) are being accomplished, with 0.25 to 0.41 mm (0.010 to 0.016 in.) being the more common honing stock allowances. In high production, prehone machine finishes, 3 to 6 mm (125 to 250 min.), are usually specified. This comparatively rough finish allows the abrasive sticks to aggressively remove material without glazing over. Initial contact with this rough surface helps to break the glaze formed on the abrasive sticks as the finish becomes smoother. The abrasive sticks are selected to be self-dressing. The correct combination of motions and pressures causes the abrasive grains to fracture and form new cutting edges at a proper rate.

Introduction to Cast Irons / 809 Table 17

Summary of short-term rust preventatives Thickness

Type

mils

Removal method

Oil

Nondrying mineral oil viscosity determines coating thickness.

5.1–7.6

0.2–0.3

Solvent

Petroleum-based inhibitors and film formers dissolved in petroleum solvents

5.1–50.8

0.2–2.0

Seldom required: solvent rinsing, emulsion spray, or vapor degreasing Seldom required: solvent rinsing or alkaline washing

Emulsified

Wax

Coating structure

Petroleum-based coating modified to form a stable emulsion with water Wax layer applied heated or in a volatile solvent

mm

5.1

0.2

38.1–76.2

1.5–3.0

Removal seldom necessary. Solvent rinsing Solvent rinsing or alkaline cleaning

Uses

Highly finished automotive parts. Galvanized irons External surfaces of machinery parts and tooling. Can be used outdoors Same as above but suggested for indoor use Highly finished parts stored for prolonged periods

Source: Ref 59

Rotational speeds used on cast iron vary from 30 m/min (100 ft/min) for harder types up to 75 m/min (250 ft/min) for the soft grades. Reciprocation speeds are usually between 15 and 23 m/min (50 and 75 ft/min). Lapping is an abrasive machining process used for refining the finish and accuracy of a surface, rather than a method of stock removal. Lapping uses a loose abrasive, frequently suspended in oil or water, between the work and a lap. In this process, size is seldom changed by more than a few ten-thousandths of an inch. The lap may be made of lead, copper, gray iron, or cloth. The term lapping is also used to refer to finishing with a very fine abrasive paper or bonded abrasive. Iron castings are seldom finished by lapping, but gray iron is often used for lapping plates to finish other materials.

Coatings Coatings are applied to iron castings to provide both a pleasing appearance and functional resistance to deterioration from corrosion, erosion, and wear. The proper selection of a coating hinges on knowing what coatings are available, what coatings are suitable, what limitations are imposed by the casting design, and how well the coating will adhere to the casting. Generally, the selection process is greatly enhanced by a close liaison with coating manufacturers. Surface Engineering, Volume 5 of ASM Handbook, provides many articles on cleaning and coating of cast irons. Surface cleaning is a vital first step in virtually all coating processes. It will produce a more suitable surface finish (roughness) for adhesion of the coating as well as remove debris from the surface. Cleaning methods include:     

Mechanical cleaning Nonmechanical cleaning Molten salt baths Pickling Chemical cleaning (alkalines, emulsions)

solvents,

Short-term corrosion or rust prevention for iron castings can be achieved with temporary coatings such as oil, solvent, emulsified, or wax types, as described in Table 17. These coatings are used primarily to protect parts that are to be stored in a sheltered area for periods ranging from a few days to two years. The heavier emulsified- and wax-type coatings are generally used for longer-term storage, while the two lighter coating types, oil and solvent, are used for short-term protection of machined surfaces. Electroplating of casting has employed cadmium, chromium, copper, nickel, tin, and zinc. Occasionally, for special purposes, castings can be electroplated with gold, lead, and rhodium. Alloy coatings, such as brass, bronze, and tin-base compositions, are also used on iron castings for decorative effects. The function of the casting and regulations will limit the materials that may be used. Surface Engineering, Volume 5 of ASM Handbook, provides detailed information on electroplating of cast irons. Thermal spraying is one of several methods of applying coatings to protect substrate materials that are exposed to severe conditions or harsh environments. Thermal spraying is also capable of improving the properties beyond those of the bulk material properties of substrate materials. It differs from hardfacing or fusion coating in that the latter is actually fused to the base metal (forming a metallurgical bond), while thermally sprayed materials mostly adhere through a mechanical bond. Despite the fact that there is a mechanical bond, very high bond strengths can be achieved. Some processes are capable of achieving bond strengths that exceed 69 kPa (10,000 psi). Thermal spraying is capable of producing a wide range of coatings. If a material has a melting point and if it does not sublimate or decompose at temperatures close to its melting point, it can be applied by thermal spray as long as it is available in wire or powder form. Thermal spraying consists of a group of various processes in which powders or wires of metallic or nonmetallic materials are deposited onto a substrate. The common element is that

a heat source is used to convert powders or wires into a spray of molten (or sometimes semimolten) particles. This heat source is either electrical or chemical (combustion).

Materials Selection Traditional selection of a material is based on evaluation of mechanical and physical properties, method of producing the casting, and machining characteristics. The grid in Fig. 32 provides a guide to the advantages of various forms of cast irons (Ref 65). Global environmental regulations along with energy considerations have an increasing impact on the overall cost of manufacture. The cupola method of melting, which was a continuous melting process, created major air pollution problems and has been replaced by electric furnace melting, which has lower energy costs and is well suited for controlled production batches. The advent of ductile iron, with its wide variety of mechanical properties along with better castability than gray iron, has become the primary alloy in the 21st century in the United States. India and China still produce sizeable tonnages of gray iron and malleable iron castings and often compete favorably with ductile iron in high-production quantities. The traditional method of producing iron castings used sand molds. This has been replaced with shell patterns backed by sand, which eliminates the need to recycle or dispose of large quantities of used sand. Shell mold castings have improved dimensional characteristics over sand castings, and investment-cast ductile iron is gaining wide use for smaller parts. Traditional uses of gray iron castings include automotive, machine tool bases, cookware, and sanitary applications such as sewage. Ductile iron has taken over most of the automotive and sanitary applications, but malleable iron still has a place in automotive and water system valves. Although limited in number, the possible volumes for austempered ductile iron (ADI) applications could be significant. Automotive cylinder liners have been successfully used in production for many years. Because of its remarkable noise-damping properties, ADI has been applied by several automotive manufacturers to bearing supports and collars. Because of its noise damping and wear resistance, ADI has been used by some heavy equipment manufacturers for light-duty gears. Austempered ductile iron is an excellent brake rotor material; however, it must be restricted to low-speed and/ or well-ventilated discs due to the instability of the microstructure at elevated temperatures. Austempered ductile iron camshafts have demonstrated excellent contact fatigue properties, and ADI performs well in wear applications where the bending stresses are low and the parts are loaded in compression.

810 / Cast Irons  Agricultural:

Farming and agricultural applications for ADI include plow points, till points, trash cutters, seed boots, ammonia knives, gears, sprockets, knotter gears, ripper points, tractor wheel hubs, rasp bars, disk parts, bell cranks, lifting arms, and a great variety of parts for planters, plows, sprayers, and harvesters.  Defense: The defense industry has been relatively slow to adopt ADI; however, some of the applications include track links, armor, ordnance, and various hardware for trucks and armored vehicles.  Gears: Gears represent some of the bestknown, most widely publicized, and highpotential uses of ADI.  Crankshafts: Crankshafts are another potentially significant application for ADI. The first commercially produced ADI part was a small crankshaft for a hermetically sealed refrigerator compressor. It was cast by Wagner Castings Company (United States) for Tecumseh Products. Production of that part was initiated in 1972; since that time, millions of those crankshafts have been produced.

ACKNOWLEDGMENTS

Fig. 32

Chart showing advantages of various forms of cast irons. Source: Ref 65

The development and commercialization of ADI has provided the design engineer with a new group of cast ferrous materials that offer the exceptional combination of mechanical properties equivalent to cast and forged steels and production costs similar to those of conventional ductile iron. In addition to this attractive performance:cost ratio, ADI also provides the designer with a wide range of properties, all produced by varying the heat treatment of the same castings, ranging from 10% elongation with 870 MPa (125 ksi) tensile strength to 1750 MPa (250 ksi) tensile strength with 1 to 3% elongation. Although initially hindered by lack of information on properties and successful applications, ADI has become an established alternative in many applications that were previously the exclusive domain of steel castings, forgings, weldments, powdered metals, and aluminum forgings and castings. Austempered ductile iron components are now evident in nearly every manufacturing segment. Some examples include the following:







 Heavy truck and bus components: Heavy

truck applications include suspension components such as spring hanger brackets, shock brackets, U-bolt plates, wheel hubs, brake calipers and spiders, knuckles, sway bar components, pintle hooks, and gears for trailer landing gear. Powertrain-related ADI heavy truck and bus components include engine brackets and mounts, timing gears, cams,



annular gears, differential gears and cases, clutch collars, accessory brackets, and pulleys. Light auto and truck components: Light vehicle ADI applications include suspension components, knuckles, spindles, hubs, tow hooks, hitch components, differential gears and cases, engine and accessory brackets, camshafts, engine mounts, crankshafts, and control arms. Constant-velocity joints for four-wheel drive General Motors vehicles have been produced in ADI since 1978 and currently run at volumes of over 5000 per day. Construction and mining components: This segment includes all manner of collets (collars), ring carriers, wear plates, sprockets, covers, arms, knuckles, shafts, rollers, track components, tool holders, digger teeth, cutters, mill hammers, cams, sway bars, sleeves, pavement breaker bodies and heads, clevises, and conveyor components. Miscellaneous industrial: Miscellaneous industrial applications include brackets, lever arms, knuckles, shafts, cams, sway bars, sleeves, clevises, conveyor components, jack components, bushings, rollers, molding line components, fixtures, gears, sprockets, deck plates, and all sorts of power transmission and structural components. Railroad: The railroad industry uses ADI for suspension housings, top caps and friction wedges, track plates, repair vehicle wheels, nipper hooks, and car wheels.

We acknowledge the support of the American Foundry Society (AFS) and the work of George M. Goodrich in adapting portions of Iron Casting Engineering Handbook (ISBN 0-87433-260), 2006, edited by him and revised by the AFS Cast Iron Division, 5A Committee. AFS may be contacted at www.afsinc.org. We also thank Ed Fairman, former Chair of the ASM Handbook Committee, for his assistance. REFERENCES 1. “Standard Specification for Ferritic Malleable Iron Casting,” A 47/A 47M, ASTM International 2. “Standard Specification for Pearlitic Malleable Iron Casting,” A 220/A 220M, ASTM International 3. “Standard Test Method for Evaluating the Microstructure of Graphite in Iron Castings,” A 247, ASTM International 4. “Standard Specification for Austenitic Gray Cast Iron,” A 436, ASTM International 5. “Standard Specification for Austenitic Ductile Iron Castings,” A 439, ASTM International 6. “Standard Specification for CorrosionResistant High-Silicon Iron Castings,” A 518, ASTM International 7. “Standard Specification for AbrasionResistant Cast Irons,” A 532/A 532M, ASTM International 8. “Standard Specification for Austenitic Ductile Iron Castings for Pressure-Containing Parts Suitable for Low-Temperature Service,” A 571, ASTM International

Introduction to Cast Irons / 811 9. H.T. Angus, Cast Iron: Physical and Engineering Properties, 2nd ed., Butterworths, London, 1976 10. I. Jimbo and A.W. Cramb, Metall. Trans. B, Vol 24, 1993, p 5 11. K. Grutter and B. Marrincek, Arch. Eisenhu¨ttenwes., No. 9–10, 1954, p 447 12. A. Goto, T. Aizawa, and S. Okada, Imono, April 1973, p 295 13. Malleable Iron Castings, Malleable Founders Society, Cleveland, OH, 1960 14. H.G. Ghirshovitch, Ed., Spravotchnik po Tchugunomu Litiu, Mashinostrojenie, Leningrad, 1978 (in Russian) 15. C.H. Walton and T.J. Opar, Ed., Iron Casting Handbook, Iron Casting Society, 1981 16. R. Rohrig and D. Wolters, Legiertes Gusseisen, Vol 1, Giesserei-Verlag, Dusseldorf, 1970 17. J.D. Mullins, in Ductile Iron Handbook, AFS, Des Plaines, IL, 1992, p 20 18. R.L. Carden, BCIRA J., Vol 10, May 1962, p 325 19. R.C. Weast, Ed., CRC Handbook of Chemistry and Physics, 67th ed., CRC Press, Boca Raton, FL, 1986 20. J.H. Awbery and A.R. Challoner, J. Iron Steel Inst., Vol 154, 1946, p 84 21. J.R. Pattison and P.W. Willows, J. Iron Steel Inst., Vol 183, 1956, p 390 22. S. Umino, Sci. Rep. Tohoku Imperial Univ., Series 1, Vol 15, Nov 1926, p 597 23. Handbook of Chemistry and Physics, 36th ed., Chemical Rubber Publishing Co., Cleveland, 1954, p 2085 24. I.P. Egorencov, Liteinoe Proizvod. (Russia), No. 7, July 1955, p 20 25. E. Piwowarsky, Hochwertiges Gusseisen, Springer-Verlag, Berlin, 1942 26. H. Esser and E.F. Baerlecken, Arch. Eisenhuttenwes., Vol 14, 1940–1941, p 617 27. S. Umino, Sci. Rep. Tohoku Imperial Univ., Series 1, Vol 16, Oct 1927, p 775 28. E.Z. Schmidt, Metallkunde, Vol 7, 1915, p 164 29. K.J. Upadhya, D.M. Stefanescu, K. Lieu, and D.P. Yeager, AFS Trans., Vol 97, 1989, p 61

30. W. Schmidt, Metallurgie, Vol 7, March 1910, p 164 31. E. Nechtelberger, The Properties of Cast Iron up to 500
C, Technicopy Limited, Stonehouse, Gloucestershire, England, 1980 32. F. Cverna, Ed., Thermal Properties of Metals, ASM International, 2002, p 14–18 33. Physical Constants of Some Commercial Steels, BCIRA, Butterworths, London, 1953 34. H. Masumoto, Sci. Rep. Tohoku Univ., Vol 16, 1927, p 417 35. G.N.J. Gilbert, Engineering Data on Gray Cast Irons, British Cast Iron Research Association, 1968 36. J. Sissener, W. Thury, R. Hummer, and E. Nechtelberger, AFS Cast Met. Res. J., Vol 8, 1972, p 178 37. R.D. Schelleng, AFS Cast Met. Res. J., Vol 3, 1967, p 30 38. K.B. Palmer, BCIRA J., Vol 8, 1960, p 266 39. J.W. Donaldson, J. Iron Steel Inst., Vol 128, 1933, p 255 40. J.W. Donaldson, Proc. Inst. Br. Foundrymen, Vol 32, 1938–1939, p 125 41. A.M. Plesinger, Fonderie, Vol 157, 1959, p 75 42. S. Shao, S. Dawson, and M. Lampic, paper presented at the Materials for Lean Vehicles Conference, The Institute of Materials, U.K., 1997 43. Properties and Applications of Ni-Resist Austenitic Cast Iron, International Nickel Co., 1965 44. L. Sofroni and D.M. Stefanescu, Special Cast Iron, Editura Tehnica, Bucharest, 1974 (in Romanian) 45. J.H. Partridge, J. Iron Steel Inst., Vol 112, 1925, p 191 46. Die Verschleissfesten Ni-Hard Werkstoffe, International Nickel, Dusseldorf, 1967 47. I. Iitaka and K. Sekiguchi, “Reports of the Casting Research Laboratory,” No. 3, Waseda Univ., 1952, p 23 48. C. Moosbrugger, Ed., Electrical and Magnetic Properties of Metals, ASM International, 2000

49. “Symbols and Definitions Related to Magnetic Testing,” A 340, ASTM International, 1996 50. M. Decrop, Fonderie, Vol 209, 1963, p 244–256 51. K.M. Guggenheimer, H. Heitler, and K. Hoselitz, J. Iron Steel Inst., Vol 158, 1948, p 192–199 52. J. Crangle and W. Sucksmith, J. Iron Steel Inst., Vol 168, 1951, p 141–151 53. A.G. Fuller, AFS Trans., Vol 85, 1977, p 509 54. J. Fowler, D.M. Stefanescu, and T. Prucha, AFS Trans., Vol 92, 1984, p 361–372 55. A.G. Fuller, P.G. Emerson, and G.F. Sergeant, AFS Trans., Vol 88, 1980, p 21–50 56. Castings, Nondestructive Evaluation and Quality Control, Vol 17, ASM Handbook, ASM International,1989, p 511–535 57. C.F. Watson, Introduction to Heat Treating of Cast Irons, Heat Treating, Vol 4, ASM Handbook, ASM International, 1991, p 667–669 58. J.S. Vanick, Engineering Properties of Heat-Treated Cast Irons, AFS Trans., Vol 54, 1946, p 460–464 59. G.M. Goodrich, Ed., Iron Casting Engineering Handbook, American Foundary Society, 2006, p 279, 377 60. T.L. Burkland, Heat Treatment of Ductile Iron—Effect on Prior Structure and Case Histories, ASM Source Book on Ductile Iron, American Society for Metals, 1977, p 203–207 61. Y.H. Lee and R.C. Voigt, The Hardenability of Ductile Iron, AFS Trans., Vol 97, 1989, p 915–938 62. J.S. Vanick, Progress in Heat Treatment of Alloy Cast Iron, Foundry, Vol 71, July 1943, p 178–179; Aug 1943, p 96–97 63. C.R. Austin, How Heat Treatment Affects High-Strength Irons, Am. Machinist, Vol 89, Oct 25, 1945, p 118–120 64. The Procedures Handbook of Arc Welding, Lincoln Electric, p 8.1–2 65. “Rio Tinto: Ductile Iron Data for Design Engineers,” QIT America (now Rio Tinto), Ductile Iron Society, www.ductile.org, accessed 2008

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 812-834 DOI: 10.1361/asmhba0005294

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Cast Iron Foundry Practices FOUNDRY PRACTICES critical to the production of cast irons include melting, alloying, molten metal treatment, pouring, and the design of feeding systems (gating and risering) to allow proper filling of the casting. Moldmaking is also an integral part of foundry operations. Other production steps may include blast cleaning, repair welding, grinding, machining, heat treatment, and inspection of castings. This article briefly reviews these production stages of iron foundry casting, with particular emphasis on the melting practices, molten metal treatment, and feeding of molten metal into sand molds. Sand is the dominant mold medium for casting of irons, and green (clay-bonded) sand molding occupies a significant majority of cast iron parts (Fig. 1). The choice of molding method or casting process depends on a number of factors, including production quantities, dimensional tolerances or complexity, patternmaking requirements, and costs. Green sand molding is well suited to very high production castings with reasonable quality levels. Chemically bonded molding is much slower than the green sand moldmaking, but it uses less expensive pattern materials and produces castings with more precise reproduction of dimensional tolerances. A general flow chart of cast iron foundry processes (with sand molding operations, as an example) is presented in Fig. 2 for gray, ductile, and malleable irons. One major indistinction in the flow chart is that ductile and malleable irons require a molten metal treatment with magnesium

Fig. 1

that promotes nodularizing of the graphite during solidification. Molten metal treatments for cast iron also include fluxing and inoculation. Fluxes are inorganic compounds such as limestone and calcium fluoride that refine the molten metal by removing dissolved gases and various impurities that may be detrimental to the lining of some furnaces. Inoculation is the late addition of ferrosilicon alloys in the molten iron to produce changes in graphite distribution and to reduce the chilling tendency. These treatments are discussed in more detail in this article for gray, ductile, and malleable irons. Foundry practices are also described for compacted graphite irons and high-alloy irons.

Cast Iron Melting Practice Feedstock for cast irons includes a mix of pig iron, steel, and scrap iron. Various types of furnaces have been used for cast iron melting (Fig. 3). In terms of tonnage, however, cupola and induction furnace melting are the primary melting methods used by iron metalcasting facilities (Fig. 4). Electric induction furnaces and some resistance heating are also used for holding furnaces between melting and casting lines for temperature control or superheating. Electric arc furnaces (EAFs) are used by a small number of iron facilities. Only cupola and induction melting are addressed in this article. The choice of melting method depends on production volumes, ferrous scrap availability, and technical expertise available in the melt

Relative percentage of cast irons produced by different molding and casting processes. Source: Ref 1

department. At higher melt rates, the inherent lower cost per ton operating costs of the cupola—using coke as a primary fuel source— outweighs the higher capital and maintenance costs associated with the cupola operation and its accompanying support facilities. A cupola can also use a wider variety of lower-cost ferrous scrap than is suited to induction melting. Induction melting, however, has the advantage of being able to quickly switch between different ferrous alloys and to allow metallurgical analysis to be performed and chemistries adjusted prior to removing iron from the induction melting furnace. In terms of energy use, Ref 5 makes recommendations on how to use each melting method more efficiently. The induction melting process primarily melts scrap iron and does not change the chemistry of the metal melted. Induction melting requires that carbon and other alloys be added to the furnace to ensure appropriate chemistries. There are fewer chemical reactions to manage in induction furnaces than in cupola furnaces, making it easier to achieve melt composition. However, induction melting is more sensitive to the quality of charge materials when compared to cupola or EAFs, limiting the types of scrap that can be melted. The inherent induction stirring provides excellent metal homogeneity. Induction furnaces include induction heel melters and batch melters (Fig. 5). Electric induction furnaces first used for melting purposes were primarily heel-style furnaces using line-frequency power supplies. The control technology available in the 1950s did not allow for variable-frequency power supplies, requiring induction melting furnaces to run at 60 Hz line frequency. This limited furnace designs to larger, coreless melt furnaces with low power densities. Power density is the power rating of a furnace in kW/ton of furnace capacity, which typically ranges from 200 to 250 kW/ton. Furnace designs must also take into consideration the mixing of iron in the furnace to ensure a homogeneous mixture without excessive stirring. The medium-to-large-capacity melt furnaces were also primarily heel melters, meaning that they maintained a heel of iron in the furnace at all times, requiring holding power when not melting. The iron maintained in the heel was typically 60 to 80% of the furnace capacity. The energy losses associated with holding iron between melts, as well as the larger overall furnace sizes, resulted in high overall energy-consumption rates.

Cast Iron Foundry Practices / 813

Fig. 2

Flow chart of cast iron processing with sand molding. (a) Annealing required for malleable irons. Adapted from Ref 2

In batch melting, the furnace empties after each melt cycle, reducing the holding power requirements. As more sophisticated solid-state power supplies with increasingly higher power ratings became available, the batch furnace increased in numbers. Over time, methods were developed to increase the frequency of the power supplies, allowing for increased power densities and smaller furnace sizes. Small furnaces with very high power densities of 700 to 1000 kW/ton can now melt a cold charge in 30 to 35 min. Another inherent advantage of the batch induction melter is that when melting a magnetic charge such as solid scrap iron, the coil efficiency can be as high as 95%, compared to 80% when heating the molten bath in a heel melter. Hysteresis losses associated with

induction heating of a solid ferrous material are responsible for this increased coil efficiency during the first part of the melting cycle. Cupola Furnaces. The basic concept of a cupola furnace involves a process of melting in a bed of hot coke in the lower portion of the vertical shaft. The coke bed is maintained by the creation of a reducing environment, which does not rapidly consume the coke at the bottom of the cupola. The area in front of the tuyeres, called the oxidation zone, is primarily where coke is consumed to provide the energy to produce molten iron. To maintain this coke bed, the reducing environment is high in carbon monoxide. As hot gases pass up through the vertical shaft, the ferrous materials charged into the top of the shaft are heated

and melt. The carbon monoxide and carbon dioxide formed when coke is consumed reach a balance, depending on the specific operating conditions set up by the cupola operator. The coke as fuel, ferrous scrap, alloy additions, and limestone as flux are all charged in layers into the top of the vertical shaft and gradually work their way down the shaft as iron is melted and coke is consumed. The limestone as flux cleans the iron and runs out the bottom of the cupola as slag. Because the cupola melting method changes the chemistry of the metal by adding carbon to the melted iron, the cupola is ideal for melting cast iron. This carbon pickup and other alloying mechanisms allow the cupola to melt a wide variety of ferrous scrap materials.

814 / Cast Irons

Fig. 5 Fig. 3

Arrangement of (a) induction heel melter and (b) induction batch melter. Source: Ref 5

Types of cast iron melting furnaces. Source: Ref 3

Fig. 4

Percentage of cast iron melted by furnace type. Source: Ref 4

Fig. 6

Cupola melt designs. (a) Water-cooled bare shell. (b) Refractory-lined shell. Source: Ref 5

The amount of carbon pickup and other metallurgical properties of the iron can be controlled by varying the height of the coke bed. In this way, both the metallic charge makeup and the alloys added, as well as the operating

parameter of the cupola, affect the metallurgical properties of the molten iron produced. However, cupolas are capital-intensive and are more typical for larger casting facilities and pipe shops.

The energy efficiency of cupola melting ranges from 40 to over 70%, depending on technological practices to improve energy efficiency and productivity. During the 1960s, cupolas used refractory-lined shells with no water cooling, and heated blast air was only beginning to be applied to cupola melting installations. The problem with this design was that the cupola melting campaign was limited to approximately 16 h. This was due to the limited refractory life of the brick used to line the cupola shells. It was typical to use two cupolas, side-by-side, to supply iron to foundry lines 16 h each day. For longer production runs, water-walled cupolas (Fig. 6a) were developed that could run for more than two weeks continuously without major repairs. This design used a high volume of water on the outside of the steel shell without a refractory lining. However, foundries and suppliers found ways to design refractorylined cupolas that were also capable of long production runs between major repairs. The newer designs usually use a small volume of water on the outside of the shell in addition to the refractory lining. This water allowed the refractory materials in the melt zone to burn out until it felt the cooling effect of the watercooled shell. This change improved the energy efficiency of the water-cooled, refractory-lined cupolas to near the level of the older refractory-lined shells. Both cupola designs are shown in Fig. 6. Another major change in cupola design was the advent of computer modeling applied to cupola melting processes. The trial-and-error methods used in past years have been replaced with computer simulations, starting in the 1980s. These models allowed engineers to develop an ideal cupola design (physical dimensions) for a specific set of operating conditions or ferrous scrap material type. Cupola design specifications cannot be easily changed; therefore, at other than design operating conditions, they lose operating efficiency. Design

Cast Iron Foundry Practices / 815

Table 1

Neutral Al2O3

A B C D E F G

6.2–22.9 5–26 15.0 ... ... 9.7–23.4 10–20

37.9–52.2 35–45 32.0 ... 41.1–51.7 37.0–65.0 40–50

Table 3

In as-cast gray irons, most of the contained carbon is present as flakes of free graphite. Because carbon is dissolved in the molten iron in amounts of approximately 2.8 to 4.0%, gray iron has the lowest casting temperature, the least shrinkage, and the best castability of all ferrous metals.

Table 2

Loss or gain, % of weight charged

(a) In pig iron or scrap. (b) As shot or scrap with minimum thickness of 4.8 mm (3/16 in.)

The metallic charge composition is based on the estimated losses and the compositional requirements of the gray iron. Table 3

Typical variations in acid cupola slags Cupola size

MnO

FeO

19.8–44.3 1.7–3.6 5.0–15.6 30–40 1.3–4.0 5–8 38.0 2.5 8.0 ... ... ... 18.3–36.8 ... 0.4–10.8 7.7–40.7 1.0–23.4 0.9–44.4 25–38 1–5 1–8


7 to 12
10 to 20
10 to 15
15 to 25
15 to 25 Trace
10 to 20
2 to 5
2 to 5
5 to 10 +40 to 60

Silicon(a) Manganese(a) Ferrosilicon (lump) Ferromanganese (lump) Spiegeleisen Phosphorus Ferrochromium (lump) Nickel (shot or ingot) Copper(b) Alloys in briquets Sulfur

The essential purpose of melting is to produce molten iron of the desired composition and temperature. For gray iron, this can be accomplished with various types of melting equipment. Cupolas and induction furnaces tend to be the types most commonly found in the gray iron foundry. The cupola was traditionally the major source of molten iron, but gradual acceptance of electric melting has reduced its dominance. Gray iron can be melted with a single furnace or a combination of furnaces (duplexing).

Basic constituents CaO + MgO

Charge constituent

Melting Practice

Acid slag compositions

Acid Acid constituent slag SiO2

Charge Composition for Cupolas. The composition of gray iron melted in a cupola depends on the composition of all materials charged and the degree of oxidation each element experiences during melting. The approximate gain or loss that various charge constituents experience during the melting process is shown as follows:

Gray Iron Foundry Practice

parameters, which include production rates, dictate ideal design specifications, such as the cupola shell diameter, stock height (height of the charge above the tuyeres), tuyere sizes, and blast volumes. Cupola slag is usually acidic. Acidity is determined by the ratio of CaO and SiO2 in the slag. Tables 1 and 2 give the compositions of several acid cupola slags with a predominance of acid SiO2 and a ratio of basic constituents ([CaO + MgO]/SiO2) of less than 1. Limestone is added to the cupola charge to control the acid/base ratio of the slag and to flux away the coke ash. This fluxing action intensifies the combustion and carburization action of the coke. The acid/base ratio of the slag affects the final composition of the iron. An increase in basicity reduces the final sulfur and increases the carbon in the iron. A decrease in silicon can accompany this increase in basicity because of increased oxidation losses. Therefore, control of the fluxing materials charged can influence the resulting composition of cupola-melted iron.

Slag composition, %

Description

mm

in.

Steel, %

Coke, %

Average fluidity Average fluidity Viscous slag Fluid and corrosive Overblown cupola 150 m3/min/m2(490 ft3/ min/ft2)

2135 1220 1220 2135 533

84 48 48 84 21

33 22 22 14 50

7.5 12.8 12.8 11.0 14.0

Flux(a), %

1.9 3.0 2.2 3.5 5.0

SiO2

Al2O3

47.1 46.2 52.1 39.6 45.6

12.1 11.0 12.4 10.1 18.5

FeO(b)

6.9 1.1 1.5 3.8 16.1

MnO

CaO

MgO

4.6 1.4 1.9 2.9 2.8

22.0 37.2 29.2 38.8 15.4

1.6 1.0 0.9 2.9 1.0

(a) Limestone. (b) Total iron as FeO including small portion of Fe2O3

Typical cupola charge compositions and final analyses for class 30 and class 40 gray iron castings Composition(a) Quantity

Material charged

Class 30 gray iron Foundry scrap (returns) Purchased cast iron scrap Rail steel Pig iron Silvery pig iron Manganese briquets Inoculant(b) Total Charge analysis Charge in melting Final analysis Class 40 gray iron Foundry scrap (returns) Rail steel Pig iron Silvery pig iron Manganese briquets Silicon briquets Inoculant(b) Total Charge analysis Change in melting Final analysis

%

30.0 22.5 27.5 15.0 5.0 ... ...

Total carbon lb

%

600 450 550 300 100 3 ... 2003

3.35 3.25 0.60 4.00 2.50 ... ...


5.0

20.10 14.62 3.30 12.00 2.50 ... ... 52.52

2.62 +0.73 3.35


5.0

27.5 60.0 5.0 7.5 ... ... ...

lb

550 1200 100 150 6 4 ... 2010

3.15 0.60 4.00 2.50 ... ... ... 1.62 +1.53 3.15

Sillicon %

1.88 2.20 0.30 2.40 10.48 ... ...

Phosphorus lb

11.30 9.90 1.65 7.20 10.48 ... 1.20 41.73

2.09
0.21 1.88 17.34 7.20 4.00 3.75 ... ... ... 32.29

1.45 0.30 2.40 10.48 ... ... ... 1.63
0.16 1.47

%

0.16 0.22 0.03 0.12 0.10 ... ...

Sulfur lb

0.96 0.99 0.16 0.36 0.10 ... ... 2.57

0.13 +0.03 0.16 7.97 3.60 2.40 15.72 ... 2.00 0.90 32.59

0.09 0.03 0.12 0.10 ... ... ... 0.06 +0.03 0.09

%

0.11 0.10 0.03 0.03 0.05 ... ...

Manganese lb

0.66 0.45 0.16 0.09 0.05 ... ... 1.41

0.07 +0.04 0.11 0.50 0.36 0.12 0.15 ... ... ... 1.13

0.08 0.03 0.03 0.05 ... ... ... 0.05 +0.03 0.08

%

lb

0.70 0.60 0.80 0.91 0.88 ... ...

4.20 2.70 4.40 2.73 0.88 2.00 ... 16.91

0.85
0.17 0.68 0.44 0.36 0.03 0.07 ... ... ... 0.90

0.80 0.80 0.91 0.88 ... ... ...

4.40 9.60 0.91 1.32 4.00 ... ... 20.23

1.01
0.20 0.81

(a) These compositions provide only for minimum tensile strength; the proportion of charged components and the desired final analyses may vary considerably among different foundries. (b) For both class 30 and class 40 irons, the inoculant was added at the spout during tapping.

816 / Cast Irons shows metal charge compositions and final analyses for class 30 and class 40 irons. These compositions are typical and provide for minimum tensile strength only. Compositions for gray iron melted in a cupola will vary among foundries, and some experimentation is usually required in order to develop a charge composition. Storage and Handling of Charge Components. A good material storage and handling system is essential to efficient cupola operations. Metal charge components are unloaded into storage bins using a crane magnet and are transferred to a weigh hopper to designated weight values as charge components are progressively added. Coke and limestone from overhead storage bins are discharged into a weigh hopper and into the charge bucket. The charge bucket with all weighed amounts of metal, ferroalloys, coke, and flux is elevated to the charge door and dumped, either with an overhead crane or with a skip hoist up an inclined track. Charge Materials for Induction Furnaces. Compositional control of gray iron is greater in induction furnaces than cupolas because of the reduced oxidation losses during melting and the elimination of coke as fuel. The composition of the metallic charge materials is directly related to the final composition of the iron. Therefore, accurate weights and analysis of charge materials are required for good compositional control. Typical charge materials include steel scrap, cast iron scrap, foundry returns, ferrosilicon, and carbon. Pig iron is rarely used because of cost, although it does offer advantages in terms of purity. The type of scrap employed will depend on economics. The elimination of coke from the charge also reduces the sulfur absorbed by the iron; therefore, resulfurization is often desirable in electrically melted gray iron. A typical charge for class 30 iron of 4.20 carbon equivalent (CE) is:

Steel scrap Gray iron borings Foundry returns Carbon 50% ferrosilicon

41.3% 10.0% 45.0% 1.7% 2.0%

Two charge compositions and charging sequences used in a high-production foundry for melting class 30 iron in induction furnaces are given in detail in Table 4. The CE can readily be adjusted to any desired level by balancing steel scrap content and additions of carbon and ferrosilicon. Alloying additions are absorbed rapidly at 90 to 95% recovery, and the ease of meeting chemical and mechanical specifications, including those difficult or impractical in cupola melting, is valuable. Carbon additions are made with pelletized petroleum coke of controlled grain size to give high and reproducible recoveries. Because little sulfur is introduced in induction melting, the sulfur content of the melted iron is considerably lower than that from a cupola, and thus, less manganese is required for balance with sulfur. Arc Furnace Melting. In the past, arc furnaces have been used in gray iron foundries for special irons and for batch or jobbing operations; very few have been used as holders or superheaters for cupolas. However, it has become apparent that electric melting can compete with cupola melting on a straight economic basis, so interest in the arc furnace as a primary melter has grown. In 1969, arc furnaces were installed in a major high-production foundry, and others are now in operation. An arc furnace is essentially a refractory hearth on which material can be melted by heat from electric arcs. The molten bath is relatively wide compared to its depth, so that bulky charge material can be handled. Reactions between slag and molten metal are efficient, because the interface area is large and the slag is at least as hot as the metal.

Thermal efficiency on meltdown is high, approximately 80%, because the arcs are surrounded by charge material. Efficiency of superheating molten metal is relatively low, approximately 20 to 30%, because the bath receives heat only from the surface. Arc furnaces are sized in terms of shell diameter; this determines the bath capacity, depending on the thickness of refractory used. Typically, a 2.7 m (9 ft) furnace would hold a 9 to 11 tonne (10 to 12 ton) bath, and a 3.4 m (11 ft) furnace would hold a 20 to 23.5 tonne (22 to 26 ton) bath. Melting or production rate in tons per hour is a function of power input, and examination of over 100 furnaces making tonnage plain carbon steel shows that, on the average, production of one ton per hour requires 1000 kVA in transformer capacity. Power factor depends on the design of the electrical circuit and on the mode of operation of the power input; typically, it is 0.80.

Desulfurization Methods Excess free sulfur in gray iron increases chill depth, decreases fluidity, and decreases the effectiveness of inoculation. Several successful desulfurization techniques have been developed in the search for methods of obtaining low-sulfur base irons. In earlier years, the removal of some excessive sulfur in gray iron was accomplished by the forehearth or ladle addition of sodium carbonate (soda ash), but refractory attack made it impractical to reach very low sulfur levels. Calcium carbide has proved to be the most effective desulfurizing agent for low sulfurs. Lime is also an effective desulfurizing agent, but it requires more volume, time, and agitation. Small percentages of fluorspar and various salts improve the desulfurizing effectiveness of lime. Metal stirring or agitation is essential for promoting optimal desulfurization. The following stirring methods have been found to be effective:  Carbide is injected with inert gas through

Table 4 Typical charge compositions and sequences for producing class 30 gray iron in an induction furnace Weight of material and sequence of charge(a) First bucket Charge material

kg

Second bucket lb

kg

Third bucket lb

kg

lb

3.30C-1.80Si-0.60Mn(b) (44% returns, 32% turnings, 12% plate, 10% borings) Steel plate ... ... ... ... Steel turnings 2315 5100 ... ... Cast iron returns ... ... 1635 3600 Coke 115 250 ... ... Ferrosilicon ... ... 100 220 Cast iron borings ... ... 680 1500

860 ... 1540 ... ... ...

1900 ... 3400 ... ... ...

3.30C-1.85Si-0.60Mn(c) (55% returns, 32% turnings, 10% plate) Steel plate ... ... 365 Steel turnings 2315 5100 ... Cast iron returns ... ... 1995 Coke 115 250 ... Ferrosilicon ... ... 45

365 ... 1995 ... 45

800 ... 4400 ... 100

800 ... 4400 ... 100

(a) Charge is made after three transfer ladles (7210 kg, or 15,900 lb) of hot metal have been tapped from the furnace. No hot metal is tapped during the charging sequence. All three buckets are charged in sequence, and the charge is completely melted before another charge is added. Maximum power is used after charging of the first bucket, to help dissolve the coke. (b) Total weight of charge 7245 kg (15,970 lb). Elements may vary by þ0.10% each. Buckets are loaded in the following order. First bucket: steel turnings, coke, steel turnings. Second bucket: gray iron borings, ferro
silicon, gray iron returns. Third bucket: gray iron returns, steel plate. (c) Total weight of charge 7235 kg (15,950 lb). Elements may vary by þ0.10%
each. Buckets are loaded in the following order. First bucket: steel turnings, coke, steel turnings. Second bucket: steel plate, ferrosilicon, gray iron returns. Third bucket: gray iron returns, ferrosilicon, steel plate

graphite or refractory lances that supply the carbide and simultaneously stir the melt.  Inert gas is bubbled through a porous plug in the bottom of a ladle, with desulfurizing agent fed onto the surface with a vibrating feeder.  The shaking ladle system places a ladle of iron in a gyrating shaking mechanism that stirs in the desulfurizing agent.  The refractory stirrer system rotates a refractory paddle in the bath, stirring the carbide or lime into contact with the iron. With any of these stirring methods, additions of 0.50 to 1.0% CaC2 have desulfurized irons of 0.07 to 0.10% S, as-melted, down to a final 0.01 to 0.02% S. A temperature loss of 30 to 55  C (50 to 100  F) can be expected. Lime alone or with various salts can be used as the desulfurizing agent. The efficiency is lower than that with carbide, but there are fewer environmental problems. Injection and porous plug desulfurizing systems can be either batch type, treating

Cast Iron Foundry Practices / 817 such as in thin sections, at corners, and along edges. Tensile strength can be improved through inoculation. This is particularly true for the low-CE irons, which are selected for applications requiring a higher tensile strength (tensile strength decreases as the CE increases) (Ref 9). However, low-CE irons are also the grades that are most susceptible to carbide formation. Inoculation helps overcome this problem by minimizing the chill-forming tendency of the iron, thus allowing low-CE irons to be poured in thin sections. Irons that are stored in holding furnaces or in pouring systems for an extended period of time are also more susceptible to chill formation. This susceptibility can be attributed to the reduction in nuclei in the melt that takes place during extended holding. This effect is accelerated if holding occurs at high temperatures. Melting method influences white iron formation; electric-melted irons are generally more prone to carbide formation than cupola-melted irons. Types of Inoculants. Graphite- or ferrosilicon-based alloys can be used to inoculate gray iron. The graphite used must be highly crystalline to give the best effect. Examples of highly crystalline graphite include some naturally occurring graphite and graphite electrode scrap. Amorphous forms of carbon, such as metallurgical coke, petroleum coke, and carbon electrode scrap, are not suitable for inoculation. Graphite is rarely used by itself and is usually added in combination with a crushed ferrosilicon. Because inconsistent results have been obtained with graphite, careful addition methods and relatively higher temperatures are needed to ensure its complete solution. Graphite has been found to promote extremely high eutectic cell counts. Ferrosilicon alloys are also used to treat gray iron. They are typically based on 50 or 75% ferrosilicon and act as carriers for the inoculating (reactive) elements, which include aluminum, barium, calcium, cerium or other rare earths, magnesium, strontium, titanium, and

individual ladles, or a continuous system, with the cupola stream flowing through a special forehearth outfitted with a porous plug and a carbide feed (or injection lance). The spent carbide or lime overflows or is raked off, and the desulfurized metal flows continuously into the holder or duplex electric furnace. Disposal of the carbide slag requires attention to safety and environmental considerations.

Inoculation Inoculation is defined as the late addition of certain silicon alloys to molten iron to produce changes in graphite distribution, improvements in mechanical properties, and a reduction of the chilling tendency that are not explainable on the basis of composition change with respect to silicon (Ref 6). Graphite, added alone or in combination with ferrosilicon, will also produce these changes without significantly altering the chemistry of the iron. It is recognized that two irons with the same apparent composition can have dramatically different microstructures and properties if one is inoculated and the other is not. A great deal of research has been dedicated to determining the mechanism behind inoculation (Ref 7, 8). Although many theories exist, no conclusions have been reached regarding possible mechanisms. The purpose of an inoculant is to increase the number of nuclei in molten iron so that eutectic solidification, specifically graphite precipitation, can begin with a minimum amount of undercooling. When undercooling is minimized, there is a corresponding reduction in the tendency to form eutectic carbide or white iron, which is referred to as chill. Instead, a more uniform microstructure consisting of small type A graphite flakes is produced. These microstructural changes can result in improved machinability and mechanical properties. Irons are inoculated for various reasons. The primary reason is to control chill in areas of castings that experience rapid solidification,

Table 5 Performance category of inoculant

Standard

Intermediate

High

Stabilizing

zirconium. The silicon in the alloys does not cause significant chill reduction unless added to a level that markedly increases the CE. Because ferrosilicon dissolves readily, it helps to distribute the reactive elements uniformly throughout the melt. It is important to note that the reactive elements, in addition to reacting with iron, react readily with sulfur and oxygen. Their addition, therefore, can result in dross formation. The quantity of dross produced is directly proportional to the amount of reactive elements in the alloy. Compositions of some of the commercially available ferrosilicon inoculants are given in Table 5. As indicated in Table 5, it is convenient to group inoculants into four performance categories: standard, intermediate, high potency, and stabilizing. The calcium-bearing alloys fall into the standard category. Improvement in chill reduction is obtained by pairing calcium with barium (Ref 10). This type of inoculant falls into the intermediate group. The strontium or calcium-cerium alloys are the strongest chill reducers. Figure 7 shows the performance obtainable with these three inoculant groups. It also shows that as the amount of inoculant increases, there is a reduction in chill until a point of diminishing returns is reached. It is not correct to assume that if a little is good, a lot is better. Addition levels above those needed to control chill and produce the desired mechanical and microstructural changes result in higher costs and can lead to a variety of problems, including inclusion defects caused by inoculant dross, hydrogen pinholing, and shrinkage. Aluminum is found in all ferrosilicon inoculants, but the amount varies considerably because the effectiveness of aluminum in controlling chill in gray iron is still under debate. Aluminum has been linked to hydrogen pinhole formation in castings produced in green sand molds, so it is best to keep its level rather low. Stabilizing inoculants are also available. They are designed to promote pearlite and, at the same time, provide graphitization during solidification. They are useful in producing highstrength castings with a minimum of chill, and

Compositions of ferrosilicon inoculants for gray cast iron Composition(a), % Si

Al

Ca

Ba

Ce

TRE(b)

Ti

Mn

Sr

Others

46–50 74–79 74–79 46–50 60–65 70–74 ... 42–44 50–55 50–55 36–40 46–50

0.5–1.25 1.25 max 0.75–1.5 1.25 max 0.8–1.5 0.8–1.5 ... ... ... ... ... 0.50 max

0.60–0.90 0.50–1.0 1.0–1.5 0.75–1.25 1.5–3.0 0.8–1.5 ... 0.75–1.25 5–7 0.5–1.5 ... 0.10 max

... ... ... 0.75–1.25 4–6 0.7–1.3 0.75–1.25 ... ... ... ... ...

... ... ... ... ... ... ... ... ... ... 9–11 ...

... ... ... ... ... ... ... ... ... ... 10.5–15 ...

... ... ... ... ... ... ... 9–11 9–11 9–11 ... ...

... ... ... 1.25 max 7–12 ... ... ... ... ... ... ...

... ... ... ... ... ... ... ... ... ... ... ...

73–78

0.50 max

0.10 max

...

...

...

...

...

6–11

0.50 max

0.50 max

...

...

...

...

...

... ... ... ... ... ... ... ... ... ... ... 0.60– 1.0 0.60– 1.0 ...

(a) All compositions contain balance of iron. (b) TRE, total rare earth elements

... 48–52 Cr

Fig. 7

General classification of inoculants showing chill reduction in iron with a carbon equivalent of 4.0

818 / Cast Irons they help to eliminate thick sections (Ref 11). Stabilizing inoculants normally employ chromium as the pearlite stabilizer. Because these alloys can be difficult to dissolve, they are not suggested for mold addition. Inoculant Evaluation. A variety of methods can be employed to evaluate the effectiveness of an inoculant. Because chill reduction is often of primary concern, one way to test an alloy is by pouring chill bars or wedges. The depth of chill produced in the treated iron can be compared to that of the base iron. Details of chill testing can be found in ASTM A 367. Inoculation is known to increase the eutectic cell count of gray iron. Therefore, eutectic cell count can be used as an indication of the nucleation state of the melt. Samples for this type of evaluation are typically polished through 400grit paper and then etched with Stead’s reagent (Ref 12). Results should be interpreted with caution because no standard relationship exists between cell count and chill depth. For example, strontium-bearing ferrosilicon inoculants reduce chill to very low levels without dramatically increasing the eutectic cell count. Evaluation of the microstructure and mechanical properties provides information regarding the performance of an inoculant. The presence of type A graphite rather than carbides and under-cooled graphite, along with attainment of the desired tensile properties, indicates that inoculation has been successful. Cooling curves can also be used to evaluate inoculant performance; the relative amount of undercooling can be used as an indicator of effectiveness. Each technique of alloy evaluation has advantages and disadvantages. The use of a suitable combination of these techniques is suggested for obtaining an adequate assessment of inoculant performance in a particular base iron under a given set of foundry conditions. Addition Methods. Ladle inoculation is a common method of treating gray iron. In this method, the alloy is added to the metal stream as it flows from the transfer ladle into the pouring ladle. A small heel of metal should be allowed to accumulate in the bottom of the ladle prior to the addition. This allows the

Fig. 8

Schematic showing the principle of stream inoculation. Source: Ref 14

inoculant to be mixed and evenly distributed in the iron. Addition of the alloy to the bottom of an empty ladle is not recommended, because this may cause sintering and a reduction in inoculant effectiveness. Problems can also arise if the alloy is added to a full ladle, because the material can become entrapped in the slag layer that forms on the surface. The amount of inoculant needed in this treatment normally varies between 0.15 and 0.4%, depending on the potency of the inoculant. If graphite alone is used, the addition level is approximately 0.1 to 0.2%. Excessive additions should be avoided for the reasons cited earlier. Inoculants for this method typically have a 6 or 12 mm ¼ or ½ in.) maximum size. Minimum size has not been found to be as critical, although excessive fines should be avoided because they can float on the surface and lose their effectiveness through oxidation. All additions should be weighed or measured accurately, and the use of proper metal temperatures ensures good dissolution. Minimizing the time between treatment and pouring helps avoid loss of the inoculating effect, which is known as fade. The maximum effect of an inoculant is realized immediately after the alloy is dissolved in the metal. More than half of the effect of inoculation can be lost, because of fade, in the first 5 min after the addition. Complete loss can occur if the iron is held for 15 to 30 min (Ref 13). Although the composition of the iron does not change dramatically, an increase in chilling tendency occurs, along with an accompanying decrease in mechanical properties. All inoculants fade to some degree. Methods that can help prevent or minimize the loss of the inoculation effect have been gaining acceptance. By adding inoculant late in the production process, the effect of time can be greatly reduced. Stream and mold inoculation are two late methods of iron treatment that are believed to promote more uniform quality from casting to casting. Stream inoculation requires that the alloy be added to the stream of metal flowing from the pouring ladle into the mold. One of the electropneumatic devices used to sense when the metal flow starts and stops is shown schematically in Fig. 8. This device ensures that the alloy is dispensed in such a manner that the last metal entering the mold is treated similarly to the first metal. The same inoculants used to treat iron in the ladle can be used for stream inoculation, but less of a performance distinction has been observed among them. Graphite is not recommended, because of its relatively poor solution characteristics. A uniform and consistent size seems to be an important factor in stream inoculation. Too large a particle can cause plugging of the equipment and incomplete dissolution. A maximum particle size of 8 to 30 mesh and a minimum size of approximately 100 mesh are recommended. Addition levels range from 0.10 to 0.15%. Mold inoculation involves placement of the alloy in the mold, such as in pouring basins,

at the base of the sprue, or in suitable chambers in the runner system (Ref 15). Inoculants for this method can be crushed material, powder bonded into a pellet, or precast slugs or blocks. As in stream inoculation, alloy dissolution rate is an important factor. Crushed alloy for this application is typically 20 to 70 mesh in size, and the addition rate can be as little as 0.05%. The precast and bonded alloys are designed to dissolve at a controlled rate throughout the entire pouring cycle (Ref 16). Mold inoculation is often used as a supplement to ladle inoculation. There are several advantages of late inoculation over ladle inoculation. As previously stated, fading is virtually eliminated, and because the castings are inoculated to the same extent, there is greater consistency in structure from casting to casting. It has also been observed that late inoculation is more successful in preventing carbide formation in thin sections, thus eliminating heat treatment.

Gray Iron Alloying Minor alloying of gray iron is done to increase strength and promote pearlite. The minor elements normally used in gray iron alloying are chromium, copper, nickel, and molybdenum. Alloying practices vary considerably, depending mainly on the method of melting, the composition and product form of the alloy being added, the amount of iron being alloyed (the entire charge or only portions of the charge), and the ladling practice in the specific foundry. In cupola melting, nickel, copper, chromium, and molybdenum are added to the molten metal during or after tapping, rather than being added to the cupola charge. If the ladle that receives the molten metal at the spout will pour the castings without a transfer, the alloying metal may be added to the stream at the spout. This practice generally provides reasonably thorough mixing, although the exact amount of iron tapped into the receiving ladle is sometimes difficult to control. A more common practice is to add the alloying metals to the stream from the receiving ladle (or forehearth, if used) as it is transferred to the pouring ladle, because it may not be desirable to alloy all of the metal in the receiving ladle or forehearth. The amount of mixing obtained by adding to the stream is usually sufficient, although to obtain more thorough mixing the alloyed metal is sometimes poured from the transfer ladle to another transfer ladle. The practices described previously may also be employed if the melting is done in an arc or induction furnace. However, if the entire charge of an arc or induction furnace is to be alloyed, the alloying metals may be added to the furnace, usually just a few minutes prior to tapping. Because of the stirring action in these furnaces (particularly in an induction furnace), losses of the alloying metals from oxidation may be excessive if too much time elapses between making the additions and tapping.

Cast Iron Foundry Practices / 819 An advantage of adding the alloying metals to the charge in the furnace is stability of temperature. When alloying metals are added cold to the ladle after tapping, there can be a significant decrease in temperature. The amount of temperature drop increases with the amount of cold metal added. Ladle additions of cold metal that total more than 3% are not recommended. Alloying metal should never be placed in an empty ladle; the bottom of the ladle should be covered with at least 25 mm (1 in.) of molten metal before any alloying metal is added. Preferably, alloy is added as a small amount at one time. Methods of adding the cold alloying metals to a molten charge vary widely. The simplest method is to put the alloy in paper bags and toss them into the stream, or into the ladle as it fills. Mechanical devices are also used to add alloying metals. When the alloying metals are added at the spout, and are in shot or granulated form, a funnellike device is sometimes mounted above the spout so that the funnel outlet is directed at the stream. The funnel portion is large enough to hold the maximum amount of shot or granulated alloy that would be required for one ladle. This dispensing device is provided with a shuttlelike control so that the flow of alloy can be started, stopped, and regulated as needed. Alloys in granulated or shot form are sometimes added from a simple sheet metal dipperlike container with a long handle that permits the operator to reach over the ladle of molten iron and pour the alloy into the stream. Although the amount of alloy is measured, the rate at which it is discharged into the stream is controlled by the operator. Another simple means of adding shot or granulated material is to pour it into a long pipe or tube that is closed at one end (an ordinary piece of downspout may be used). The tube must be long enough for the operator to hold the open end over the stream and let the alloy pour out. Nickel is easily added, and losses are so insignificant that no allowance for loss is made. If nickel is to be added to the stream at the spout, or to a ladle, it is added as shot, an alloy that usually contains approximately 92% Ni. The addition is calculated on the nickel content. For instance, if it is desired to develop an alloy iron containing 1% Ni, approximately 5 kg (10.8 lb) of 92% Ni shot will be added to each 455 kg (1000 lb) of iron. When nickel is added to the molten charge of an arc or induction furnace, it is feasible to use larger pieces, because there will be more time for the nickel to become dissolved and distributed. Copper is also simple to add, and losses are so low that no allowance for loss is made. Scrap copper wire, clippings, and various other forms of copper scrap are most commonly used. The melting temperature of copper is considerably lower than that of iron, and the specific gravity of copper is reasonably near that of iron, so copper is easily dissolved and distributed in molten iron.

Casting

Chromium is added as granulated ferrochromium, which usually contains 65 to 70% Cr. Chromium additions are subject to some loss, which must be allowed for when adding the ferroalloy. Most foundry operations allow for a loss of 5%, but experience in a particular foundry may prove the need for slightly increasing or decreasing that allowance. Ferrochromium dissolves more slowly than nickel or copper, and hence it must be given more time or more agitation, or both. Molybdenum is also added as a granulated ferroalloy, generally one that contains 60 to 70% Mo. As for chromium, a 5% allowance for loss is made in most foundry operations.

Gray iron expands slightly because of the formation of graphite during eutectic solidification. This is most pronounced in higher-CE irons, in which more graphite is precipitated. This expansion stresses the molding material, causing an enlargement of the mold cavity if the sand is insufficiently compacted. This can lead to shrinkage defects. Therefore, it is important when producing gray iron castings that the mold hardness be sufficient to withstand the eutectic expansion of the gray iron. If the dimensional accuracy and surface finish requirements of the casting exceed the limits of the green sand molding process, other processes can be used, such as shell molding, chemical bond molding, and permanent molding. Each process creates a rigid mold surface that with-stands the eutectic expansion of gray iron, thus improving dimensional accuracy. In addition, an excellent surface finish can be obtained with each of these processes. Shell molds are produced from a mixture of sand and a thermosetting resin. Chemically bonded molds are produced from a mixture of sand and a resin binder that requires a catalyst for room-temperature curing. Permanent molds are often produced from gray iron and generally require some form of coating to assist in the removal of the casting. The gating and feeding system serves the same purpose in the casting of gray iron as in the other metal casting systems:

Pouring Metal temperature, cleanliness, and delivery technique are the essential variables to be controlled when pouring. Removal of slag and dross from the liquid iron surface reduces the possibility of inclusion-type defects. A variety of metal-pouring techniques are employed in an iron foundry. Each technique should enable the operator to maintain a constant flow of metal sufficient to keep the gating system full of metal until the pour is completed. The molten iron should possess sufficient superheat to enable the cavity to be completely filled with iron without the formation of temperature-related defects. The fluidity of the iron is directly proportional to the amount of superheat contained in the iron. Therefore, thinner sections that require greater fluidity for successful casting also require more superheat than thicker sections. The liquids of gray iron is determined by the composition of the iron. For hypoeutectic irons, the liquids is inversely proportional to the CE of the iron. Therefore, the higher-strength irons with lower CE require higher pouring temperatures to maintain the necessary level of superheat. Table 6 lists the typical pouring temperatures required for the various classes of gray iron. Iron containing insufficient superheat can lead to partially filled cavities, misruns, blowholes, and chill. An excess of superheat can lead to shrinkage, metal penetration, veining, and scabbing. The amount of superheat required will vary for individual castings, and some experimentation will be required to determine the optimal pouring temperature for each casting.

 Filling the mold cavity without turbulence  Preventing slag, dross, or mold material

from entering the mold

 Preventing the introduction of air or mold

gases into the stream of metal

 Producing heat-transfer characteristics that

will aid in the progressive solidification of the casting  Enabling production of the casting with the use of a minimum amount of metal Gating and feeding practice for gray iron is less complicated than for other materials. The precipitation of graphite during eutectic solidification causes an expansion, which offsets some of the contraction usually experienced during the solidification of a metal. In fact, a class 20 iron precipitates enough graphite to create sufficient expansion so that feeding is not required. As the CE of the iron decreases, less graphite is

Table 6 Typical pouring temperatures for some classes of gray iron Pouring temperature Small castings Approximate liquidus temperature Class

30 35 40 45



C

1150 1175 1200 1220



F

2100 2150 2190 2230

Large castings

Thin sections

Thick sections

Thin sections

Thick sections









C

1400 1425 1450 1470



F

2550 2600 2640 2680

C

1370 1400 1420 1445



F

2500 2550 2590 2630

C

1345 1370 1395 1415



F

2450 2500 2540 2580

C

1315 1345 1365 1390



F

2400 2450 2490 2530

820 / Cast Irons precipitated, and feeding requirements increase. A class 50 iron requires approximately 4% volumetric feeding to offset the contraction of the iron. Dimensional Control. The production of gray iron castings with good dimensional stability requires that the foundry worker produce a mold rigid enough to withstand eutectic expansion and that the pattern be constructed in such a way as to compensate for the metal contraction during cooling. The contraction depends on the amount of graphite precipitated during eutectic solidification. Irons with a higher CE will precipitate more graphite, resulting in less shrinkage. Typical shrinkage values of the various grades of iron are: Class

30 35 40 45 50 55

Carbon equivalent, %

Pattern shrinkage allowance, mm/m (in./ft)

... 3.7–4.1 3.5–3.9 3.45–3.8 3.3–3.6 ...

8.3 (1/10) 10.4 (1/8) 10.4 (1/8) 10.4 (1/8) 13.0 (5=32) 13.0 (5=32)

Ductile Iron Foundry Practice Ductile iron has been known only since the late 1940s, but it has grown in relative importance and currently represents approximately 20 to 30% of the cast iron production of most industrial countries. Ductile iron is also known as nodular iron or spheroidal graphite iron. Unlike gray iron, which contains graphite flakes, ductile iron has an as-cast structure containing graphite particles in the form of small, rounded, spheroidal nodules in a ductile metallic matrix. Therefore, ductile iron has much higher strength than gray iron and a considerable degree of ductility; both of these properties of ductile iron, as well as many of its others, can be further enhanced by heat treatment. Ductile iron also supplements and extends the properties and applications of malleable irons. It has the advantage of not having to be cast as a white iron and then annealed for castings having section thicknesses of approximately 6 mm (¼ in.) and above, and it can be manufactured in much thicker section sizes. However, it cannot be routinely produced in very thin sections with as-cast ductility, and such sections usually need to be heat treated to develop ductility. It has the advantage, in common with gray iron, of excellent fluidity, but it requires more care to ensure sound castings and to avoid hard edges and carbides in thin sections, and it usually has a lower casting yield than gray iron. Compared to steel and malleable iron, it is easier to make sound castings, and a higher casting yield is obtained; however, more care is often required in molding and casting. Ductile iron is made by treating low-sulfur liquid cast iron with an additive containing magnesium (or occasionally cerium), and it is usually inoculated just before or during casting with a silicon-containing alloy. There are many variations in

commercial treatment practice. In general, the composition range is similar to that of gray iron, but there are a number of important differences.

Ductile Iron Melting Practice The spheroidal form of graphite that characterizes ductile iron is usually produced by a magnesium content of approximately 0.04 to 0.06%. Magnesium is a highly reactive element at molten iron temperatures, combining readily with oxygen and sulfur. For magnesium economy and metal cleanliness, the sulfur content of the iron to be treated should be low (preferably <0.02%). This is readily achieved in an electric furnace by melting charges based on steel scrap or special-quality pig iron supplied for ductile iron production, together with ductile iron returned scrap. Low sulfur content can also be achieved by melting in a basic cupola, but acid cupola-melted iron has a higher sulfur content and normally needs to be desulfurized before treatment by continuous or batch desulfurization in a ladle or special vessel. Treatment of acid cupola-melted iron with magnesium without prior desulfurization is not recommended, because the iron consumes more magnesium and produces excessive magnesium sulfide slag, which is difficult to remove thoroughly and may lead to dross defects in castings. To produce ductile iron with the best combination of strength, high ductility, and toughness, raw materials must be chosen that are low in many trace elements, particularly those that promote a pearlitic matrix structure. A low manganese content is also needed to achieve as-cast ductility and to facilitate successful heat treatment in order to produce a ferritic structure. For this purpose, it is necessary to use deep-drawing or other special grades of steel scrap and pig iron of a quality produced specially for ductile iron production. International Standard ISO/DIS 9147 specifies two grades of pig iron for ductile iron production: grade 3.1 (nodular [spheroidal graphite, or SG] base) and grade 3.2 (nodular [SG] base, higher manganese). The compositions of these grades are: Composition Grade

3.1 3.2

C

Si

Mn

P

S

3.5–4.6 3.5–4.6

<3.0 <4.0

<0.1 <0.1–0.4

<0.08 <0.08

0.03 max 0.03 max

The contents of elements that affect the formation of nodular graphite and promote the

Table 7

formation of carbide are low according to the application of the grade concerned. Higher-strength grades of ductile iron can be made with common grades of constructional steel scrap, pig iron, and foundry returns, but certain trace elements, notably lead, antimony, and titanium, are usually kept as low as possible in order to achieve good graphite structure. Their undesirable effects can be offset by the addition of a small amount of cerium to give a residual cerium content of 0.003 to 0.01%. An important factor in controlling raw materials is to exclude aluminum, which may promote unsoundness and dross defects. Table 7 lists the typical minor element contents of three raw materials used in ductile iron manufacture.

Ductile Iron Composition Control Carbon. In electric furnace melting practice, carbon is derived from pig iron, carburizers, and cast iron scrap. The carburization of steel scrap charges is achieved by adding low-sulfur graphite or graphitized coke, and the rate of solution and the recovery of carbon increase with the purity of the carbon source used. In cupola melting, carbon is also derived from the coke charged. The optimal range for this element is usually 3.4 to 3.8%, depending on silicon content. Above this range, there is a danger of graphite flotation, especially in heavy sections, and of increased casting expansion during solidification, leading to unsoundness, particularly in soft green sand molds. Below this range, unsoundness may occur because of lack of feeding, and at very low carbon contents, carbides may appear in the castings, particularly in thin sections. Silicon enters ductile iron from raw materials, including cast iron scrap, pig iron, and ferroalloys, and to a small extent from silicon-containing alloys added during inoculation. The preferred range is approximately 2.0 to 2.8%. Lower silicon levels lead to high ductility in heat treated irons but to a danger of carbides in thin sections as-cast, while high silicon accelerates annealing and helps prevent carbides in thin sections. As the silicon content rises, the ductile-to-brittle transition temperature in ferritic iron increases. Hardness, proof stress, and tensile strength also increase. Carbon Equivalent. The carbon, silicon, and phosphorus contents can be considered together as a CE value, which can be a useful guide to foundry behavior and some properties. There are several CE formulas that are useful in assessing the casting properties and solidified

Typical minor element contents of raw materials for ductile iron Minor elements, %

Raw material

High-purity pig iron Deep-drawing steel Constructional steel

Mn

S

P

Ni

Cr

Cu

Mo

Al

Sn

As

B

V

0.04 0.26 0.4–0.9

0.015 0.015 0.023

0.013 0.016 0.015

0.06 0.01 0.08

0.01 0.01 0.2

0.02 0.02 0.08

0.01 0.01 0.02

<0.005 <0.005 <0.005

<0.01 <0.01 0.02–0.04

<0.01 <0.01 0.01–0.04

0.0005 0.0006 0.0008

<0.01 0.01 0.02

Cast Iron Foundry Practices / 821 structure of the iron. When CE = C% + 1=3 (%Si + %P) = 4.3, the iron will be of wholly eutectic composition and structure, and the deviation of the CE value from this value is a measure of the relative amount of eutectic. If CE < 4.3, there will be a proportion of dendrites; if CE > 4.3, there will be primary graphite nodules in the structure. The degree of saturation, Sc, is sometimes used, particularly in the German literature, to express the nearness to the eutectic composition. The value of Sc can be determined from Eq 1: Sc ¼

Actual %C 4:23
0:3ð% Si þ PÞ

(Eq 1)

When Sc < 1, the iron is hypoeutectic and will contain primary dendrites. When Sc > 1, there will be primary graphite in the structure. The CE liquidus (CEL) is a measure of the liquidus temperature, which has a minimum value at the eutectic composition; that is, CEL = %C + %Si/4 + %P/2. Maximum fluidity occurs when this value is reached. The CEL is the only carbon equivalent that can be conveniently measured in a shopfloor thermal analysis test for metal composition, used on untreated iron prior to magnesium treatment. It is usual to aim for a CEL value of approximately 4.4 to 4.5, and values much higher than this are restricted in order to avoid graphite flotation. Figure 9 is an example of a nomogram constructed to provide guidance on an optimal range of composition. Manganese. The main source of manganese is steel scrap used in the charge. This element should be limited in order to obtain maximum ductility. In as-cast ferritic irons, it should be 0.2% or less. In irons to be heat treated to the ferritic condition, it should be 0.5% or less.

Fig. 9

In irons to be used in the as-cast pearlitic condition, it may be present up to 1%. Manganese is subject to undesirable microsegregation. This is especially true in heavy sections, in which manganese encourages grain-boundary carbides that promote low ductility, low toughness, and persistent pearlite. Magnesium. The magnesium content required to produce spheroidal graphite usually ranges from 0.04 to 0.06%. If the initial sulfur content is below 0.015%, a lower magnesium content (in the range of 0.035 to 0.04%) may be satisfactory. Compacted graphite structures with inferior properties may be produced if the magnesium content is too low, while too high a magnesium content may promote dross defects. Sulfur is derived from the charged metallic raw materials. In cupola melting, it is also absorbed from the coke. Before magnesium treatment, the sulfur content should be as low as possible, preferably below 0.02%. The final sulfur content of ductile iron is usually below 0.015%, but if cerium is present, it may be higher because of the presence of cerium sulfides in the iron. Excessive final sulfur contents are usually associated with magnesium sulfide slag and dross. When using cupola-melted iron, it is common to desulfurize the metal before magnesium treatment (usually with lime or calcium carbide, either continuously or in batches) to levels of 0.02% or less. Before treatments in the mold or in the metal stream, it is advantageous to reduce the initial sulfur to 0.01% or less. Cerium can be added to neutralize undesirable trace elements that interfere with the formation of spheroidal graphite and to aid inoculation. It may then be desirably present to the extent of 0.003 to 0.01%. In irons of very low minor element content, cerium may be

Typical carbon and silicon ranges for ductile iron castings. Source: Ref 17

undesirable and may promote chunky nonspheroidal graphite, especially in thick sections; the deliberate addition of impurities may be necessary to avoid this effect. Cerium is added as a minor constituent of a number of magnesium addition alloys and inoculants to improve graphite structure. It is removed during the remelting of scrap ductile iron. Aluminum. The presence of even trace amounts of aluminum in ductile iron may promote subsurface pinhole porosity and dross formation and should therefore be avoided. The most common sources of aluminum are contaminants in steel and cast iron scrap, notably aluminum pistons from scrap automobile engines. Another source is aluminum-containing inoculants. Aluminum as low as 0.01% may be sufficient to cause pinholes in magnesium-containing ductile iron. It is advisable to use inoculants of low aluminum content and to practice late metal stream inoculation with very small quantities of inoculant. Phosphorus is normally kept below 0.05% because it promotes unsoundness and lowers ductility. Minor Elements Promoting Nonspheroidal Graphite. Lead, antimony, bismuth, and titanium are undesirable elements that may be introduced in trace amounts with raw materials in the charge, but their effects can be neutralized by a cerium addition. The effect of excess antimony on the structure of ductile iron is shown in Fig. 10. During solidification of the casting, antimony segregates at the metal/graphite interface, forming a thin shell around the nodules. This shell will slow down or stop carbon diffusion between the matrix and the nodules. Figure 10 shows that after the formation of the antimony shell, carbon is prevented from diffusing from the liquid to the solid. Much of the carbon is trapped in the last solidifying liquid in the cell boundary. When the last of the liquid solidifies, the carbon forms lamellar graphite. Minor Elements Promoting Pearlite. Nickel, copper, manganese, tin, arsenic, and antimony all promote pearlite and are listed here in order of increasing potency. They can enter the iron as trace constituents of raw materials. Copper up to 0.3% and tin up to 0.1% can be deliberately used as ladle additives when fully pearlitic structures are required. Almost all trace elements promote pearlite formation, and their effects are cumulative; therefore, a charge of high purity is essential for achieving fully ferritic structures as-cast or with minimal annealing. Minor Elements Promoting Carbides. Chromium, vanadium, and boron are all carbide promoters. Manganese may accentuate the carbide-stabilizing effects of these elements, especially in heavy sections in which segregation promotes grain-boundary carbides. These elements are controlled by careful selection of metallic raw materials for melting. Alloying Elements to Promote Hardenability. Nickel up to approximately 2% and molybdenum up to approximately 0.75% are the usual

822 / Cast Irons elements added to promote hardenability when heat treatment is to be used. Small amounts of manganese and copper also promote hardenability, but they are normally used only in combination with other elements. Copper has limited solubility and must be kept below 1.5%. Alloying Elements to Achieve Special Properties. Austenitic matrix structures are achieved by the addition of 20% Ni (more when resistance to heat, corrosion, or oxidation is required). Up to 5% Cr can be added to such irons. Nickel contents to 36% produce irons of controlled low-expansion properties. Up to 10% Mn in austenitic irons leads to low magnetic permeability and allows a lower nickel content to be used to achieve a stable austenite. Silicon contents to 6% produce ferritic matrix structures with reduced growth, scaling and thermal distortion, and cracking at elevated temperatures. The addition of up to 2% Mo to pearlitic, ferritic, and austenitic irons improves creep and elevated-temperature strength.

Magnesium Treatment Treatment to produce ductile iron involves the addition of magnesium to change the form of the graphite, followed by or combined with inoculation of a silicon-containing material to ensure a graphitic structure with freedom from carbides. For magnesium treatment, the iron will normally be at 1450 to 1510 C (2640 to 2750 F). Many methods of treatment are practiced; these are discussed more fully in Ref 18 to 20. Three of the most important methods are outlined in this section.

Fig. 10

Metallic Magnesium Treatment. The reaction between metallic magnesium and molten iron is violent. The magnesium is vaporized and burns vigorously in air. Special enclosed reaction vessels are needed, and one of the best established processes uses a converter (Fig. 11) in which ingots of magnesium are introduced into the bottom of the liquid metal in a tiltable vessel under atmospheric pressure. Other methods have involved introducing magnesium powder, rods, or wire into the molten metal with atmospheric or pressurized vessels designed to exclude air and to prevent the ejection of molten metal and fumes during the solution of the magnesium. Magnesium particles have been made into compacts with iron powder or swarf, which reduces somewhat the violence of the reaction. Such materials can be added in special vessels or by plunging into the ladle in a refractory bell. In another technique, coke impregnated with magnesium is plunged into the liquid iron. Magnesium Alloy Addition. A nickel alloy with 14 to 16% Mg can be added to the ladle during filling or by plunging. The reaction is spectacular but not violent, and a very consistent recovery is obtained. A disadvantage lies in the accompanying increase in nickel and the cost of the alloy. Other nickel alloys containing much lower magnesium contents (down to 4%) have also been used and involve a much quieter reaction. Most alloys used to introduce magnesium into molten iron are based on ferrosilicon containing 3 to 10% Mg. The reaction varies from fairly violent (with 10% Mg) to quiet (with 3% Mg). The alloy can be plunged in a refractory bell or added in the ladle using

Lamellar graphite in a ductile iron resulting from excess antimony. See text for discussion. Original Magnification: 500x. Courtesy of B.V. Kovacs, Kosmar Enterprises

a number of different techniques, including pouring the molten iron onto the alloy in the bottom of the ladle. When this method is used, the alloy is often placed into a specially designed pocket and covered with a layer of steel turnings (this is known as the sandwich process). Treatment usually takes place in ladles having a height-to-diameter ratio of approximately 2:1 to contain any splashes of metal. The use of a cover with a tundish through which the iron can be poured reduces any ejection of metal, fume, and flame and improves the yield of magnesium (Fig. 12). The Flotret process is a method of treating the molten metal as it fills the ladle. The magnesium alloy is placed in a cavity in a closed vessel through which the molten metal is poured between the furnaces and the ladle. This method can be used when pouring the metal from one ladle to another. Another design of flow-through vessel is used in the Imconod process. In-the-Mold Treatment (In-Mold Process). Magnesium-containing alloys can be added in the mold. The alloy, usually of the ferrosilicon-magnesium type, is placed in a specially designed chamber or enlargement of the running system before the mold is closed, and the alloy dissolves as the metal is cast. This method

Fig. 11 position

Schematic of the Fischer converter. (a) Vessel in filling position. (b) Vessel in treating

Cast Iron Foundry Practices / 823 contains the reaction in the mold and avoids ejection of fume and flare. Low-sulfur iron (0.01% S) is used to avoid problems resulting from sulfide inclusions in the castings.

Control of Magnesium Content In all methods of magnesium treatment, accurate measurement of the weight of metal treated and the amount of additive is essential. The initial sulfur content of the iron and the treatment temperature must also be known, because they influence the final magnesium content. The recovery of magnesium can be expressed as: Recovery %   ð% final Mg contentÞ ¼ 3= ð% added MgÞ
4ð% initial S contentÞ  100

Inoculation Following magnesium treatment, the iron is usually subjected to final inoculation, sometimes referred to as postinoculation. Inoculation is commonly carried out in the ladle using a granular inoculant, which may be commercial ferrosilicon containing 75% Si or one of a wide range of proprietary alloys, usually containing 60 to 80% Si. The amount of inoculant added usually ranges from approximately 0.25 to 1.0%. A higher percentage of silicon in the magnesium addition alloy may permit less inoculation. The inoculant can be added during reladling, stirred into the metal, placed on the bottom of the ladle before filling, or plunged in a refractory bell, as late as possible before casting. Effective stirring is necessary, and

one way of achieving this is by bubbling air or nitrogen through the melt using a porous plug in the bottom of the ladle. Inoculation reduces undercooling during solidification and helps to prevent carbides in the structure, especially in thin sections. It increases the number of graphite nodules, thus improving homogeneity, assisting in the formation of ferrite, and promoting ductility. It assists in reducing annealing time and reduces hardness. The effect of an inoculant is greatest as soon as it has dissolved, after which the effect fades over a period of 20 to 30 min. Both the initial potency and the rate of fading are influenced by trace elements, including calcium, aluminum, cerium, strontium, barium, and bismuth. Figure 13 shows how the number of nodules

(Eq 2)

The magnesium sulfide produced by the reaction flows into the slag on the top of the ladle and is removed by skimming or by use of a teapot ladle. Typical values of recovery at a treatment temperature of 1450  C (2640  F) are:  50% for a 16% Ni-Mg alloy added in the

ladle

 40% for a 9% Mg ferrosilicon added using

the sandwich process

 60% for a 5% Mg ferrosilicon added using

the sandwich process

 50% for pure magnesium added in the

converter

 45% for the in-mold or Flotret process

During magnesium treatment, there is usually a decrease in metal temperature of 35 to 50  C (65 to 90  F).

Fig. 12

Treatment ladle with tundish cover used in the magnesium treatment of ductile iron

Fig. 13

Variations in nodule number with time for 22 mm (0.875 in.) bars of cerium-free ductile irons after addition of various inoculants. Source: Ref 21

824 / Cast Irons in test bars decreased with time after ladle inoculation for ten different inoculants. Late addition of an inoculant as the metal is being cast is much more effective. It can be achieved by placing granular inoculant or shaped inoculant particles in the mold in an enlargement of the runner or a special chamber in the running system. Alternatively, fine granular inoculant can be added to the metal pouring stream, either by dispensing it to the sprue or by encasing it in hollow wire and passing this into the sprue. These late methods of inoculation require only 1=5 to 1=10 as much inoculant as ladle inoculation and are sometimes also used to supplement ladle inoculation, particularly when fading has occurred through long holding times in the ladle. The superior effect of late inoculation in increasing nodule number is illustrated in Fig. 14, which also shows that the advantage is greatest in thinner-section castings. Late inoculation is sometimes practiced in addition to other methods of inoculation as a means of intensifying the inoculation effect or as a safe-guard against fading of inoculation in the ladle, especially when very thin castings are being made. When magnesium treatment is done in the mold, the magnesium alloy usually contains enough silicon to produce the required inoculation effect, but additional inoculant can be mixed with the magnesium alloy.

Casting and Solidification Ductile iron shares with gray irons the properties of very good fluidity and castability, but there are important differences in the casting and solidification characteristics.

Gating Systems. Ductile iron normally contains magnesium, and oxidation of this element, together with silicon, may give rise to small inclusions and dross in both the ladle and the gating system. This dross must not enter the casting where it can cause laps, rough surfaces, or pin-holes. Other recommendations include:  Metal held in the ladle must be skimmed

clean; the use of teapot ladles is advisable.

 Magnesium content should be kept to a

minimum, preferably less than 0.06%.

 A gating system of adequate size should be

used, designed to achieve rapid mold filling with minimum turbulence of metal.  A slightly pressurized gating ratio is desirable. For example, suitable ratios of cross sections of sprue-runner-ingate may be 4:8:3 to ensure a short filling time and a full sprue during casting.  Ingates should be placed in the drag, as near the bottom of the casting as possible.  A ceramic filter in the gating system removes small slag inclusions and assists smooth metal flow. It has been shown to reduce dross defects and to improve the mechanical properties of the iron. Pouring Temperature and Speed. Excessive temperature gradients in the casting and pouring of cold metal are undesirable and may promote carbide formation in thin sections and casting extremities. These can be avoided by pouring at not less than 1315  C (2400  F) for castings of 25 mm (1 in.) section and above, by pouring at up to 1425  C (2600  F) for castings of 6 mm (¼ in.) section, and by modifying the gating system to include such features as flowoffs and runner bar extensions. There are no absolute rules for the speed of pouring, but Fig. 15 illustrates typical values found to be successful. Solidification and Feeding (Risering). The change from flake to nodular graphite is accompanied by important differences in solidification behavior. In hypereutectic irons, graphite segregation and flotation occur more easily than in

gray iron. Such segregation can aggravate dross problems and give rise to local high-carbon regions that adversely affect mechanical properties and the appearance of machined surfaces. During solidification, the precipitation of graphite in a nodular form causes ductile iron castings to expand to a greater degree and with more force than gray iron. Consequently, ductile iron castings require more rigid molds and more attention to feeding. In green sand molds, in particular, ductile iron castings are likely to be oversized compared with the pattern and compared with gray iron castings. It is necessary to make dense molds. Flasks require good weighting and clamping, especially for larger castings, and they may need internal bars. The use of chemically hardened or dry sand molds is a great advantage, provided that they are adequately cured (Ref 23). It is sometimes possible to make sound ductile iron castings of simple shapes without the use of risers by taking advantage of graphite expansion, which occurs during eutectic solidification. The requirements are:  A strong, well-compacted, chemically bonded

sand mold contained in a rigid molding flask

 A CE of approximately 4.3 and a carbon

content of approximately 3.6%

 A low casting temperature, probably below

1350  C (2460  F)

 A gating system and running speed to mini-

mize temperature gradients developed in the mold Nevertheless, for most purposes, ductile iron castings will require the use of risers to ensure fully sound castings, even in rigid molds. The principles of directional solidification should be employed in casting design with regard to the placement of gates and the locations of risers. There are three stages of solidification volume change, and rational risering practice takes account of them. They are:  A volume decrease of the liquid metal as

soon as it has been poured, while it cools to the temperature of eutectic solidification  A volume increase during eutectic solidification, when graphite nodules grow and exert considerable expansion pressure on the mold (depending on casting design and mold strength, this can lead to oversized castings)  A volume decrease during the last stage of solidification (secondary shrinkage)

Fig. 14

Nodule numbers in ductile irons inoculated with foundry-grade ferrosilicon (75% Si) added either to the ladle or to the metal stream. Source: Ref 21

Fig. 15 Ref 22

Example of pouring rates found to be successful for ductile iron castings. Source:

For the most efficient use of risers to achieve sound castings, advantage is taken of the second-stage volume increase to reduce the amount of feed metal required. This is done with pressure control, in which riser (feeder) necks are designed to freeze during solidification, causing the solidification volume increase to compensate for any casting expansion and late-stage volume decrease or secondary shrinkage.

Cast Iron Foundry Practices / 825 There are a number of approaches to the feeding of ductile iron castings. A pressure control approach relates the cooling modulus of the casting to those of the riser and the riser neck. A typical nomogram is shown in Fig. 16, in which the modulus of the riser neck, Mn, is related to the significant modulus of the casting, Ms. More advanced calculations of feeding requirements for ductile iron take into account variables affecting mold rigidity, metal composition (including CE and carbon content), and pouring temperature. These can be combined to provide a factor that can be included in a nomogram such as that shown in Fig. 17. The risers themselves can promote or accentuate hot spots and excessive mold wall movement, so the use of smaller risers by the application of exothermic feeder sleeves can be an advantage and can improve yield, often to more than 80%. Nomograms and computer programs for the use of such sleeves are generally supplied by manufacturers. Excessive ferrostatic head should be avoided because this increases the pressure on the mold, leading to casting swelling and further feed metal requirement. Several computer-aided methods are available for determining the best risering system, but for their use to be successful, all of the production variables affecting the shrinkage behavior of ductile iron must be controlled. Ductile iron risers are more difficult to remove than those of gray iron. Notched riser necks and the use of breaker cores can facilitate breaking off, and mechanical devices are available to assist the breaking or levering off of risers. However, abrasive cutting wheels sometimes must be used to avoid damage to the castings. The change in shape of graphite from flake to nodular makes ductile iron more prone to carbide formation in thin sections, and it may be difficult to obtain graphite throughout sections less than 6 mm (¼ in.) thick. There is also a tendency for inverse chill to occur in the heat

A system for calculating riser dimensions. Ms, volume/effective cooling surface of largest section of castings; Mn, circumference of riser neck/cross section area of riser neck. Modulus of riser = 1.2 Mn. Source: Ref 24

centers of round and symmetrical sections. This is related to the cooling conditions of such sections and is most likely to be cured by adjusting the system and by using late inoculation in the metal stream or in the mold.

Ductile Iron Production Quality Tests Metal Composition. Before treatment with magnesium, it is necessary to know the sulfur content of the liquid iron. This is best determined rapidly by taking a chilled coin sample for spectroscopic analysis. If required, the sulfur and the carbon content can be determined rapidly by the combustion of crushed solid samples taken from the pins attached to the test coin. Spectroscopic analysis for other elements can be carried out either before or after magnesium treatment, but it is unlikely that a completely white sample suitable for analysis will be obtained after inoculation or if a heavy addition of a silicon-containing magnesium alloy has been made. Final silicon content after inoculation may need to be determined chemically with drillings taken from a sample cast after treatment. A reliable determination of carbon in treated ductile iron cannot be obtained chemically except on a solid sample (obtained, for example, from pins attached to a coin sample, because of loss of carbon during drilling and the likely presence of graphite in the coin). Rapid shopfloor control of CE, carbon, and silicon can be carried out by thermal analysis of the metal in the same manner as for gray iron, but this must be done before magnesium treatment.

Other properties of the iron can be deduced from thermal analysis data, including the undercooling of the eutectic, which is a guide to nodule number and chilling tendency. However, in general, such developments are still taking place. For this reason, nodularity and nodule number are generally determined by other methods. Nodularity and Nodule Number. Figure 18 shows a test coupon of a design recommended by the American Foundrymen’s Society Committee 12K. Such a coupon generally enables rapid determination of the success of the magnesium treatment, freedom from undesirable trace elements, and adequate postinoculation in producing an iron with a good nodular graphite structure and of adequate nodule number to ensure freedom from carbides and to yield correct properties. The mold is made in hardened sand and consists of a central block with two ears, which are broken off as soon as the casting has solidified and the ears have cooled to a dull red or black heat. After cooling to room temperature, an ear is broken or cut in two, and the cross section is ground, polished, and examined under the microscope. Alternatively, small samples for metallographic examination can be cut from selected castings or from other types of agreed test bar. The number of nodules in a specified area is counted, usually using a magnification of 100; the specified area is typically a square millimeter or a square inch of the field of view or along a specified length of line. The shape of the nodules is evaluated in accordance with the classification in ISO 945 or ASTM A 247. Nodule number and

Fig. 16

System of calculating feeder dimensions. First, calculate significant modulus, Ms. Second, read off from Ms the values of Mf, Mn, D, and a:b. Third, select feeding requirements, depending on mold type and pouring temperature. Fourth, select Mi/Ms based on foundry experience. Source: Ref 25

Fig. 17

826 / Cast Irons perfection of shape can also be evaluated using quantitative metallographic instruments. Good practice requires a nodularity sample to be taken from each batch of iron treated and poured or, when using automated holding and pouring of metal, at regular intervals of time. For most purposes, 85 to 100% of the graphite should be in a fully nodular form. Samples from castings or test bars will be periodically examined to verify the graphite and matrix structure of the iron produced, to evaluate ferrite and pearlite contents, and to verify freedom from undesirable carbides, inclusions, and dross as a general control of the production process. Mechanical Test Coupons. To ensure that the castings meet the specified requirements, mechanical properties are checked periodically on testpieces machined from cast test coupons. Recommended test bars cast separately or attached to castings are in the form of keel blocks. Y-blocks, or horizontal bars with knockoff heads and are indicated in most specifications. If the castings are to be used in the heat treated condition, the test bars will be heat treated with the castings they represent before mechanical test specimens are obtained. The actual properties of castings are sometimes checked by machining test bars from the castings and obtaining agreement between the foundry and its customer.

Compacted Graphite Iron Foundry Practice Compacted graphite (CG) irons are characterized by a graphite structure that is intermediate between those of gray and ductile irons. The graphite exists as blunt flakes that are interconnected.

Fig. 18

Chemical Composition. The range of acceptable carbon and silicon contents for the production of CG iron is rather wide, as shown in Fig. 19. Nevertheless, the optimum CE must be selected as a function of section thickness, in order to avoid carbon flotation when CE is too high or excessive chilling tendency when CE is too low. The manganese content can vary between 0.1 and 0.6%, depending on whether a ferritic or pearlitic structure is desired. Phosphorus content should be kept below 0.06% in order to obtain maximum ductility from the matrix. The initial sulfur level should be below 0.025%, although techniques for producing CG iron from base irons with higher sulfur levels are now available. The change in graphite morphology from the flake graphite in the base iron to the compacted graphite in the final iron is achieved by liquid treatment with different minor elements. These elements may include one or more of the following: magnesium, rare earths (cerium, lanthanum, praseodymium, etc.), calcium, titanium, and aluminum. The amounts and combinations to be used are a function of the method of liquid treatment, base sulfur, and section thickness. Compacted graphite iron has a strong ferritization tendency. Copper, tin, molybdenum, and even aluminum can be used to increase the pearlite/ferrite ratio. Again, the optimum amounts of these elements for a particular matrix structure are largely a function of section size. Melting Practice. The melting aggregates to produce CG iron are, in principle, the same as those used for ductile iron. Induction furnaces, cupolas, and arc furnaces have been reported to be adequate. Similar requirements of raw materials, superheating, and desulfurization

Test coupon recommended by American Foundrymen’s Society Committee 12K for rapid examination of ductile iron microstructures. Dimensions are given in inches.

before melt treatment apply. If a ferritic structure in the as-cast condition is desired, a pure pig iron with low manganese, phosphorus, and sulfur contents is recommended. If some pearlite is acceptable, steel scrap can be used. Melt Treatment. The most important methods used for manufacturing CG iron can be classified as:  Controlled undertreatment with magnesium-

containing alloys

 Treatment with alloys containing both com-

pactizing elements (magnesium, rare earths, and calcium) and anticompactizing elements (titanium and aluminum)  Treatment with a rare-earth-base alloy or magnesium/rare earth alloys  Treatment of a base iron containing rather high amounts of anticompactizing elements (sulfur and aluminum) with alloys containing compactizing elements (magnesium and cerium) Compositions of typical treatment alloys used in the production of iron are listed in Table 8. As shown in Fig. 20, CG iron can be obtained when treating a ductile-type base iron with a Mg-Fe-Si alloy (e.g., alloy 1 in Table 8) if residual magnesium is controlled in the range of 0.013 to 0.022% (Ref 28). This type of control is difficult to achieve when ladle treatment is used; overtreatment (resulting in too much spheroidal graphite) or undertreatment (resulting in flake graphite) can occur. Nevertheless, instances in which this process has been applied are reported in Ref 36. Better control of residual magnesium can be achieved if the in-mold process is used. Indeed,

Fig. 19

Optimum range for carbon and silicon contents for compacted graphite (CG) iron. Source: Ref 26

Cast Iron Foundry Practices / 827 good-quality CG iron has been produced using alloys similar to alloy 1 (Table 8) from iron with 0.008 to 0.017% base sulfur. Residual magnesium was 0.015 to 0.025% in section sizes of 8 to 63 mm (0.3 to 2.5 in.) (Ref 27). As the base sulfur content increases, higher residual magnesium is required. The treatment of iron with alloys containing compactizing elements balanced by anticompactizing elements balanced by anticompactizing elements has much wider industrial application. A treatment alloy containing approximately 5% Mg and 0.3% Ce balanced by approximately 9% Ti (e.g., alloy 2 in Table 8) is used to obtain 0.015 to 0.035% residual Mg, 0.06 to 0.13% residual Ti, and low levels of cerium in the final cast iron (Ref 29). As shown in Fig. 21), using a magnesium-titanium combination rather than magnesium alone dramatically increases the range of residual magnesium over which CG can be obtained. This treatment alloy was later improved by the inclusion of 4 to 5.5% Ca (alloy 3 in Table 8) to extend the working range of residual magnesium and the tolerable range of base sulfur (Ref 30). A wide range of carbon and silicon contents Table 8

(3.15 to 4% and 1.7 to 3%, respectively) can be used to produce CG iron in section thicknesses of 12 to 65 mm (0.5 to 2.5 in.). Base iron sulfur contents can vary from 0.02 to 0.04%, resulting in final sulfur levels of 0.01 to 0.02% (Ref 38). Excessive sulfur has been shown to promote a gray rim (flake graphite, or FG), while too low a sulfur content promotes higher nodularity. The amount of magnesium-titanium alloy used for CG treatment depends on the base sulfur content, but it is not critical. As for ductile iron, a ladle postinoculation is indicated. It appears important to keep treatment at approximately 1400  C (2550  F) because lower temperatures result in higher nodularity (Ref 37). The main disadvantage of this method seems to be titanium contamination (0.1 to 0.5% Ti) of returns and castings. Alloy 4 in Table 8 can also be included in the category of alloys that rely on a balance of compactizing and anticompactizing elements. Additions of 0.2% of this alloy resulted in CG structures in bar castings ranging in thickness from 25 to 127 mm (1 to 5 in.) (Ref 31). A rather high chilling tendency was observed. The treatment of liquid iron with rare earth alloys was actually one of the first processes to

Nominal compositions of typical treatment alloys for compacted graphite iron Composition, % Compactizing elements

Anticompactizing elements

Neutral elements

Alloy number

Mg

Ce

La

TRE(a)

Ca

Ti

Al

Si

Fe

Reference

1 2 3 4 5 6 7 8 9

5 5 5 ... ... ... ... 3.7 4.3

... 0.3 0.3 24 30 16 2.9 0.8 0.7

... ... ... 14.4 50 80 26.5 0.5 1.8

... 0.3 0.3 48 80 96 29.4 1.7 2.9

1 <1 4.5 7.5 ... ... 0.63 1.05 0.66

... 9 9 ... ... ... ... ... ...

<1.2 <1.5 1.2 4.3 ... ... 0.13 0.88 0.9

45 52 50 33.2 ... ... 30.5 45.3 45.3

bal bal bal bal bal bal bal bal bal

27, 28 29 30 31 32, 33 34 34 35 35

(a) TRE, total rare earth elements

Fig. 21

Fig. 20

The influence of residual magnesium on graphite shape. Source: Ref 28, 36

Range of residual magnesium that produces compacted graphite. SG, spheroidal graphite; CG, compacted graphite; FG, flake graphite. Source: Ref 37

gain industrial application. Cerium-mischmetal (for example, alloy 5 in Table 8) (Ref 32, 33) is used for production of medium- and heavysection CG iron castings. Residual cerium contents of 0.013 to 0.075% are required, depending on base sulfur. The dependency of alloy addition on base sulfur level is shown in Fig. 22. Although rare earth treatment offers a number of advantages over magnesium treatment, including lower fading time and no smoke during the reaction time, the high chilling tendency in thin sections is a drawback. Enhanced postinoculation with ferrosilicon does not seem to solve the problem entirely, because it results in increased nodularity. Alloys containing a higher lanthanum/rare earth ratio than typical mischmetal has may alleviate the chilling tendency (alloys 6 and 7 in Table 8) (Ref 34). Postinoculation is generally required when silicon-free alloys are used. Good results have also been claimed for a Ce-Ca-Si alloy of undisclosed composition (Ref 39). With low base sulfur, good CG structure could be achieved when at least 0.016% Ce and 0.24% Ca were added to the melt. When base sulfur was raised to 0.059%, the cerium and calcium contents had to be increased to 0.091 and 0.53%, respectively. Fading times of the order of 22 min for a 500 kg (1100 lb) ladle, as well as good response to the recycling of returns, were also claimed as advantages of this process. Another way to avoid the chilling problems associated with mischmetal treatment is to use a magnesium/rare earth/ferrosilicon alloy such as alloy 8 or 9 in Table 8. When these alloys were used in conjunction with the in-mold process (Ref 35), a residual magnesium, cerium, and lanthanum content of 0.018 to 0.028% was required for CG structure when the base sulfur was 0.016%, which correlates with data given in Ref 27. Finally, another group of treatment methods addresses the problem of high levels of anticompactizing elements, such as sulfur or aluminum, in the basic iron. A special alloy with undisclosed composition, containing high levels of magnesium and cerium and conventional quantities of aluminum and calcium, is claimed to allow CG iron production at virtually all sulfur levels. Regardless of the treatment alloy used, for a specific alloy addition and base sulfur content there is an optimum range of treatment temperatures. For example, for an iron with base sulfur of 0.12 to 0.13% and a treatment addition of 1.75% alloy, the treatment temperature must be between 1468 and 1510 C (2675 and 2750  F). Too low a temperature results in increased nodularity; an excessive temperature causes premature alloy dissolution, and the iron remains gray with flake graphite. Another approach to producing CG iron is to substitute aluminum for most of the silicon in the iron. This allows the use of a magnesium-ferrosilicon alloy, because aluminum has an anticompactizing effect and can counterbalance the

828 / Cast Irons influence of magnesium. Thus, a correct balance between aluminum and magnesium should result in CG. This idea has been applied using both ladle and in-mold treatment (Ref 40). A 5% magnesium-ferrosilicon alloy with 0.3% Ce was used. Good CG structures with no chill on wedge tests were achieved for aluminum levels of 2.68 to 4.48% by using the trigger treatment method. A 1.4% addition of magnesium-ferrosilicon alloy resulted in residual magnesium of 0.016 to 0.028%, and less than 20% nodularity was achieved over this range. No postinoculation was used. It is interesting to note that holding times in excess of 30 min have proved to be acceptable, even when the treated metal was held in the induction furnace. Excellent results were obtained when the in-mold process was used with a chamber area to result in 0.013 to 0.023% residual Mg and 0.022 to 0.032% residual Ce, for irons containing 1.62 to 4.43% Al. Regarding postinoculation, it is generally accepted that CG iron exhibits a high chilling tendency, especially in sections thinner than 6 mm (¼ in.). To counteract this negative effect, it is necessary to use either silicon in the treatment alloy or some type of postinoculation. Most of the time, 0.2 to 0.3% ferrosilicon containing 75% Si is used for postinoculation. Nevertheless, in thin sections of CG iron castings, it is difficult to counteract chilling effects properly by postinoculation, because increasing the amount of inoculant normally results in spheroidal graphite (SG) rather than CG. It has been suggested that 0.1 to 0.2% Al (Ref 41) or a ferrosilicon-titanium-aluminum alloy of undisclosed composition (Ref 42, 43) can be used as postinoculants to decrease chill without increasing nodularity. Process Control. Production of good-quality CG iron requires careful control at every stage. The chemical composition must be kept within rather rigid limits, particularly in regard to

Fig. 22

The correlation between base sulfur content and amount of cerium-mischmetal treatment alloy required to produce compacted graphite. Source: Ref 33

sulfur content. Sulfur content must be determined with 0.001% accuracy before melt treatment (Ref 36) because excess sulfur can result in undertreatment, while too low a sulfur content can produce iron with excessive nodularity. Infrared methods are recommended for determining sulfur content. Carbide-stabilizing elements must also be checked accurately, because they increase the already high chilling tendency of CG iron. For example, chromium should be kept under 0.01%. A variety of methods can be used to control melt treatment. When cerium-containing treatment alloys are used, a classic wedge test approach is recommended (Ref 36). Wedge tests are poured before treatment, after treatment with mischmetal, and after postinoculation, with required fractures being dark gray, completely white, and light gray, respectively. Cooling curve analysis is a good way to predict microstructure before pouring and thus allow corrections if necessary. Typical cooling curves for flake, compacted, and nodular graphite irons and for white iron are shown in Fig. 23. It can be seen in Fig. 23(a) that hypoeutectic CG iron has a temperature of eutectic undercooling (TEU) lower than that of FG or SG iron, although its temperature of eutectic recalescence (TER) is between the other two, Nevertheless, depending on the level of postinoculation, it is sometimes possible for CG to have a higher TEU or a lower TER than SG, although this is the exception rather than the rule. It is apparent that the easiest way to identify a good CG structure is based on the value of DT = TER
TEU. The value of DT should be in excess of 25 to 30  C (77 to 86  F) for rare-earth-treated iron and in excess of 10  C (50  F) for magnesium-treated irons (Ref 45). For hypereutectic composition (Fig. 23b), the undercooling can be higher for either SG or CG iron, so simple cooling curve analysis cannot be used to differentiate between these two irons. Improved microstructure predictions can be made by computer-aided differential thermal analysis (CADTA), numerical processing of cooling curve data. Based on first derivative curves obtained using CADTA, hypoeutectic CG differs from other irons by the fact that it has only one slope on the second part of the curve (after recalescence), while the other irons have two. For hypereutectic irons, the first derivative curve of CG iron exhibits two maxima, compared with only one for SG iron. When large quantities of CG iron are poured, it may be possible to run a metallographic sample, as is done in some SG iron foundries, immediately after melt treatment and before pouring. To date, a universally accepted method for estimating the amount of CG in cast iron is not available. Certainly the simplest method is one in which the nodularity is determined by the ratio between the number of type I and II (nodular) graphite particles (ASTM A 247) and the total number of graphite particles (Ref 46):

Nodularity ð%Þ ¼

type I þ type II type I þ type II þ type III þ type IV

(Eq 3)

The amount of CG is considered to be the difference between 100% and the percent of nodularity. Equation 3 is likely to be satisfactory for graphite shape estimation in production conditions, but it neglects the fact that several CG particles can belong to the same graphite aggregate, or that graphite considered to be spheroidal is in fact part of a CG aggregate. Another method suggests the use of different standard micrographs (Ref 29, 31). Molding Materials. As for ductile iron, all molding materials that can produce rigid molds can be used for CG iron. These include bentonite-bonded sands, cement-bonded sands, and resin-bonded sands. Compacted graphite iron melts are more sensitive to sulfur pickup from the mold than are SG iron melts, so unlike SG iron, CG iron must not be overtreated (Ref 36). Particular attention should be paid when using reclaimed resinbonded sand with paratoluosulfonic acid (PTS) as a hardening catalyst. While such a new sand contains 0.086% S, at an amount of 80% reclaimed sand, the sulfur content increases to 0.31%. This amount can cause sulfur pickup in the casting of up to 0.05% S in the surface area, resulting in a flake graphite rim up to several millimeters in thickness. To avoid graphite deterioration in the surface layer, especially for fatiguestressed components, it may be useful to substitute phosphoric acid in part for PTS. Also, the application of protective mold coatings with lime, magnesia, or talc base is recommended (Ref 36). Quality Control. Typical quality-control procedures may include verification of structure, mechanical properties, physical properties, casting soundness, and dimensional accuracy. Structure can be assessed by classic metallographic techniques on coupons but also to a certain extent by ultrasonic velocity measurements on castings. While FG is easily separated from CG using ultrasound, it is virtually impossible to use ultrasound to differentiate between CG and low-nodularity SG (Ref 27). Nevertheless, good correlations between mechanical properties and ultrasonic velocity have been reported. When ultrasound is applied to castings, the ultrasonic velocity for CG structures is independent of the shape of the casting, but it should be calibrated for the section thickness. For example, for 30 mm (1.2 in.) diameter bars, the ultrasonic velocity range associated with CG is 5.2 to 5.45 km/s (3.2 to 3.4 mi/s), but for very large castings such as ingot molds, the ultrasonic velocity for good CG structures lies between 4.85 and 5.1 km/s (3.0 and 3.15 mi/s). On the other hand, resonant frequency (sonic testing) must be calibrated for a particular casting design, for which examples of satisfactory and unsatisfactory structures must be previously checked to provide a calibration range.

Cast Iron Foundry Practices / 829

Fig. 23

Cooling curves. (a) For hypoeutectic irons. (b) For hypereutectic irons. TEU, temperature of eutectic undercooling; TER, temperature of eulectic recalescence. Source: Ref 44

Separately cast samples are recommended for determination of mechanical properties. They can be 30 mm (1.2 in.) diameter cylindrical bars poured in a vertical position or keel (U) blocks with 25 mm (1 in.) wide legs. For larger castings, however, sample bars cast on the appropriate location on the casting are preferred (Ref 36, 47).

Malleable Iron Foundry Practice In malleable iron, the major part of the carbon in the microstructure occurs as irregularly shaped nodules of graphite. This form of graphite is called temper carbon because it is formed in the solid state during heat treatment. The iron is cast as a white iron of a suitable chemical composition. After the castings are removed from the mold, they are given an extended annealing heat treatment, starting at a temperature above 885  C (1625  F). Detailed information on this annealing process can be found in the article “Malleable Iron Castings” in this Volume. Chemical Composition. Malleable irons are produced from base metal in the following ranges of composition: Carbon Silicon Manganese Sulfur Phosphorus Boron Aluminum

2.00–3.00% 1.00–1.80% 0.20–0.50% 0.02–0.17% 0.01–0.10% 0.0005–0.0050% 0.005–0.0150%

Each foundry establishes its own melting practice, one that permits it to produce iron with characteristics suitable to the type of castings produced in that foundry and the heat treating equipment used in that foundry. Melting Practice. Foundries having low tonnage requirements generally charge, melt down, refine, and superheat the iron in batches. This practice is commonly referred to as cold melting.

Such installations use electric induction, electric arc, or air (reverberatory) furnaces for melting a properly proportioned charge of white iron returns, scrap iron and steel, pig iron, and ferroalloys. Air furnaces are fired with powdered coal, natural gas, or oil. The charge is laid in the hearth of the furnace and is melted by passing the flame over it. After the melt has been superheated, a sample has been chemically analyzed, and the bath has been corrected by alloy additions, if necessary, the melt is tapped and transferred by ladle to molds. The tapping temperature, usually 1475 to 1600  C (2700 to 2900  F), depends on the composition of the iron, the fluidity of the iron required to fill the thinnest sections to be cast, and the facilities for metal transfer. Ordinarily, only one heat per day is melted in batch air furnace operations. Where larger tonnages are needed, or where a continuous supply of molten iron is required to pour conveyorized molding lines, electric melting furnaces or duplexing systems are employed. With electric installations, either multiple batch heats can be arc melted or line-frequency coreless induction furnaces can be used. The choice of an electric melting system for a malleable iron foundry involves the same considerations as for a gray iron foundry. In coreless induction melting, a portion of the superheated melt is periodically tapped and replenished at once with new charge. Cupola furnaces are also used to melt iron without duplexing for some malleable castings, especially pipe fittings. Furnace linings are generally either silica or superduty refractories. Metal superheating temperatures range from 1475 to 1600  C (2700 to 2900  F), depending on the necessity for reladling, the composition of the iron, and the section thickness of the castings to be produced. Duplexed malleable iron is first melted in cupolas or in arc or induction furnaces, and then is refined in arc, induction, or air furnaces.

Usually, primary melting units are operated to provide molten metal for duplexing at relatively low temperatures (1425 to 1500  C, or 2600 to 2750  F), because primary melters generally are not efficient superheaters. Final superheating temperatures vary from 1475 to 1600  C (2700 to 2900  F), depending on the needs of individual foundries. Minor corrections in composition can be made in the duplexing furnace, but basic process control is generally accomplished in the primary melting unit. Control of Melting. Metallurgical control of the melting operation is based on the requirement for a molten iron of a certain composition that will:  Solidify white in the castings to be produced  Anneal on an established time-temperature

cycle set to minimum values in the interest of economy  Produce the desired hardenability after annealing Changes in melting practice or composition that would satisfy the first and third of these requirements generally oppose the satisfaction of the second, while attempts to improve annealability beyond an optimum level may result in difficulty with mottle (primary graphite) and low hardenability. One high-production malleable iron foundry operates to a composition standard of: Carbon Silicon Manganese Sulfur Chromium Boron Aluminum

2.50–0.10% 1.52 þ 0.08%
[(%S  1.7) + 0.15%] + 0.10.
0.05% 0.17% max 0.09% max 0.0030 þ 0.0005%
0.0030
þ 0.0020%

A minimum carbon content is required in the interest of mechanical quality and annealability, because decreasing carbon content reduces fluidity of the molten iron, increases shrinkage

830 / Cast Irons during solidification, and reduces annealability. A maximum carbon content is imposed by the requirement that the casting be white as-cast. The silicon content is limited to ensure proper annealing during a short-cycle, high-production annealing process and to avoid formation of primary graphite during solidification. Manganese and sulfur contents are balanced to ensure that all sulfur is combined with manganese and that only a safe, minimum quantity of excess manganese is present in the iron. An excess of either sulfur or manganese will retard annealing in the second stage and therefore increase annealing costs or decrease casting quality. The chromium content is kept low because chromium is a carbide stabilizer and because it retards second-stage annealing. Boron and aluminum are required for two reasons: to provide rapid annealing and to control nitrogen and oxygen activity.

Foundry Practice for High-Nickel Ductile Irons High-nickel ductile irons are austenitic alloys containing 18 to 37% Ni. Commonly referred to as the ductile Ni-Resists, some of these irons also contain chromium (up to 5.5%). Additional information on the production of ductile Ni-Resist irons can be found in Ref 48. Melting Practice. In the past, high-nickel ductile irons were generally cupola melted, but melting is now done almost exclusively in electric furnaces. Coreless induction furnaces are the most widely used in some high-production foundries and in foundries producing very large castings. Choice of furnace linings is usually based on what other alloys are being melted in the shop. Acid, neutral, or basic linings are used. Selection of charge materials is more critical in melting the higher-nickel alloys of the ductile iron versions because they are more sensitive to the tramp elements that affect graphite structure. The high nickel content causes the materials to be more prone to hydrogen gas defects, so charge materials should be thoroughly dry and melting times should be quick. The molten iron should be superheated only to the temperature necessary to treat and pour, and for as short a time as possible. Pouring is generally done above 1400  C (2550  F) to keep the molten iron surface clean and free of oxide. Magnesium treatment of the high-alloy ductile irons is normally accomplished with nickel-magnesium alloys. The nickel-magnesium treatment alloy is often added in the furnace; there is no concern for pyrotechnics. The same rules for balancing sulfur and setting residual magnesium levels in standard ductile irons apply to nickel-alloyed irons. These irons are more susceptible to the formation of intercellular carbides when magnesium levels exceed those needed to balance sulfur and to

nodularize the graphite. Although these irons respond like standard ductile irons to most inoculants, foundry-grade ferrosilicon, containing 75 to 85% Si, is most often used for postinoculation. Postinoculation is performed when tapping the furnace. Stream inoculation, where feasible, is also recommended for improved machinability. The carbon content of type D-5S alloy must be monitored carefully, because section sensitivity is high. Although the ASTM International specification allows up to 2.4% C, sections over 25 mm (1 in.) are susceptible to exploded and chunky graphite formation. Carbon levels of 1.6 to 1.8% are recommended for heavy-section castings. Molding and Casting Design. Conventional ferrous molding sands are used, including green sand, shell mold, and chemically bonded sands. The same precautions taken for high-strength irons apply to these alloys. Solidification should progress from thin to thick sections without interruption. Abrupt changes in section thickness should be avoided. Riser location should allow convenient access for riser removal. The shrinkage allowance is generally 21 mm/m (¼ in./ft), or 2.1%. Shakeout. High-nickel ductile irons can develop large thermal stresses upon cooling because of a relatively low thermal conductivity, combined with high elevated-temperature strength and high thermal expansion rates. Consequently, care should be taken not to shake out too hot, and mold cooling is recommended for intricately shaped castings and castings of widely varying section thickness.

Foundry Practice High-Silicon Ductile Irons High-silicon ductile irons, which are generally used for elevated-temperature service, contain 4 to 6% Si, often in conjunction with 0.5 to 2.0% Mo. Melting Practice. For high-silicon ductile irons, standard ductile iron melting practices apply. Cupola melting is acceptable, but these irons are commonly electric melted. Acid, neutral, or basic linings are used. Conventional ductile iron charge materials are also used, but care should be taken to minimize chromium, manganese, and phosphorus. Manganese content should not exceed 0.5%, and it is preferable to keep it below 0.3%. Chromium should be no more than 0.1% and preferably below 0.6%. These elements promote as-cast embrittlement due to carbides and pearlite, which form networks in the interdendritic regions that are difficult to break down through subsequent heat treatment and that degrade toughness and machinability. They are a particular problem in turbocharger applications, in which minimum ductility requirement must be met. Consequently, a larger proportion of pig iron in the charge is common.

Silicon additions can be made in the ladle, but it is highly recommended that all silicon be added to the furnace, except for that required for magnesium treatment and postinoculation. The same is true for molybdenum additions; either ferro-molybdenum or molyoxide briquettes can be used. Conventional magnesium treatment techniques are used, and the same rules for balancing sulfur and setting residual magnesium levels apply to high-silicon irons. These irons are more susceptible to the formation of intercellular carbides when magnesium levels exceed those needed to balance sulfur and to nodularize the graphite. Although these irons respond like standard ductile irons to most inoculants, foundry-grade ferrosilicon, containing 75 to 85% Si, is usually used for postinoculation. Pouring Practice. Due to increased silicon contents, higher pouring temperatures (>1425  C, or 2600  F) are required to develop a clean melt surface in the ladle. Therefore, pouring temperatures somewhat higher than those for ductile irons should be used to minimize dross and slag defects. Molds, Patterns, and Casting Design. These alloys are normally sand cast; sand molds can be made of green sand, shell mold, and air-set chemically bonded sands suitable for gray and ductile irons. As with any graphitic iron, high mold rigidity is necessary to minimize mold wall movement and shrinkage porosity. Softer molds will produce bigger castings, but in general, little or no allowance for shrinkage is required in the pattern. As with conventional ductile irons, the highly reactive magnesium dissolved in these irons, along with the high levels of silicon, gives rise to the formation of dross and inclusions due to the reaction with oxygen in the air during pouring and filling of the mold. Consequently, clean ladles with teapot designs or skimmers are recommended. Furthermore, measures must be taken in designing the runners and gating system to minimize turbulence and to trap dross before it enters the mold cavity. Shakeout Practice. As mentioned previously, room-temperature impact resistance is low; therefore, riser and gate removal is somewhat easier with high-silicon ductile irons than with standard ductile iron grades. These irons are quite ductile at elevated temperatures, and they should be allowed to cool before cleaning. They do exhibit a ductility trough in the temperature range of 315 to 425  C (600 to 800  F), and riser and gate removal is aided when performed upon cooling through this temperature range.

Foundry Practice for High-Silicon Gray Irons High-silicon gray irons, which are generally used for corrosion-resistant applications, contain 14 to 17% Si. They may also contain 3.25 to 5% Cr and 0.4 to 0.6% Mo.

Cast Iron Foundry Practices / 831 Melting Practice. Induction melting is the preferred method for high-silicon gray irons. Induction melting permits the very tight control of chemistry that is needed in melting these materials in order to minimize scrap losses due to cracking. The alloys have a very low tolerance for hydrogen and nitrogen, so charge materials must be carefully controlled to minimize the levels of these gases. Vacuum treatment can be used to increase strength and density. The melting point of the eutectic 14.3% Si cast iron is approximately 1180  C (2160  F), and the alloy is generally poured at approximately 1345  C (2450  F). Because of the very brittle nature of high-silicon cast iron, castings are usually shaken out after cooling to ambient temperature. However, some casting geometries demand hot shakeout while the castings are still red hot, so that the castings can be immediately stress relieved and furnace cooled to prevent cracking. Molding and Casting Design. High-silicon gray irons are routinely cast in sand molds, investment molds, and permanent steel molds. Cores must have good collapsibility to prevent fracture during solidification. Flash must be minimized, because it will be chilled iron and can readily become an ideal nucleation site for cracks that propagate into the casting. Casting design for high-silicon gray irons requires special considerations. To avoid cracking, sharp corners and abrupt changes in section size must be avoided. Casting designs should have tapered ingates to permit easy removal of gates and risers by impact and to minimize the amount of grinding required. Unless vacuum degassing is employed, conventional risers are not generally used or desired. High-silicon gray irons are generally cast in sections ranging from 4.8 to 38 mm (3=16 to 1.5 in.). Thin sections fill well because of the excellent fluidity of these cast irons, but they tend to chill and become extremely brittle white iron. Sections over 38 mm (1.5 in.) are prone to segregation and porosity, which reduce strength and corrosion resistance. Castings are generally designed with a patternmaker’s rule of 18 mm shrink to the meter (7=32 in. shrink to the foot), or 1.8%.

Foundry Practice for High-Alloy White Irons High-alloy white cast irons are an important group of materials whose production must be considered separately from that of ordinary types of cast irons. In these cast iron alloys, the alloy content is well above 4%, and consequently they cannot be produced by ladle additions to irons of otherwise standard compositions. They are usually produced in foundries that are specially equipped for the heat treatment and other thermal processing unique to the production of highly alloyed irons. These iron alloys are most often melted in electric

furnaces, specifically electric arc furnaces and induction furnaces, in which precise control of composition and temperature can be achieved. The high-alloy white irons consist of two major groups. The nickel-chromium white irons, commonly referred to as Ni-Hards, contain 3 to 7% Ni and approximately 1.5 to 11% Cr. The second group, the high-chromium white irons, contain 11 to 28% Cr and 0.5 to 3.5% Mo. Both groups are used for abrasionresistant applications.

Nickel-Chromium White Irons Melting Practice. A principal advantage of the nickel-chromium irons is that they can be melted in a cupola; however, better control of composition and temperature are achieved with electric furnace melting. Consequently, electric arc furnace melting and induction furnace melting are most common. The nickel-chromium irons are readily made in either acid-, neutral-, or basic-lined electric furnaces. They are normally dead melted, and there is usually no reason to use oxygen lancing except as a means of slightly reducing carbon content. Acid linings are generally more economical than basic linings. The type of lining affects silicon and chromium losses. Very little adjustment to slag composition is necessary in acid melting, which is used for the majority of current production. Normal charge materials are various kinds of steel scrap, foundry returns, or returns of similar alloy from service. For foundries that lack the facilities for rapid melt analysis, it may be necessary to select steel scrap rather carefully to control residual levels of alloying elements that affect austenite retention (e.g., manganese and copper). Chromium is obtained in the form of highcarbon ferrochrome and is generally added near the end of the heat to avoid excessive oxidation losses. Carbon is obtained from electrode graphite, petroleum coke, and other sources. Pig iron is also used to carburize the melt and can be an additional source of silicon. Silicon and manganese are added as ferroalloys. Molybdenum is usually added as ferromolybdenum; however, with arc furnaces, molybdenum oxide briquettes may be used. Sulfur is limited to 0.06%, and phosphorus is kept to 0.10%, as specified by ASTM A 532. High superheating temperatures have not been necessary when melting in induction furnaces, which have good stirring action, and a final temperature of 1480  C (2700  F) is usually adequate for thick-section castings. Final temperatures of up to 1565  C (2850  F) are commonly used in arc furnace practice to ensure homogeneity of the bath composition and to accelerate the solution of carbon, added after meltdown, and late alloy additions. No particular problems have been associated with high superheat temperatures, but it is more difficult to predict melting losses.

As in steel melting, deoxidation of the bath with aluminum was common but it is no longer practiced by most producers, with no adverse effects on the soundness or mechanical properties of the casting. A late addition of foundrygrade ferrosilicon is often made to improve toughness. Composition Control. The elements of importance in nickel-chromium irons include carbon, nickel, silicon, chromium, manganese, copper, and molybdenum. Carbon is varied according to the properties needed for the intended service. Carbon contents of 3.2 to 3.6% are prescribed when maximum abrasion resistance is desired. When impact loading is expected, carbon content should be held in the range of 2.7 to 3.2%. Nickel content increases with section size or cooling time of the casting to inhibit pearlitic transformation. For castings 38 to 50 mm (1.5 to 2 in.) thick, 3.4 to 4.2% Ni is sufficient to suppress pearlite formation upon mold cooling. Heavier sections may require nickel levels up to 5.5% to avoid the formation of pearlite. It is important to limit nickel content to the level needed for control of pearlite; excess nickel increases the amount of retained austenite and lowers hardness. Silicon is needed for two reasons. A minimum amount of silicon is necessary to improve the fluidity of the melt and to produce a fluid slag, but of equal importance is its effect on as-cast hardness. Increased levels of silicon, 1 to 1.5%, have been found to increase the amount of martensite and the resulting hardness. Late additions of ferrosilicon (0.2% as 75% Si-grade ferrosilicon) have been reported to increase toughness. Higher silicon contents can promote pearlite and may increase the nickel requirement. Chromium, primarily added to offset the graphitizing effects of nickel and silicon in types A, B, and C alloys, ranges from 1.4 to 3.5%. Chromium content must increase with increasing section size. In type D alloy, chromium levels range from 7 to 11% (typically 9%) for the purpose of producing eutectic carbides of the M7C3 chromium carbide type, which are harder and less deleterious to toughness. Manganese is typically held to a maximum of 0.8%, even though 1.3% is allowed according to ASTM A 532. While it provides increased hardenability to avoid pearlite formation, manganese is a more potent austenite stabilizer than nickel, and it promotes increased amounts of retained austenite and lower as-cast hardness. For this reason, higher manganese levels are undesirable. The level of manganese present should be a factor in determining the nickel content required to avoid pearlite in a given casting. Copper increases both hardenability and the retention of austenite, so it must be controlled for the same reason that manganese must be limited. Copper should be treated as a nickel substitute and, when properly included in the calculation of the amount of nickel required to

832 / Cast Irons inhibit pearlite, it reduces the nickel requirement. Molybdenum is a potent hardenability agent in nickel-chromium irons and is used in heavy-section castings to augment hardenability and inhibit pearlite. Pouring Practices. High pouring temperatures aggravate shrinkage under feeder heads and other hot spots and can lead to microshrinkage and coarse dendritic structures. Careful control of pouring temperature is particularly important in the consistent production of thick-section castings. Low pouring temperatures are necessary not only to avoid shrinkage defects but also to avoid problems such as metal penetration and burned-on sand. Low pouring temperatures are also effective in controlling dendrite size and the coarseness of the eutectic carbide structure. The eutectic temperature for the various nickel-chromium iron grades is approximately 1200  C (2190  F), and solidification generally begins at 1200 to 1280  C (2190 to 2340  F), depending on composition. When selecting a pouring temperature, common rules apply. Pouring temperatures are seldom lower than 100  C (180  F) above the liquids temperature. Higher pouring temperatures may be necessary when pouring smaller castings. Of course, casting configurations must be considered when selecting optimum pouring temperatures. Molds, Patterns, and Casting Design. Nickel-chromium irons may be sand cast or chill cast in permanent molds. The chill-cast micro-structures develop higher hardness, strength, and impact toughness than the sandcast structures because of the finer carbides. Whenever practical, the wearing faces of the casting should be cast against a chill to improve abrasion resistance and toughness. Sand molds may be made either of green sand or of dry sand, oil sand, or steel casting sand. Air-setting and thermal-setting resin sands are also commonly used. Molds must be rigid to minimize shrinkage defects. White irons are subject to hot tearing unless suitable precautions are taken. When castings are extracted from the mold, they are occasionally found to be cracked. Stresses large enough to cause fracture can arise when either the mold or the cores are excessively strong and there is inadequate mold or core breakdown to allow normal thermal contraction with cooling. Large cores, such as those used in pump volutes, and even small cores used to form bolt holes can cause cracking of this type. Shrinkage allowance is 13 to 21 mm/m (5=32 to ¼ in./ft); a commonly used figure is 16 mm/m (3=16 in./ft). The actual pattern allowance depends on the choice of alloy and the geometry of the casting. Patterns should employ generous radii and avoid sharp section changes to avoid initiating cracking upon solidification and subsequent cooling. Nickel-chromium irons exhibit relatively high liquid-to-solid shrinkage (5%), so large gates and risers are required for feeding. Special

consideration should be given to the design of end gates and the positioning of risers so that they can be readily removed. Risers cannot be removed by burning and are often notched or necked down to facilitate removal by impact. They are most easily removed after the castings have cooled to room temperature, that is, when martensite is present. Martensitic transformation begins at temperatures below 230  C (450  F). The shakeout practice is a critical step in the successful production of nickel-chromium iron castings. High residual stresses and cracking can result from extracting castings from a mold at too high a temperature. Cooling all the way to room temperature in the mold is desirable because, as stated previously, transformation to martensite occurs below 230  C (450  F) during the last stages of cooling. This precaution is mandatory in heavy-section castings; the volume increase that occurs with martensite formation results in the development of high transformation stresses in castings having steep temperature gradients due to rapid cooling. Thin-section castings and castings having simple shapes are less susceptible to stresses and are often removed from the mold once the castings have reached black heat. With these alloys, slow cooling favors higher as-cast hardness.

High-Chromium White Irons Melting Practice. The high level of carbon pickup with the high chromium contents in these alloys precludes the use of cupola melting. Consequently, electric arc and induction furnaces are normally used. The high-chromium irons are readily made in acid-, neutral-, or basic-lined electric furnaces. They are normally dead melted, and there is usually no reason to use oxygen lancing except as a means of slightly reducing carbon content. Despite rapid refractory wear due to chromium-bearing slag, acid linings are generally more economical than basic linings. Very little adjustment to slag composition is necessary in acid melting, which is used for the majority of current production. Viscous or nonfluid slags should be avoided. Normal charge materials are various kinds of steel scrap, foundry returns, or returns of similar alloy from service. For foundries that lack the facilities for rapid melt analysis, it may be necessary to select steel scrap rather carefully in order to control the total content of alloying elements that affect hardenability and austenite retention. Some furnace operators prefer to add a portion of high-carbon ferrochrome to the charge, but it is generally added near the end of the heat to avoid excessive oxidation losses. Carbon is obtained from electrode graphite, petroleum coke, and other sources. If pig iron is used to carburize the melt, it should have a low silicon content. The ideal silicon content is 0.6%. Less than 0.4% Si in the bath can cause difficulties with viscous slag, while higher silicon contents can promote pearlite. Selection of an incorrect

grade of ferrochromium with high silicon is a common cause for an excessively high silicon content. Manganese can range from 0.5 to 1.5%, according to ASTM A 532. Manganese on the high side of this range can erode acid furnace linings. Therefore, manganese content should be limited to 1% in meltdown; the remainder can be added to the ladle as crushed ferroalloy. Molybdenum is usually added as ferromolybdenum; however, with arc furnaces, molybdenum oxide briquettes may be used. Sulfur is limited to 0.06%, and phosphorus is kept to 0.10%, as specified by ASTM A 532. High superheating temperatures have not been necessary when melting in induction furnaces, which have good stirring action, and a final temperature of 1480  C (2700  F) is usually adequate for thick-section castings. Final temperatures of up to 1565  C (2850  F) are commonly used in arc furnace practice to ensure homogeneity of the bath composition and to accelerate the solution of carbon, added after meltdown, and late alloy additions. No particular problems have been associated with high superheat temperatures, but it is difficult to predict melting losses. As in steel melting, deoxidation of the bath with aluminum was common in the past, but this practice has been discontinued by most producers with no adverse effects on the soundness or mechanical properties of the casting. Titanium is sometimes added to limit dendrite size. There have been reports that bath additions of aluminum and titanium aggravate feeding problems. For heats in which hardenability is marginal, there is evidence that these elements tend to promote pearlite with mold cooling or in heat treated castings. Pouring Practice. High pouring temperatures aggravate shrinkage under feeder heads and other hot spots and can lead to microshrinkage and coarse dendritic structures. Careful control of pouring temperature is particularly important in the consistent production of thick-section castings. Low pouring temperatures are necessary to avoid shrinkage defects and prevent problems such as metal penetration and burned-on sand. Low pouring temperatures are also effective in controlling dendrite size and the coarseness of the eutectic carbide structure. The eutectic temperature for the various high-chromium iron grades ranges from 1230 to 1270  C (2245 to 2315  F), and solidification generally begins at temperatures up to 1350 C (2460  F), depending on composition. Pouring temperatures are seldom lower than 100  C (180  F) above the liquidus temperature. Castings thicker than 102 mm (4 in.) are generally poured between 1345 and 1400  C (2450 and 2550  F). Higher pouring temperatures may be necessary for smaller castings. Casting configurations also must be considered when selecting optimum pouring temperatures. In the ladle or during pouring, the metal appears cold and sluggish because an oxide film readily forms on the surface, but the metal is actually quite fluid and may be poured into

Cast Iron Foundry Practices / 833 intricate shapes. This rather viscous surface oxide is less liable to cause surface defects on castings that are poured quickly. Even when the metal is poured on the cold side, flasks have to be clamped, or weighted, to prevent the metal from running out at the parting line. Molds, Patterns, and Casting Design. Molds must be rigid to minimize shrinkage defects. Molds may be made of green sand, dry sand, oil sand, or steel casting sand. Airsetting and thermal-setting resin sands are also commonly used. Because white irons are subject to hot tearing, cores should break down readily. Unless suitable precautions are taken, the high-chromium irons are somewhat more prone to crack in the foundry than are white irons of lower alloy content. Castings are occasionally found to be cracked when extracted from the mold. Stresses large enough in magnitude to cause fracture can arise when either the mold or cores are excessively strong and there is inadequate mold or core breakdown to allow normal thermal contractions upon cooling. Large cores such as those used in pump volutes, and even small cores used to form bolt holes, can cause cracking of this type. Shrinkage allowance is 18 to 21 mm/m (7=32 to ¼ in./ft). The actual pattern allowance depends on the choice of subsequent heat treatment procedures, if any, and the geometry of the casting. Patterns should employ generous radii and avoid sharp section changes to avoid initiating cracking on solidification and subsequent cooling. High-chromium irons exhibit relatively high liquid-to-solid shrinkage, so large gates and risers are required for feeding. Special consideration should be given to the design of end gates and to the positioning of risers so that they can be readily removed. Risers are often notched or necked down to facilitate removal by impact. The shakeout practice is probably the most critical step in the successful production of high-chromium iron castings. A frequent cause of high residual stresses and of cracking is the common practice of extracting castings from the mold at too high a temperature. Cooling all the way to room temperature in the mold is desirable and can be a requirement to avoid cracking, especially if martensite forms during the last stages of cooling. This precaution is mandatory in heavy-section castings to be used in the as-cast condition where the desired moldcooled structure is a mixture of austenite and martensite. For irons that are heat treated, the desired mold-cooled structure is often pearlite. This softer structure facilitates removal of gates and risers, minimizes the transformational and thermal stresses that cause cracking, and shortens the response to heat treatment. Careful design of alloy composition ensures the development of a substantially pearlitic structure in the casting after mold cooling, yet provides enough hardenability to prevent pearlite formation

during subsequent heat treatment. Heavy-section castings made pearlitic by such an alloy content can often be removed from the mold once the castings have reached black heat. ACKNOWLEDGMENTS The information in this article is largely taken from:  S.F. Carter, Cupolas, Casting, Vol 15, ASM  







 

Handbook (formerly Metals Handbook, 9th ed.), ASM International, 1988, p 383–392 Casting, Section 23, Metals Handbook Desk Edition, American Society for Metals, 1985, p 23–1 to 23–52 D.B. Craig, M.J. Hornung, and T.K. McCluhan, Gray Iron, Casting, Vol 15, ASM Handbook (formerly Metals Handbook, 9th ed.), ASM International, 1988, p 629–646 R.B. Gundlach, High-Alloy Graphitic Irons, Casting, Vol 15, ASM Handbook (formerly Metals Handbook, 9th ed.), ASM International, 1988, p 698–701 R.B. Gundlach, High-Alloy White Irons, Casting, Vol 15, ASM Handbook (formerly Metals Handbook, 9th ed.), ASM International, 1988, p 678–685 I.C.H. Hughes, Ductile Iron, Casting, Vol 15, ASM Handbook (formerly Metals Handbook, 9th ed.), ASM International, 1988, p 647–666 Melting of Gray Iron, Forging and Casting, Vol 5, Metals Handbook, 8th ed., American Society for Metals, 1970, p 335–360 D.M. Stefanescu, R. Hummer, and E. Nechtelberger, Compacted Graphite Irons, Casting, Vol 15, ASM Handbook (formerly Metals Handbook, 9th ed.), ASM International, 1988, p 667–677

REFERENCES 1. Background Information for Proposed Standards, EPA-453/R-2-013, National Emissions Standards for Hazardous Air Pollutants (NESHAP) for Iron and Steel Foundaries, prepared by RTI International for U.S. Environmental Protection Agency, Dec 2002 2. The Foundry Industry: Review of Process Energy Use, Markets, and Information Resources, Energy Center of Wisconsin, May 1997 3. Seminar on Energy Conservation in Iron Casting Industry, United Nations Industrial Development Organization (UNIDO), 1998 4. Energy and Environmental Profile of the U.S. Metalcasting Industry, Energeties, Inc., U.S. Department of Energy, Office of Industrial Technologies, Sept 1999 5. J.F. Schifo and J.T. Radia, Theoretical/Best Practice Energy Use in Metalcasting Operations, Department of Energy, May 2004 6. N.C. McClure et al., Inoculation of Gray Cast Iron, Trans. AFS, Vol 65, 1957, p 340–351

7. H.W. Lownie, Theories of Gray Cast Iron Inoculation, Trans. AFS, Vol 54, 1946, p 837–844 8. B. Lux, Nucleation of Eutectic Graphite in Inoculated Gray Iron by Saltlike Carbides, Mod. Cast., May 1964, p 222–232 9. C.F. Walton, The Gray Iron Castings Handbook, Gray Iron Founder’s Society, 1958, p 119 10. P.J. Bilek, J.M. Dong, and T.K. McCluhan, The Roles of Calcium and Aluminum in the Inoculation of Gray Iron, Trans. AFS, Vol 80, 1972, p 183–188 11. J. Briggs, R.W. Newman, and M.D. Bryant, Control of Inoculation—How Much? What Size? When?, Proc. AFS-CMI Conf., American Foundrymen’s Society, Feb 1979 12. “Counting Eutectic Cells in Flake Graphite Iron Castings,” Broadsheet 94-1, BCIRA, 1974 13. A.G. Fuller, Fading of Inoculants, Proc. AFS-CMI Conf., American Foundrymen’s Society, Feb 1979 14. G.F. Sergeant, Late Metal Stream Inoculation BCIRA Developments, Proc. AFS-CMI Conf., American Foundrymen’s Society, Feb 1979 15. G.-Fr. Hillner and K.H. Kleeman, Mold Inoculation of Gray and Ductile Cast Iron—New Solutions to Old Problems, Trans. AFS, Vol 83, 1975, p 167 16. R.E. Eppich, Solid Inserts, Proc. AFS-CMI Conf., American Foundrymen’s Society, Feb 1979 17. H.E. Henderson, Compliance with Specifications for Ductile Iron Castings Assures Quality, Met. Prog., Vol 99, May 1966, p 82–86; Ductile Iron—Our Most Versatile Ferrous Material, Gray, Ductile and Malleable Iron Castings Capabilities, STP 455, American Society for Testing and Materials, 1969, p 29–53 18. H.J. Heine, “An Overview of Magnesium Treatment Processes Which Have Stood the Test of Time in America,” Paper 9, BCIRA Conf., April 1987 19. Ductile Iron Molten Metal Processing, 2nd ed., American Foundrymen’s Society, 1986 20. W.A. Henning, Magnesium Treatment Methods, Ductile Iron Handbook, American Foundrymen’s Society, 1992, p 155–166 21. Modern Inoculating Practices for Gray and Ductile Iron, Conf. Proc., American Foundrymen’s Society/Cast Metals Institute, 1979 22. J.V. Anderson and S.I. Karsay, Pouring Rate, Pouring Time and Choke Design for SG Iron Castings, Br. Foundryman, Vol 78 (No. 10), Dec 1995, p 492–498 23. P.J. Rickards, Factors Affecting the Soundness and Dimensions of Iron Castings Made in Cold-Curing Chemically Bonded Sand Moulds, Br. Foundryman, Vol 75 (No. 11), Nov 1982, p 213–223 24. H. Roedter, An Alternative Method of Pressure-Control Feeding for Ductile Iron Castings, Foundry Trade J. Int., Vol 9 (No. 31), Sept 1986, p 174–179

834 / Cast Irons 25. R. Hummer, Feeding Requirements and Dilation during Solidification of Spheroidal Graphite Cast Iron—Conclusions for Feeder Dimensioning, Giesserei-Prax., No. 17/18, Sept 16, 1985, p 241–254 26. H.H. Cornell and C.R. Loper, Jr., Variables Involved in the Production of Compacted Graphite Cast Iron Using Rare Earth-Containing Alloys, Trans. AFS, Vol 93, 1985, p 435 27. J. Fowler, D.M. Stefanescu, and T. Prucha, Production of Ferritic and Pearlitic Grades of Compacted Graphite Cast Iron by the In-Mold Process, Trans. AFS, Vol 92, 1984, p 361 28. W. Du¨nki, Gusseisen mit Vermiculargraphit-Herstellung und Eigenschaften, Escher Wyss Mittellungen, Vol 53 (No. 1/2), 1980, p 215 29. E.R. Evans, J.V. Dawson, and M.J. Lalich, Compacted Graphite Cast Irons and Their Production by a Single Alloy Addition, Trans. AFS, Vol 84, 1976, p 215 30. M.J. Lalich, Compacted Graphite Cast Iron—Its Properties and Production with a New Alloy, Mod. Cast, July 1976, p 50 31. T. Kimura, C.R. Loper, Jr., and H. Cornell, Rare Earth Silicide Additions to Cast Iron, Trans. AFS, Vol 88, 1980, p 67 32. W. Thury, R. Hummer, and E. Nechtelberger, Austrian Patent 290, 592, 1968 33. J. Sissener, W. Thury, R. Hummer, and E. Nechtelberger, Cast Iron with Vermicular Graphite, AFS Cast Met. Res. J., 1972, p 178 34. D.M. Stefanescu, R.C. Voigt, and C.R. Loper, Jr., The Importance of the Lanthanum/Rare Earth Ratio in the Production of Compacted

35.

36.

37.

38.

39.

40.

41.

Graphite Cast Irons, Trans. AFS, Vol 89, 1981, p 119 C.G. Chao, W.H. Lu, J.L. Mercer, and J.F. Wallace, Effect of Treatment Alloys and Section Size on the Compacted Graphite Structures Produced by the In-the-Mold Process, Trans. AFS, Vol 93, 1985, p 651 E. Nechtelberger, H. Puhr, J.B. von Nesselrode, and A. Nakayasu, “Cast Iron with Vermicular/Compacted Graphite—State of the Art Development, Production, Properties Applications,” paper presented at the International Foundry Congress (Chicago), April 1982 G.F. Sergeant and E.R. Evans, The Production and Properties of Compacted Graphite Irons, Br. Foundryman, Vol 71, 1978, p 115 K.P. Cooper and C.R. Loper, Jr., A Critical Evaluation of the Production of Compacted Graphite Cast Iron, Trans. AFS, Vol 86, 1978, p 267 W. Simmons and J. Briggs, Compacted Graphite Irons Produced with a CeriumCalcium Treatment, Trans. AFS, Vol 90, 1982, p 367 F. Martinez and D.M. Stefanescu, Properties of Compacted/Vermicular Graphite Cast Irons in the Fe-C-Al System Produced by Ladle and In-Mold Treatment, Trans. AFS, Vol 91, 1983, p 593 D.M. Stefanescu, F. Martinez, and I.G. Chen, Solidification Behavior of Hypoeutectic and Eutectic Compacted Graphite Cast Irons—Chilling Tendency and Eutectic Cells, Trans. AFS, Vol 91, 1983, p 205

42. D.M. Stefanescu and C.R. Loper, Recent Progress in the Compacted/Vermicular Graphite Cast Iron Field, Giesserei-Prax., No. 5, 1981, p 73 43. D.M. Stefanescu, I. Dinescu, S. Craciun, and M. Popescu, “Production of Vermicular Graphite Cast Irons by Operative Control and Correction of Graphite Shape,” Paper 37, 46th International Foundry Congress (Madrid), 1979 44. D.M. Stefanescu, Solidification of Flake, Compacted/Vermicular and Spheriodal Graphite Cast Irons as Revealed by Thermal Analysis and Directional Solidification Experiments, The Physical Metallurgy of Cast Iron, H. Fredriksson and M. Hillert, Ed., Elsevier, 1985, p 151 45. D.M. Stefanescu, C.R. Loper, Jr., R.C. Voigt, and I.G. Chen, Cooling Curve Structure Analysis of Compacted Vermicular Graphite Cast Irons Produced by Different Melt Treatments, Trans. AFS, Vol 90, 1982, p 333 46. C.R. Loper, M.J. Lalich, H.K. Park, and A.M. Gyarmaty, “Microstructure-Mechanical Property Relationship in Compacted/ Vermicular Graphite Cast Iron,” Paper 35, 46th International Foundry Congress (Madrid), 1979 47. J.B. von Nesselrode, Gusseisen mit Vermicular-graphit, ein Werkstoff fur Zylinderko¨pfe, Giesserei-Prax, No. 23/24, 1979, p 445 48. W.M. Spear, Production of Ductile NiResist, Ductile Iron Handbook, American Foundrymen’s Society, 1992, p 52–65

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 835-855 DOI: 10.1361/asmhba0005323

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Gray Iron Castings CAST IRONS are alloys of iron, carbon, and silicon in which more carbon is present than can be retained in solid solution in austenite at the eutectic temperature. In gray cast iron, the carbon that exceeds the solubility in austenite precipitates as flake graphite. Gray irons usually contain 2.5 to 4% C, 1 to 3% Si, and additions of manganese, depending on the desired microstructure (as low as 0.1% Mn in ferritic gray irons and as high as 1.2% in pearlitics). Sulfur and phosphorus are also present in small amounts as residual impurities. Certain important but low-tonnage specialty items in this family of cast metals (notably the austenitic and other highly alloyed gray irons) are not dealt with here; instead the emphasis is on the properties of gray irons used most often and in the largest tonnages. The basic metallurgy of gray cast irons is discussed in the article “Introduction to Cast Irons” in this Volume.

Classes of Gray Iron A simple and convenient classification of the gray irons is found in ASTM specification A 48/A 48M (Ref 1), which designates the castings in terms of the minimum tensile strength of a separately cast test bar, expressed in ksi or separately as MPa. The designation also indicates the size of the test bar (A, B, C, or S). In customary units, the designations range from class 20 (minimum tensile strength 20 ksi) to class 60 (minimum tensile strength 60 ksi). The metric classification is from class 150 (minimum tensile strength 150 MPa) to class 400 (minimum tensile strength 400 MPa). The international classification standard ISO 185 (Ref 2) similarly designates gray iron from 100 to 350 by minimum tensile strength (MPa) of separately cast samples. The standard also allows manufacturer and purchaser to agree to use cast-on samples and hardness values as criteria. These cast-on mandatory values are dependent on the relevant wall thickness. The strength scale by no means connotes a scale of ascending superiority. In many applications, strength is not the major criterion for the choice of grade. For example, for parts such as clutch plates and brake drums, where resistance to heat checking is important, low-strength grades of iron are the superior performers.

Similarly, in heat shock applications such as ingot or pig molds, a class 60 iron would fail quickly, whereas good performance is shown by class 25 iron. In machine tools and other parts subject to vibration, the better damping capacity of low-strength irons is often advantageous. Generally, it can be assumed that the following properties of gray cast irons increase with increasing tensile strength from class 20 to class 60 and from 150 to class 400:
All strengths, including strength at elevated

temperature


Ability to be machined to a fine finish
Modulus of elasticity
Wear resistance

On the other hand, the following properties decrease with increasing tensile strength, so that low-strength irons often perform better than high-strength irons when these properties are important:





Machinability Resistance to thermal shock Damping capacity Ability to be cast in thin sections

The SAE International standard J431 (Ref 3) classifies sand-molded gray cast irons used by the automotive and allied industries. It defines iron grades by a test bar tensile strength/Brinell hardness ratio, hardness grades by hardness alone, and casting grades by a combination of the iron grade, hardness grade, and special requirements.

Applications Gray iron is used for many different types of parts in a very wide variety of machines and structures. Automotive applications include pistons, cylinders, cylinder blocks and heads for gasoline and diesel engines, brake drums and discs, and hardened iron for camshafts. The design of parts intended to be produced as gray iron castings must be evaluated for the specific service conditions before being approved for production. Often, a stress analysis of prototype castings helps establish the appropriate class of gray iron, mold design, as well as any proof test requirements or other acceptance criteria for production parts.

Castability Successful production of a gray iron casting depends on the fluidity of the molten metal and the cooling rate (which is influenced by the minimum section thickness and section thickness variations). Casting design is often described in terms of section sensitivity. This is an attempt to correlate properties in critical sections of the casting with the combined effects of composition and cooling rate. All these factors are interrelated and may be condensed into a single term, castability, which for gray iron may be defined as the minimum section thickness that can be produced in a mold cavity with given volume/area ratio and mechanical properties consistent with the type of iron being poured. Fluidity. Scrap losses resulting from misruns, cold shuts, and round corners are often attributed to the lack of fluidity of the metal being poured. Mold conditions, pouring rate, and other process variables being equal, the fluidity of commercial gray irons depends primarily on the amount of superheat above the freezing temperature (liquidus). As the total carbon content decreases, the liquidus temperature increases, and the fluidity at a given pouring temperature therefore decreases. Fluidity is commonly measured as the length of flow into a spiral-type fluidity test mold. The relation between fluidity and superheat is shown in Fig. 1 for four unalloyed gray irons of different carbon contents. The significance of the relationships between fluidity, carbon content, and pouring temperature becomes apparent when it is realized that the gradation in strength in the ASTM classification of gray iron is due in large part to differences in carbon content (3.60 to 3.80% for class 20; 2.70 to 2.95% for class 60). The fluidity of these irons thus resolves into a measure of the practical limits of maximum pouring temperature as opposed to the liquidus of the iron being poured. These practical limits of maximum pouring temperature are largely determined by three factors:
The ability of both mold and cores to with-

stand the impact of molten iron, an ability that decreases as the pouring temperature increases, thereby favoring low pouring temperatures

836 / Cast Irons Superheat above liquidus, ⬚F 100

200

300

400

500

600

1.25

50

1.0

40

30

0.75 %C 2.13 2.52 3.04 3.60

0.5

%Si 2.07 2.00 2.10 2.08

20

10

0.25

0

Length of spiral, in.

Length of spiral, m

0 1.5

0

50

100

150

200

250

300

0 350

Superheat above liquidus, ⬚C

Fig. 1

Fluidity versus degree of superheat for four gray irons with varying carbon contents

Table 1 Superheat above liquidus for 2% Si irons of various carbon contents poured at 1455  C (2650  F) Liquidus temperature Carbon, %

2.52 3.04 3.60



C

1295 1245 1175



F

2360 2270 2150

Superheat above liquidus 

C

160 210 280



F

290 380 500


The fact that metal tap temperatures seldom

exceed 1550  C (2825  F). Because ladling and reladling to the point of pouring generally accounts for temperature losses of 55 to 85  C (100 to 150  F), the final pouring temperatures seldom exceed 1450 to 1495  C (2640 to 2720  F), and in most instances, maximum pouring temperatures in the range of 1410 to 1450  C (2570 to 2640  F) are considered more realistic.
The necessity to control the overall thermal input to the mold in order to control the final desired microstructure It can be seen from Table 1 that because of differences in liquidus temperature, the amount of superheat (and therefore fluidity) varies with carbon content when various compositions are cast from the same pouring temperature.

Microstructure The usual microstructure of gray iron is a matrix of pearlite with graphite flakes dispersed throughout. Foundry practice can be varied so that nucleation and growth of graphite flakes occur in a pattern that enhances the desired properties. The amount, size, and distribution of graphite are important. Cooling that is too

rapid may produce so-called chilled iron, in which the excess carbon is found in the form of massive carbides. Cooling at intermediate rates can produce mottled iron, in which carbon is present in the form of both primary cementite (iron carbide) and graphite. Very slow cooling of irons that contain large percentages of silicon and carbon is likely to produce considerable ferrite and pearlite throughout the matrix, together with coarse graphite flakes. See the article “Metallography and Microstructures of Cast Iron” in Ref 4 for details and micrographs of these microstructures. Flake graphite is one of seven types (shapes or forms) of graphite established in ASTM A 247 (Ref 5). Flake (lamellar) graphite, which is a characteristic of gray iron, is subdivided into five types (patterns) that are designated by the letters A through E (see the article “Introduction to Cast Irons” in this Volume and also Ref 5). Graphite size is established by comparison with an ASTM International size chart, which shows the typical appearances of flakes of eight different sizes at 100 magnification. Type A flake graphite (random orientation) is preferred for most applications. In the intermediate flake sizes, type A flake graphite is superior to other types in certain wear applications, such as the cylinders of internal combustion engines. Type B flake graphite (rosette pattern) is typical of fairly rapid cooling, such as is common with moderately thin sections (approximately 10 mm, or 3/8 in.) and along the surfaces of thicker sections, and sometimes results from poor inoculation. The large flakes of type C flake graphite are typical of kish graphite that is formed in hypereutectic irons. These large flakes enhance resistance to thermal shock by increasing thermal conductivity and decreasing elastic modulus. On the other hand, large flakes are not conducive to good surface finishes on

machined parts or to high strength or good impact resistance. The small, randomly oriented interdendritic flakes in type D flake graphite promote a fine machined finish by minimizing surface pitting, but it is difficult to obtain a pearlitic matrix with this type of graphite. Type D flake graphite may be formed near rapidly cooled surfaces or in thin sections. Frequently, such graphite is surrounded by a ferrite matrix, resulting in soft spots in the casting. Type E flake graphite is an interdendritic form, which has a preferred rather than a random orientation. Unlike type D graphite, type E graphite can be associated with a pearlitic matrix and thus can produce a casting whose wear properties are as good as those of a casting containing only type A graphite in a pearlitic matrix. There are, of course, many applications in which flake type has no significance as long as the mechanical property requirements are met. Solidification of Gray Iron. In a hypereutectic gray iron, solidification begins with the precipitation of kish graphite in the melt. Kish grows as large, straight, undistorted flakes or as very thick, lumpy flakes that tend to rise to the surface of the melt because of their low density. When the temperature has been lowered sufficiently, the remaining liquid solidifies as a eutectic structure of austenite and graphite. Generally, eutectic graphite is finer than kish graphite. In hypoeutectic iron, solidification begins with the formation of proeutectic austenite dendrites. As the temperature falls, the dendrites grow, and the carbon content of the remaining liquid increases. When the increasing carbon content and decreasing temperature reach eutectic values, eutectic solidification begins. Eutectic growth from many different nuclei proceeds along crystallization fronts that are approximately spherical. Ultimately, the eutectic cells meet and consume the liquid remaining in the spaces between them. During eutectic solidification, the austenite in the eutectic becomes continuous with the dendritic proeutectic austenite, and the structure can be described as a dispersion of graphite flakes in austenite. After solidification, the eutectic cell structure and the proeutectic austenite dendrites cannot be distinguished metallographically except by special etching or in strongly hypoeutectic iron. With eutectic compositions, obviously, solidification takes place as the molten alloy is cooled through the normal eutectic temperature range, but without the prior formation of a proeutectic constituent. During the solidification process, the controlling factor remains the rate at which the solidification is proceeding. The rapid solidification favored by thin section sizes or highly conductive molding media can result in undercooling. Undercooling can cause the solidification to start at a temperature lower than the expected eutectic temperature for a given composition (Fig. 2). This can result in a modification of the carbon form from A to E type or can completely suppress its formation and form primary carbides instead.

Gray Iron Castings / 837

Fig. 2

Undercooling from rapid cooling of a eutectic composition

Room-Temperature Structure. Upon cooling from the eutectic temperature, the austenite will decompose, first by precipitating some of the dissolved carbon and then, at the eutectoid temperature, by undergoing complete transformation. The actual products of the eutectoid transformation depend on rate of cooling as well as on composition of the austenite, but under normal conditions the austenite will transform either to pearlite or to ferrite plus graphite. Transformation to ferrite plus graphite is most likely to occur with slow cooling rates, which allow more time for carbon migration within the austenite; high silicon contents, which favor the formation of graphite rather than cementite; high values of carbon equivalent; and the presence of fine undercooled (type D) flake graphite. Graphite formed during decomposition is deposited on the existing graphite flakes. When carbon equivalent values are relatively low or when cooling rates are relatively fast, the transformation to pearlite is favored. In some instances, the microstructure will contain all three constituents: ferrite, pearlite, and graphite. With certain compositions, especially alloy gray irons, it is possible to produce a martensitic matrix by oil quenching through the eutectoid transformation range or an austempered matrix by appropriate isothermal treatment (Ref 6). These treatments are often done deliberately in a secondary heat treatment where high strength or hardness is especially desired, such as in certain wear applications. The secondary heat treatment of gray iron castings is of great value in producing components that must be hard when machining requirements prohibit the use of components that are cast to final shape in white iron.

Section Sensitivity In practice, the minimum thickness of section in which any given class of gray iron may be poured is more likely to depend on the cooling rate of the section than on the fluidity of the metal. For example, although a plate 300 mm (12 in.) square by 6 mm (0.24 in.) thick can be poured in class 50 as well as in class 25 iron, the former casting would not be gray iron because the cooling rate

Fig. 3

Effect of section thickness on hardness and structure. Hardness readings were taken at increasing distance from the tip of a cast wedge section. Composition of iron: 3.52% C, 2.55% Si, 1.01% Mn, 0.215% P, and 0.086% S. Source: Ref 7

would be so rapid that massive carbides would be formed. Yet it is entirely feasible to use class 50 iron for a diesel engine cylinder head that has predominantly 6 mm (0.24 in.) wall sections in the water jackets above the firing deck. This is simply because the cooling rate of the cylinder head is reduced by the “mass effect” resulting from enclosed cores and the proximity (often less than 12 mm, or 0.47 in.) of one 6 mm (0.24 in.) wall to the other. Thus, the shape of the casting has an important bearing on the choice of metal specification. It should be recognized that the smallest section that can be cast gray, without massive carbides, depends not only on metal composition but also on foundry practices. For example, by adjusting silicon content or by using graphitizing additions called inoculants in the ladle, the founder can decrease the minimum section size for freedom from carbides for a given basic composition of gray iron. The mass effect associated with increasing section thickness or decreasing cooling rate is much more pronounced in gray iron than in cast steel. The mass effect in cast steel results in increased

grain size in heavy sections. This also applies to gray iron, but the most important effects are on graphite size and distribution, and on amount of combined carbon. For any given gray iron composition, the rate of cooling from the freezing temperature to below approximately 650  C (1200  F) determines the ratio of combined to graphitic carbon, which controls the hardness and strength of the iron. For this reason the effect of section size in gray iron is considerably greater than in the more homogeneous ferrous metals in which cooling rate does not affect the form and distribution of carbon throughout the metal structure. Typical Effects of Section Size. When a wedge-shaped bar with approximately a 10 taper is cast in a sand mold and sectioned near the center of the length, and Rockwell hardness determinations are made on the cut surface from the point of the wedge progressively into the thicker sections, the curves so determined show to what extent continually increasing section size affects hardness (Fig. 3). Progressing along the curve from the left in Fig. 3, the following metallographic

838 / Cast Irons for a part, corrections are made to the molten metal prior to pouring.

Prevailing Sections

Fig. 4

Effect of section diameter on tensile strength at center of cast specimen for five classes of gray iron

Table 2 Minimum prevailing casting sections recommended for gray irons Minimum thickness

V/A ratio(a)

ASTM A48 class

in.

mm

in.

20 25 30 35 40 50 60

1/8 1/4 3/8 3/8 5/8 3/4

3.2 6.4 9.5 9.5 15.9 19.0 25.4

0.06 0.12 0.17 0.17 0.28 0.33 0.42

1

mm

1.5 3.0 4.3 4.3 7.1 8.4 10.7

(a) Volume/area (V/A) ratios are for square plates.

Fig. 5

Standard W2 wedge block used for measuring depth of chill (ASTM A 367). Dimensions in

inches

constituents occur. The tip of the wedge is white iron (a mixture of carbide and pearlite) with a hardness greater than 50 HRC. As the iron becomes mottled (a mixture of white iron and gray iron), the hardness decreases sharply. A minimum is reached because of the occurrence of fine type D flake graphite, which usually has associated ferrite in large amounts. With a slightly lower cooling rate, the structure becomes fine type A flake graphite in a pearlite matrix, with the hardness rising to another maximum on the curve. This structure is usually the most desirable for wear resistance and strength. With increasing section thickness beyond this point, the graphite flakes become coarser, and the pearlite lamellae become more widely spaced, resulting in slightly lower hardness. With further increase in wedge thickness and decrease in cooling rate, pearlite decomposes progressively to a mixture of ferrite and graphite, resulting in softer and weaker iron. The structures of most commercial gray iron castings are represented by the right-hand downward-sloping portion of the curve in Fig. 3, beyond 5 mm (0.2 in.) wedge thickness, and

increasing section size is normally reflected by the gradual lowering of hardness and strength. However, thin sections may be represented by the left-hand downward-sloping portion. Figure 4 shows the effect of section size on the average tensile strength of five classes of irons. Section sensitivity effects are used in the form of a wedge test in production control. It should be recognized that the smallest section that can be cast gray, without massive carbides, depends not only on metal composition but also on foundry practices. For example, by adjusting silicon content or by using graphitizing additions called inoculants in the ladle, the founder can decrease the minimum section size for freedom from carbides for a given basic composition of gray iron to judge the suitability of an iron for pouring a particular casting. In this test, a wedge-shaped casting is poured and then evaluated upon solidification. The standard W2 wedge block specified in ASTM A 367 (Ref 8) is shown in Fig. 5 (five test wedge sizes are available). The evaluation consists of measuring the length of the chilled zone. The measurement, usually made in 0.8 mm (1/32 in.) increments, is related to empirically determined data obtained from a “good” casting. If the evaluation indicates an excessive sensitivity

Although the ASTM size B test bar (30.5 mm, or 1.20 in., nominal diameter at midlength) is the bar most commonly used for all gray irons from class 20 to class 60, ASTM specification A 48 (Ref 1) provides a series of bar sizes from which one that approximates the cooling rate in the critical section of the casting can be selected. In practice, it is customary to be somewhat more definite regarding the prevailing values of minimum casting section considered feasible for the various ASTM classes of cast iron. As summarized in Table 2, these minimum prevailing sections include the requirement of freedom from carbidic areas. In a platelike section, occasional thinner walls (such as ribs) are of no importance unless they are very thin or are appended to the outer edges of the casting. The relationship between the controlling section in the casting and the appropriate test bar diameter is established on the basis of the volume-to-surface-area ratio of the relative sections, because this is the primary factor in determining the cooling rate of a casting. Unfortunately, all surfaces of a typical casting are not equally efficient in dissipating heat, so the selection of a suitable test bar size often requires knowledge of heat transfer in the mold when the casting is made. Foundries can provide this information, but, in some complex cases, the appropriate test bar size must be established by a trial in which the properties of the iron in the casting are determined and compared with those of the test bar. This may include cutting tensile tests or even fatigue specimens from actual castings, but Brinell hardness testing usually provides a satisfactory and nondestructive correlation (Ref 9). Mechanical properties of class 30 and class 50 gray irons in various sections are shown in Fig. 6. For class 30 iron, the combined carbon content and hardness are still at a safe level in sections equivalent to a 10 mm (0.4 in.) plate, which has a volume/area (V/A) ratio of approximately 5 mm (0.20 in.). For class 50 iron, however, both combined carbon and Brinell hardness show marked increases when the thickness of the equivalent plate section is decreased to approximately 15 mm (0.6 in.), with a V/A ratio of approximately 7 mm (0.27 in.). These results are consistent with the minimum prevailing casting sections recommended in Table 2. The hazards involved in pouring a given class of gray iron in a plate section thinner than recommended are discovered when the casting is machined. Typical losses as a result of specifying too high a strength for a prevailing section of 9.5 mm (3/8 in.) are given as follows (rejections were for “hard spots” that made it impossible to machine the castings by normal methods):

Gray Iron Castings / 839 production of a 25 mm (1 in.) diameter singlethrow crankshaft for an air compressor. This shaft was hard at the extreme ends when poured in class 50 iron. The difficulty was corrected by flowing metal through each end into flow-off risers that adequately balanced the cooling rate at the ends with the cooling rate at the center. In sum, the selection of a suitable grade of gray iron for a specific casting necessarily requires an evaluation of the size and shape of the casting as related to its cooling rate, or V/A ratio. For a majority of parts, this evaluation need be no more than a determination of whether or not the V/A ratio of the casting exceeds the minimum V/A ratio indicated for the grade considered.

Test Bar Properties

Fig. 6

Mechanical properties of class 30 and class 50 gray iron as a function of section size. Composition of the class 30 iron: 3.40% C, 2.38% Si, 0.71% Mn, 0.423% P, and 0.152% S; for the class 50 iron: 2.96% C, 1.63% Si, 1.05% Mn, 0.67% Mo, 0.114% P, and 0.072% S. Source: Ref 10

Mechanical property values obtained from test bars are sometimes the only available guides to the mechanical properties of the metal in production castings. When test bars and castings are poured from metal of the same chemical history, correlations can be drawn between the thermal history of the casting and that of the test bar. The strength of the test bar gives a relative strength of the casting, corrected for the cooling rate of the various section thicknesses. Through careful analysis of the critical sections of a casting, accurate predictions of mechanical behavior can be achieved. Test bars that are attached as an appendage to a casting may be unsatisfactory unless their size and location in the mold are carefully selected. The cooling rate of an attached test bar and the casting where it is attached are interdependent, and this affects the attainment of both a sound casting and a sound test bar. A more dependable test is obtained by pouring a separate test bar of appropriate size from the same metal, just before or after pouring the casting. Because of the small amount of solidification shrinkage in gray iron, separately cast test bars are sound without having large feed heads (risers) and provide similar conditions to the casting. Accurate results in the tensile testing of gray iron requires that the specimen be accurately machined and tested in equipment that provides precise axial loading. If any bending is present in the tensile test load application, the results will be below the correct strength of the iron. An industry-wide investigation has also proven that the factors involved in achieving full potential tensile strength in gray iron tensile testing include test specimen machining. The six most important factors are (Ref 11):
Use a tapered shoulder whenever possible,

Class

Rejections, %

35 45 55

Negligible 25 80–100

In marginal applications, a higher class of iron may sometimes be used if the casting is cooled slowly (in effect, increasing the section thickness) by judicious placement of flow-offs and risers. An example is the successful

not threaded. A 30 taper is preferred. This may require machining special grips for the tensile testing machine.
Cut bars from the bottom of the test casting.
Make sure the gage length is parallel, not tapered.

840 / Cast Irons
There should be no tool marks or undercuts

on the gage length or shoulder.
The finish of the test bar should be smooth.
If the test bars require aging, age them uniformily. Usual Tests. Tensile tests on bars that are cast specifically for such tests are the most common methods used for evaluating the strength of gray iron. Yield strength, elongation, and reduction of area are seldom determined for gray iron in standard tension tests. Hardness tests, on either test bars or castings, are used as an approximate measure of strength and sometimes as an indication of relative machinability. Unfortunately, the general relationship between hardness and tensile strength

for all gray irons is a very broad one. This is because variations in the amount and shape of the flake graphite in gray iron influence the strength more than the hardness. The general relationship between gray iron strength and hardness (Ref 12) is shown in Fig. 7, along with those for malleable and ductile irons and for steel. Whereas steels have a fixed tensilestrength-to-Brinell-hardness ratio of approximately 500 to 1, and malleable and ductile irons have a ratio of approximately 400 to 1, gray irons exhibit considerable variation and have wide limits. Examples of such deviations, produced largely by changing foundry variables, are shown in Fig. 8. This figure (Ref 14) shows that permanent mold gray irons with a very fine graphite structure exhibit higher tensile-

strength-to-Brinell-hardness ratios than do gray irons of similar composition cast in sand and having a coarser graphite. From the standpoint of machinability, a low hardness for a given tensile strength is desirable (Ref 9). When the processing variables are controlled, the relationships between Brinell hardness and tensile strength generally follow the pattern reproduced in Fig. 9, which shows the variation of tensile strength with Brinell hardness for a series of gray irons produced by a single foundry. The data in Fig. 9 are from ASTM size A and B test bars poured in a series of inoculated gray irons. The successful use of Brinell hardness as a measure of strength depends on whether it can be proved suitable for the application, which may involve service tests or mechanical property tests on specimens cut from production castings. Testing Precautions. In the assessment of mechanical properties for a series of heats, precautions should be taken to ensure minimum statistical variation between bars. By its nature, gray iron behaves as a brittle material in tension, with no measurable elongation after fracture. This characteristic can be exaggerated by imposing a nonaxial load during tensile testing, resulting in statistical variations, which may not be a true measure of the quality of the iron. To overcome this tendency, many shops use selfaligning nonthreaded grips in the performance of tensile tests on gray iron test bars.

Specifications

Fig. 7

Tensile strength relationship to hardness for steel, malleable and ductile irons, and gray irons. Effect of carbon equivalent for gray iron is seen. The MacKenzie relationship is shown. Source: Ref 12, 13

ASTM A 48/A 48M (Ref 1) is typical of specifications based on tensile properties determined by separately cast test bars. In practice, one of three different standard sizes of separately cast test bars is used to evaluate the properties in the controlling section of the castings. After manufacturer and purchaser agree on a

Fig. 8

Tensile strength relationship to Brinell hardness for annealed gray iron is influenced by the shape of graphite. Iron cast in sand has mainly type A graphite, while permanent mold castings have type D graphite. The SAE minimum is shown for comparison. Source: Ref 14

Fig. 9 Ref 14

Relationship between tensile strength and Brinell hardness for a series of inoculated gray irons from a single foundry. Open circles represent unalloyed gray iron, and closed circles represent alloyed gray iron. Source:

Gray Iron Castings / 841 controlling section of the casting, the size of test bar that corresponds, approximately, to the cooling rate expected in that section is designated by a letter (Table 3). Most gray iron castings for general engineering use are specified as either class 25, 30, or 35 (in customary units). The customary units and SI units are regarded as separate classes, and combining values from the two unit systems may result in nonconformity. Classes 20 to 60 are the customary unit classes, and classes 150 to 400 are the metric classes. Specification A 48/A 48M is based entirely on mechanical properties, and the composition that provides the required properties can be selected by the individual producer. A manufacturer whose major production is medium-section castings of class 35 iron will Table 3 Test bars designed to match controlling sections of castings (ASTM A 48/A 48M) Wall thickness of controlling section in.

<0.25 0.25–0.50 0.51–1.00 1.01–2 >2

Nominal diameter of test bar at midlength(a)

mm

Test bar

in.

mm

<5 5–14 15–25 26–50 >50

S A B C S

(b) 0.88 1.20 2.00 (b)

(b) 22.4 30.5 50.8 (b)

(a) See Ref 1 for range of acceptable diameters and for lengths. (b) All dimensions of test bar by agreement between manufacturer and purchaser. Source: Ref 1.

find that the same composition will not meet the requirements for class 35 in a heavy-section castings requiring a 50 mm (2 in.) test bar. It will qualify only for some lower class, such as 25 or 30. As the thickness of the controlling section increases, the composition must be adjusted to maintain the same tensile strength. International standard ISO 185 (Ref 2) provides eight classifications of gray iron based on tensile strength and six grades based on Brinell hardness. The tensile strength is determined by separately cast samples or cast-on samples cut from the casting as agreed upon by the manufacturer and purchaser. If a caston sample is used, the relevant wall thickness must be agreed upon. SAE standard J431, “Automotive Gray Iron Castings” (Ref 3), describes requirements that are based on hardness, tensile strength, and specific requirements of sand castings for the automotive industry. To address the variations in the relationship between tensile strength and hardness, test bar tensile-strength-to-Brinellhardness ratio (t/h) requirements are established so that a consistent t/h relationship can be used for each grade to predict and control material performance. The new SAE material designation (for example, G7H16c) is composed of an iron grade (G7, meaning the lower limit of t/h is 0.070 MPa/MPa), a hardness grade (H16, meaning the lower hardness limit expressed

Table 4 Estimated minimum transverse and tensile properties of SAE J431 automotive gray irons Transverse strength, min Transverse deflection, min Tensile strength, min SAE casting grade Former SAE grade Hardness(a), HB

G9H12 G9H17 G10H18 G11H18 G11H20 G12H21 G13H19

G1800 G2500 G3000 G3000 G3500 G4000 G4000

122 173 184 184 204 214 194

kN

lbf

mm

in.

MPa

ksi

7.65 8.90 9.79 9.79 10.90 11.56 11.56

1720 2000 2200 2200 2450 2600 2600

3.6 4.3 5.1 5.1 6.1 6.9 6.9

0.14 0.17 0.20 0.20 0.24 0.27 0.27

124 170 198 217 239 272 268

18.0 24.6 28.7 31.5 34.7 39.4 38.9

(a) Statistically defined specified minimum. Source: Ref 3

Table 5 SAE grade

Typical automotive applications of gray cast iron Previous designation

Typical uses

G9H12

G1800

G9H17

G2500

G9H17a

G2500a

G10H18

G3000

G11H20

G3500

G11H20b

G3500b

G11H20c G12H21 G11H24d

G3500c G4000 G4000d

Miscellaneous soft iron castings (as-cast or annealed) in which strength is not a primary consideration Small cylinder blocks, cylinder heads, air-cooled cylinders, pistons, clutch plates, oil pump bodies, transmission cases, gearboxes, clutch housings, and light-duty brake drums Brake drums and clutch plates for moderate service requirements; here, high-carbon iron is desired to minimize heat checking. Passenger car and light-duty truck cylinder blocks, cylinder heads, flywheels, differential carrier castings, pistons, medium-duty brake drums and clutch plates Medium- and heavy-duty truck and tractor diesel cylinder blocks and heads, heavy flywheels, tractor transmission cases, heavy gearboxes Brake drums and clutch plates for heavy-duty service where both resistance to heat checking and higher strength are definite requirements Brake drums for extra-heavy-duty service Extra-heavy-duty diesel engine castings, liners, cylinders, and pistons Camshafts of hardenable gray iron

Source: Ref 3

in MPa is 1600), and a special requirement (c, meaning the application is for brake drums, discs, or special service clutch plates, and specific requirements for total carbon and microstructure are defined). Table 4 shows a near equivalence of the new and old SAE grades. The SAE standard itself must be consulted for equivalency information for engineering purposes. As a reference, Table 5 shows automotive applications of gray irons with the previous SAE grade and a possible current grade. The ISO 185 standard has an informative annex with approximate cross references to other standards. For example, ISO 185:2004 grade ISO 185/JL/150, EN 1561 grade EN-GJL-150, ASTM A 48/A 48M grade 150, JIS G5501, grade FC150, and SAE J431 G9H12 are similar designations. Because of the minimum hardness and t/h ratio requirements of SAE J431, several SAE grades may satisfy other designation systems based only on tensile strength. The earlier ISO 185:1988 had designated hardness grades without specifying the thickness, as in the current ISO standard. This points out the need to examine standards and the relevant version of the standard specified when determining the equivalency of gray irons. Other ASTM specifications for specific applications and products are A 159 (automotive), A 126 (valves, flanges, and pipe fittings), A 74 (soil pipe and fittings), A 278/A 278M (pressurecontaining parts for temperatures up to 350  C, or 660  F), A 319 (nonpressure-containing parts for elevated-temperature service), and A 436 (austenitic gray irons for heat, corrosion, and wear resistance). ASTM A 823 addresses statically cast permanent mold gray iron castings. These are classified by tensile strength, Brinell hardness, heat treatment (annealed or normalized), and whether the casting is cored or solid.

Properties Compressive Strength. When gray iron is used for structural applications such as machinery foundations or supports, the engineer begins his calculations based on the compressive strength of the material. Table 6, which summarizes typical values for mechanical properties of the various grades, shows the high compressive strength of gray irons. Figure 10 compares the stress-strain curves in tension and compression for a class 20 and a class 40 gray iron. The compressive strength of gray iron is typically three to four times that of the tensile strength. If nonstatic loads are involved, the problem is one of dynamic stresses, which requires the consideration of fatigue and damping characteristics. Tensile strength is considered in selecting gray iron for parts intended for static loads in direct tension or bending. The bending strength is comparable to the tensile strength. Such parts include pressure vessels, autoclaves, housings

842 / Cast Irons Table 6 Typical mechanical properties of as-cast standard gray iron test bars Tensile strength

Torsional shear strength

Compressive strength

Reversed bending fatigue limit(a)

Modulus (E)

ASTM A 48/A 48M

ISO 185

MPa

ksi

MPa

ksi

MPa

ksi

MPa

ksi

GPa

20 25 30 35 40 50 60 150 200 225 250 275 300 350

... ... ... ... ... ... ... /JL/150 /JL/200 /JL/225 /JL/250 /JL/275 /JL/300 /JL/350

152 179 214 252 293 362 431 150–250 200–300 225–325 250–350 275–375 300–400 350–450

22 26 31 36.5 42.5 52.5 62.5 ... ... ... ... ... ... ...

179 220 276 334 393 503 610 170 230 260 290 320 345 400

26 32 40 48.5 57 73 88.5 ... ... ... ... ... ... ...

572 669 752 855 965 1130 1293 600 720 780 840 900 960 1080

83 97 109 124 140 164 187.5 ... ... ... ... ... ... ...

69 79 97 110 128 148 169 70 90 105 120 130 140 145

10 11.5 14 16 18.5 21.5 24.5 ... ... ... ... ... ... ...

66–97 79–102 90–113 100–119 110–138 130–157 141–162 78–103 88–113 95–115 103–118 105–128 108–137 123–143

103 ksi

9.6–14 11.5–15 13–16.4 14.5–17.2 16–20 18.8–22.8 20.4–23.5 ... ... ... ... ... ... ...

Hardness, HB

156 174 210 212 235 262 302 212 223 ... 241 ... 262 277

(a) From ISO 185:2000 (Ref 2) annex based on 35–50% of tensile strength

and other enclosures, valves, fittings, and levers. Depending on the uncertainty of loading, safety factors of 2 to 12 have been used in figuring allowable design stresses. Elasticity and Deformation. The elastic properties of gray iron require special consideration because they are different from those of other engineering materials (Ref 16). The ratio of stress (load per unit of area) to strain (elongation per unit length) is called the modulus of elasticity and is commonly symbolized by the letter E. The stress-strain relationship is seen in the stress-strain curves (Fig. 10, 11). The relationship between stress and strain in gray iron is not constant but is a curve of continually decreasing slope. It is different for different grades of iron and is influenced by heat treatment. The modulus of elasticity is the slope at any point on the curve of a given grade of gray iron, and it is not constant. Although gray iron is considered a brittle metal, it can be permanently deformed a measurable amount. Annealed grades can exhibit 0.5% of plastic strain before breaking (Ref 2). A portion of the total strain is elastic, and the balance is plastic (Fig. 12). The difference between plastic and elastic strain can be demonstrated by loading and unloading tensile specimens that are equipped with strain gages. The curves in Fig. 13 were made in a cyclic tensile test of a class 30 pearlite gray iron and show the appreciable plastic extension (permanent deformation) at stresses above 75% of the ultimate strength. When the metal has been stressed sufficiently to cause permanent deformation, its stress-strain relationship on subsequent loading is nearly linear up to the maximum stress that had been previously applied. It should be noted that this effect is only significant at high stress levels. It should also be noted that the value of the slope (E ) depends on the history of the sample. Modulus of Elasticity. Gray irons with a higher stress-to-strain ratio, that is, a higher modulus of elasticity, are desirable for structures where rigidity is important and where deflection due to loading should be minimized. A lower-modulus gray iron is preferred for

vibration damping and for severe heat shock applications where stresses result from differential thermal expansion. Other property requirements may interfere with the selection of an iron on the basis of elastic modulus alone. For example, high-modulus gray iron does not damp vibration as well as does a low-modulus iron, and low-modulus iron may not have sufficient strength or be machined with sufficient smoothness for some applications. These conflicting requirements may be compromised by selecting an intermediate class of iron or, in some applications, can be resolved by design. Increased stiffness may be combined with good vibration damping in a machine tool base by using a high-damping iron in a more rigid shape. Because of the curvature in the stress-strain relationship for gray iron, the slope of the curve cannot be accurately expressed as a single value for the modulus of elasticity. Two approximate values for the modulus are of practical use in gray iron casting design (Fig. 14). These are the secant modulus and the tangent modulus (Ref 9). The more commonly quoted value is the secant modulus, which represents the slope of a straight line from the origin at no load to the stress-strain curve at 25% of the ultimate tensile strength. This provides a modulus value that is conservative for most engineering work, because design loads are seldom as high as one-fourth of the ultimate strength, and the deviation from a linear relationship is less than 0.01% at this load (Ref 9). The tangent modulus is based on the slope of the stress-strain curve at no load. This is useful in the design of precision machinery where applied stresses are low. The tangent modulus can also be determined by measuring the resonant frequency of vibration of an appropriately sized rod with a frequency metering system. The modulus of elasticity is affected by structure, composition, and processing of the gray iron. Increasing graphite content and longer graphite flake length reduce the modulus. The modulus of elasticity decreases with increasing carbon equivalent. The presence of the usual alloying elements increases the

modulus of elasticity as well as the strength. The modulus of elasticity is decreased for increasing section size of the gray iron and is also lowered by annealing the iron (Ref 16). The modulus of gray iron (Table 6) varies considerably more than do the moduli for most other metals. Thus, in using observed strain to calculate stress, it is essential to measure the modulus of the particular gray iron specimen being considered. A significant range in modulus values is experienced because of both section size and chemical analysis variations (Fig. 15). In addition, the modulus experiences a linear reduction with increasing temperature. The rate of reduction can be reduced through alloy additions (Fig. 16). The numerical value of the modulus in torsion is always less than it is in tension, as is the case with steel. Torsional Shear Strength. As shown in Table 6, most gray irons have high torsional shear strength. Many grades have torsional strength greater than that of some grades of steel. This characteristic, along with low notch sensitivity, makes gray iron a suitable material for shafting of various types, particularly in the grades of higher tensile strength. Most shafts are subjected to dynamic torsional stresses, and the designer should carefully consider the exact nature of the loads to be encountered. For the higher-strength irons, stress-concentration factors associated with changes of shape in the part are important for torque loads as well as for bending and tension loads. Hardness of gray iron, as measured by Brinell or Rockwell testers, is an intermediate value between the hardness of the soft graphite in the iron and that of the harder metallic matrix. Variations in graphite size and distribution will cause wide variations in hardness (particularly Rockwell hardness) even though the hardness of the metallic matrix is constant. To illustrate this effect, the matrix microhardnesses of five types of hardened iron are compared with Rockwell C measurements on the same iron in Table 7. If any hardness correlation is to be attempted, the type and amount of graphite in the irons being compared must be constant. Rockwell hardness tests are considered appropriate only

Gray Iron Castings / 843

Fig. 10

Comparison of stress-strain curves in tension and compression for a class 20 and a class 40 gray iron. Source: Ref 15

for hardened castings (such as camshafts), and even on hardened castings, Brinell tests are preferred. Brinell tests (making a larger impression that will average the graphite and matrix) must be used when attempting any strength correlations for unhardened castings. The compressive strength correlates well, because the hardness test is a compression measure.

Fatigue Limit in Reversed Bending Because fatigue limits are expensive to determine experimentally, the designer usually has

incomplete information on this property. Fatigue life curves at room temperature for a gray iron under completely reversed cycles of bending stress are shown in Fig. 17(a), in which each point represents the data from one specimen. The effects of temperature on fatigue limit and tensile strength are shown in Fig. 17 (b) and (c), respectively. Axial loading or torsional loading cycles are frequently encountered in designing parts of cast iron, and, in many instances, these are not completely reversed loads. Types of regularly repeated stress variation can usually be expressed as a function of a mean stress and a

stress range. Wherever possible, the designer should use actual data from the limited information available. Without precisely applicable test data, an estimate of the reversed bending fatigue limit of machined parts may be made by using approximately 35% of the minimum specified tensile strength of the particular grade of gray iron being considered. This value is probably more conservative than an average of the few data available on the fatigue limit for gray iron. In ISO 185:2000 (Ref 2) annex, bending fatigue strengths of 35 to 50% of tensile strength are given as an informative property value (Table 6). The fatigue limit for reversed tension-compression is given as 26% of tensile strength. An approximation of the effect of range of stress on fatigue limit may be obtained from diagrams such as Fig. 18. Tensile strength is plotted on the horizontal axis to represent fracture strength under static load (which corresponds to a 0 stress range). Reversed bending fatigue limit is plotted on the ordinate for 0 mean stress, and the two points are joined by a straight line. The resulting diagram yields a fatigue limit (maximum value of alternating stress) for any value of mean stress. Few data available are applicable to design problems involving dynamic loading where the stress cycle is predominantly compressive rather than tensile. Some work done on aluminum and steel indicates that for compressive (negative) mean stress, the behavior of these materials could be represented by a horizontal line beginning at the fatigue limit in reversed bending, as indicated in Fig. 18. Gray iron is probably at least as strong as this for loading cycles resulting in negative mean stress, because it is much stronger in static compression than in static tension. It is therefore a natural assumption that the parallel behavior shown in Fig. 18 is conservative. If, prior to design, the real stress cycle can be predicted with confidence and enough data are available for a reliable S-N diagram for the gray iron proposed, the casting may be dimensioned to obtain a minimum safety factor of 2 based on fatigue strength. (Some uses may require more conservative or more liberal loading.) The approximate safety factor is best illustrated by point “P” in Fig. 18. The safety factor is determined by the distance from the origin to the fatigue limit line along a ray through the cyclic-stress point, divided by the distance from the origin to that point. In Fig. 18, this is OF/OP. On this diagram, point “P0 ” represents a stress cycle having a negative mean stress. In other words, the maximum compressive stress is greater than the tensile stress reached during the loading cycle. In this instance, the safety factor is the distance OF0 /OP0 . However, this analysis assumes that overloads will increase the mean stress and alternating stress in the same proportion. This may not always be true, particularly in systems with mechanical vibration in which the mean stress may remain

844 / Cast Irons

Fig. 11

Tensile stress-strain curves for class 20 to 50 gray iron casting. Source: Ref 17

emphasized that the number of cycles of alternating stress implied in Fig. 18 is the number normally used to determine fatigue limits, that is, approximately 10 million. Fewer cycles, as encountered in infrequent overloads, will be safer than indicated by a particular point plotted on a diagram for infinite life. Too few data are available to draw a diagram for less than infinite life. Fatigue Notch Sensitivity. In general, very little allowance need be made for a reduction in fatigue strength caused by notches or abrupt changes of section in gray iron members. The low-strength irons exhibit only a slight reduction in strength in the presence of fillets and holes. That is, the notch sensitivity index approaches 0; in other words, the effective stress-concentration factor for these notches approaches 1. This characteristic can be explained by considering the graphite flakes in gray iron to be internal notches. Thus, gray iron can be thought of as a material that is already full of notches and that therefore has little or no sensitivity to the presence of additional notches resulting from design features. The strength-reducing effect of the internal notches is included in the fatigue limit values determined by conventional laboratory tests with smooth bars. High-strength irons usually exhibit greater notch sensitivity but probably not the full theoretical value represented by the stress-concentration factor. Normal stressconcentration factors (Ref 20) are probably suitable for high-strength gray irons.

Pressure Tightness

Fig. 12

Tensile stress-strain curves for pearlite gray iron casting. Total strain is composed of plastic and elastic deformation components. Source: Ref 17

constant. For this condition, the vertical line through “P” would be used; that is, DK/DP would be the factor of safety. Most engineers use diagrams such as Fig. 18 mainly to determine whether a given condition of mean stress and cyclic stress results in a design safe for

infinite life. The designer can also determine whether variations in the mean stress and the anticipated alternating stress will place his design in the unsafe zone. Usually, the data required to analyze a particular set of conditions are obtained experimentally. It is

Gray iron castings are widely used in pressure applications such as cylinder blocks, manifolds, pipe and pipe fittings, compressors, and pumps. An important design factor for pressure tightness is uniformity of section. Parts of relatively uniform wall section cast in gray iron are pressuretight against gases as well as liquids. Most trouble with leaking castings is encountered when there are unavoidable, abrupt changes in section. Shrinkage, internal porosity, stress cracking, and other defects are most likely to occur at junctions between heavy and light sections. Watertight castings are considerably less challenging than gastight castings. A slight sponginess or internal porosity at heavy sections will not usually leak water, and even if there is slight seepage, internal rusting will soon plug the passages permanently. For gastightness, however, castings must be quite sound. Lack of pressuretightness in gray iron castings can usually be traced to internal porosity, which also is called internal shrinkage. In gray iron, this seems to be a phenomenon distinctly different from the normal solidification shrinkage that often appears on the casting surface as a sink or draw, which can be cured by risering. Internal porosity or shrinkage (very difficult to prevent, even by the use of very heavy risers) is usually associated with poor

Gray Iron Castings / 845

Fig. 14

Two methods of establishing the modulus of elasticity, E, for gray iron. Tangent modulus is line tangent at no load. Secant modulus drawn to 25% of ultimate tensile strength is more commonly used. Source: Ref 9

Fig. 13

Cyclic tensile stress-strain curves for class 30 gray iron. Permanent deformation results for application, removal, and reapplication of load. Source: Ref 17

Fig. 15

Interrelationship of mechanical properties, section diameter, carbon equivalent, and liquidus temperature of gray iron. Data are from one foundry and are based on dry sand molding and ferrosilicon inoculation.

feeding and lack of directional solidification. It can also be associated with heavy inoculation with calcium to promote a high graphite cell count. On the other hand, the use of strontium-bearing inoculants does not reduce cell size and can help control internal shrinkage in marginally gated castings.

Visible internal porosity may appear at centers of mass when the phosphorus content exceeds 0.25%. In critically gated castings, visible internal porosity may appear when the phosphorus content is as low as 0.09%. Chromium and molybdenum accentuate this effect of phosphorus, while nickel has a slight

mitigating influence. The effect of phosphorus may be caused in part by the fact that lowering phosphorus content also lowers the effective carbon equivalent (CE). Lowering CE by reducing carbon or silicon, or both, instead of phosphorus, may similarly reduce the leakage of pressure castings, but other foundry

846 / Cast Irons Table 7 Influence of graphite type and distribution on the hardness of hardened gray irons Type of graphite

A A A D D

Total carbon, %

Conventional hardness(a), HRC

Matrix hardness(b), HRC

3.06 3.53 4.00 3.30 3.60

45.2(c) 43.1 32.0 54.0 48.7

61.5 61.0 62.0 62.5 60.5

(a) Measured by conventional Rockwell C test. (b) Hardness of matrix measured with superficial hardness tester and converted to Rockwell C. (c) Although this value was obtained in the specific test cited, it is not typical of gray iron of 3.06% C. Ordinarily, the hardness of such iron is 48 to 50 HRC.

Fig. 16

Elastic modulus as a function of temperature in various alloyed gray cast irons. Source: Ref 18

problems (such as increasing amounts of normal shrinkage) would be encountered. ASTM A 278 (Ref 21) for pressure castings requires a CE of 3.8% (max), a phosphorus content of 0.25% (max), and a sulfur content of 0.12% (max) for castings to be used above 230  C (450  F). In addition to composition control, good overall foundry practice is required for consistently producing pressuretight castings. Sand properties and gating must be controlled to avoid sand inclusions. Pouring temperature must be adequate for good fluidity, and heavy sections should be fed wherever possible. Mold properties have been found to interact with composition to influence internal soundness. Generally, molds rammed tightly produce the soundest castings because of freedom from mold wall movement during solidification.

Impact Resistance Where high impact resistance is needed, gray iron is not recommended. Gray iron has considerably lower impact strength than cast carbon steel, ductile iron, or malleable iron. However,

many gray iron castings need some impact strength to resist breakage in shipment or use. There is incomplete agreement on a standard method of impact testing for cast iron. Two methods that have been used successfully are given in ASTM A 327 (Ref 22). The round bar test is used for gray and white irons, and the standard Charpy test is used for malleable and ductile irons.

Machinability The machinability of most gray cast irons is superior to that of most other cast irons of equivalent hardness, as well as to that of virtually all steel. The flake graphite introduces discontinuities in the metal matrix, which act as chip breakers. The graphite itself serves as a lubricant for the cutting tool. However, economical cutting depends on more than inherent machinability alone. Often, trouble in machining gray iron can be traced to one or more of several factors: the presence of chill at corners and in light sections; the presence of adhering sand on the surface of the casting; swells, which are usually the result of soft molds;

shifted castings; shrinks; and phases included in the matrix as a result of melting practice. Chill at corners and in light sections is more likely to be encountered with small castings, with higher-strength irons, and with designs that have light sections in the cope (or top) of the mold. Most foundries control iron with a chill test that gives an indication of the tendency of the iron to form white or mottled iron in light sections. The founder may treat iron with a small amount (0.5 to 2.5 kg/Mg, or l to 5 lb/ton) of a graphitizing alloy such as calcium-bearing ferrosilicon or other proprietary inoculant, which effectively decreases chill. Inoculation to achieve control of the tendency to chill usually does not result in significant changes in the composition or the physical properties of the iron, although it does produce changes in the mechanical properties. Light sections (5 mm, or 3/16 in.) usually cannot be cast in gray iron of higher than class 25. Class 30 iron can be cast in 6 mm (1/4 in.) sections. These values are different for different designs, depending on how the casting is made and gated. The important thing to understand is that the cooling rate in the mold at the time of freezing determines whether the iron will be gray, white, or mottled. If the thin section is in the drag or near the gate, the flow of hot metal heats the mold, thereby decreasing the rate of cooling and enhancing the formation of the gray iron. If chill is encountered, it is generally best to correct the trouble at its origin. It is usually uneconomical to anneal castings to remove chill because, in addition to heat treating for 2 h at 900  C (1650  F) for unalloyed irons, recleaning may be required for removing scale. Distortion beyond tolerance often occurs, and there are sacrifices in hardness and strength. Adhering sand usually can be removed by effective cleaning, but sand present as the result of penetration of the iron into the mold wall is extremely difficult to blast clean. This is a foundry defect that is best corrected at the source. Slowing the speed of machining and increasing the rate of feed is the best approach to salvaging castings of this type. Carbide tools are better than high-speed tools for resisting the extreme abrasion.

Gray Iron Castings / 847

Fig. 17

Effect of temperature on fatigue behavior and tensile strength of a gray iron (2.84% C, 1.52% Si, 1.05% Mn, 0.07% P, 0.12% S, 0.31% Cr, 0.20% Ni, 0.37% Cu). (a) Reversed bending fatigue life at room temperature. (b) Reversed bending fatigue limit at elevated temperatures. (c) Tensile strength at elevated temperatures. Source: Ref 19

Fig. 18

Diagram showing safe and unsafe fatigue zones for cast iron subjected to ranges of alternating stress superimposed on a mean stress. Example point “P” shows conditions of tensile (positive) mean stress; “P0 ” shows compressive (negative) mean stress. The safety factor is represented by the ratio of OF to OP or OF 0 to OP 0 . For conditions of constant mean tensile stress, DK/DP is the safety factor.

Swells are most troublesome in operations such as broaching and in other setups tooled for high production. The additional metal often places an excessive load on the tool, which may chip or dull but not actually fail until some time after the troublesome parts have been machined. In highly automated machining centers, this casting defect can result in significant statistical variation in dimensions and high scrap rates. Shifted castings are similar to swells in their action on cutting tools. Shifts or swells also may cause excessive tool loading if the locating points are affected. It is important to consider the positions of such locating points when designing the castings and also to avoid

indiscriminate grinding of locating points in the foundry cleaning room. Shrinks are not often present but can be troublesome when encountered in operations such as drilling. For example, the drill may tend to drift from its intended path to follow the shrink, which offers less resistance to the drill. Sometimes, a drill may break because it encounters a region of higher hardness, which often is associated with an area of shrink. Cast iron is the easiest metal to cast without internal shrink. Eutectic freezing is accompanied by expansion due to the precipitation of low-density graphite, which aids in obtaining internal soundness.

Machinability rating is complex, and criteria such as power per unit volume in unit time are of greatest importance in selecting a machine tool and the size of its motor. Machinability ratings based on power and force required, surface finish and accuracy, and tool life under standard test conditions are helpful but are not readily interpreted into the economics of machining. One way to indicate the effect on machinability of changing from one grade of iron to another is shown in Table 8, which is based on an experiment in which metal was removed from four types of gray iron using single-point carbide tools. For each type of iron, cutting speed was varied until the removal of 3300 cm3 (200 in.3) of iron resulted in a 0.75 mm (0.030 in.) wear land on the tool. The values of cutting speed thus obtained serve as qualitative evaluations of machinability but not as qualitative indexes. Optimum cutting speeds, tool materials, cutting fluids, feeds, and finish requirements must be studied for any given machine tool setup. Tool life is an important factor, because the machine must be stopped to change tools and the tools must be resharpened. Progress has been made in decreasing both machine downtime for tool changes and resharpening cost by the use of solid carbide inserts or bits and by other means. Annealing. The greatly improved machinability obtainable in gray iron by annealing has been advantageous to automotive and other industries for many years. Annealing is usually of the subcritical type, such as 1 h at 730 to 760  C (1350 to 1400  F). Some employ a cycle of heating to 785 to 815  C (1450 to 1500  F) and cooling at 22  C/h (40  F/h) to approximately 595  C (1100  F). These treatments graphitize the carbide in the pearlite and result in a ferritic

848 / Cast Irons Table 8 Machinability of gray iron Tensile strength Microstructure

Acicular iron Fine pearlite, alloy Ferrite (annealed) Coarse pearlite, no alloy

Cutting speed(a)

ASTM class

MPa

ksi

Hardness, HB

m/min

ft/min

50 40 ... 35

407 310 108 241

59 45 15.7 35

263 225 100 195

46 95 293 99

150 310 960 325

(a) Cutting speed at which removal of 3280 cm3 (200 in.3) produced 0.75 mm (0.030 in.) wear land on single-point carbide tools. Source: Ref 23

Table 9 Effect of annealing on hardness and strength of class 35 gray iron Tensile strength Condition

MPa

ksi

Hardness, HB

As-cast Annealed

268 165

38.9 23.9

217 131

Composition of iron: 3.30% total C, 2.22% Si, 0.027% P, 0.18% S, 0.61% Mn, 0.03% Cr, 0.03% Ni, 0.14% Cu, Mo nil. Annealing treatment consisted of 1 h at 775  C (1425  F), followed by cooling in the furnace to 540  C (1000  F).

metal removal. In all probability, accelerated wear tests between clean surfaces (with or without the presence of a lubricant) are predominantly tests of galling or scuffing. Corrosive wear combines abrasion or adhesion with the electrochemical action of the environment. In engine cylinders, it is caused by condensed acidic products of combustion during low-temperature operation. Usually, this kind of wear cannot be corrected by modifications in ordinary types of gray iron.

Fig. 19

Structure of class 35 iron, as-cast (left) and after annealing

Resistance to Wear matrix. Finding it uneconomical to graphitize primary carbide, most users try to avoid obtaining it. Annealing for improved machinability is most economical when the casting is small and the amount of machining is large. The annealing treatments described result in sacrifices in hardness and strength. A typical class 35 iron will be downgraded to approximately class 20 in strength by this treatment. In applications in which wear resistance is important, such as cylinder blocks, gray iron is not annealed because of the unsatisfactory performance obtained with a ferritic matrix. Table 9 and Fig. 19 show the changes in hardness, strength, and structure of class 35 gray iron obtained by annealing.

Wear Gray iron is widely used for machine components that must resist wear. Different types of iron, however, exhibit great differences in wear characteristics. These differences do not correlate with the commonly measured properties of the iron. In the discussion that follows, the general conclusions are related to metal-to-metal wear during sliding contact under conditions of normal lubrication. Although most of the supporting illustrations are for engine cylinders, the results have wider applicability. The published data on gray iron wear are somewhat inconsistent; accelerated-wear tests often do not correlate with field service experience, nor does field experience in one application necessarily agree with that in another application. In many applications, properties of the surface are at least as important as properties of the metal; for instance, wear resistance may frequently be enhanced by lapping the

wear areas. Relative hardness between mating parts also may be important to optimum wear resistance. Frequently, a hardness difference no greater than 10 points on the Brinell scale is considered optimum. For components in sliding contact, such as engine cylinders, valve guides, and latheways, the recognized types of wear after annealing are cutting wear, abrasive wear, adhesive wear, and corrosive wear. In well-designed machinery, wear proceeds slowly and consists of combinations of very mild forms of these four types of wear. The predominant characteristic of this so-called normal wear is the development of a glazed surface during break-in. The primary objective in any wear application is to establish conditions that produce and maintain this glaze. A secondary objective is to minimize normal wear. Cutting wear is caused by the mechanical removal of surface metal as a result of surface roughness and is similar to the action of a file. It usually occurs during the breaking in of new parts. Abrasive wear is caused by the cutting action of loose, abrasive particles that get between the contacting faces and act like a lapping compound. Under some circumstances, abrasive particles embedded in one or both of the contacting faces can produce a similar action. Adhesive wear is caused by metal-to-metal contact, resultant welding, and the breaking of those welds. When this happens on a large scale (a process known as galling), the metal is smeared, and severe surface damage results. Even in properly operating equipment, some wear occurs by adhesion on a microscopic scale. An intermediate form of adhesive wear, called scuffing, is less destructive than galling but still results in an abnormally high rate of

Resistance to normal wear, like resistance to scuffing, is affected by both graphite form and matrix microstructure (and possibly by the composition of the iron). In contrast to clearly defined limits for tensile strength and hardness, wear behavior is typically evaluated by comparison of materials. The relative performance of the materials is evaluated in service or experimentally in tests that vary from real load simulations to simpler pin-on-disc or abrasion tests. For instance, unalloyed pearlitic compact graphite iron (CGI) has approximately half the wear of unalloyed pearlitic gray iron when exposed to scarring or abrasion conditions (Ref 9). The average scar width in a low-friction-wear pin-on-disc test was 45% less for pearlitic CGI than pearlitic gray iron (Ref 24). Weight loss is used as a wear criterion. In similar abrasion tests, CGI was reported to lose 40 to 55% less weight than gray iron (Ref 25, 26). Effect of Matrix Microstructure. Gray iron is used for wear resistance in both the as-cast and hardened conditions. To show the effects of some of the matrix variations on wear resistance, comparable engine wear tests were made on seven types of gray iron cylinders, all of which apparently wore in a normal manner. These data are summarized in Tables 10 and 11. By present standards, hardening is the process that can produce the most significant improvement in resistance to normal wear. Although iron E (Table 11) showed exceptionally good wear properties, it was less resistant than the hardened iron G and at the same time could not be cast readily into the required shape. Thus, when evaluated on the basis of combined properties, the hardened iron G was twice as wear resistant as iron D, the as-cast iron that had

Gray Iron Castings / 849 Table 10

Effect of matrix microstructure on resistance of gray iron to normal wear Rate of wear

Iron(a)

No. of tests

Type and size of graphite

A B C D

28 19 4 6

A, A, A, A,

size size size size

3 to 4 4 5 to 6 7 to 8

E F G

30 3 2

A, size 5 to 6 A, size 3 to 4 Not shown

Matrix

mm/1000 h

in./1000 h

Lamellar pearlite Lamellar pearlite plus phosphide Fine lamellar pearlite Very fine lamellar pearlite plus undetermined constituent Same as iron D Lamellar pearlite plus ferrite Martensite

0.14 0.03 0.022 0.025

0.0057 0.0012 0.0009 0.00097

0.016 0.050 0.012

0.00062 0.002 0.00048

(a) Compositions are given in Table 11.

Table 11

Compositions of irons reported in Table 10 Composition, %

Iron

A B C D E F G

TC

Mn

Si

Cr

Mo

P

3.00–3.30 3.40 max 2.85–3.30 2.80–3.10 3.10–3.40 3.30–3.40 3.00–3.20

0.90–1.10 0.90 max 1.00 max 0.80 max 0.80–1.00 0.90–1.10 0.90–1.10

1.15–1.35 1.50 max 1.25–1.75 4.6–5.0 3.30–3.70 1.75–2.00 1.15–1.35

... ... 0.30–0.40 1.90–2.20 1.20–1.50 ... ...

... ... 0.25–0.35 ... ... 0.40–0.50 ...

0.20 max 0.35–0.50 0.20 max 0.20 max 0.40 max 0.20 max 0.20 max

S

0.12 0.13 0.12 0.12 0.12 0.12 0.12

max max max max max max max

Others

0.80–1.10 Cu ... 1.00–1.50 Ni ... ... ... ...

TC, total carbon

tensile strength of a gray iron. Figure 20 shows the creep-rupture properties of gray iron with molybdenum (Fig. 20a) and with molybdenum and chromium (Fig. 20b). As the figure shows, creep-rupture strength is improved with the addition of chromium and molybdenum. Dimensional stability of gray iron is degraded as temperature and exposure time increase. Dimensional stability at elevated temperatures is affected by factors such as growth, scaling, and creep rate. Growth of the iron is the result of the breakdown of the pearlite to ferrite and graphite. To prevent this growth, alloying elements such as copper, molybdenum, chromium, tin, vanadium, and manganese must be added because they will stabilize the carbide structure. Examples of these effects are seen in Fig. 21. Alloying additions are generally recommended at temperatures above 400  C (750  F). Chromium and Cr-Ni-Mo additions are most effective in retarding growth. Creep. A gray iron part that operates at elevated temperature will deform by creep if the load is great enough. Table 12, which gives the creep characteristics of several gray irons at 370  C (700  F) and at stresses ranging from 72 to 165 MPa (10.5 to 24 ksi), suggests the usefulness of molybdenum additions for improving the creep properties of gray iron. Creep-rupture properties are improved with additions of chromium and molybdenum (Fig. 20).

Dimensional Stability If the part is to operate below 480  C (895  F), two sources of dimensional inaccuracy must be considered: residual stresses and machining practice. Residual stresses are present in all castings in the as-cast condition and are caused by:
Differences in cooling rate between sections

Fig. 20

Influence of chromium and molybdenum on stress to produce rupture in 100 h in gray iron at various temperatures. (a) Unalloyed base iron. (b) 0.6% Cr alloyed base iron. Source: Ref 28

the best combination of wear resistance and castability. From other extensive testing, it has been concluded that hardened iron has as much as five times more wear resistance than pearlitic as-cast iron of the same general composition (Ref 27). Another significant advantage of hardening is that it greatly reduces variations in matrix microstructure and thus provides a superior iron of more dependable performance. In most applications such as valve guides, latheways, and various sliding members in machines and engines, satisfactory wear resistance is obtained with properly specified and controlled as-cast gray iron. Hardened iron is used to obtain maximum wear resistance in severe wearing applications such as high-speed diesel cylinder

sleeves, camshafts, gears, and similar heavily loaded wearing surfaces.

Elevated-Temperature Properties The most demanding applications of gray irons at elevated temperatures are those in which dimensional accuracy is important. Less rigorous applications at high temperatures are those in which a load is to be carried but in which appreciable distortion can be tolerated. In this case, the important properties include tensile strength, fatigue properties, and creep-rupture strength at elevated temperatures. Figures 17(b) and (c) show the effects of temperature on the fatigue limit and

of the same casting because of different cross sections or locations in the mold
Resistance of the mold to contraction of the casting during cooling
High-energy cleaning, such as with shot, which can induce compressive stresses. Residual stresses as high as 220 MPa (32 ksi) have been reported in gray iron wheels. Stresses of 170 MPa (25 ksi) have been observed in localized areas of other gray iron castings. Residual stresses in engine cylinder blocks have been measured as high as 130 MPa (19 ksi). Figure 22 shows two examples of residual stress in light castings, as determined with resistance strain gages. Only a small percentage of castings are stress relieved before machining, chiefly those requiring exceptional accuracy of dimensions or those with a combination of high or nonuniform stress associated with either low section stiffness or an abrupt change in section size.

850 / Cast Irons

Growth of four gray iron alloys produced from the same base iron (3.3% C, 2.2% Si) and tested at 455  C (850  F) in air. Source: Ref 18, 29

Fig. 21

Table 12

Fig. 22

Results of creep tests of gray iron at 370  C (700  F) Tensile strength

Iron composition(a), %

3.4 C, 1.5 Si 2.95 C, 2.46 Si 3.2 C, 2.60 Si, 1.5 Ni 2.75 C, 2.10 Si, 0.83 Mo 2.72 C, 2.5 Si, 0.83 Mo 2.72 C, 2.5 Si, 0.83 Mo 2.72 C, 2.5 Si, 0.83 Mo

Hardness, HB

197 237 230 241 241 241 241

MPa

217 310 307 410 362 362 362

Creep stress

Deformation rate(b)

ksi

MPa

ksi

31.5 45 44.5 59.5 52.5 52.5 52.5

72 72 72 72 119 134 165

10.5 10.5 10.5 10.5 17.3 19.5 24.0

0–150 h 150–450 h

12.1 5.3 ... 0.8 10.0 16.7 30.0

12.1 5.3 8.2(c) 0.8 0.0 3.7 6.7

450–900 h

5.9 5.3 5.3 2.2 0.0 0.0 1.6

900–1200 h 1200–2000 h

3.4 3.4 0.0 0.0 0.0 0.0 0.0

3.0 0.0 0.0 0.0 ... ... ...

(a) Also approximately 0.2% P (max), 0.1% S (max), and 0.7% Mn, except the 3.2C-2.60Si-1.5Ni iron, which had 0.15% S. (b) In mm/m h. (c) 0 to 500 h. 103 in./in.



Castings of class 40, 50, and 60 iron are more likely to contain high residual stresses. Figure 23 shows stress relief at seven temperatures between approximately 310 and 600  C (600 and 1100  F). The range from 480 to 600  C (900 to 1100  F) is recommended. For best results, castings should be held at temperature 1 h or more and then cooled in the furnace to below 450  C (840  F). High-strength and alloyed gray irons require temperatures approaching 600  C (1100  F). Machining Practice. In castings in the as-cast condition, residual tension and compression stresses are balanced, which makes the castings dimensionally stable at room temperature. However, when part of the surface is removed in machining, the balance of forces is altered, which can lead to distortion. If the casting is of relatively stiff section or has properly designed stiffness ribs, there may be no noticeable change in dimensions. Distortion will be most evident in castings of low stiffness

Location and magnitude of residual stresses in two gray iron castings

 h, or

from which a large volume of highly stressed metal has been removed. Because the surface of a casting is often the principal site of residual stresses, a large proportion of the stress is relieved by rough machining, with consequent maximum distortion. If, before final machining, the casting is relocated carefully and properly supported in the machine tool fixtures, acceptable dimensional accuracy will usually be obtained in the finished piece. In designing fixtures, it must be recognized that the usual gray iron (class 30) has a modulus of elasticity of approximately 97 GPa (14  106 psi), which means that deflection under tool bit loads will be approximately twice as great in a part made of this cast iron as in a steel part having the same section. It is difficult to make general statements about the dimensional stability that can be achieved in a gray iron casting without stress relieving. However, it is well known that automotive engine blocks are taken directly from

the foundry without stress relieving and are machined to tolerances of  þ 0.005 mm (þ 0.0002 in.) in parts such as crankshaft bear ings, camshaft bearings, and cylinder bores. Therefore, if a casting is properly designed, is cast under controlled conditions, and is allowed to cool sufficiently in the mold before shakeout, and if proper machining practice is followed, extremely high dimensional stability can be obtained in many applications without stress relieving. On the other hand, it is frequently economical to stress relieve complex castings (other than engine blocks) that are produced in small quantities and that must be machined to precise dimensions.

Effect of Shakeout Practice Shakeout practice may be influential in establishing the patterns of residual stresses in castings, because most residual stresses are basically caused by differences in cooling rate, and thus in contraction behavior, between light and heavy sections. In addition to its effect on residual stress, shakeout practice may influence both microstructure and hardness of gray iron castings. If the iron is austenitic at the time the mold is dumped, higher hardness and residual stress may result. Many different microstructures may be obtained; the austenite in thin sections may transform in the mold, whereas in heavier sections, which cool more

Gray Iron Castings / 851

Fig. 23

Stress relief of gray iron at various temperatures. Source: Ref 30

slowly, transformation may be delayed until air cooling after shakeout. In general, the effect of shakeout practice on hardness is negligible in unalloyed irons. Alloyed irons are the most sensitive. However, the martensitic irons that contain enough alloying elements to reach full hardness with slow cooling in the mold will display little if any effect of shakeout practice.

Alloying to Modify As-Cast Properties The term alloying as used here does not include inoculation because, by definition, the effect of inoculation on the mechanical properties of an iron is greater than can be explained by the change in chemical composition. Strength and hardness, resistance to heat and oxidation, resistance to corrosion, electrical and magnetic characteristics, and section sensitivity can be changed by alloying, which can extend the application of gray iron into fields where costlier materials have traditionally been used. There has also been considerable replacement of unalloyed gray iron by alloyed iron to meet increased service demands and to provide increased safety factors. The use of alloyed iron often depends, in practice, on the relative production requirements in a given foundry. When the applications requiring alloyed iron in a given plant are a small fraction of total production, manufacturing policy may dictate the use of unalloyed iron for all castings in order to

achieve maximum uniformity of production practice. Continuous production of 450 to 1350 kg (1000 to 3000 lb) heats of alloyed iron is usually needed for economical utilization. Manganese, chromium, nickel, vanadium, and copper can also be used to strengthen cast irons. In many irons, a combination of elements will provide the greatest increase in strength. To develop resistance to the softening effect of heat and protect against oxidation, chromium is the most effective element. It stabilizes iron carbide and therefore prevents the breakdown of carbide at elevated temperatures; 1% Cr gives adequate protection against oxidation up to approximately 760  C (1400  F) in many applications. For temperatures above 760  C (1400  F), the chromium content should be greater than 15% for long-term protection against oxidation. This percentage of chromium suppresses the formation of graphite and makes the alloy solidify as white cast iron. For corrosion resistance, chromium, copper, and silicon are effective. Additions of 0.2 to 1.0% Cr decrease the corrosion rate in seawater and weak acids. The corrosion resistance of iron to dilute acetic, sulfuric, and hydrochloric acids and to acid mine water can be increased by the addition of 0.25 to 1.0% Cu. For sulfur and acid corrosion, 15 to 30% Cr is effective. Silicon additions in the range of 14 to 15% give excellent corrosion resistance to sulfuric, nitric, and formic acids; however, both high-chromium and high-silicon irons are white, and the high-silicon irons are extremely brittle. The electrical and magnetic properties of cast iron can be modified slightly by minor

additions of alloying elements, but a major change in characteristics can be accomplished by the use of approximately 15% Ni or of nickel plus copper, which results in an austenitic iron that is virtually nonmagnetic. The austenitic gray irons also have good resistance to oxidation and growth at temperatures up to approximately 800  C (1500  F). Molybdenum is an effective alloying addition for retaining strength in heavy sections. It is normally added in amounts of 0.5 to 1.0%, but the low end of this range applies chiefly when molybdenum is added in combination with other elements. In casting in thin sections, nickel is the most effective in combating the tendency to form chilled iron. Base Irons. The selection of alloying elements to modify as-cast properties in gray iron depends to a large extent on the composition and method of manufacture of the base iron. For example, a foundry producing a base iron containing 2.3% Si and 3.4% total C for automotive castings may add 0.5 to 1.0% Cr if required to make heavier castings with the same hardness and strength as the normal castings. However, a foundry producing a base iron with 1.7% Si and 3.1% C for a heavy casting would add 0.5 to 0.8% Si to decrease hardness and chill when pouring this iron in light castings. Depending on the strength desired in the final iron, the CE of the base iron may vary from approximately 4.4% for weak irons to 3.0% for high-strength irons. The method of producing the base iron will affect mechanical properties and the alloy additions to be made, because factors such as type and percentage of raw materials in the metal charge, amount of superheat, and cooling rate of the iron after pouring affect the properties. The base iron used for alloying will vary considerably from foundry to foundry, as will the alloying elements selected to give the desired mechanical properties. However, parts produced from different base irons and alloy additions can have the same properties and performance in service.

Heat Treatment Gray iron, like steel, can be hardened by rapid cooling or quenching from a suitable elevated temperature. The quenched iron may be tempered by reheating in the range from 150 to 650  C (300 to 1200  F) to increase toughness and relieve stresses. The quenching medium may be water, oil, hot salt, or air, depending on composition and section size. Heating may be done in a furnace for hardening throughout the cross section, or it may be localized as by induction or flame so that only the volume heated above the transformation temperature is hardened. In the range of composition of the most commonly used unalloyed gray iron castings, that is, approximately 1.8 to 2.5% Si and 3.0 to 3.5% T total C, the transformation range is approximately 760 to 845  C (1400 to 1550  F). The higher temperature must

852 / Cast Irons Table 13 Hardness of quenched samples of gray iron Plain cast iron

Cr-Ni-Mo cast iron

Combined carbon, %

Hardness, HB

Combined carbon, %

Hardness, HB

As-cast

0.69

217

0.70

255

After quenching 650 (1200) 675 (1250) 705 (1300) 730 (1350) 760 (1400) 790 (1450) 815 (1500) 845 (1550) 870 (1600)

from,  C ( F) 0.54 207 0.38 187 0.09 170 0.09 143 nil 137 0.05 143 0.47 269 0.59 444 0.67 477

0.65 0.63 0.59 0.47 0.45 0.42 0.60 0.69 0.76

250 241 229 217 197 207 444 514 601

Condition

Note: Specimens were 30.5 mm diam bars, 51 mm long (1.2 in. diam bars, 2 in. long) quenched in oil from temperatures shown. Source: Ref 31

Table 14 Hardenability data for gray irons quenched from 855  C (1575  F) See Table 15 for compositions. Distance from quenched end

Hardness, HRC

mm

/16 in.

increments

Plain iron

Mo (A)

Mo (B)

NiMo

CrMo

Cr-NiMo

3.2 6.4 9.5 12.7 15.9 19.0 22.2 25.4 28.6 31.8 34.9 38.1 41.3 44.4 47.6 50.8 54.0 57.2 60.3 63.5 66.7

2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42

54 53 50 43 37 31 26 26 25 23 22 22 21 20 19 17 18 18 19 22 20

56 56 56 54 52 51 51 49 46 46 45 43 43 40 39 39 36 40 38 38 35

53 52 52 51 50 49 46 45 45 44 43 44 44 41 40 40 41 40 37 36 35

54 54 53 53 52 52 52 52 52 51 47 47 47 45 45 45 44 45 45 42 42

56 55 56 55 55 54 54 54 53 50 50 49 47 47 44 41 38 36 34 35 32

55 55 54 54 53 53 52 53 52 51 50 50 49 48 50 47 46 45 46 46 45

1

Source: Ref 31

be exceeded in order to harden the iron during quenching. The proper temperature for hardening depends primarily on silicon content, not carbon content; silicon raises the critical temperature. During the heating of unalloyed gray iron for hardening, graphitization of the matrix frequently begins as the temperature approaches 600 to 650  C (1100 to 1200  F) and may be entirely completed at a temperature of 730 to 760  C (1350 to 1400  F). This latter range is used for maximum softening. The changes in combined carbon content and hardness that occur upon heating and quenching of both alloyed and unalloyed gray iron are shown in Table 13. Ordinarily, gray iron is furnace hardened from a temperature of 860 to 870  C (1575 to 1600  F). This results in a combined carbon content of approximately 0.7% and a hardness of approximately 45 to 52 HRC

Table 15 Compositions of irons for which hardenability data are given in Table 14 Composition, % Iron

TC(a)

CC(b)

GC(c)

Mn

Si

Cr

Ni

Mo

P

S

Plain Mo(A) Mo(B) Ni-Mo Cr-Mo Cr-Ni-Mo

3.19 3.22 3.20 3.22 3.21 3.36

0.69 0.65 0.58 0.53 0.60 0.61

2.50 2.57 2.62 2.69 2.61 2.75

0.76 0.75 0.64 0.66 0.67 0.74

1.70 1.73 1.76 2.02 2.24 1.96

0.03 0.03 0.005 0.02 0.50 0.35

... ... Trace 1.21 0.06 0.52

0.013 0.47 0.48 0.52 0.52 0.47

0.216 0.212 0.187 0.114 0.114 0.158

0.097 0.089 0.054 0.067 0.071 0.070

(a) TC, total carbon. (b) CC, combined carbon. (c) GC, graphite carbon. Source: Ref 31

(415 to 514 HB) in the as-quenched condition. The actual hardness of the martensitic matrix is 62 to 67 HRC, but the presence of graphite causes a lower indicated hardness (Table 7). Temperatures much above this are not advisable because the as-quenched hardness will be reduced by retained austenite. Oil is the usual quenching medium for through hardening. Quenching in water may be too drastic and may cause cracking and distortion unless the castings are massive and uniform in cross section. Hot oil and hot salt are sometimes used as quenching media to minimize distortion and quench cracking. Water is often used for quenching with flame or induction hardening if only the outer surface is to be hardened. Hardenability of unalloyed gray iron is approximately equal to that of low-alloy steel. Hardenability can be measured using the standard end-quench hardenability test employed for steels. The hardenability of cast iron is increased by the addition of chromium, molybdenum, or nickel. Gray iron can be made air hardenable by the addition of the proper amounts of these elements. Some typical data on the hardenability of plain and alloyed irons are shown in Table 14. Compositions of the irons are listed in Table 15. Mechanical Properties. As-quenched gray iron is brittle. Tempering after quenching improves strength and toughness but decreases hardness. A temperature of approximately 370  C (700  F) is required before the toughness (impact strength) approaches the as-cast level. The tensile strength after tempering may be from 35 to 45% greater than the strength of the as-cast iron. Changes in properties brought about by quenching and tempering are shown in Fig. 24. Heat treatment is not ordinarily used commercially to increase the overall strength of gray iron castings, because the strength of the as-cast metal can be increased at less cost by reducing the silicon and total carbon contents or by adding alloying elements. When gray iron is quenched and tempered, it usually is done to increase resistance to wear and abrasion by increasing hardness. A structure consisting of graphite embedded in a hard martensitic matrix is produced by heat treatment. Localized heat treatment such as flame or induction hardening can be used in some applications in which alloy iron or chilled iron has traditionally been used,

often with a savings in cost. Heat treatment can be used when chilling is not feasible, as with complicated shapes or large castings, or when close tolerances that can be attained only by machining are required. Heat treatment extends the field of application of gray iron as an engineering material. The hardness of quenched and tempered gray iron measured with either a Brinell or Rockwell C tester is a composite hardness that superimposes the effects of graphite type, distribution, and size on the hardness of the metal matrix. The true hardness of the matrix, measured with a microhardness test, is generally from 8 to 10 HRC points higher than the hardness indicated by the conventional Rockwell C test. This is shown in Fig. 25, which compares composite hardness with matrix hardness. The composite hardness was obtained by conventional Rockwell testing; the matrix hardness was measured with a Vickers indentor and a 200 g load, and the values were converted to equivalent Rockwell hardness. The hardness of the matrix and the change in matrix hardness with tempering temperature are approximately the same as in an alloy steel containing approximately 0.7% C. After tempering at 370  C (700  F) for maximum toughness, the hardness of the metal matrix is still approximately 50 HRC. Where toughness is not required and a tempering temperature of 150 to 260  C (300 to 500  F) is acceptable, the matrix hardness is equivalent to 55 to 60 HRC. High matrix hardness and the presence of graphite result in a surface with good wear resistance for some applications, for instance, farm implement gears, sprockets, diesel cylinder liners, and automotive camshafts. Dimensional changes resulting from the hardening of gray iron are quite uniform and predictable if the prior structure and composition are uniform. Such dimensional changes can be allowed for in machining before heat treatment. Complex shapes or nonuniform sections may be distorted as a result of the relief of residual casting stresses or a differential rate of cooling and hardening during quenching, or both. The former condition can be minimized by stress relieving at approximately 565 to 590  C (1050 to 1100  F) before machining. The latter can be minimized by either marquenching or austempering; both of these processes are used for cylinder liners where out-of-roundness must

Gray Iron Castings / 853

Fig. 25

Effect of tempering temperature on the hardness of quenched gray iron (3.30% C,

2.35% Si)

Fig. 24

Changes in mechanical properties of gray iron as a function of tempering temperature

be held to a minimum. Through hardening is employed for gears, sprockets, hub bearings, and clutches. Localized Hardening. In parts requiring only localized areas of hardness, conventional induction hardening or flame hardening may be used. For flame hardening, it is generally desirable to alloy the iron with small amounts of either chromium or molybdenum to stabilize the iron carbide and thus prevent graphitization. Also, adequate localized hardening is enhanced when the structure contains little or no free ferrite prior to hardening. Compared to a pearlitic matrix, it may take from two to four times as long at temperature to condition a ferritic matrix for hardening. Even though the diffusion of carbon into austenite from the adjacent graphite is quite rapid, the attainment of a carbon concentration that can produce full

hardness may require 1 to 10 min (or more) for a previously ferritic matrix, depending on ferrite composition, austenitizing temperature, and graphite distribution and spacing. Water may be used as the quenching medium when the depth of hardening is shallow and the part is progressively heated and quenched. When only small areas are being hardened, the parts may be dropped into an oil quench. Localized hardening has been used for camshafts, gears, sprockets, and cylinder liners.

Physical Properties Patternmakers’ rules (shrink rules) allow 1% linear contraction upon solidification and cooling of gray iron. (The comparable allowances for other cast ferrous metals are 0% for

annealed ductile iron, 0.7% for as-cast ductile iron, 1½ to 2% for white iron, 2% for cast carbon steel, and 2½% for cast alloy steel.) Often, in practice this allowance is inaccurate, necessitating more exact evaluation of the effects of mass, composition, shape, mold, and core. Generally, contraction is less with an increase in mass and, in gray irons, increases with increasing tensile strength. Three contraction phenomena are exhibited during the cooling of a cast iron from the molten state to room temperature: in the liquid, contraction of approximately 2.8% from 1425  C (2600  F) to the liquidus at approximately 1200  C (2200  F); contraction during solidification; and subsequent to solidification, contraction of approximately 3.0% from 1150  C (2100  F) to room temperature. Shrinkage in volume during solidification ranges from negative shrinkage in “soft” irons to +1.94% in an iron containing approximately 0.90% combined carbon. The white irons undergo 4.0 to 5.5% contraction in volume within the same temperature range. Coefficient of thermal expansion of gray irons is approximately 13 mm/m  C (7.2 min./ in.  F) in the range from 0 to 500  C (32 to 930  F) and approximately 10.5 mm/m  C (5.8 min./in.  F) in the range from 0 to 100  C (32 to 212  F). The coefficient of thermal expansion is highly dependent on the matrix structure. The coefficients of thermal expansion of ferritic and martensitic irons are slightly higher than those of pearlitic irons. For the temperature range from 1070  C (1960  F) to room temperature, the coefficient of thermal expansion varies from 9.2 to 16.9 mm/ m  C (5.1 to 9.4 min./in.  F). At approximately room temperature, the commonly used figure of 10 mm/m  C (5.5 min./in.  F) is accurate enough for moderate changes in temperature.

















854 / Cast Irons Table 16 Thermal conductivity values for various gray irons compared with those for several steels and other irons Thermal conductivity at indicated mean temperature, W/m

K

Metal

200  F (95  C)

400  F (205  C)

800  F (425  C)

Steels 1020 4135 2335

51.9 43.3 38.1

48.5 41.5 38.1

41.5 38.1 36.3

57.1 55.4 48.5 46.7 36.3 36.3

51.9 51.9 ... ... ... 36.3

... 46.7 ... 41.5 ... 36.3

Ductile irons Ferritic, 2.35% Si Pearlitic, alloyed

41.5 26.0

38.1 27.7

32.9 ...

High-alloy irons 36% Ni steel (Invar) 36% Ni cast iron

10.4 39.8

... ...

... ...

Gray irons C, % 3.50 3.93 3.58 3.16 2.92 2.90

Si, % 2.25 1.40 1.90 1.54 1.75 1.51

Carbon equivalent, % 4.25 4.40 4.21 3.67 3.50 3.40 (alloyed)

Source: Ref 32

Table 17 Relative damping capacities of some common structural alloys Material

Relative damping capacity

Gray iron, coarse flake Gray iron, fine flake Malleable iron Ductile iron Pure iron Eutectoid steel White iron Aluminum

100–500 20–100 8–15 5–20 5 4 2–4 0.4

for bases and supports, as well as for moving parts. It reduces or eliminates parasitic vibration. The noise level of a machine operating on a gray iron base is materially reduced. High damping capacity is especially desirable in structures and parts in which vibration can cause stresses in excess of those that result from direct loading. Table 17 compares the relative damping capacity of gray iron with the relative damping capacities of other common structural alloys. ACKNOWLEDGMENT

Density of gray irons at room temperature varies from approximately 6.95 g/cm3 for open-grained high-carbon irons to 7.35 g/cm3 for close-grained low-carbon irons. The density of white iron is approximately 7.70 g/cm3. Thermal conductivity of gray iron is approximately 46 W/m K. See also Table 16. Electrical and Magnetic Properties. The resistivity of gray iron, compared with that of other ferrous metals, is relatively high, apparently because of the amounts and distribution of the graphite. Increases in total carbon content and in silicon content increase resistivity. The magnetic properties of gray iron vary within wide limits, ranging from those of irons having low permeability and high coercive force (suitable for permanent magnets) to those of irons having high permeability, low coercive force, and low hysteresis loss (suitable for electrical machinery). The highest magnetic induction and permeability are found in annealed white irons, such as malleable cast iron. Flake graphite, as in gray iron, does not affect hysteresis loss but prevents the attainment of high magnetic induction by causing small demagnetizing forces. Damping capacity is the capability of a material to quell vibrations and to dissipate the energy as heat, or simply the relative capability to stop vibrations or ringing. Because gray iron possesses high damping capacity, it is well suited



This article is based on the article “Gray Iron” revised by Charles V. White, Properties and Selection: Irons, Steels, and High-Performance Alloys, Vol 1, ASM Handbook. We thank George Goodrich and the American Foundry Society for their assistance in this revision. REFERENCES 1. “Standard Specification for Gray Iron Castings,” A 48/A 48M-03, Annual Book of ASTM Standards, 01.02, ASTM International, 2005 2. Grey Cast Irons—Classification, 2nd ed., ISO 185, International Standards Institute, 2005 3. “Automotive Gray Iron Castings,” J431, SAE Ferrous Materials Standards Manual, SAE International, 2004 4. J.M. Radzikowska, Metallography and Microstructures of Cast Iron, Metallography and Microstructures, Vol 9, ASM Handbook, ASM International, 2004, p 565–587 5. “Evaluating the Microstructure of Graphite in Iron Castings,” A 247, Annual Book of ASTM Standards, 01.02, ASTM International, 2005 6. M.D. Van Maldegiam and K.B. Rundman, On The Structure and Properties of Austempered Gray Cast Iron, Trans. AFS, 1986, p 249

7. R. Schneidewind and R.G. McElwee, Composition and Properties of Gray Iron, Parts I and II, Trans. AFS, Vol 58, 1950, p 312–330 8. “Chill Testing of Cast Iron,” A 367, Annual Book of ASTM Standards, 01.02, ASTM International, 2005 9. G.M. Goodrich, Ed., Iron Castings Engineering Handbook, American Foundry Society, 2006 10. R.A. Flinn and R.W. Kraft, Improved Test Bars for Standard and Ductile Grades of Cast Iron, Trans. AFS, Vol 58, 1950, p 153–167 11. W.F. Shaw, “Factors Involved in Achieving Full Potential Tensile Test Strength in Gray and Ductile Iron Tensile Testing,” ICRI Report 5529, 1989 12. L.G. Carmack, “Control of Physical Properties in Gray Iron Castings Through the Use of Arc Furnaces and the Holding Furnace,” 35th Electric Furnace Conference, Dec 8, 1977 (Chicago IL) 13. J.T. MacKenzie, The Brinell Hardness of Gray Cast Irons and Its Relationship to Some Other Properties, ASTM Proc., Vol 46, 1946, p 1025 14. D.E. Krause, Gray Iron—A Unique Engineering Material, Gray, Ductile, and Malleable Iron Castings—Current Capabilities, STP 455, American Society for Testing and Materials, 1969, p 3–28 15. C.F. Walton and T.J. Opar, Iron Castings Handbook, Iron Casting Society, 1981, p 235 16. E. Plenard, Elastic Properties of Cast Iron, Recent Research on Cast Iron, Gordon and Breach, New York, 1968 17. Atlas of Stress-Strain Curves, 2nd ed., ASM International, 2002 18. R.B. Gundlach, The Effects of Alloying Elements on the Elevated-Temperature Properties of Gray Irons, Trans. AFS, 1983, p 389 19. W.L. Collins and J.O. Smith, Fatigue and Static Load Tests of a High-Strength Cast Iron at Elevated Temperatures, Proc. ASTM, Vol 41, 1941, p 797–807 20. R.E. Peterson, Stress Concentration Factors, John Wiley, New York, 1974 21. “Gray Iron Castings for Pressure-Containing Parts for Temperatures Up to 650  F (350  C),” A 278/A 278M, Annual Book of ASTM Standards, 01.02, ASTM International, 2005 22. “Impact Testing of Cast Irons,” A 327, Annual Book of ASTM Standards, 01.02, ASTM International, 2005 23. U.S. Air Force Machinability Report, Vol 1, 1950, p 135 24. J.P. Hrusovsky, “Formation, Production and Properties of Compacted Graphite Iron,” Ph.D. thesis, Case Western Reserve University, Cleveland, OH, 1982 25. K. Bobylev, A.S. Glinkin, N.N. Aleksandrov, and V.I. Popov, Compacted Graphite Iron for D50 Cylinder Liners, Russ. Cast. Prod., Nov 1976, p 445–446

Gray Iron Castings / 855 26. I. Riposan, M. Chisamera, and L. Sofroni, Contribution to the Study of Some Technological and Applicational Properties of Compacted Graphite Cast Iron, Trans. AFS, Trans, Vol 93, 1985, p 35–48 27. G.P. Phillips, Hardened Gray Iron—An Ideal Material for Diesel Engine Cylinder Sleeves and Liners, Foundry, Vol 80, Jan 1952, p 88–95, 222, 224, 226, 228 28. G.K. Turnbull and J.F. Wallace, Molybdenum Effect on Gray Iron Elevated-Temperature Properties, Trans. AFS, 1959, p 35

29. J.E. Bevan, “Effect of Molybdenum on Dimensional Stability and Tensile Properties of Pearlitic Gray Irons at 600 to 850  F (315 to 455  C),” internal report, Climax Molybdenum Company 30. J.H. Schaum, Stress Relief Heat Treatment of Gray Cast Iron, Trans. AFS, Vol 61, 1953, p 646–650 31. G.A. Timmons, V.A. Crosby, and A.J. Herzig, Trans. AFS, Vol 49, 1941, p 397 32. C.F. Walton, Gray and Ductile Iron Castings Handbook, Iron Founders’ Society, 1971

SELECTED REFERENCES
C.E. Bates, Alloy Element Effects on Gray Iron

Properties, Part II, Trans. AFS, 1986, p 889


J.W. Grant, Comprehensive Mechanical Tests

of Two Pearlitic Gray Irons, BCIRA J., 1951, p 861
R.B. Gundlach, Thermal Fatigue Resistance of Alloyed Gray Irons for Diesel Engine Components, Trans. AFS, 1979, p 551
K.B. Palmer, Creep Tests on a Flake Graphite Cast Iron at 400  C, BCIRA J., 1959, p 839

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 856-871 DOI: 10.1361/asmhba0005324

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Ductile Iron Castings* George M. Goodrich, American Foundry Society

DUCTILE IRON, also known as nodular iron or spheroidal graphite iron, is second to gray iron in the amount of casting produced. Ductile iron accounts for 34% of all castings, ferrous and nonferrous (Ref 1). While gray iron has graphite flakes, ductile iron has an as-cast structure containing graphite particles in the form of small, rounded, spheroidal nodules in a ductile metallic matrix. Ductile iron contains trace amounts of magnesium (or cerium) that, by reacting with the sulfur and oxygen in the molten iron, precipitate carbon in this spherical form. These spheres improve the stiffness, strength, and shock resistance of ductile iron over gray iron. Different grades are produced by controlling the matrix structure around the graphite, either as-cast or by subsequent heat treatment. Ductile iron shares and supplements applications with malleable irons. Malleable iron is cast as a white iron and then annealed to form a temper graphite structure. The energy costs of annealing must be balanced with the cost of additional alloying elements in ductile iron to choose the most economical material at any time. For castings having section thicknesses of approximately 6 mm (0.25 in.) and above, ductile iron can be manufactured in much thicker section sizes than malleable. See the article “Malleable Iron Castings” in this Volume. However, it cannot be routinely produced in very thin sections with as-cast ductility, and such sections usually need to be heat treated to develop ductility. Ductile iron has the advantage, in common with gray iron, of excellent fluidity, but it requires more care to ensure sound castings and to avoid hard edges and carbides in thin sections, and it usually has a lower casting yield than gray iron. Compared to steel and malleable iron, it is easier to make sound castings, and a higher casting yield is obtained; however, more care is often required in molding and casting. Ductile iron is used in the transportation industry for applications such as crankshafts because of its good machinability, fatigue strength, and high modulus of elasticity, and in heavy-duty gears because of its high yield strength and wear resistance. Ductile iron is

stronger and more shock resistant than gray iron, so although it is more expensive by weight than gray iron, it may be preferred and an economical choice because a lighter casting can perform the same function. The second-largest end use for ductile iron is for pressurized water and wastewater systems. Since its introduction over 50 years ago, ductile iron has become the industry standard. Properties of ductile iron are compared with other cast irons in this article to aid the designer in materials selection. Austempered Ductile Iron (ADI). This heat treatment of ductile iron has grown in application because it almost doubles the tensile strength of ductile iron while retaining toughness. Most applications for ADI are in cars, trucks, and railroad and military vehicles.

Classes and Grades of Ductile Iron The common grades differ primarily by the matrix structure that contains the spherical graphite. These differences are the result of differences in composition, the metal processing, and the cooling rate of the casting after it is cast and/or as the result of heat treatment. Minor differences in composition or the addition of alloys may be used to enhance the desired microstructure. Five grades of ductile iron are designated by their tensile properties (tensile strength, yield strength, elongation) in ASTM specification A 536 (Ref 2), as shown in Table 1. The property values are stated in customary units; 60-40-18 grade designates minimum mechanical properties of 60 ksi tensile strength (414 MPa), 40 ksi yield strength (276 MPa), and 18% elongation. ISO standard 1083 (Ref 3) has a similar classification for spheroidal graphite cast irons, based on the tensile strength (MPa) and elongation, such as JS/400-18/S. The designations identify how specimens are made to determine the mechanical properties (separately cast, the “S”; cast-on; or samples cut from casting). Designations in Table 1 are simplified, and only some of the “S” values are given. The standard

must be consulted for design and purchasing. The designation can be augmented to include room-temperature and low-temperature impact resistance values. There are other ISO standards for spheroidal graphite cast irons. The Japanese spheroidal graphite iron castings standard JIS G5502 (Ref 5) is based on ISO 1083. It presents grades for separately cast samples, cast-on samples (suffix “A”), and impact strength at low temperature (suffix “L”), for example, FCD400-18AL. The common grades of ductile iron can also be specified by Brinell hardness, although the appropriate microstructure for the indicated hardness is also a requirement. This method had been used in SAE specification J434 JUN86 for automotive castings and similar applications. The current SAE J434 FEB2004 (Ref 4) includes the SI-based designation, with the former style designations listed in parentheses, as in Table 1. The new designation relates to tensile strength, with SI being the primary units. ISO standard 1083 offers hardness classes (with an “HBW” suffix) in an annex applicable upon agreement of manufacturer and purchaser. The hardness values given in Table 1 are for information only. Other specifications for special applications, such as low-temperature service, not only specify tensile properties but also have limitations in composition. There are numerous standards addressing ductile iron pipe. In the United States, ASTM International and American National Standards Institute/American Water Works Association standards specify the base material, coatings, and form of the pipe. The Ductile Iron Pipe Research Association (Ref 6) can be consulted for more on this specific product form. ASTM A 897/A 897M (Ref 7) was adopted as a standard to designate ADI in 1990; the grades are based on tensile properties in either customary (grade 130-90-09) or SI units (900650-09). The grades in either unit system are considered as separate grades (Table 2). Other national and international ADI standards, the Japanese JIS G5503 (Ref 8) and the European EN 1564 (Ref 9), are listed, showing grades based on SI tensile properties (Table 2).

*This article is adapted from the Iron Casting Engineering Handbook, edited by George M. Goodrich and Published by the American Foundry Society, 2006.

Ductile Iron Castings / 857 Table 1

Factors Affecting Mechanical Properties

Ductile iron properties of various industry and international standards Tensile strength

Impact energy(a)

0.2% offset yield strength

Mean

ksi

MPa

ksi

Elongation (min), %

J

ISO standard 1083(b) 350-22/S 350 51 400-18/S 400 58 500–7/S 500 73 550-5/S 550 80 600-3/S 600 87 700-2/S 700 102 800-2/S 800 116 900-2/S 900 130

220 240 320 350 370 420 480 600

32 35 46 51 54 61 70 87

22 18 7 5 3 2 2 2

17 14 ... ... ... ... ... ...

12 10 ... ... ... ... ... ...

ASTM A 536 (United States)(c) 60-40-18 414 60 276 60-42-10 415 60 290 65-45-12 448 65 310 70-50-05 485 70 345 80-55-06 552 80 379 80-60-03 555 80 415 100-70-03 689 100 483 120-90-02 827 120 621

40 42 45 50 55 60 70 90

18 10 12 5 6 3 3 2

... ... ... ... ... ... ... ...

SAE J434 (United States)(d) D400 (D4018) 400 58 250–275 D450 (D4512) 450 65 285–310 D500 (D5506) 500 73 320–345 D550 (D5504) 550 80 350–380 D700 (D7003) 700 102 425–450 D800 800 116 455–480

40 45 50 55 65 70

18 12 6 4 3 2

...

...

...

Grade

DQ&T(e)

MPa

...

...


lbf

Individual


lbf

Hardness, HB

14 11 ... ... ... ... ... ...

10 8.1 ... ... ... ... ... ...

<160 130–180 170–230 180–250 190–270 225–305 245–335 270–360

... ... ... ... ... ... ... ...

... ... ... ... ... ... ... ...

... ... ... ... ... ... ... ...

... ... ... ... ... ...

... ... ... ... ... ...

120 80 54 40 27 ...

90 60 40 30 20 ...

...

...

...

...

ft

J

ft

Structure

Ferrite Ferrite Ferrite + pearlite Ferrite + pearlite Pearlite + ferrite Pearlite Pearlite or tempered Pearlite or tempered

... ... ... ... ... ... ... ...

... ... ... ... ... ... ... ...

143–170 156–217 187–229 217–269 241–302 255–311

Ferrite Ferrite + pearlite Ferrite + pearlite Ferrite + pearlite Pearlite Pearlite or tempered martensite By Tempered agreement martensite

(a) ISO mean values from three tests. Measured from separately cast V-notched samples at room temperature (RT). SAE J434 are typical values for separately cast Charpy unnotched samples at RT. (b) Grades are listed in a simplified format. See text. Customary units are for information and are not part of Ref 3. Mechanical properties of test specimens from cast-on samples vary with wall thickness. (c) Customary units are the primary units, and SI unit conversion varies slightly for general-use grades and special-application grades in Ref 2. (d) Ref 4 lists both grade designations. SI units are primary, and customary units are derived. Yield strength varies with wall thickness within range given in SI units. Hardness values are listed as a guideline. (e) Quenched and tempered grade; minimum properties are agreed upon by producer and purchaser.

Table 2 Austempered ductile iron properties of various industry and international standards Tensile strength

Yield strength, 0.2% offset

MPa

ksi

MPa

ASTM A 897/A 897M-03 130-90-09 ... 900-650-09 900 150-110-07 ... 1050-750-07 1050 175-125-04 ... 1200-850-04 1200 200-155-02 ... 1400-1100-01 1400 230-185-01 ... 1600-1300-01 1600

130 ... 150 ... 175 ... 200 ... 230 ...

JIS G 5503-1995 (Japan) FCAD 900-4 900 FCAD 900-8 900 FCAD 1000-5 1000 FCAD 1200-2 1200 FCAD 1400-1 1400 EN 1564:1997 EN-GJS-800-8 EN-GJS-1000-5 EN-GJS-1200-2 EN-GJS-1400-1

Grade

800 1000 1200 1400

Impact energy(a)

ksi

Elongation (min), %

J

... 650 ... 750 ... 850 ... 1100 ... 1300

90 ... 110 ... 125 ... 155 ... 185 ...

9 9 7 7 4 4 2 2 1 1

... 100 ... 80 ... 60 ... 35 ... 20

75 ... 60 ... 45 ... 25 ... 15 ...

269–341 269–341 302–375 302–375 341–444 341–444 388–477 388–477 402–512 402–512

... ... ... ... ...

600 600 700 900 1100

... ... ... ... ...

4 8 5 2 1

... ... ... ... ...

... ... ... ... ...

... ... ... 341 401

... ... ... ...

500 700 850 1100

... ... ... ...

8 5 2 1

... ... ... ...

... ... ... ...

260–320 300–360 340–440 380–480

ft


lbf

Hardness HBW(b), kg/mm2

(a) Unnotched Charpy bars tested at room temperature. Value is the minimum average for the three highest values of four bars tested. (b) Hardness Brinell tungsten carbide ball indenter

Microstructure. Ductile iron is characterized by having all of its graphite occur in microscopic spheroids. Although this graphite constitutes approximately 10% by volume in ductile iron, its compact spherical shape minimizes the effect on mechanical properties, compared to the effect that is produced by the flake graphite found in gray cast iron. The graphite in commercially produced ductile iron is not always in perfect spheres. It can occur in a somewhat irregular form, but if it has the same shape as type II in ASTM standard A 247, the properties of the iron will be similar to the iron with perfectly spherical graphite. Of course, further degradation of the graphite shape can influence mechanical properties. The shape of the graphite is established when the iron solidifies. The difference between the various grades of ductile iron is in the microstructure of the matrix around the graphite. This matrix will vary with the chemical composition, metal processing, and cooling rate of the casting. The casting can be cooled in the mold, producing an as-cast structure, or the matrix structure and hardness can be changed by heat treatment. The high-ductility-grade castings can be produced as-cast, subcritically annealed, or fully annealed, producing a matrix structure that is made up of carbon-free ferrite. The intermediate grades are often used in the as-cast condition, without heat treatment, and have a mixed-matrix structure of ferrite and pearlite. The ferrite usually occurs as rings around the graphite spheroids, and because of this it is called bull’s-eye ferrite. The highstrength grades can be given a normalizing heat treatment to make the matrix pearlitic, they can be quenched and tempered to form a matrix of tempered martensite, or they can be made bainitic or to have an austempered (ADI) structure. Note that ADI had often been referred to as bainitic ductile iron, but ADI contains no bainite. In steel, bainite is an acicular ferrite and carbide mix. No carbides will exist in ADI unless it has been overtempered (Ref 10). All these matrix forms are obtained through heat treatment or by alloying. Much of the high-strength ductile iron is moderately alloyed to form an as-cast matrix of pearlite. Composition. The chemical composition of ductile iron, the molten-metal processing (including the magnesium treatment and inoculation), and the cooling rate (section size) of the casting have a direct effect on the properties by influencing the type of graphite and matrix structure that is formed. The spheroidal graphite shape is most often produced by treating a low-sulfur- (usually <0.02% S) base iron (which is a gray iron) with a low-magnesium or low-magnesium-plus-cerium alloy. This can also be done with a high-sulfur-base iron and a high-magnesium-content metal or alloy. All of the regular grades of ductile iron can be

858 / Cast Irons made from the same base iron, provided the chemical composition is appropriate, so that the desired matrix microstructure can be obtained by controlling the cooling rate of the casting after it is poured or by subsequent heat treatment. Extensive mathematical studies (Ref 11) have developed detailed equations that relate the composition of ductile iron to its mechanical properties. Most commercial irons are sufficiently established so that a generalized factor will suffice. In commercial practice, the softer (highductility) grades of ductile iron are usually made from a composition that is low in pearlite-stabilizing elements, such as copper and manganese. This is done to minimize the combined carbon content in the as-cast condition or to shorten the annealing cycle, if one is used. The higher-strength grades are made of a composition that will provide sufficient hardenability to respond satisfactorily to the cooling method being used. Castings with heavier metal sections will be alloyed for this purpose. For most casting requirements, the chemical composition of the iron is primarily a matter of facilitating production and minimizing defects. The typical range in carbon and silicon contents (Ref 12) is indicated in Fig. 1 for casting sections from 12 mm to approximately 38 mm (½ to 1.5 in.). Note that for castings with controlling sections outside this range, other limits for carbon and silicon will apply. Three sides of the preferred range are bounded by requirements for satisfactory casting production. The fourth side concerns the influence of a higher-silicon content on the impact-transition temperature; iron specifications for low-temperature impact service usually include maximum silicon and phosphorus content requirements. An increased silicon content tends to lower the hardness of the iron and increase the ductility if it promotes ferrite in preference to pearlite. However, in an entirely ferritic iron, such as those that are annealed, an increase in silicon content causes an increase in strength and hardness. Nickel strengthens ferrite with an effect similar to that of silicon. In the higher-strength pearlitic irons, alloy additions of copper, molybdenum, and nickel tend to increase the amount and hardness of the pearlite that is formed, thus tending to increase strength and decrease ductility. Chromium is not normally used as an alloy in ductile iron because of its tendency to form stable primary carbides. The presence of approximately 0.05% Sn is very effective in eliminating the formation of free ferrite, such as bull’s-eye ferrite, around the graphite spheroids in pearlitic irons. Section Effect. The spheroidal form of graphite in ductile iron results in less section sensitivity than other types of cast iron. However, very thin sections may solidify with some carbide or as white iron, unless precautions with inoculation and composition are taken. The matrix, and even the graphite structure, may be affected in very heavy sections due to

very slow cooling. Adjustments can be made to the composition or the heat treatment employed to produce the desired matrix structure in thin or heavy sections. The very slow cooling of thick metal sections can influence the casting properties in three ways:  Cause graphite flotation in high-carbon-

equivalent iron: The flotation of graphite to the top (cope side) of heavy-section castings is avoided by maintaining a lower total carbon content than is used for smaller castings (Fig. 2).  Form a large cell size and increase element segregation: The formation of large graphite nodules with a large cell (grain) size can result in increased segregation and possibly carbides or flake graphite at the cell

boundaries. This reduces the strength and ductility of the iron and increases its impact transition temperature (Ref 13). The increased segregation at the cell boundaries of carbide-stabilizing elements (chromium and molybdenum) can form carbides that may not be removed by a normal annealing heat treatment.  Result in the formation of ferrite with a relatively low hardness and strength: Slowly cooled heavy sections will have a lower hardness and strength, unless the analysis is adjusted to compensate for the slower cooling, or unless the cooling rate of the casting is increased through the transformation temperature range. Heavier castings are often alloyed and/or heat treated to overcome this effect and achieve a higher strength.

Fig. 1

Typical carbon and silicon range for ductile iron castings. Source: Ref 12

Fig. 2

Appearance of graphite (carbon) flotation on machined surfaces of blocks cast from ductile iron with 4.9% C equivalent

Ductile Iron Castings / 859 The effect of section thickness up to 152 mm (6 in.), with different alloy contents (Table 3), on the hardness of ductile iron is demonstrated (Ref 14) in Fig. 3 and 4. The irons in Fig. 3 show the relatively small effects in the as-cast condition. The effect of section size is more marked in the alloyed irons after normalizing (Fig. 4). Higher silicon content, up to a point, increases the tendency to form ferrite and results in a lower hardness. Increasing contents of alloying elements, such as nickel, copper, molybdenum, and small amounts of tin, produce more and finer pearlite, and this increases the hardness. However, copper is probably the most commonly used alloying element to Table 3 Alloying elements of ductile iron in Fig. 3 and 4 Alloying elements, % Iron No.

1 2 3 4 5 6 7 8

Ni

Mo

V

1.0 2.0 2.0 2.0 2.0 3.75 3.75 3.75

... ... 0.25 0.55 0.55 ... 0.26 0.55

... ... ... ... 0.25 ... ... ...

promote pearlite, because of its effectiveness and relatively low cost. As section size decreases, the solidification and cooling rates in the mold increase. This results in a fine-grained structure that can be annealed more rapidly. In thinner sections, however, carbides may be present, which will increase hardness, decrease machinability, and lead to brittleness. To achieve soft ductile structures in thin sections, heavy inoculation, probably at a late stage, is desirable to promote graphite formation through a high nodule number. As the section size increases, the nodule number decreases, and microsegregation becomes more pronounced. This results in a large nodule size, a reduction in the proportion of as-cast ferrite, and increasing resistance to the formation of a fully ferritic structure upon annealing. In heavier sections, minor elements, especially carbide formers such as chromium, titanium, and vanadium, segregate to produce a segregation pattern that reduces ductility, toughness, and strength (Fig. 5). The effect on proof strength (0.1% offset yield) is much less pronounced. It is important for heavy sections to be well inoculated and to be made from a composition low in trace elements.

properties. The relationship between tensile properties and hardness is dependable when the microstructure and chemical analysis are typical. This relationship is not dependable if, for example, the graphite is very irregular, if the matrix contains primary carbides, or there are alloys present. The presence of unusual constituents in the microstructure, such as primary carbides or the occurrence of other forms of graphite, can affect some properties quite differently than others. The hardness of all graphitic irons is essentially the hardness of the matrix metal reduced to a somewhat lower value by the presence of the graphite. Graphite in a spheroidal form does influence the hardness values obtained with conventional testers, but not as much as graphite in flake form. A martensitic ductile iron with an actual matrix hardness of Rockwell C 63 to 65 will indicate a hardness of 55 to 58. This effect presents no problem if it is recognized. The Brinell test is preferred for determining the hardness of ductile iron castings, and typical values for different matrix structures are listed in Table 1 and for austempered ductile iron in Table 2.

Tensile Properties Hardness Properties Because of the minimum influence of the spheroidal graphite on mechanical properties, the hardness value of ductile iron is a very useful piece of information and, with some reservations, can be directly related to other

The tensile properties of tensile strength, yield strength and percent elongation are often used to designate the grades. The minimum values for these properties are typically established by the specification. The component designer must be aware that higher strength does make a grade universally better. Because of the nominal and consistent influence of spheroidal graphite, the tensile properties and the Brinell hardness of ductile iron are related. The relationship between tensile properties and hardness depends on the alloys used and the microstructure obtained:  Ferritic matrix irons, possibly annealed,

Fig. 3

have a very low combined carbon content. Hardness and strength are dependent on hardening of the ferrite by the elements dissolved in it, silicon being the most important and common. Nickel is also a common ferrite strengthener.  Pearlitic matrix irons have lamellar carbide layers as the principal hardening agent, even though they may contain some free ferrite.  A uniform matrix of tempered martensite produced by heat treatment has a somewhat higher strength-to-hardness relationship.  The acicular or bainitic matrix irons have a similar relationship but generally have a lower ductility at a given strength.

Effect of alloying elements, as listed in Table 3, on the hardness of as-cast ductile irons. Source:

Ref 14

Fig. 4

Effect of alloying elements, as listed in Table 3, on the hardness of normalized ductile irons. Source: Ref 14

Fig. 5

Effect of cast section size on the properties of ductile iron

The general relationships between properties for the first two conditions are shown in Fig. 6. The left side (low hardness) of this graph represents irons with an entirely ferritic matrix. The right side represents the very fine pearlitic matrix that is obtained in a normalizing heat treatment.

860 / Cast Irons

Fig. 6

General relation between hardness and tensile properties of ductile irons in the as-cast, annealed, or normalized condition with a ferrite or pearlite microstructure. Source: Ref 15

Fig. 8

Typical stress-strain curves for pearlitic (curve 1) and ferritic (curve 2) ductile iron castings. Curve 1 is an as-cast pearlitic iron with ultimate tensile strength of 745 MPa (108 ksi). Curve 2 is an annealed ferritic with ultimate tensile strength of 400 MPa (58 ksi). Curve 3 (dashed) is the 0.2% offset yield strength. PL, proportionality limits. Source: Ref 16

Fig. 7

Typical relation between hardness and tensile properties of ductile irons that have been quenched and tempered so as to have a matrix of tempered martensite. Source: Ref 14

The central portion indicates the properties of irons with a matrix of ferrite and pearlite or of coarse pearlite. The relationship of properties for quenched and tempered ductile iron with a matrix of tempered martensite is shown (Ref 14) in Fig. 7. Ductile iron can be heat treated to have an acicular matrix, or it can be alloyed so that an acicular matrix structure is formed during the mold cooling of lighter sections, and by normalizing heavier-sectioned castings. The tensile test for ductile iron is similar to those for other ductile metals. Yield strength is established by the 0.2% offset method from a stress-strain curve, because ductile iron does not exhibit an obvious yield point. The modulus

of elasticity can also be determined from this curve. The percent elongation is usually measured over a 50 mm (2 in.) gage length or 4D (diameters) in accordance with ASTM A 370 E 8 after the testpiece has been broken. This includes the fracture. Typical stress-strain curves for ductile iron using the 0.2% offset method to establish the yield strength are shown (Ref 4) in Fig. 8. Tensile test samples for ductile iron are machined from the castings themselves or test castings called “Y” blocks or keel blocks (because of their shape). Round tension specimens are also used. These test castings are designed to provide a sound test bar and are made in several sizes so that they may be related in cooling rate to the controlling section of the castings. The test casting size, test bar dimension, and the testing procedure are established in the relevant specifications. Test bar coupons or blocks can be attached to the casting, or a test bar can be cut out of the casting, but these procedures must be done properly to avoid obtaining erroneous data. Attached coupons can affect the solidification of the casting and result in unsoundness in the test bar or in the casting. Tensile tests should only be conducted on sound metal to evaluate its mechanical properties. Defects in the test bar will produce erratic results. Nondestructive test methods are more effective for evaluating unsoundness. Test samples cut out of a casting should be taken from the highly stressed portion of the casting. This section is often at the

surface of the casting. Microstructure and hardness provide a means of correlating the properties in the casting and in separately cast test bars. The relationship between hardness, strength, and elongation for several A 536-grade ductile irons is shown in Fig. 9. The loss of strength and ductility in heavier sections can be minimized by using an alloyed iron having a bainitic matrix (Ref 17). The tensile properties are better with increasing nodularity, smaller nodule size, smaller matrix grain size, and a lower total carbon content. The properties of ductile iron are also affected by processing considerations, including charge materials, melting and holding of the molten iron, magnesium treatment, inoculation, and shakeout temperatures. Reduced mold cooling time and a hot shakeout temperature can increase strength and hardness because the castings are effectively normalized by this treatment (Ref 8). However, this practice can produce stresses in the casting, which, if not relieved, may affect the performance of the piece. Modulus of Elasticity. Ductile iron exhibits a constant modulus of elasticity (E) up to its proportionality limit. This limit is shown for ferritic and pearlitic ductile irons in Fig. 8. The slope of the stress-strain curve is not affected by the matrix structure but only by the amount and form of graphite present. The modulus of elasticity varies from approximately 159 to 176 GPa (23 to 25.5  103 ksi), with the lower values for irons of higher total carbon

Ductile Iron Castings / 861

Fig. 10

Correlation between ductile iron nodularity and dynamic modulus of elasticity. Source:

Ref 18

Fig. 9

Strength and ductility versus hardness ranges for A 356 standard-grade ductile irons

content, because the increased amount of graphite reduces the load-bearing cross section of the iron. If the graphite nodules become irregularly shaped, their span increases, and the modulus of elasticity is decreased (Ref 18), as illustrated in Fig. 10. The percentage of nodularity is seen in the micrographs in Fig. 11. The modulus of elasticity tends to be increased by the presence of primary carbides. In general, E is the same in compression and tension. The SAE International (Ref 4) informational appendix notes 166 GPa (24  103 ksi) as a typical test value and suggests that in highstrain applications, some designers reduce this by up to 15%. Dynamic elastic properties can be calculated from resonant frequency determinations (Ref 19), as in Table 4. The limit of proportionality in tension, based on a 0.005% proof stress, is

approximately two-thirds of the 0.2% offset yield strength (Ref 20) because of the gradual curvature of the stress-strain relationship near the yield strength. The proportional limit in compression is consistently higher than in tension by 10 to 30% (Ref 20).

Shear and Torsional Properties The ultimate strength in torsion or shear is generally approximately 90% of the tensile strength. Somewhat larger differences occur for the higher-strength, lower-ductility ductile irons (Ref 16, 21). The yield strength and limit of proportionality in torsion is approximately 75% of that in tension (Ref 16). From torsion tests, the shear modulus, or modulus of rigidity, G, was determined (Ref 21) to be

63 to 64 GPa (9.1 to 9.3 103 ksi). It is usually calculated as 0.39 times E (Ref 16). Poisson’s ratio (n) for ductile iron in tension and in compression is approximately 0.28, with values of 0.275 to 0.285 often quoted. Modulus of rigidity G is related to n and to E by the formula E = 2G(1 + n). All ductile irons deform considerably under torsional shear stresses, and it is difficult to obtain satisfactory test data to failure. It is estimated that torsional strength is approximately 0.9 times the tensile strength and that values measured for limit of proportionality and proof strength in torsion are generally between 0.7 and 0.775 times the corresponding values in tension. Damping Capacity. The damping capacity for ductile iron is substantially lower than for gray iron but higher than for steel (Table 4). It varies by grades of iron and somewhat with stress level. Figure 12 illustrates the specific damping capacity of ferritic and pearlitic ductile irons at various stress levels compared to class 40 gray iron and a low-carbon steel (Ref 20). The damping capacity is increased slightly with a higher carbon content in the ductile iron. Ferrite and as-quenched martensite matrices usually provide higher damping than do ductile irons with structures of pearlite or tempered martensite.

Compressive Properties The compressive yield strength of ductile iron, as measured by the 0.2% offset method, is somewhat higher than the yield strength in tension. This difference can be as much as 20% higher in compression because of the greater influence of the graphite in tension than in compression. The expected relationship between Brinell hardness and the compression yield strength (Ref 22) is shown in Fig. 13. The ultimate strength in compression is not determined for ductile metals. The increase in yield strength in compression over tension is greater in ductile irons with a ferritic matrix

862 / Cast Irons

Fig. 11

Microstructure of ductile irons of varying degrees of nodularity. (a) 99% nodularity. (b) 80% nodularity. (c) 50% nodularity. All unetched. Original Magnification. 36

Table 4 Average dynamic elastic properties from resonant frequency measurements Property

Young’s modulus, GPa (103 ksi) Shear modulus, GPa (103 ksi) Poisson’s ratio Damping, log decay  104

Ductile iron(a)

SAE 1018 steel(b) L

T

Gray iron, class 30

172 (25) 211 (31) 212 (31) 122 (18)

67 (9.7)

82 (12)

82 (12)

49 (7.1)

0.283 8.316

0.283 1.31

0.299 1.23

0.254 68.67

(a) Ferrite-Pearlite, 156–241 HBN. (b) Cold rolled, L, longitudinal; T, transverse. Source: Ref 19

Fig. 12

Fig. 13

Relation between the hardness of ductile iron and the compressive yield strength at 0.2% offset. Source: Ref 22

than in irons with a pearlitic structure. The 0.1% proof stress in compression exceeds that in tension by 20 MPa (3.0 ksi) for ferritic irons. For pearlitic irons and those that have been quenched and tempered, the excess in compression is approximately 12 MPa (1.8 ksi). Mixed

Damping capacity of various cast irons and mild steel as a function of surface stress. Source: Ref 19

structures fall between these values. The limit of proportionality in compression is 80% of the 0.1% proof stress (Ref 14) in compression. Combined Stresses. Since ductile iron has a linear stress-strain relationship and exhibits definite yield strength, both in tension and in compression, the theory of yielding and other conventional mathematical methods of stress analysis can be applied to ductile iron castings. In critical applications, cognizance should be given to the possible effects of multidirectional stresses; because castings are seldom simple in shape, they often have involved stress patterns under load. The yielding characteristics of metal in a complex shape can be expected to be different than in a tensile test specimen. Stress analysis techniques can be used very effectively on iron castings. The measurement of strain and its distribution, either in models or in full-sized components, when under actual

or simulated service, provides the most dependable basis for efficient design. Combined stress systems are also encountered in fatigue. Table 5 provides average properties for tensile, compressive, and torsional loads.

Fatigue Properties Metals can fracture because of cyclic loading or repeated and varied stresses that, at the maximum, are considerably below the yield strength. This type of fracture is called a fatigue failure. The fracture resembles that of a brittle material, even when it occurs in very ductile metals. A failure by fatigue is related directly to the stress and the number of cycles. In the unnotched condition, the fatigue limit for a ferritic iron is approximately 0.5 times the tensile strength obtained in a Wo¨hler

Ductile Iron Castings / 863 Table 5 Average mechanical properties of ductile irons heat treated to various strength levels Determined for a single heat of ductile iron, heat treated to approximate standard grades. Properties were obtained using test bars machined from 25 mm (1 in.) keel blocks. Nearest standard grade

Ultimate strength

Yield strength

Modulus

Hardness, HB

MPa

ksi

MPa

ksi

Elongation in 50 mm (2 in.), %

Tension 60-40-18 65-45-12 80-55-06 120-90-02

167 167 192 331

461 464 559 974

66.9 67.3 81.1 141.3

329(a) 332(a) 362(a) 864(a)

47.7(a) 48.2(a) 52.5(a) 125.3(a)

15.0 15.0 11.2 1.5

169 168 168 164

24.5 24.4 24.4 23.8

0.29 0.29 0.31 0.28

Compression 60-40-18 65-45-12 80-55-06 120-90-02

167 167 192 331

... ... ... ...

... ... ... ...

359(a) 362(a) 386(a) 920(a)

52.0(a) 52.5(a) 56.0(a) 133.5(a)

... ... ... ...

164 163 165 164

23.8 23.6 23.9 23.8

0.26 0.31 0.31 0.27

Torsion 60-40-18

167

472

68.5

195(b)

28.3(b)

...

65-45-12

167

475

68.9

207(b)

30.0(b)

...

80-55-06

192

504

73.1

193(b)

28.0(b)

...

120-90-02

331

875

126.9

492(b)

71.3(b)

...

63 65.5(c) 64 65(c) 62 64(c) 63.4 64(c)

9.1 9.5(c) 9.3 9.4(c) 9.0 9.3(c) 9.2 9.39(c)

GPa

(103 ksi)

Poisson’s ratio

... ...

...

(a) 0.2% offset. (b) 0.0375% offset. (c) Calculated from tensile modulus and Poisson’s ratio in tension

Fig. 15

Notched and unnotched fatigue properties of an annealed ductile iron with tensile strength of 454 MPa (65.8 ksi). Endurance ratio is 0.41, and notch sensitivity ratio is 1.67. Dimensions given in inches. Source: Ref 23

Fig. 14

Effect of cast section size on the fatigue properties of pearlitic and ferritic ductile irons

reversed-bending test. The fatigue limit decreases with increasing hardness to approximately 0.4 times the tensile strength in hardened irons up to 740 MPa (107 ksi) and may be even lower in stronger irons. Ductile irons are notch sensitive, and in a Wo¨hler specimen 10.6 mm (0.42 in.) in diameter with a 45 notch of root radius measuring 0.25 mm (0.01 in.), the fatigue limit decreases to approximately 0.63 times the unnotched value for ferritic irons and 0.6 times the unnotched value for hardened and tempered irons. The fatigue strength decreases as the cast section size is increased, as shown in Fig. 14

for both ferritic and pearlitic irons, notched and unnotched. As the size of the component subjected to reversed-bending fatigue increases, there is a decrease in fatigue strength up to approximately 50 mm (2 in.) in diameter. This effect is not thought to occur in axial fatigue in tension/compression. Stress concentration, such as at notches, is detrimental to fatigue life, but the speed of the loading cycle and the occurrence of rest periods are not significant. As the maximum stress is reduced, the number of cycles necessary to produce a failure becomes much larger. The highest stress at which the number of cycles for failure approaches infinity (generally in excess of 10 million cycles) is called the endurance limit. The endurance ratio is the

Fig. 16

Effect of tensile strength and matrix structure on endurance ratio for ductile iron. Source:

Ref 24, 25

relationship between the endurance limit and the tensile strength. Most of the data on fatigue strength for ductile iron have been obtained with the rotatingbeam-type test, where the maximum stress alternates between tension and compression. A typical S-N curve on which the stress is plotted against the number of cycles to failure (Ref 23) is shown in Fig. 15. From these results, the unnotched endurance limit is indicated to be 193 MPa (28 ksi). This is the maximum stress at which a fatigue failure should not occur under similar conditions in any number of cycles. Since this particular ductile iron has a tensile strength of 454 MPa (68.5 ksi), its endurance ratio is 0.41. The effect of stress raisers on the endurance limit has been evaluated by the use of notched test bars. For this purpose, a notch is usually turned into the circumference of an oversized bar, so that the base of the notch leaves a cross section equal to the regular unnotched bars to which the notched bars are compared. The ratio of the unnotched to the notched endurance limit is termed the notch sensitivity factor or the dynamic stress-concentration factor. Several investigations have obtained endurance ratios of from 0.33 to 0.52 for the different types of ductile iron. The endurance limit increases with tensile strength, but, as with other ferrous metals, the increase is less than proportional. The relationship between tensile strength and the endurance ratio is different for the annealed ferritic irons than for the irons with a matrix of pearlite or tempered martensite (Ref 24, 25) (Fig. 16). In fact, the as-cast irons have the highest endurance ratios. Typical fatigue properties for three common types of ductile iron are summarized in Table 6. The test results (Ref 24, 26) in Table 7 indicate the influence of heat treatment on endurance properties. Although tensile strength increases with hardness, no increase in the notched endurance limit occurs at the higher hardness values. Chemical composition, including alloys, within the usual commercial ranges, has not

864 / Cast Irons Table 6 Reversed-bending fatigue properties of three standard grades of ductile iron 45 V-notched

Unnotched Tensile strength, St

Endurance limit, Se

Type

MPa

ksi

MPa

ksi

Endurance ratio(a)

64-45-12 80-55-06 120-90-02(d)

490 620 930

71 90 135

210 275 338

30.5 40 49

0.43 0.44 0.36

Endurance limit, Sn MPa

ksi

Endurance ratio(b)

Notch sensitivity factor(c)

145 165 207

21 24 30

0.30 0.27 0.22

1.4 1.7 1.6

(a) Sc/St. (b) Sn/St. (c) Sc/Sn. (d) Oil quenched from 900  C (1650  F) and tempered at 595  C (1100  F)

Table 7 Influence of heat treatment on endurance properties of ductile iron Property

Brinell hardness No. Tensile strength, MPa (ksi) Elongation % Endurance limit Unnotched MPa (ksi) 45 V-notched MPa (ksi) Endurance ratio(d) Notch sensitivity(e)

As-cast

Quenched and tempered(a)

Quenched and tempered(b)

As-cast

Normalized(c)

255 664 (96.3) 3

295 928 (134.5) 4

350 1030 (149.3) 1.5

267 659 (95.6) 2

325 1041 (151) 4

278 (40.3) 178 (25.8) 0.42 1.56

340 (49.3) 208 (30.2) 0.37 1.63

340 (49.3) 194 (28.2) 0.33 1.76

301 (43.7) 208 (30.2) 0.46 1.44

340 (49.3) 208 (30.2) 0.33 1.63

(a) Oil quenched from 900  C (1650  F). Tempered 2 h at 600  C (1110  F). (b) Oil quenched from 900  C (1650  F). Tempered 2 h at 650  C (1020  F). (c) Normalized from 900  C (1650  F) in 22 by 22 by 216 mm (7/8 by 7/8 by 8½ in.) pieces. (d) Endurance ratio = Endurance limit/Tensile strength. Notch sensitivity = Unnotched endurance/Notched endurance

Table 8 Influence of decreasing nodularity on tensile and rotating-beam properties of pearlitic ductile iron Tensile strength Iron No.

1 2 3 4 5 6

Unnotched endurance limit

Nodularity, %

MPa

ksi

MPa

ksi

Endurance ratio

Notch factor

98 97 95 77 44 50

908 906 868 784 841 561

131.6 131.3 125.9 113.7 93.0 81.3

278 286 263 239 217 208

40.3 41.4 38.1 34.7 31.4 30.2

0.31 0.32 0.30 0.31 0.34 0.37

2.00 1.95 1.89 1.73 1.65 1.69

Source: Ref 27

been found to be significant to fatigue properties, other than through the effect on microstructure and tensile properties. Fatigue properties depend to some extent on the graphite structure. Nonspheroidal graphite lessens the fatigue strength but to a lesser extent than it decreases the tensile strength (Ref 27). This behavior is given by Table 8 for a series of ductile irons with a pearlitic matrix and decreasing amounts of nodularity (roundness) of the graphite. The tensile strength decreased by more than one-third and the ductility by one-half when the nodularity was reduced to 50%. However, the fatigue strength still remained at 75% of the value for the optimum graphite structure. Accordingly, the endurance ratio increased from 0.31 to 0.37. Table 8 also shows the effect of a very sharp 0.25 mm (0.01 in.) radius V-notch on the fatigue behavior. The fatigue strength of the iron with optimum nodularity was decreased to half of its unnotched value by the presence of this sharp notch. This is indicated by a notch factor of 2. In the case of the ductile iron with only 50% nodularity, the notched specimens exhibited only approximately 60% of the fatigue resistance of unnotched specimens, indicated by a notch factor of 1.69. Accordingly,

the notched specimens with excellent nodularity had a fatigue limit of 139 MPa (20.2 ksi) compared to 123 MPa (17.9 ksi) for the notched specimens with 50% nodularity. Fatigue data are plotted on Goodman diagrams (Ref 28), as shown in Fig. 17, to allow fatigue strength data obtained on tension-compression stress reversals to be used for alternating stress about various mean stresses other than zero. The diagrams show the fatigue behavior of several strength levels of ductile iron in bending and under uniaxial tensioncompression stresses, respectively. The minimum tensile properties of these irons are listed in the table accompanying Fig. 17. The relationship between rotary-bending fatigue and hardness for ductile iron is provided (Ref 29) in Fig. 18 and 19. Figure 18 indicates that the fatigue limit increases with increasing Brinell hardness, although at a lesser rate at higher hardness levels. The scatter in fatigue limit values demonstrates that hardness can be an unreliable criterion for accurate prediction of fatigue strength. The rotary-bending fatigue limit increases with matrix microhardness up to approximately 500 HV (Vickers hardness) but decreases with higher hardness matrices (Fig. 19).

Minimum properties of ductile irons in curves Tensile strength

Yield strength

Iron No.

MPa

ksi

MPa

ksi

Elongation, %

1 2 3 4 5

380 420 500 600 700

55 61 72 87 102

250 280 350 420 500

35 41 61 61 72

17 12 7 2 2

Fig. 17

Goodman diagrams for ductile irons. (a) Endurance limits for bending stresses. (b) Endurance limits for tension-compression stresses. Source: Ref 28

As indicated by the data in Table 8 and Fig. 16 through Fig. 19, the fatigue strength of ductile iron is affected by the graphite nodularity and the matrix structure. It is apparent from these data and data in Tables 6 and 7 that the endurance ratio generally decreases (although the fatigue strength increases) for the higherstrength ductile irons. The presence of nonmetallic inclusions and rough cast surfaces also

Ductile Iron Castings / 865

Fig. 19 Fig. 18

Brinell hardness versus fatigue limit for ductile iron shows data scatter that make a prediction of fatigue unreliable. Source: Ref 29

lower fatigue strength. To the extent that the matrix structure is affected, the presence of alloying elements affects the fatigue behavior. Corrosive atmosphere or environment can reduce both tensile and fatigue properties (Ref 30) and gives the added complication of a frequency dependence for the number of cycles to failure. Phenomena of stress-corrosion cracking and corrosion fatigue need to be investigated for application involving static or cyclic stresses in environments corrosive to ductile iron. The fatigue limit of ductile iron can be increased, particularly in bending and torsion tests, by shot peening, surface rolling, and flame and induction hardening. These treatments both increase the strength and induce compressive stresses at the surface but should be evaluated for effect and used under close control. Shot peening and surface rolling will significantly increase the endurance limit of ductile iron components, especially when applied to stressed radii or notches. More than a 60% increase in the endurance limit was obtained with a rolling pressure that was insufficient to depress the surface a measurable amount (Ref 31). See Ref 32 for more information on these topics related to ductile iron.

Fracture Toughness Toughness is simple as a concept but is complex as a measured characteristic of metal. The toughness of a material is a measure of the work required to fracture it. This work is physically related to the resistance to crack initiation and growth or propagation. Fracture generally occurs by either cleavage or void coalescence. A cleavage or brittle fracture occurs with a minimum of deformation, giving a faceted, shiny fracture surface. Very little energy is dissipated in such a fracture, so the fracture toughness is low. By contrast, ductile fracture occurs as a result of void formation at the interface between the two phases, followed by void growth, void coalescence, and fracture. The dull, grayish fractured surface will be rough or dimpled, and the energy dissipation due to the plastic deformation results in a high fracture toughness.

Relation between micro-Vickers hardness of matrix and bending fatigue. Source: Ref 29

Cleavage and void coalescence may be thought of as competing modes of fracture. Where plastic deformation is inhibited, brittle or cleavage fracture will occur. All iron and steel, except the austenitic types, have a yield strength that increases significantly with decreasing temperature (Ref 33) (Fig. 20). At sufficiently low temperatures, plastic deformation is restricted, and brittle fracture results in a low fracture toughness value. The temperature at which this ductile-to-brittle change occurs is called the transition temperature. Austempered ductile iron exhibits a good combination of low transition temperature and high impact energy that can be obtained in irons containing retained austenite (Ref 34). Many figures have been reported for the stressintensity factor, KIc, obtained under plane-strain linear-elastic conditions. However, because of the ductility of ductile irons, most investigations of ferritic and pearlitic/ferritic irons have been conducted at subzero temperatures to obtain the necessary criteria of elastic behavior. pffiffiffiffiffi Values ranging from 25 to 54 MPa m (23 to pffiffiffiffiffiffi 49 ksi in:) have been reported, with the higher values for ferritic irons. In general, the value of KIc has been observed to increase with testing temperature. Figure 21 shows the low-energy brittle fracture and upper-shelf value of ductile fracture for various microstructures. Planestrain fracture toughness (KIc) at ambient temperature and upper-shelf regime is estimated at pffiffiffiffiffiffi pffiffiffiffi 88 to 90 MPa m (80 to 81 ksi in:) for ferritic ductile irons based on the review (Ref pffiffiffiffi 35). Upper-shelf KIc values of 100 MPa m pffiffiffiffiffiffi (90 ksi in:) or more are also considered possible for ferritic ductile irons, as compared with pffiffiffiffiffi ffi pffiffiffiffi 75 MPa m ð68 ksi in:Þ for malleable irons. A high strain rate also inhibits plastic flow by minimizing thermally assisted flow processes. This causes the material to behave as if it were at a lower temperature and enhances the possibility of a brittle fracture. A triaxial state of stress, such as occurs at the tip of a crack or notch, will greatly reduce the shear stress, which causes plastic flow and induces a brittle fracture. Impact Tests. There are three different tests used to determine the fracture toughness of materials. The Charpy test and the dynamic tear test use a V-notched specimen that is impact loaded in three-point bending. The energy absorbed in fracturing a series of test specimens

Fig. 20

Influence of low temperature on the tensile properties of annealed ductile iron containing 2.95% Si. Source: Ref 33

Fig. 21

Effect of microstructure on fracture toughness of ductile iron. These results are generally based on data in literature, with some approximation where data were unavailable. Source: Ref 35

is determined at various temperatures. The fracture mechanics test uses a sharply notched (usually by fatigue precracking) specimen with the load applied slowly. The critical stress intensity, KIc, at which crack extension occurs, is measured in these tests. The specimen used in the fracture mechanics test is much larger than the impact specimens and provides more restraint to plastic deformation. Charpy tests use small coupons (10 by 10 by 55 mm, or 0.4 by 0.4 by 2.2 in.) that are generally V-notched (but can also be U-notched) and impact loaded in three-point bending. Impact tests measure the energy absorbed during the fracturing of the testpiece, and, due to the small size, the three tests are generally run at a particular test temperature. Since most of the energy is used to deform the sample, the greater its ductility, the higher will be the impact test results. The dynamic tear test uses a much larger sized sample (181 by 41 by 16 mm, or 7.1 by 1.6 by 0.63 in.) and tends to give more reproducible results than the Charpy test. Complex stress states, due to the shape of the notch, high strain rates, and lower temperatures,

866 / Cast Irons

Fig. 22

Influence of loading rate on the energy absorbed in fracturing an annealed ductile iron containing 2.95% Si. Testpieces were standard Charpy type with a 0.25 mm (0.010 in.) notch but were also precracked 0.76 mm (0.030 in.) by fatigue. Black circles are shear fracture; open circles are cleavage fracture. Source: Ref 36

Fig. 24

Effect of matrix structure on impact transition curves of ductile iron unnotched specimens. Source: Ref 38

reduce both the absorbed energy and the amount of ductility during fracturing for a given material. The results from one impact test are not directly comparable to another test, nor are the data directly applicable in design. The effect of strain rate on the transition behavior of annealed ferritic ductile iron is illustrated in Fig. 22. This plot shows the notch toughness in in-lb/in.2 (J/cm2) for precracked Charpy Vnotch specimens over the transition temperature range. The type of fracture obtained on each test is indicated. The more severe (triaxial) stress state is imposed by a notched specimen, and this reduces impact resistance when compared to unnotched specimens (Ref 33). The effect of the notch both increases the transition temperature and decreases the upper-shelf energy. Typically, in an unnotched impact test, corresponding transition temperatures are approximately 50  C (90  F) lower than in a notched test. The transition temperature is usually considered to occur when the fracture

Fig. 23

Charpy V-notch impact energies of ductile irons. Source: Ref 36

appearance is 50% ductile or shear and 50% brittle or cleavage fracture. The effect of test temperature is shown in both Fig. 21 and 23. The impact properties of ductile irons are improved by the graphite being more nodular. The transition temperature and the upper-shelf energy are lowered by both an increasing volume of graphite and increasing nodule count. The larger size of graphite nodules in heavy casting sections is not detrimental to mechanical properties. When the nodularity is good, the matrix structure and composition determine the impact behavior. Silicon has a dual effect on impact properties because higher silicon favors a ferritic matrix, and the ferrite decreases the transition temperature and increases the upper-shelf energy. However, lower-silicon irons (2%) with a ferritic matrix have a lower transition temperature with the same or better upper-shelf energy as highersilicon irons that are also ferritic (Ref 37). A phosphorus content above 0.1% lowers the toughness sharply, and for maximum impact resistance, the content should be maintained below 0.03%. The effect of other elements is primarily exhibited in their influence on the matrix structure. The matrix effects on toughness are noted (Ref 36, 38) in Fig. 23 and 24. In general, higher manganese, molybdenum, and copper contents, and the presence of tin, lower the toughness (Ref 39). Nickel is often added to lower the ductile-to-brittle transition temperature and increase toughness. Standard ductile iron specifications for low-temperature impact service establish a maximum allowable content for silicon and phosphorus because of their adverse effect on the transition temperature. Minor elements that stabilize carbides can also be detrimental to impact properties. Ductile iron impact properties are significantly affected by heat treatment. A full anneal tends to dissipate segregation and minimize the effect of minor elements. This is because heat treatment

determines the matrix structure, strength, ductility, and ferritic grain size, all of which influence toughness. The highest toughness indicated in Fig. 25 is for an annealed or subcritically annealed ductile iron. The higher-strength, pearliticmatrix irons have a ductile-to-brittle transition temperature in the notched impact test that is above room temperature. A series of ductile iron samples of the same composition were given a variety of heat treatments (Ref 40). For a higher-strength iron, a microstructure of martensite, tempered so as to contain spheroidized carbides, provides somewhat better impact properties than does a pearlitic microstructure of similar hardness. The J-series samples (Fig. 25) were quenched from a temperature just above the critical range to produce a matrix of martensite with a minimum carbon content. When tempered to a spheroidized condition, this provided good impact properties for ductile iron with yield strength of over 550 MPa (80 ksi) and may be compared to the low-temperature impact properties of ductile irons that are fully annealed. The C-series samples were quenched and tempered in a conventional manner. The G samples were also of a lower-carbon martensite but were tempered to a higher hardness and strength. Hardened ductile irons, particularly those with high silicon and phosphorus contents are subject to embrittlement when tempered in the temperature range of 350 to 500  C (660 to 935  F) (Ref 41). Avoiding this chemistry or adding a small amount (0.05%) of molybdenum will reduce the effect. Tempering for shorter times, 1/2 to 1 h compared to 12 to 24 h, increases toughness at equivalent strength ranges. Little difference in impact resistance results from stress relieving as-cast ferritic or ferriticpearlitic ductile irons. Increasing the test or service temperatures will generally increase ductile iron impact properties (Ref 42). Whereas ferritic ductile irons display significantly higher impact

Ductile Iron Castings / 867 Table 9 Fracture toughness values for ductile iron Yield strength Type of iron

Ferritic 3% Si

Ferritic(a) 1.55 Si, 1.5 Nl. 0.12 Mn Ferritic(b)

Pearlitic(b) 0.5 Mo 80-60-03

D7003

Ni-Resist D-5B Sample(a)

A B C J G E

F(b)

Heat treatment

As-cast Subcritical annealed (700  C, or 1300  F) Quenched and tempered (quenched: 900  C, or 1650  F; tempered: 680  C, or 1250  F) Quenched and tempered (quenched: 860  C, or 1580  F; tempered: 650  C, or 1200  F) Quenched and tempered (quenched: 860  C, or 1580  F; tempered: 480  C, or 900  F) Normalized and tempered (normalized: 900  C, or 1650  F; tempered: 635  C, or 1175  F) Austempered (370  C, or 700  F) Tensile strength

Yield strength

Sample(a)

MPa

ksi

MPa

ksi

Elongation in 50 mm (2 in.), %

Hardness, HRC

A B C J G E F(b)

535 430 635 750 1050 780 930

78 62 92 108 152 113 135

340 315 525 580 795 470 655

49 49 76 84 115 68 95

13.0 23.6 8.8 9.4 4.1 8.2 11.5

7 1 19 18 29 21 ...

(a) Analysis: 3.65% total carbon, 0.52% Mn, 0.065% P, 2.48% Si, 0.08% Cr, 0.78% Ni, 0.15% Cu. (b) Analysis not available; data from unpublished work conducted at International Harvester Company

Fig. 25

Effect of heat treatment on Charpy V-notch impact properties of ductile iron. Souce: Ref 40

strength than pearlitic irons, their maximum impact value is attained at a relatively moderate temperature and decreases with a further increase in temperature. Pearlitic ductile irons, on the other hand, achieve their maximum impact strength at approximately 200  C (400  F) and maintain that level to approximately 500  C (900  F). The difference in impact strength between ferritic and pearlitic ductile irons is reduced with increasing temperature (Ref 42). Fracture of Sharply Notched Specimens. The toughness of ductile iron can also be

Temperature 

C



F

KIcðaÞ pffiffiffiffi pffiffiffiffi MPa m ksi in.

MPa

ksi

...

...

40

40

35.2

32.0

... ... 269 310 324

... ... 39 45 47

107 160 107 160 24 75 55 67 73 100

30.3 46.0 42.8 48.3 59.3

27.5 41.8 38.9 43.9 53.9

331 372 385 483 493 503 432 458 476 717 727 740 324 325

48 54 55.8 70 71.5 73 62.6 66.4 69.0 103.9 105.4 107.3 47.0 47.1

24 75 55 67 73 100 24 75 12 10 55 67 24 75 19 2 48 54 24 75 19 3 59 75 24 75 59 75

48.3 61.5 53.8 48.3 50.5 22.0 27.1 25.4 26.2 51.7 47.9 50.5 64.1 67.1

43.9 55.9 48.9 43.9 45.9 20.0 24.6 23.1 23.8 47.0 43.5 45.9 58.3 61.0

(a) All samples ware 25 mm (lin.) thick except for the 3% Sl Iron. These ware, respectively, 21.1, 21.1, and 31.8 mm (0.83, 0.83, and 1.25 in.) thick. (b) Composition was 3.6% C, 2.5% Sl. 0.38% Ni, 0.35% Mo, unless shown otherwise. Source: Ref 43–47

evaluated by special fracture mechanics tests in which all or a significant part of the fracture is brittle. This method can be directly related to service in some cases. It permits an estimate of the size of crack that will cause catastrophic brittle fracture (critical crack size) for a given service stress condition and material toughness, KIc. For a service stress below the yield stress of the metal, the critical crack size approximately equals (KIc/YS)2, where YS is the yield strength. The experimental procedure for measuring the critical stress intensity, KIc, using a linear elastic fracture mechanics approach has been specified in ASTM E 399. Where significant plastic deformation precedes fracture, E 399 requires the use of very large specimens, which is impractical. More recently, two elastic-plastic analyses have been developed for use on materials that experience significant ductility prior to fracture. These approaches, crack opening displacement and J-integral, have been used along with E 399 to determine KIc values for various ductile irons. These results are compiled in Table 9, and only valid plane-strain KIc values have been included (Ref 43–47). Both the graphite and matrix structure influence the fracture toughness. More spherical graphite increases the toughness of ferritic ductile irons. Increasing the nodule count allows a completely ductile fracture and therefore higher KIc value to occur at low temperatures near 75  C (100  F). However, no significant effect of nodule count is noted at higher temperatures. Calculations of the critical defect size that will result in catastrophic failure at a given stress level show that larger defects can

be tolerated in ferritic than in pearlitic irons (Ref 45). The critical crack size for ferritic ductile iron is approximately 25 mm (1 in.), even at temperatures as low as 45  C (50  F). This is quite favorable in comparison to other materials and proves that ferritic ductile iron has excellent toughness for its strength. The higher-strength, heat treated ductile irons have smaller critical crack sizes and lower fracture toughness values (Ref 43, 48, 49). Fracture toughness is improved by higher carbon (Ref 45) and lower silicon and phosphorous contents. (Ref 45, 48). The effect of silicon is to increase the transition temperature and uppershelf KIc values, although the upper-shelf critical crack size remains unchanged (Ref 47). The Charpy values of 14 to 20 J (10 to 15 ft lbf) are usually observed for ferritic ductile irons. The blunt V-notched Charpy specimen obscures the benefits of the coarse microstructure of ductile iron on the fracture toughness. However, fatigue precracked Charpy results compared to those using a standard Charpy specimen show only a slight decrease in values for ferritic ductile iron, while significantly lower results are observed for carbon steels. Therefore, the comparison of carbon steel to ductile iron using critical stress-intensity values (KIc) normalized to their respective yield strengths (KIc/YS)2 show a much smaller difference in toughness than the usual impact test results.




Physical Properties The physical properties of all cast irons were discussed at some length in the article “Introduction to Cast Irons” in this Volume. A summary of ductile and austempered ductile properties is included here. Among the physical properties of ductile iron, density is affected only by the relative amounts of the microconstituents present, not by their form or distribution, whereas thermal and electrical conductivity are markedly affected by the form and distribution of the various phases, especially graphite, which has properties very different from those of the various metallic phases. Density. For most ductile irons, density at room temperature is approximately 7.1 g/cm3 (0.26 16/in.3). Density is largely affected by carbon content and by the degree of graphitization and any amount of microporosity. A lowcarbon pearlitic ductile iron may have a density as high as 7.4 g/cm3 (0.27 lb/in.3), and a highcarbon ferritic ductile iron may have a density as low as 6.8 g/cm3 (0.25 lb/in.3). High-nickel compositions, such as the NiResist ductile irons, have slightly greater density, approximately 7.4 to 7.7 g/cm3 (0.27 to 0.28 lb/in.3). Thermal Properties. Specific heat of ductile iron is relatively unaffected by composition, but it will vary with the temperature range considered for unalloyed ductile iron:

868 / Cast Irons Temperature 

Specific heat



C

20–200 20–300 20–400 20–500 20–600 20–700

F

J/kg

70–390 70–570 70–750 70–930 70–1110 70–1290


(K)

Btu/lb

461 494 507 515 536 603






F

0.110 0.118 0.121 0.123 0.128 0.144

The melting point varies with silicon content, and the melting range varies with carbon content. The closer the value of the carbon equivalent is to the eutectic composition, the narrower the melting range. Unalloyed and low-alloy ductile irons melt in the range of 1120 to 1160  C (2050 to 2120  F); austenitic highnickel ductile irons melt at approximately 1230  C (2250  F). The heat of fusion for all ferritic and pearlitic grades of ductile iron is approximately 210 to 230 kJ/kg (90 to 99 Btu/lb). The coefficient of linear thermal expansion (CTE), like specific heat, is given as a constant over a given temperature range, even though the CTE actually varies with temperature. Because it varies with temperature, different values of expansion coefficient are needed Table 10 1563 cal











Some physical properties expected of ductile iron grades covered by BS EN

Thermal conductivity, 100  C (212  F)

Grade

for different temperature ranges. The value of the expansion coefficient is ordinarily determined by measuring the change in length of a standard-sized specimen as it is heated and cooled over a specific temperature range. Coefficients of linear thermal expansion for ferritic and pearlitic ductile irons are given in Tables 10 and 11. Values for highly alloyed ductile irons may be considerably different from those given. The CTE of austenitic ductile (Ni-Resist) irons for 20 to 205  C (70 to 400  F) varies from 14.4 to 18.7  106/ C (8.0 to 10.4  106/ F). The coefficient of linear thermal expansion for austempered ductile iron is related to the proportion of ferrite and retained austenite in the austempered ductile iron. For A879 grades, the CTE varies from 13.5 to 14.6  106/ C (7.5 to 8.1  106/ F) for an unspecified temperature range. The thermal conductivity of ferritic ductile iron is approximately 36 W/m K (250 Btu in./ft2 h  F) over the temperature range of 20 to 500  C (70 to 930  F). There is a slight, negative temperature dependence for thermal conductivity, causing it to drop with increasing temperature. Graphite shape and alloy content have greater influences on thermal conductivity than does temperature; for instance, increasing

Ferritic grades 350/22; 350/22L40 400/18; 400/18L20 420/12







W/m K (Btu/ft  F) at 500  C (930  F)

Specific heat capacity, 20–700  C (68–1290  F) J/kg


K

Btu/lb






F

Coefficient of thermal expansion at 20–400  C (68–750  F), 106/K

Electrical resistivity, mV/m

36.5 (21.1)

35.8 (20.7)

603

0.144

12.5

0.500

36.5 (21.1)

35.8 (20.7)

603

0.144

12.5

0.500

36.5 (21.1)

35.8 (20.7)

603

0.144

12.5

0.500

Intermediate grades 450/10 36.5 (21.1) 500/7 36.5 (21.1) 600/3 32.8 (18.9)

35.8 (20.7) 34.9 (20.2) 32.2 (18.6)

603 603 603

0.144 0.144 0.144

12.5 12.5 12.5

0.500 0.510 0.530

Pearlitic as-cast and normalized grades 700/2 31.4 (18.2) 30.8 (17.8) 800/2 31.4 (18.2) 30.8 (17.8) 900/2 31.4 (18.2) 30.8 (17.8)

603 603 603

0.144 0.144 0.144

12.5 12.5 12.5

0.540 0.540 0.540

Hardened and tempered grades 700/2 33.5 (19.4) 800/2 33.5 (19.4) 900/2 33.5 (19.4)

603 603 603

0.144 0.144 0.144

12.5 12.5 12.5

>0.540 >0.540 >0.540

32.9 (19.0) 32.9 (19.0) 32.9 (19.0)

the nickel-plus-silicon content of ferritic ductile iron reduces thermal conductivity. The graphite shape does have a differing influence on the thermal and electrical conductivity. As show in Fig. 26, normalizing to steel, gray irons with a flake graphite have high thermal conductivity and lower electrical conductivity. As the graphite becomes spherical in ductile iron, both properties approach the steel value. Electrical Conductivity. Ductile irons have higher electrical conductivity and lower thermal conductivity than gray irons. The influence of the matrix structure and silicon content on electrical resistivity (the reciprocal of conductivity) is seen in Fig. 27. Magnetic properties should be determined separately for each application in which they are important. Although the properties of cast irons do not approach those of permanent magnet alloys on the one hand, or of silicon steels on the other, cast irons often are used for parts that require known magnetic properties. Cast iron can be cast into intricate shapes and sections more easily than most permanent magnet alloys. The magnetic properties of cast irons depend largely on structure; the influence of alloying elements is indirect, derived mainly from the influence exerted on structure. A ferritic structure exhibits low hysteresis loss and high permeability, whereas a pearlitic structure exhibits high hysteresis loss and low permeability. Free cementite produces an iron with low permeability, magnetic induction, and remanence, together with high coercive force and hysteresis loss. Spheroidal graphite slightly increases hysteresis loss in pearlitic irons but considerably reduces hysteresis loss in ferritic irons. Low-phosphorus ferritic ductile iron that is essentially free of cementite and very low in combined carbon has the highest permeability and lowest hysteresis loss of all cast irons.

(a) These tests were reported in 1986 for irons specified by BS 2789, which was replaced by BS EN 1563. ISO 1083 is not equivalent.

Table 11

Coefficients of thermal expansion for ferritic and pearlitic ductile irons Coefficient of thermal expansion

Temperature range 

C

20–110 20–200 20–300 20–400 20–500 20–600 20–700 20–760 20–870



Ferritic ductile iron

F

70–230 70–390 70–570 70–750 70–930 70–1110 70–1290 70–1400 70–1600

mm/m


K

11.2 12.2 12.8 13.2 13.5 13.7 13.8 14.8 15.3

min./in.




6.23 6.78 7.12 7.34 7.51 7.62 7.67 8.23 8.51

Pearlitic ductile iron



F

mm/m


K

10.6 11.7 12.4 13.0 13.3 13.6 ... 14.8 15.3

min./in.




5.89 6.51 6.89 7.23 7.39 7.56 ... 8.23 8.51



F

Fig. 26

Effect of graphite shape on the thermal and electrical conductivity of gray and ductile irons relative to steel. Source: Ref 50

Ductile Iron Castings / 869

Austempered Ductile Iron (ADI)

Fig. 27

Influence of matrix structure and silicon content on the electrical resistivity of ductile iron at room temperature. Source: Ref 50

Fig. 28

Tensile properties of austempered ductile iron (ADI) and conventional ductile iron based on minimum values complying to ASTM International standards. Source: Ref 10

In alloyed ductile irons, copper and nickel decrease permeability and increase hysteresis loss. When nickel is present in sufficient quantities to produce an austenitic matrix, ductile iron becomes paramagnetic. Manganese and chromium reduce magnetic induction,

permeability, and remanent magnetism and increase hysteresis loss. Silicon has little effect on magnetic properties in pearlitic ductile iron but slightly increases maximum permeability and reduces hysteresis loss in ferritic ductile iron.

When compared to conventional ductile iron, ADI exhibits over twice the strength for a given level of ductility (Fig. 28). Austempered ductile iron is 10% less dense than steel; therefore, for a given shape, the part will weigh 10% less. This has led to the substitution of ADI for parts traditionally made of plain carbon and lowalloy steels in weight-sensitive applications. Furthermore, ADI is three times as strong as the best forged or cast aluminum, and it weighs only 2.4 times as much. As previously discussed, the ausferrite microstructure of ADI is more wear resistant than its bulk hardness would indicate. When considering component weight required as a function of yield strength, ADI is very competitive with most common engineering materials (Fig. 29). Table 2 shows the typical mechanical properties of ADI. Although the Charpy impact resistance of ADI is lower than a comparable forged steel, its fracture toughness will generally be very competitive. Additionally, the design engineer should know that ausferrite microstructures respond to surface finishing (shot peening, fillet rolling, and grinding) with vary large increases in surface compressive stresses. Not only does ausferrite undergo a mechanical deformation, the stresses are further increased by a strain transformation that takes place when the ausferrite is exposed to high normal forces. This results in significant increases in bending strength. Applications. The development and commercialization of ADI has provided the design engineer with a new group of cast ferrous materials that offer the exceptional combination of mechanical properties equivalent to cast and forged steels, and production costs similar to those of conventional ductile iron. In addition to this attractive performance-to-cost ratio, ADI also provides the designer with a wide range of properties, all produced by varying the heat treatment of the same castings, ranging from 10% elongation with 870 MPa (125 ksi) tensile strength to 1750 MPa (250 ksi) with 1 to 3% elongation. Austempered ductile iron has become an established alternative in many applications that were previously the exclusive domain of steel castings, forgings, weldments, powdered metals and aluminum forgings and castings. Austempered ductile iron components are now evident in nearly every manufacturing segment. ACKNOWLEDGMENTS We acknowledge the support of the American Foundry Society (AFS) and the work of George M. Goodrich in adapting portions of Iron Casting Engineering Handbook (ISBN 0-87433260), 2006, edited by him and revised by the AFS Cast Iron Division, 5A Committee. The AFS may be contacted at www.afsinc.org.

870 / Cast Irons

Fig. 29

Weight-to-yield-strength ratio for various structural metals. Ratio for forged steel taken as 1.0. ADI, austempered ductile iron. Courtesy of Applied Process, Inc.

Some portions of this article were adapted from “Ductile Iron,” revised by Lyle R. Jenkins of the Ductile Iron Society and R.D. Forrest of Pechiney Electrometalluie (Ref 51), and from the existing article (Vol 15) by I.C.H. Hughes, BCIRA. REFERENCES 1. Metalcasting: Industry of the Future 2002 Annual Report, U.S. Department of Energy, Feb 2003 2. “Ductile Iron Castings,” A 536, Annual Book of ASTM Standards, ASTM International, 2005 3. “Spheroidal Graphite Cast Irons—Classification,” ISO 1083, International Organization for Standardization, 2004 4. “Automotive Ductile (Nodular) Iron Castings,” J434 FEB2004, SAE Ferrous Materials Standards Manual, SAE International, 2004 5. “Spheroidal Graphite Iron Castings,” JIS G5502, Japanese Standards Association, 2001 6. Ductile Iron Pipe Research Association, Birmingham, AL, www.dipra.org, accessed May 2008 7. “Austempered Ductile Iron Castings,” A 897/A 897M-03, Annual Book of ASTM Standards, ASTM International, 2005 8. “Austempered Spheroidal Graphite Iron Castings,” JIS G5503, Japanese Standards Association, 1995 9. “Founding. Austempered Ductile Cast Irons,” BS EN 1564, British Standards Institution, 1997 10. Ductile Iron Data for Design Engineers, QIT America (now Rio Tinto), 1990; available online at www.ductile.org, accessed June 2008 11. J. Zotos, “Evaluation of Cast Shell Alloys Listing Mathematical Models,” CR68-07/1, AMMRC, Watertown, MA

12. H.E. Henderson, Compliance with Specifications for Ductile Iron Castings Assures Quality, Met. Prog., Vol 89, May 1966, p 82–86; Ductile Iron—Our Most Versatile Ferrous Material, Gray, Ductile and Malleable Iron Castings—Current Capabilities, STP 455, American Society for Testing and Materials, 1969, p 29–53 13. G. Jolley and C.N.J. Gilbert, Segregation in Nodular Iron and Its Influence on Mechanical Properties, Br. Foundryman, Vol LX (No. 3), March 1967, p 79–92 14. C. Isleib and R. Savage, Ductile Iron— Alloyed and Normalized, AFS Trans., Vol 65, 1957, p 75 15. G.M. Goodrich, Ed., Iron Casting Engineering Handbook, American Foundry Society, 2003 16. G.N.J. Gilbert, Behavior of Cast Irons Under Stress, Engineering Properties and Performance of Modern Iron Castings, British Cast Iron Research Association, 1970, p 41 17. N.L. Church, “Bainitic Ductile Irons,” Technical Paper 816-T-OP, The International Nickel Company, Suffern, NY, Sept 1972 18. B.V. Kovacs, Quality Control and Assurance by Sonic Resonance in Ductile Iron Castings, AFS Trans., Vol 85, 1977, p 499 19. Dynamic Elastic Properties of Malleable Iron, Malleable Research and Development Foundation, Cleveland, OH, 1966 20. S. Palmer, What Is Required of Iron Castings as Engineering Components?, Engineering Properties and Performance of Modern Iron Castings, British Cast Iron Research Association, Birmingham, England, 1970, p9 21. The Ductile Iron Process, Miller and Company, 1972 22. R.A. Flinn, P.K. Trojan, and D.J. Reese, The Behavior of Ductile Iron Under Compressive Stress, Trans. ASTM, 1960

23. K.B. Palmer and G.N.J. Gilbert, The Fatigue Properties of Nodular Cast Iron, J. Res., Vol 5 (No. 1), Aug 1953, p 2–15 24. G.N.J. Gilbert, Tensile and Fatigue Tests on Normalized Pearlitic Nodular Irons, J. Res., Vol 6 (No. 10), Feb 1957, p 498–504 25. R.C. Haverstraw and J.F. Wallace, Fatigue Properties of Ductile Iron, Gray and Ductile Iron News, Gray and Ductile Iron Founders’ Society, Inc., Aug 1966, p 5–19 26. G.N.J. Gilbert and K.B. Palmer, Tensile and Fatigue Tests on Hardened and Tempered Nodular Irons, J. Res., Vol 5 (No. 11), April 1955, p 604–615 27. A.G. Fuller, Effect of Graphite Form on Fatigue Properties of Pearlitic Ductile Irons, AFS Trans., Vol 85, 1977, p 527 28. E. Moal and M. Briner, The Behavior of Ductile Iron Subjected to Alternating Stresses, Giesserei-Prax., July 1970, p 193 29. M. Sofue, S. Okada, and T. Sasaki, HighQuality Ductile Cast Iron with Improved Fatigue Strength, AFS Trans., Vol 86, 1978, p 173 30. K.B. Palmer, Fatigue Properties of Cast Iron, Engineering Properties and Performance of Modern Iron Castings, British Cast Iron Research Association, Birmingham, England, 1970, p 97 31. G.N.J. Gilbert, The Influence of Surface Rolling on the Fatigue Properties of Flake and Nodular Graphite Cast Irons, J. Res., Vol 5 (No. 8), Oct 1954, p 447–464 32. S. Lampman, Fatigue and Fracture Properties of Cast Irons, Fatigue and Fracture, Vol 19, ASM Handbook, ASM International, 1996, p 665–679 33. “Fracture Resistance of Malleable Iron,” Bulletin 78, Malleable Research and Development Foundation, Dayton, OH, 1969 34. Austempered Ductile Irons:Your Means to Improved Performance, Productivity, and Cost, Proceedings of the First International Conference (Rosemont, IL), ASM International, 1984 35. J.V. Anderson and S.I. Karsay, Pouring Rate, Pouring Time and Choke Design for SG Iron Castings, Br. Foundryman, Vol 78 (No. 10), Dec 1985, p 492–498 36. G. Sandoz, H. Bishop, and W. Pellini, Notch Ductility of Nodular Irons, Trans. ASM, Vol 46, 1954, p 418 37. G. Cox and J. O’Loughlin, Low-Silicon, Nickel-Alloyed Ferritic SG Irons for use at Sub-Zero Temperatures, Founding Trade J., Dec 16–23, 1971, p 833 38. Y. Tanaka and K. Ikawa, “Grain Refining Heat Treatment of Spheroidal Graphite Cast Iron and Its Mechanical Properties,” Paper 5, 44th International Foundry Congress, Sept 1977 39. C. Dinges and W. Gruver, Temper Embrittlement of Ductile Iron, AFS Trans., Vol 74, 1966, p 437

Ductile Iron Castings / 871 40. C. Vishnevsky and J.F. Wallace, The Effect of Heat Treatment on the Impact Properties of Ductile Iron, Gray Iron News, Gray Iron Founders’ Society, July 1962, p 5–10 41. R. Barton, “Recent Work and Experience with Nodular Iron at BCIRA,” Sixth Annual Ductile Iron Conference of the Gray and Ductile Iron Founders’ Society (Cincinnati, OH), March 1966 42. J.A. Dilewijns, Properties of Nodular Iron at Temperatures up to 500F with Emphasis on Impact Strength, Cast Met. Res. J., June 1973, p 64 43. R.K. Nansted et al., Fracture Toughness Testing of Nodular Cast Irons, AFS Trans., Vol 82, 1974, p 473 44. R.K. Nansted et al., Static and Dynamic Toughness of Ductile Cast Iron, AFS Trans., Vol 83, 1975, p 245

45. S.R. Holdsworth and G. Jolley, Fracture Toughness Tests on Pearlitic Spheroidal Graphite Cast Iron, Br. Foundryman, p 77 46. D.C. Veroma, J.T. Berry, and A.A. Iseng, A Study of the Fracture Behavior of Pearlitic Gray Iron, ASME-MPC II Publication, Dec 1979, p 126 47. W.L. Bradley and H.E. Mead, Fracture Toughness Studies of Nodular Iron Using a J-Integral Approach, ASME-MPC II Publication, Dec 1979, p 69–88 48. F.J. Worzola et al., Toughness of Malleable and Ductile Iron, AFS Trans., Vol 84, 1976, p 675 49. A. Lazaridis et al., Fracture Toughness of Ductile Cast Irons, AFS Trans., Vol 79, 1971, p 351 50. C.F. Walton, Ed., Gray and Ductile Iron Casting Handbook, Gray and Ductile Iron Founders’ Society, 1971

51. L.R. Jenkins and R.D. Forrest, Ductile Iron, Properties and Selection: Irons, Steels, and High-Performance Alloys, Vol 1, ASM Handbook, ASM International, 1990, p 33–55 SELECTED REFERENCES  Ductile Iron Handbook, American Foundry-

men’s Society, 1993

 G.M. Goodrich, Ed., Iron Casting Engineer-

ing Handbook, American Foundry Society, 2003

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 872-883 DOI: 10.1361/asmhba0005325

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Compacted Graphite Iron Castings COMPACTED GRAPHITE (CG) IRONS are the newest member of the cast iron family to be developed as a distinct category. Compacted (vermicular) graphite irons have inadvertently been produced in the past as a result of insufficient magnesium or cerium levels in melts intended to produce spheroidal graphite ductile iron (SG); however, it has only been since 1965 that CG iron has occupied its place in the cast iron family as a material with distinct properties requiring distinct manufacturing technologies. The first patent was obtained by R.D. Schelleng (Ref 1). Also known as vermicular graphite, the CG irons are more widely produced in Europe than in the United States or elsewhere. This article reviews the graphite morphology, chemical composition requirements, castability, and properties of CG irons. In general, the property values of CG irons (both mechanical and physical) fall between those of gray flake graphite irons (FG) and ductile irons (SG). This position gives CG a desirable combination of properties for certain applications of high interest. Compact graphite iron has higher strength, stiffness, and fatigue strength than FG with better thermal conductivity and machinability than SG. It also does not need the high alloy content of ductile iron.

article). Similarities between the solidification patterns of flake and compacted graphite iron explain the good castability of the latter compared to that of ductile iron. In addition, the interconnected graphite provides better thermal conductivity and damping capacity than SG.

Chemical Composition Carbon Equivalent. The characteristic properties of CG irons have been demonstrated over a rather wide range of carbon equivalent (CE) values, extending from hypoeutectic

Graphite Morphology The shape of compacted graphite is rather complex. An acceptable CG iron is one in which there is no flake graphite in the structure and for which the amount of spheroidal graphite is less than 20%; that is, 80% of all graphite is compacted (vermicular) (ASTM A 247, type IV or similarly form 2, JIS G 5502 or form II, ISO 945(E). Typical CG iron microstructures are shown in Fig. 1. It can be seen that although the two-dimensional appearance of compacted graphite is that of flakes with a length-to-thickness ratio of 2 to 10 (Fig. 1a), the three-dimensional SEM structures (Fig.1b,c,d) show that graphite does not appear in flakes but rather in clusters interconnected within the eutectic cell. This graphite morphology allows better use of the matrix, yielding higher strength and ductility than gray irons containing flake graphite (gray irons are referred to as FG irons in this

Fig. 1

Typical microstructures of compacted graphite (CG) irons. (a) Light micrograph. Etched with nital. (b) Tensile load fracture surface. Ion bombardment etched. SEM. Original magnification: 65. (c) and (d) True shape of graphite in CG irons revealed in deep etch. SEM. Original magnification: 395

Compacted Graphite Iron Castings / 873 (CE = 3.7) to hypereutectic (CE = 4.7), with carbon contents of 3.1 to 4.0% and silicon in amounts of 1.7 to 3.0% (Ref 2, 4). At constant silicon levels, a lower CE slightly increases the chilling tendency and results in lower nodularity. At constant CE, higher silicon increases nodularity (Ref 5). The optimal carbon and silicon contents can be selected from Fig. 2. The optimum CE must be selected as a function of section size. For a given section size, too high a CE will result in graphite flotation, as in the case of spheroidal graphite cast iron, while too low a CE may result in increased chilling tendency. For wall thicknesses ranging from 10 to 40 mm (0.4 to 1.6 in.), eutectic composition (CE = 4.3) is recommended in order to obtain optimum casting properties. Manganese and Phosphorus Contents. The manganese content of CG iron can vary

Fig. 2

Optimal range for carbon and silicon composition for compacted graphite (CG) iron. Source. Ref 5

between 0.1 and 0.6%, depending on whether a ferritic or pearlitic structure is desired. The phosphorus content should be less than 0.06% in order to obtain maximum ductility from the matrix. Sulfur Content. Although CG iron has been produced from base irons having sulfur contents as high as 0.07 to 0.12% (Ref 6), it is probably more economical to desulfurize the iron to a level of 0.01 to 0.025% before liquid treatment. The higher the sulfur content, the more alloy is required for melt treatment. Also, the risk of missing the composition window for CG iron is increased, because the residual treatment elements must be balanced with the residual sulfur. Typical residual sulfur levels after treatment are 0.01 to 0.02% (Ref 4). Melt Treatment Elements. The change in graphite morphology from the flake graphite in the base iron to the compacted graphite in the final iron is achieved by liquid treatment with different minor elements. These elements may include one or more of the following: magnesium, rare earths (cerium, lanthanum, praseodymium, etc.), calcium, titanium, and aluminum. The amounts and combinations to be used are a function of the method of liquid treatment, base sulfur, section thickness, and so forth. For example, Fig. 3 shows a typical in-mold treatment method where SG iron flows over granular magnesium ferrosilicon alloy, to provide inoculation and the proper amount of magnesium, and is then filtered. Alloying elements, such as copper, tin, molybdenum, and even aluminum, can be used to change the as-cast matrix of CG iron from ferrite to pearlite. Typical ranges are 0.48% Cu or 0.033% Sn (Ref 7), 0.5 to 1% Mo (Ref 8), and up to 4.55% Al (Ref 9, 10). Regardless of the alloying elements used, the high ferritizing tendency of CG iron should be taken into account. Figure 4 shows that even strong pearlite promoters such as copper or tin show reduced effectiveness in CG iron (Ref 10). The same is true when treating with cerium-mischmetal. A fully pearlitic structure could not be obtained even at 1.7% Cu when using high-purity charge

materials. Tin, however, is effective. As shown in Fig. 5, 95% pearlite was obtained with 0.13% Sn. To produce a pearlitic matrix, therefore, it may be necessary to add higher-thanusual levels of alloying elements or to make multiple additions of two or three elements.

Castability The fluidity of molten cast iron is broadly governed by carbon and silicon contents and temperature. In addition, solidification morphology plays a role (Ref 11). Flake graphite iron has the best fluidity and SG iron the worst. As expected, CG iron occupies an intermediate position. However, because CG irons are stronger than FG irons with the same CE, a higher CE can be used to obtain the same strength, allowing greater fluidity and easier pouring of thin sections. Shrinkage Characteristics. Obtaining sound castings in CG irons, free from external and internal shrinkage porosity, is easier than in SG irons and slightly more difficult than in FG irons. This is because the tendency for mold wall movement to occur lies between that of SG and FG irons. Figure 6 shows the results of experiments that demonstrate these effects using a 76 mm (3 in.) diameter spherical test (Ref 3). For cast iron of eutectic composition cast in sand molds, it is reported that the shrinkage volume is 4.1% for FG, 4.8% for CG, and 7.0% for SG irons (Ref 12). A higher-grade FG iron will have a shrinkage volume very close to that of CG iron. In relative numbers, solidification expansion has been found to be 4.4 for SG iron and 1 to 1.8 for CG iron if FG iron is 1 (Ref 13). Because of the rather low shrinkage of CG iron, it can be cast riserless. Expensive pattern changes are therefore not necessary when converting from FG iron to CG iron, because the

Fig. 4

Fig. 3

Placement of alloy reaction chamber and filter for in-mold treatment. Ductile base iron flows over granular magnesium ferrosilicon alloy, inoculating the melt.

Effect of copper, nickel, and tin on the type of matrix in the composition range between compacted graphite and flake graphite iron of 25 mm (1 in.) wall thickness. Source: Ref 10

874 / Cast Irons such loading are stress raisers that include some defects or abrupt changes in section thickness. The elongation values of approximately 1% obtainable with high-strength FG iron are insufficient for certain types of applications, such as diesel cylinder heads (Ref 16). Compacted graphite irons have strength properties close to those of SG irons, at considerably higher elongations than those of FG irons, and with intermediate thermal conductivities. Consequently, they can successfully outperform other cast irons in a number of applications. At room temperature and elevated temperatures, the main factors affecting the mechanical properties of CG irons are:  Composition  Structure (nodularity and matrix)  Section size

Fig. 7

Influence of graphite shape on the chilling tendency of cast irons. Type A, flake graphite with uniform distribution and random orientation. Type D, flake graphite with interdendritic segregation and random orientation. Source: Ref 15

Fig. 5

Effect of tin on pearlite content and tensile properties of as-cast compacted graphite iron of 25 mm (1 in.) wall thickness. Source: Ref 10

Fig. 6

Comparison of volume shrinkage and dimensional change for flake graphite (FG), compacted graphite (CG), and spheroidal graphite (SG) poured into green sand molds. A 76.4 mm (3 in.) diameter mold was used. Source: Ref 3

same gating and risering techniques can be applied. Riserless casting produces higher casting yields. Yields for large castings vary between 65 and 75%, depending on casting complexity and weight (Ref 14). Dross Formation. Magnesium, cerium, and calcium are dross-forming elements, and the problem of dross exists in CG irons just as it

does in SG irons. Lower magnesium contents should be an advantage, but the best conditions for producing dross-free CG irons are the same as for SG irons: low CE composition, low sulfur content before treatment, treatment alloy low in calcium, high pouring temperature, adequate slag traps in the gating system, and smooth, nonturbulent pouring. Chilling Tendency. When the superheated metal contacts the cold mold wall, a supercooled zone is created at the interface. Impurities in the liquid or mold wall are nucleation catalysts, and a chill zone, evidenced by small grain size in the rapidly solidified metal, is created. Although many believe that the chilling tendency of CG iron is also intermediate between that of FG (lowest) and SG (highest) irons, this is not true. Figure 7 shows the influence of nodularity on the structure of chill pins cast in air set molds (Ref 15). It can be seen that the highest chilling tendency is achieved for irons with nodularities between 6 and 64%. In other words, the chilling tendency of CG iron is higher than that of both SG and FG iron. This correlates with cooling curve data and is explained by the combination of a low nucleation rate and low growth rate occurring during the solidification of CG iron.

Mechanical Properties and Corrosion Resistance The in-service behavior of many structural parts is a function of not only their mechanical strength but also their deformation properties. Thus, it is not surprising that many castings fail not because of insufficient strength but because of a low capacity for deformation. This is especially true under conditions of rapid loading and/or thermal stress. Particularly sensitive to

In turn, the structure is heavily influenced by processing variables such as the type of raw materials, preprocessing of the melt (superheating temperature, holding time, desulfurization), liquid treatment (graphite compaction and postinoculation), and processing after solidification (shakeout time, cooling rate, and heat treating). Typical mechanical and physical properties for CG iron are seen in Table 1. The effect of the matrix structure (ferritic and pearlitic) on the properties is given.

Tensile Properties and Hardness Table 2 lists properties of CG irons in accordance with ASTM A 842, and Table 3 shows ISO 16112 grades. Because CG iron can be tailored with combinations of properties suited for the automotive industy, automotive manufacturers have established in- house standards. Compacted graphite irons exhibit linear elasticity for both pearlitic and ferritic matrices, but to a lower limit of proportionality than SG iron (Fig. 8). The ratio of yield strength to tensile strength ranges from 0.72 to 0.82, which is higher than that for SG iron of the same composition. This makes possible a higher loading capacity. The limit of proportionality is 125 MPa (18 ksi) for both ferritic and pearlitic CG irons, which is slightly lower than that of SG iron. This can be explained by the higher notching effect from the sharper-edged morphology of the compacted graphite compared to spheroidal graphite. Consequently, for the same stress, plastic deformation occurs sooner around the graphite in CG iron than in SG iron, causing earlier divergence from proportionality (Ref 14). As the hardness increases, tensile strength increases but elongation decreases, the effect of a higher pearlite/ferrite ratio (Fig. 9). The graph in Fig. 10 compares hardness with tensile strength for SG, CG, and FG irons. The average ratio of tensile strength to hardness for FG iron is 1.3, compared with 2.08 for CG iron and 2.75 for SG iron. The intermediate behavior of CG iron and the apparent effect of sulfur content

Compacted Graphite Iron Castings / 875 Table 1 Typical mechanical and physical properties of compacted graphite iron at 20  C (68  F) Property

Ferritic

Pearlitic

Ultimate strength, MPa (ksi) 0.2% yield strength, MPa (ksi) Elastic modulus, GPa (106 psi) Elongation, % Poisson’s ratio

>300 (>43.5) >200 (>29) 140–155 (20–22.5) 2–5 0.27

>400 (>58) >300 (>43.5) 145–160 (21–23.2) 0.5–2 0.26

Compression properties Strength limit, MPa (ksi) 0.2% yield strength, MPa (ksi) Elastic modulus, GPa (106 psi) Brinell hardness, 10/3000

>500 (>72.5) >290 (>42) 140–155 (20–22.5) 150–190

>1000 (>145) >415 (>60) 145–160 (21–23.2) 190–260

>175 (>25) 0.55 1.30–1.60

>200 (>29) 0.45 1.40–1.70

Tension-compression limit, MPa (ksi) Three-point bending limit, MPa (ksi) Fully reversed torsion limit, MPa (ksi)

>110 (>16) >150 (>22) >140 (>20)

>175 (>25) >200 (>29) >155 (>22.5)

Toughness values Notched Charpy impact, J (ft lbf) Unnotched Charpy impact J (ft lbf) Bending strength MPa (ksi) Shear strength, MPa (ksi)

3–5 (2.2–3.7) 15–20 (11–14.8) >600 (>87) >275 (>40)

2–4 (1.5–3.0) 7–10 (5.2–7.4) >700 (>102) >325 (>47)

38–42 (22–24) 11.5–13 (6.4–7.2) ... >125 (>18) >50 (>7.3) 7.0–7.1 5.2–5.4 (3.2–3.4) 0.45 (108  10 3) 80,000 (34) 50 (20)

35–38 (20–22) 11–13 (6.1–7.2) >200 (>29) ... ... 7.0–7.1 5.2–5.4 (3.2–3.4) 0.45 (108  10 3) 80,000 (34) 50 (20)

Tensile properties

Fatigue properties Rotating bending limit, unnotched, MPa (ksi)

 

Endurance ratio Fatigue strength reduction factor







Physical properties Thermal conductivity, 20–100  C (68–212  F), W/m. C (Btu/ft h.  F) Thermal expansion coefficient, 20–400  C (68–750  F), mm/m.  C (in./in. 1% creep strain limit, 10,000 at 350  C (660  F), MPa (ksi) at 400  C (750  F), MPa (ksi) at 500  C (930  F), MPa (ksi) Density, g/cm3 Ultrasonic velocity, km/s (miles/s) Specific heat capacity, 100  C (212  F), J/g.K (Btu/lb  F) Enthalpy, 26.6–200  C (79.9–392  F), J/kg (Btu/lb) Specific electrical resistance, 20  C (68  F), mO.cm (mO.in.)









F)




Table 2

Fig. 8

Stress-strain curves of pearlitic and ferritic compacted graphite iron castings. Straight line indicates modulus of elasticity 144 GPa (20.9  106 psi). Source: Ref 14

Properties of compacted graphite irons in accordance with ASTM A 842 Minimum tensile strength

Fig. 9

Correlation among hardness, tensile strength, and elongation in compacted graphite irons. Source: Ref 17

Minimum yield strength

Grade(a)

MPa

ksi

MPa

ksi

Elongation in 50 mm (2 in.), %

Hardness(b), HB

BID(b), mm

250(c) 300 350 400 450(d)

250 300 350 400 450

36.3 43.5 50.8 58.0 65.3

175 210 245 280 315

25.4 30.5 35.5 40.6 45.7

3.0 1.5 1.0 1.0 1.0

179 max 143–207 163–229 197–255 207–269

4.50 min 5.0–4.2 4.7–4.0 4.3–3.8 4.2–3.7

(a) Grades are specified according to the minimum tensile strength requirement given in MPa. Values in ksi are not part (b) Hardness range is nonmandatory information. Brinell impression diameter (BID) is the diameter (in mm) of the impression of a 10 mm diam ball at a load of 3000 kgf. (c) The 250 grade is a ferritic grade. Heat treatment to attain required mechanical properties and microstructure shall be the option of the manufacturer. (d) The 450 grade is a pearlitic grade usually produced without heat treatment with addition of certain alloys to promote pearlite as a major part of the matrix.

Table 3 Mechanical and physical properties of ISO 16112-grade compacted (vermicular) graphite iron Grade

GJV-300 GJV-350 GJV-400 GJV-4500 GJV-500




Tensile strength, MPa

Yield strength, MPa

Elongation, %

Modulus of elasticity, GPa

Thermal conductivity, W/m K

300–375 350–425 400–475 450–525 500–575

210–260 245–295 280–330 315–365 350–400

2.0–5.0 1.5–4.0 1.0–3.5 1.0–2.5 0.5–2.0

130–145 135–150 140–150 145–155 145–160

47 43 39 38 36




For all grades: Poisson’s ratio = 0.26, CTE = 11  10 6/K, specific heat capacity = 0.475 J/g K (100 C)

are worth noting. In general, CG iron has lower hardness than an FG iron of equivalent strength because of the higher amount of ferrite in the structure. For the same elongation, CG iron has considerably less yield strength than SG iron. Effect of Composition. The tensile properties of CG irons are much less sensitive to variations in CE than those of FG irons. Even at CE near the eutectic value of 4.3, both pearlitic and ferritic CG irons have higher strengths than does low-CE, high-duty, unalloyed FG cast iron (Fig. 11). An increase in the silicon content up to 2.6% benefits strength and hardness in both the as-cast and annealed conditions (Fig. 12). This is true even though the matrix becomes more ferritic, because silicon strengthens the ferrite. The same is true in the case of as-cast elongation, because there is a decrease in the amount of pearlite with an increase in silicon, while

876 / Cast Irons

Fig. 10

Tensile strength versus hardness for spheroidal graphite (SG), compacted graphite (CG), and flake graphite (FG) irons. Source: Ref 18

Fig. 11

Effect of carbon equivalent on the tensile strength of spheroidal, compacted, and flake graphite irons cast in 30 mm (1.2 in.) diameter bars. Source: Ref 3

elongation in the annealed condition decreases (Ref 18). Although increasing the phosphorus content slightly improves strength, a maximum of 0.04% P is desirable to avoid lower ductility and impact strength. The pearlite/ferrite ratio, and thus the strength and hardness of CG irons, can be increased by the use of a number of alloying elements, such as copper, nickel, molybdenum, tin, manganese, arsenic, vanadium, and aluminum (Ref 7, 8, 10, Ref 19–21). After annealing to a fully ferritic structure, it is possible to increase the yield point of CG iron by 24% when using 1.5% Ni (Table 4). This is because of the strengthening of the solid solution by nickel (Ref 4). However, additions of copper, nickel, and molybdenum may increase nodularity (Ref 16). Figure 13 shows some correlations of microstructure and properties for iron-carbonaluminum CG irons produced by the in-mold process. Effect of Structure. One of the most important variables influencing the tensile properties of CG irons is nodularity. As nodularity increases, higher strength and elongation are to be expected, as shown in Table 5 and Fig. 14 and 15, although nodularity must be maintained at levels under 20% for the iron to qualify as CG iron. However, spheroidal graphite contents of up to 30% and even more must be expected in thin sections of castings with considerable variation in wall thickness. As previously discussed, the pearlite/ferrite ratio can be increased by using alloying elements. Another way of increasing or decreasing this ratio is by using heat treatment.

Table 4 Effect of heat treatment and alloying with nickel on the tensile properties of compacted graphite iron measured on a 25 mm (1 in.) section size Tensile strength

Yield strength

Heat treatment

Iron matrix(a)

MPa

ksi

MPa

ksi

Elongation, %

Hardness, HB

Nickel, %

As-cast Annealed(b) Normalized(c) As-cast Annealed(b) Normalized(c)

60% F 100% F 90% P ... 100%F 90% P

325 294 423 427 333 503

47.1 42.6 61.3 61.9 48.3 73

263 231 307 328 287 375

38.1 33.5 44.5 47.6 41.6 54.4

2.8 5.5 2.5 2.5 6.0 2.0

153 121 207 196 137 235

0 0 0 1.53 1.53 1.53

(a) F, ferrite; P, pearlite. (b) Annealed, 2 h at 900  C (1650  F), cooled in furnace to 690  C (1275  F), held 12 h, cooled in air. (c) Normalized, 2 h at 900  C (1650  F), cooled in air

Fig. 12

Effect of silicon on mechanical properties of compacted graphite irons produced by the in-mold process. Carbon equivalents range from 4.33 to 4.45. Source: Ref 18

Compacted Graphite Iron Castings / 877

Fig. 13

Effect of aluminum on the structure and mechanical properties of compacted graphite irons produced by the in-mold process. Source: Ref 10

Table 5

Properties of compacted graphite iron as a function of nodularity Tensile strength

Nodularity, %

10–20 20–30 40–50

MPa

ksi

Elongation, %

320–380 380–450 450–500

46–55 55–65 65–73

2–5 2–6 3–6

Thermal conductivity, W/(m

50–52 48–50 38–42


K)

Shrinkage, %

1.8–2.2 2.0–2.6 3.2–4.6

Source: Ref 22

increased nodularity and higher chilling must be considered. This is particularly true for overtreated irons. While it is possible to eliminate the carbides that result from chilling by heat treatment, it is impossible to change the graphite shape, which remains spheroidal with the associated consequences. Other factors influencing the cooling of castings, such as shakeout temperature, can also influence properties. Thin-wall castings (2.5 to 4 mm, or 0.1 to 0.16 in.) are of particular interest to designers of engines in the transportation industry. This tendency of increased nodularity in thin sections can work to the advantage of the design. In engine block design, for instance, the thick section may contain 5 to 15% nodularity, while the thin sections (<4.5 mm, or 0.18 in.) contain 30 to 60% nodularity, depending on gating and cooling rate. The local increased strength and stiffness benefit the product (Ref 23). Modulus of Elasticity. As is evident from Fig. 8 and 18, CG irons exhibit a clear zone of proportionality, both in tension and in compression. In general, the moduli of elasticity for CG iron are similar to those of high-strength FG irons and can be even higher as nodularity increases. The elasticity modulus measured by the tangent method depends on the level of stress, as shown in Fig. 19. A comparison of the stress dependency of the elasticity modulus for different types of cast irons is shown in Fig. 20. Poisson’s ratios of 0.27 to 0.28 have been reported for CG irons (Ref 17).

Compressive and Shear Properties

Fig. 14

The ultimate tensile and yield strengths of 85–100% pearlitic cast irons increase with nodularity. Nodularity values <0% indicate the presence of flake graphite that reduces the mechanical properties significantly. Source: Ref 23

The influence of heat treatment on mechanical properties of CG irons with 20% nodularity is given in Table 4. Studies were conducted on CG iron with fixed 85 to 100% as-cast pearlite and graphite ranging from mixed flake graphite (the negative nodularity in Fig. 14) to 80% nodularity. Positive nodularity shows consistent yield strength and a gradual rise in ultimate tensile strength with increased nodularity. There is a steep drop in these tensile properties when even a small amount of flake graphite is introduced (Ref 23).

Effect of Section Size. Like all other irons, CG irons are rather sensitive to the influence of cooling rate (i.e., to section size), because it affects both the pearlite/ferrite ratio and the graphite morphology. As mentioned previously, a higher cooling rate promotes more pearlite and increased nodularity. Although CG iron is less section sensitive than FG iron (as shown in Fig. 16 for tensile strength), the influence of cooling rate may be quite significant (Fig. 17). When the section size decreases until it is below 10 mm (0.4 in.), the tendency toward

The stress-strain diagram for compression and tensile tests of CG iron is shown in Fig. 18. It can be seen that an elastic behavior occurs up to a compression stress of approximately 200 MPa (30 ksi). Some compressive properties of 179 HB as-cast ferritic CG iron are compared with those of SG iron in Table 6. It can be seen that the 0.1% proof stress in compression for CG iron is 76 MPa (11 ksi) higher than the 0.1% proof stress in tension, while for SG iron the difference is only 23 MPa (3.3 ksi). Compressive strengths up to 1400 MPa (203 ksi) have been reported for ferritic annealed CG irons (Ref 17). Shear Properties. For a pearlitic CG iron, the shear strength on 20 mm (0.8 in.) diameter specimens has been measured at 365 MPa (53 ksi), with a shear-to-tensile strength ratio of 0.97 (Ref 25). Ratios of 0.90 for SG iron and of 1.1 to 1.2 for FG iron have been reported. Materials exhibiting some ductility have ratios lower than 1.0 (Ref 17).

Impact Properties While SG iron exhibits substantially greater toughness at low pearlite contents, pearlitic CG irons have impact strengths equivalent to those of SG irons (Fig. 21). Charpy impact

878 / Cast Irons

Fig. 18

Tensile and compressive stress-strain curves for compacted graphite casting with 4.35 carbon equivalent. 0.1%, 0.2%, and 0.5% yield strengths are indicated. Proportionality limits are 201 MPa (29.1 ksi) in compression and 124 MPa (18 ksi) in tension. Source: Ref 3

Fig. 15

Correlation between nodularity and elongation for ferritic compacted graphite. Source: Ref 24

(Ref 26). In this investigation, matrix cracks were usually initiated in the ferrite by transgranular cleavage (graphite was nearly always surrounded by ferrite), although in some instances intergranular ferrite fracture appeared to be the initiating mechanism. Matrix crack propagation generally occurred by a brittle cleavage mechanism, transgranular in ferrite and interlamellar in pearlite. In general, the impact resistance of CG irons increases with carbon equivalent and decreases with phosphorus or increasing pearlite. As given in Table 7, cerium-treated CG irons seem to exhibit a higher impact energy than magnesium-titanium-treated irons. This may be attributable to TiC and TiCN inclusions present in the matrix of magnesium-titaniumtreated CG irons (Ref 16).

Fatigue Strength Fig. 17

Influence of section size on the tensile strength of ferritic and pearlitic compacted graphite irons with carbon equivalents 3.9 to 4.2%. Source: Ref 3

Fig. 16

Influence of section size on the tensile strength of compacted graphite iron compared to flake graphite iron. Source: Ref 22

energy measurements at 21  C (70  F) and 41  C ( 42  F) have shown that CG irons produced from an SG- based iron absorbed greater energy than those made from FG-based iron (Ref 18). This was attributed to the solute-hardening effects of tramp elements in the FG iron. The results from dynamic tear tests were similar, although greater temperature dependence was observed. A comparison of the dynamic tear energies of CG cast irons is presented

in Fig. 22. There are significant differences in the values obtained in the ferritic condition, but equivalent values are obtained when the matrix structure is primarily pearlitic. Studies on crack initiation and growth under impact loading conditions have shown that, in general, the initiation of matrix cracking is preceded by graphite fracture at the graphitematrix interface, fracture through the graphite, or both. The most dominant form of graphite fracture appears to be that occurring along the boundaries between graphite crystallites

Because the notching effect of graphite in CG iron is considerably lower than that in FG iron, it is expected that CG iron will have higher fatigue strengths. Table 8 lists the fatigue strengths of five CG irons. The as-cast ferritic (>95% ferrite) compacted graphite with a hardness of 150 HB had the highest fatigue strength and the highest fatigue endurance (ratio of fatigue strength to tensile strength). Fatigue properties for three of these CG irons with comparable endurance ratios are shown in Fig. 23. It is evident that pearlitic structures, higher nodularity, and unnotched samples resulted in better fatigue strength. The fatigue endurance ratio was 0.46 for a ferritic matrix, 0.45 for a pearlitic matrix, and 0.44 for a pearlitic higher-nodularity CG iron (Ref 25). With fatigue notch factors (ratio of unnotched to

Compacted Graphite Iron Castings / 879 Sonic and Ultrasonic Testing of Properties

Fig. 19

Stress dependency by elastic modulus for two heat treated compacted graphite (CG) irons. Source: Ref 16

Resonant frequency (sonic testing) and ultrasonic velocity measurements provide reliable methods for verifying the structure and properties of castings. As shown in Fig. 24, ultrasonic velocity is directly related to nodularity. Unfortunately, it is rather difficult to distinguish between CG and low-nodularity SG irons. Better results seem to be obtained when ultrasonic velocity is related to tensile strength. Figure 25 shows the correlation between tensile strength and ultrasonic velocity or resonant frequency for test bars of 30 mm (1.2 in.) diameter. When these tests are applied to castings, the ultrasonic velocity for CG structures is independent of the shape of the casting, but it should be calibrated for the section thickness. Thus, for 30 mm (1.2 in.) diameter bars, the range associated with CG is between 5.2 and 5.45 km/s (3:2 and 3.39 miles/s), but for very large castings such as ingot molds, the ultrasonic velocity for good CG structures lies between 4.85 and 5.10 km/s (3.01 and 3.17 miles/s). Sonic testing, on the other hand, must be calibrated for a particular design of casting, for which examples of satisfactory and unsatisfactory structures must be checked in advance to provide a calibration range (Ref 3).

Corrosion Resistance

Fig. 20

Influence of stress level on the elastic modulus of pearlitic flake, compacted, and spheroidal graphite irons. (a) Tensile stress. (b) Compressive stress. Source: Ref 16

Table 6 Comparison of tensile and compressive properties of compacted graphite (CG) and spheroidal graphite (SG) irons Property

Tensile strength, MPa (ksi) 0.1% proof stress, MPa (ksi) 0.2% proof stress, MPa (ksi) Compressive stress 0.1% proof stress, MPa (ksi) 0.2% proof stress, MPa (ksi)

CG iron

SG iron

380 (55) 246 (35.7) 242 (35.1)

370 (54) 224 (32.5) 236 (34.2)

420 (61) 261 (37.8) 273 (39.6)

322 (46.7) 350 (50)

247 (35.8) 250 (36.3)

284 (41.2) 284 (41.6)

Source: Ref 16

notched fatigue strength) of 1.71 to 1.79, CG iron is almost as notch sensitive as SG iron (>1.85). Gray iron is considerably less notch sensitive, with a notch factor of less than 1.5 (Ref 16). Statistical analysis of a number of experimental data allowed the calculation of a relationship between fatigue strength and tensile strength measured in MPa of CG irons (Ref 16).

Another study (Ref 27) shows similar relative results with higher stress amplitudes (Fig. 23b). A 36% decrease of the alternating bending fatigue strength has been observed on unnotched bars with casting skin compared to machined bars. This compares with a 50% reduction for FG iron of comparable strength and a 32% reduction for ferritic SG iron.

Cast irons are generally good in fresh water and atmospheric conditions. Seawater presents problems due to galvanic action between carbon and the matrix. No grade or type of cast iron will resist all corrosive environments. Many types of cast irons are used in acidic environments. Unalloyed cast irons often behave well in alkalis. Caution should be used when comparing corrosion rates for uniform corrosion, because short-term tests may not reflect long-term performance. Also, localized corrosion, such as graphitic corrosion in which the carbon is selectively leached, or stress-corrosion cracking may be more degrading forms of corrosion. Detailed information on the corrosion resistance of cast irons is available in Ref 28.

Applications The applications of CG irons stem from their intermediate position between FG and SG irons. Compared to FG irons, CG irons have certain advantages:  Higher tensile strength at the same carbon

equivalent, which reduces the need for expensive alloying elements such as nickel, chromium, copper, and molybdenum  Higher ratio of tensile strength to hardness  Much higher ductility and toughness, which result in a higher safety margin against fracture

880 / Cast Irons Modern car and truck engines require that manifolds work at temperatures of 500  C (930  F). At this temperature, FG iron manifolds are prone to cracking, while SG iron manifolds tend to warp. Compacted graphite iron manifolds warp and oxidize less and thus have a longer life. ACKNOWLEDGMENTS This article is adapted from “Metallurgy and Properties of Compacted Graphite Iron,” ASM Specialty Handbook: Cast Irons, which in turn was based on:  D.M. Stefanescu, Compacted Graphite Iron,

Properties and Selection: Irons, Steels, and High-Performance Alloys, Vol 1, ASM Handbook, ASM International, 1990, p 56–70  D.M. Stefanescu, R. Hummer, and E. Nechtelberger, Compacted Graphite Irons, Casting, Vol 15, ASM Handbook (formerly Metals Handbook 9th ed.), ASM International, 1988, p 667–677

REFERENCES

Fig. 21

Effect of pearlite content on the Charpy V-notch impact strength for as-cast compacted graphite (CG) irons compared to spheroidal graphite (SG) iron at room temperature (21  C, or 70  F). Source: Ref 18

 Lower

oxidation and growth at high temperatures  Less section sensitivity for heavy sections Compared to SG irons, the advantages of CG irons are:     

Lower coefficient of thermal expansion Higher thermal conductivity Better resistance to thermal shock Higher damping capacity Better castability, leading to higher casting yield and the capability for pouring more intricate castings  Improved machinability Compacted graphite iron can be substituted for FG iron in all cases in which the strength of FG iron has become insufficient but in which a change to SG iron is undesirable because of the less favorable casting properties of the latter. Examples include bed plates for large diesel engines, crankcases, gearbox housings, turbocharger housings, connecting forks, bearing brackets, pulleys for truck servodrives, sprocket wheels, and eccentric gears.

Because the thermal conductivity of CG iron is higher than that of SG iron, CG iron is preferred for castings operating at elevated temperature and/or under thermal fatigue conditions. Applications include ingot molds, crankcases, cylinder heads, exhaust manifolds, and brake disks. The largest industrial application by weight of CG iron produced is for ingot molds weighing up to 54 metric tonnes (60 tons). According to a number of reports summarized in Ref 29, the life of ingot molds made of CG iron is 20 to 70% longer than the life of those made of FG iron. In the case of cylinder heads, it was possible to increase engine output by 50% by changing from alloyed FG iron to ferritic CG iron (Ref 1). The specified minimum values for cylinder heads are 300 MPa (43 ksi) tensile strength, 240 MPa (35 ksi) yield strength, and 2% elongation. New emission standards and the need for higher-efficiency diesel engines have increased the requirements of cylinder liners to 340 to 400 MPa (49 to 58 ksi) with a modulus of elasticity of 130 to 150 GPa (19 to 22  106 psi) in order to withstand the peak pressures.

1. R.D. Schelleng, Cast Iron (with Vermicular Graphite), U.S. Patent 3,421,886, May 1965 2. E.R. Evans, J.V. Dawson, and M.J. Lalich, Compacted Graphite Cast Irons and Their Production by a Single Alloy Addition, Trans. AFS, Vol 84, 1976, p 215 3. G.F. Sergeant and E.R. Evans, The Production and Properties of Compacted Graphite Irons, Br. Foundryman, Vol 71, 1978, p 115 4. K.P. Cooper and C.R. Loper, Jr., A Critical Evaluation of the Production of Compacted Graphite Cast Iron, Trans. AFS, Vol 86, 1978, p 267 5. H.H. Cornell and C.R. Loper, Jr., Variables Involved in the Production of Compacted Graphite Cast Iron Using Rare EarthContaining Alloys, Trans. AFS, Vol 93, 1985, p 435 6. G.F. Ruff and T.C. Vert, Investigation of Compacted Graphite Iron Using a High Sulfur Gray Iron Base, Trans. AFS, Vol 87, 1979, p 459 7. J. Fowler, D.M. Stefanescu, and T. Prucha, Production of Ferritic and Pearlitic Grades of Compacted Graphite Cast Iron by the In-Mold Process, Trans. AFS, Vol 92, 1984, p 361 8. K.R. Ziegler and J.F. Wallace, The Effect of Matrix Structure and Alloying on the Properties of Compacted Graphite Iron, Trans. AFS, Vol 92, 1984, p 735 9. E. Nechtelberger, R. Hummer, and W. Thury, Aluminum- Alloyed Cast Iron with Vermicular Graphite, GiessereiPrax., Vol 24, 1970, p 387

Compacted Graphite Iron Castings / 881

Fig. 22

Dynamic tear energy versus temperature from 100 to 100  C ( 150 to 212  F). (a) compacted graphite (CG) iron. (b) spheroidal graphite (SG) iron. Source: Ref 18

Table 7 Impact toughness of a cerium-treated compacted graphite (CG) cast iron and two magnesium-titanium-treated CG cast irons Structural condition and graphite type

Iron

Cerium-treated

>95 % ferrite (as-cast); 95% CG, 5% SG

Magnesium (0.018%) and 100% ferrite (annealed); CG titanium (0.089%) treated Magnesium (0.017%) and Ferritic (as-cast); CG titanium (0.062%) treated

Test temperature 

C



F

Impact bend toughness(a) J

Notched-bar(b) impact toughness




ft lbf




J

ft lbf

20 20 40 20

68 4 40 68

32.1 23.7 26.5 19.5 26.7 19.7 13.5–19 10–14

6.5 4.6 5.0 5.4

4.8 3.4 3.7 4.0

20

68

6.8–10.2 5–7.5

3.4

2.5

(a) Unnotched 10 by 10 mm (0.4 by 0.4 in) (Charpy) testpiece. (b) V-notched 10 by 10 mm (0.4 by 0.4 in.) (Charpy) testpiece. Source: Ref 16

Table 8 Fatigue strengths and endurance ratios for five compacted graphite (CG) irons from rotating-bending tests Tensile strength Matrix structure

As-cast As-cast As-cast As-cast As-cast

ferrite (>95% ferrite) ferrite pearlite pearlite pearlite (70% pearlite)

SG, spheroidal graphite. Source: Ref 16

Fatigue strength

Graphite type

MPa

ksi

MPa

ksi

Fatigue endurance ratio

Hardness, HB

95% CG CG CG CG + SG CG

336 388 414 473 386

48.7 56.3 60 68.6 56.0

211 178 185 208 186

30.6 25.8 26.8 30.2 27

0.63 0.46 0.45 0.44 0.48

150 184 205 217 ...

10. F. Martinez and D.M. Stefanescu, Properties of Compacted/Vermicular Graphite Cast Irons in the Fe-C-Al System Produced by Ladle and In-Mold Treatment, Trans. AFS, Vol 91, 1983, p 593 11. R. Hummer, unpublished research at the Austrian Foundry Research Institute, A.-Nr. 31.253, 1987 12. G. Nandori and J. Dul, Untersuchungen Uber den Abklingeffekt bei Gusseisen mit Kugelgraphit Durch Messung der La¨ngen—und Temperatura¨nderung wahrend der Erstarrung, Giesserei-Prax., No. 18, 1978, p 284 13. D.M. Stefanescu, I. Dinescu, S. Craciun, and M. Popescu, “Production of Vermicular Graphite Cast Irons by Operative Control and Correction of Graphite Shape,” Paper 37, presented at the 46th International Foundry Congress (Madrid, Spain), 1979 14. E. Nechtelberger, H. Puhr, J.B. von Nesselrode, and A. Nakayasu, “Cast Iron with Vermicular/Compacted Graphite—State of the Art Development, Production, Properties, Applications,” paper presented at the 49th International Foundry Congress (Chicago, IL), April 1982 15. D.M. Stefanescu, F. Martinez, and I.G. Chen, Solidification Behavior of Hypoeutectic and Eutectic Compacted Graphite Cast Irons—Chilling Tendency and Eutectic Cells, Trans. AFS, Vol 91, 1983, p 205 16. E. Nechtelberger, The Properties of Cast Iron up to 500  C, Technicopy Ltd., 1980 17. C.F. Walton and T.J. Opar, Ed., Iron Castings Handbook, Iron Casting Society Inc., 1981, p 381–397 18. K.P. Cooper and C.R. Loper, Jr., Trans. AFS, Vol 86, 1978, p 241 19. C.R. Loper, M.J. Lalich, H.K. Park, and A.M. Gyarmaty, “Microstructure-Mechanical Property Relationship in Compacted/ Vermicular Graphite Cast Iron,” Paper 35, presented at the 46th International Foundry Congress, (Madrid, Spain), 1979 20. N.N. Aleksandrov, B.S. Milman, L.V. Ilicheva, M.G. Osada, and V.V. Andreev, Production and Properties of High-Duty Iron with Compacted Graphite, Russ. Cast. Prod., Aug 1976, p 319 21. R.B. Gundlach, Trans. AFS, Vol 86, 1978, p 551 22. Spravotchnik po Tchugunomu Ljitiu [Cast Iron Handbook], 3rd ed., li Mashinostrojenie, 1978 23. S. Dawson, “Compacted Graphite Iron: Mechanical and Physical Properties for Engine Design,” Materials in Powertrain VDI, Dresden, Germany, Oct 28–29, 1999 24. K.H. Riemer, Giesserei, Vol 63 (No. 10), 1976, p 285 25. K.B. Palmer, BCIRA J., Report 1213, Jan 1976, p 31 26. A.F. Heiber, Trans. AFS, Vol 87, 1979, p 569

882 / Cast Irons

Fig. 24

Correlation between ultrasonic velocity and nodularity. SG, spheroidal graphite, CG, compacted graphite, FG, flake graphite. Source: Ref 7

27. M. Shidida, Y. Kanayama, and H. Nakayama, Strength and Crack Growth Behaviors of Compacted Graphite Vermicular Cast Iron in Rotating Bending, 29th Japan Congress on Materials Research, 1986, p 23–28 28. T.C. Spence, Corrosion of Cast Irons, Corrosion: Materials, Vol 13B, ASM Handbook, ASM International, 2005, p 43–50 29. D.M. Stefanescu and C.R. Loper, Jr., Giesserei-Prax., No. 5, 1981, p 74

Fig. 23

Fatigue (S-N) curves for ferritic, pearlitic, and higher-nodularity compacted graphite (CG) (Table 7) from rotating-bending tests. Arrows indicate specimens did not fail. SG, spheroidal graphite. (a) Source: Ref 25. (b) Source: 27

Compacted Graphite Iron Castings / 883

Fig. 25

Tensile strength related to (a) ultrasonic velocity and (b) resonant frequency for cast irons of various graphitic structures. SG, spheroidal graphite. Source: Ref 3

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 884-895 DOI: 10.1361/asmhba0005326

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Malleable Iron Castings Revised by C.J. van Ettinger, Gieterij Doesburg

MALLEABLE IRON is a cast ferrous metal that is initially produced as white cast iron and is then heat treated to convert the carbon-containing phase from iron carbide to a nodular form of graphite called temper carbon. The history of malleable iron goes back to Europe. The French scientist and metallurgist R.A.F. de Re´aumur published the first technical discussions of malleable iron castings (720 to 1722). In America, the inventor Seth Boyden introduced malleable iron (Ref 1). Boyden knew of the whiteheart malleable iron of Europe and foresaw a potential market for such a material; he started a series of experiments to find out how to produce malleable iron. Because of the chemical composition of the pig iron used, Boyden found the fractures of his specimens “black and grey” (blackheart) because of the presence of free carbon after the heat treatment process. In 1831, he was granted the first American patent for malleable iron; subsequently, the production of castings began in his foundry in Newark, New Jersey. There are two types of ferritic malleable iron: blackheart and whiteheart. Only the blackheart type is produced in the United States. This material has a matrix of ferrite with interspersed nodules of temper carbon. Cupola malleable iron is a blackheart malleable iron that is produced by cupola melting and is used for pipe fittings and similar thin- section castings. Because of its low strength and ductility, cupola malleable iron is usually not specified for structural applications. Pearlitic malleable iron is designed to have combined carbon in the matrix, resulting in higher strength and hardness than ferritic malleable iron. Martensitic malleable iron is produced by quenching and tempering pearlitic malleable iron. Malleable iron, like ductile iron, possesses considerable ductility and toughness because of its combination of nodular graphite and low-carbon metallic matrix. Because of the way in which graphite is formed in malleable iron, however, the nodules are not truly spherical, as they are in ductile iron, but are irregularly shaped aggregates. Malleable iron and ductile iron are used for some of the same applications in which ductility and toughness are important. In many cases, the choice between malleable and ductile iron is based on economy or availability rather than on properties. In certain applications, however,

malleable iron properties have a distinct advantage. It is preferred for thin-section castings; for parts that are to be pierced, coined, or cold formed; for parts requiring maximum machinability; for parts that must retain good impact resistance at low temperatures; and for some parts requiring wear resistance (martensitic malleable iron only). Ductile iron has a clear advantage where low solidification shrinkage is needed to avoid hot tears or where the section is too thick to permit solidification as white iron. (Solidification as white iron throughout a section is essential to the production of malleable iron.) Malleable iron castings are produced in section thicknesses ranging from approximately 1.5 to 100 mm (0.06 to 4 in.) and in weights from less than 0.03 to 180 kg (1 oz to 400 lb) or more. Composition. The chemical composition of malleable iron generally conforms to the ranges given in Table 1. Alloying elements such as copper (max 1.0%), nickel (max 0.8%), and molybdenum (max 0.5%) are also sometimes present. When the alloy elements are not used, then all other elements besides carbon, silicon, manganese, and sulfur should be called trace elements. Trace elements such as chromium, tin, titanium, vanadium, aluminum, copper, and molybdenum should be kept as low as possible and as consistent as possible, because they will retard or promote the graphitization process.

Melting Practices Melting can be accomplished by batch cold melting or by duplexing. Cold melting is done in coreless or channel-type induction furnaces, electric arc furnaces, or cupola furnaces. In duplexing, the iron is melted in a cupola or electric arc furnace, and the molten metal is transferred to a coreless or channel-type induction furnace for holding and pouring. Charge materials (foundry returns, steel scrap, ferroalloys, and, except in cupola melting, carbon) are carefully selected, and the melting operation is well controlled to produce metal having the desired composition and properties. Minor corrections in composition and pouring temperature

are made in the second stage of duplex melting, but most of the process control is done in the primary melting furnace. Molds are produced in green sand, silicate CO2-bonded sand, or resin-bonded sand (shell molds) on equipment ranging from highly mechanized or automated machines to that required for floor or hand molding methods, depending on the size and number of castings to be produced. Cores are generally produced in cold-box-type bonded sand on highly automated core-shooters. Cores are usually coated with a core-wash to protect the core sand against penetration. In general, the technology of molding and pouring malleable iron is similar to that used to produce gray iron. Heat treating is done in high-production controlled-atmosphere continuous furnaces or batch-type furnaces, again depending on production requirements. After it solidifies and cools, the metal is in a white iron state, and gates, sprues, and feeders can be easily removed from the castings by impact. This operation, called spruing, is generally performed manually with a hammer because the diversity of castings produced in the foundry makes the mechanization or automation of spruing very difficult. After spruing and shotblasting, the castings proceed to heat treatment, while gates and risers are returned to the melting department for reprocessing. Malleable iron castings are produced from the white iron by a two-stage annealing process. First- and second-stage annealing processes are described in detail in the section “Control of Annealing” in this article. After heat treatment, ferritic or pearlitic malleable castings are cleaned by shotblasting, gates are removed by shearing or grinding, and, where necessary, the castings are coined

Table 1 Chemical composition of malleable iron Element

Carbon Silicon Manganese Sulfur Phosphorus

Composition, %

2.16–2.90 0.90–1.90 0.15–1.25 0.02–0.20 0.01–0.05

Malleable Iron Castings / 885 or punched. Close dimensional tolerances can be maintained in ferritic malleable iron and in the lower-hardness types of pearlitic malleable iron, both of which can be easily straightened in dies. The harder pearlitic malleable irons are more difficult to press because of higher yield strength and a greater tendency toward springback after die pressing. However, even the highest-strength pearlitic malleable can be straightened to achieve good dimensional tolerances. Control of Melting. Metallurgical control of the melting operation is designed to ensure that the molten iron will have a certain composition and will:
Solidify white in the castings to be produced
Anneal on an established time-temperature

cycle set to minimum values in the interest of economy
Produce the desired graphite distribution (nodule count) upon annealing Changes in melting practice or composition that would satisfy the first requirement listed previously are generally opposed to satisfaction of the second and third, while attempts to improve annealability beyond a certain point may result in an unacceptable tendency for the as-cast iron to be mottled instead of white. The common elements in malleable iron are generally controlled within approximately  þ0.05 to  þ0.1%, using emission spectrometers. Thermal analysis is used for controlling the white solidification of the iron. A limiting minimum carbon content is required in the interest of mechanical quality and annealability, because decreasing carbon content reduces the fluidity of the molten iron, increases shrinkage during solidification, and reduces annealability. A limiting maximum carbon content is imposed by the requirement that the casting be white as-cast. The range in silicon content is limited to ensure proper annealing during a short-cycle high-production annealing process and to avoid the formation of primary graphite during solidification. Manganese and sulfur contents are balanced to ensure that all sulfur is combined with manganese and that only a safe, minimum quantity of excess manganese is present in the iron. An excess of either sulfur or manganese will retard annealing in the second stage and therefore increase annealing costs. The chromium content is kept low because of the carbide-stabilizing effect of this element and because it retards both the first-stage and secondstage annealing reactions. A mixture of gray iron and white iron in variable proportions that produces a mottled (speckled) appearance is particularly degrading to the mechanical properties of the annealed casting, whether ferritic or pearlitic malleable iron. Primary control of mottle is achieved by maintaining a balance of carbon and silicon contents. Because economy and castability are enhanced when the carbon and silicon contents of the base iron are in the higher portions of their respective ranges, some malleable iron

foundries produce iron with carbon and silicon contents at levels that may produce mottle and then add a balanced, mild carbide stabilizer to prevent mottle during casting. Bismuth and boron in balanced amounts accomplish this control. A typical addition is 0.01% Bi (as metal) and 0.001% B (as ferroboron). Bismuth retards graphitization during solidification; small amounts of boron have little effect on graphitizing tendency during solidification but accelerate carbide decomposition during annealing. The balanced addition of bismuth and boron permits the production of heavier sections for a given base iron or the utilization of a higher-carbon higher-silicon base iron for a given section thickness. Tellurium can be added in amounts of 0.0005 to 0.001% to suppress mottle. Tellurium is a much stronger carbide stabilizer than bismuth during solidification but also strongly retards annealing if the residual exceeds 0.003%. Less than 0.003% residual tellurium has little effect on annealing but has a significant influence on mottle control. Tellurium is more effective if added together with copper or bismuth. Because tellurium is toxic, it is no longer used in the foundries. Residual boron should not exceed 0.0035% in order to avoid nodule alignment and carbide formation. Also, the addition of 0.005% Al to the pouring ladle significantly improves annealability without promoting mottle.

the metallic matrix as lamellar pearlite, tempered martensite, or spheroidite. Molten iron produced under properly controlled melting conditions solidifies with all carbon in the combined form, producing the white iron structure fundamental to the manufacture of either ferritic or pearlitic malleable iron (Fig. 2). The base iron must contain balanced quantities of carbon and silicon to simultaneously provide castability, white iron in even the thickest sections of the castings, and annealability; therefore, precise metallurgical control is necessary for quality production. Thick metal sections cool slowly during solidification and tend to graphitize, producing mottled or gray iron. This is undesirable, because the graphite formed in mottled iron or rapidly cooled gray iron is generally of the type D configuration, a flake form in a dense, lacy structure, which is particularly damaging to the strength, ductility, and stiffness characteristics of both ferritic and pearlitic malleable iron. Control of Nodule Count. Proper annealing in short-term cycles and the attainment of high levels of casting quality require that controlled distribution of graphite particles be obtained during first-stage heat treatment. With low

Microstructure Malleable iron is characterized by microstructures consisting of uniformly dispersed fine particles of free carbon in a matrix of ferrite or tempered martensite. These microstructures can be produced in base metal of essentially the same composition. Structural differences between ferritic malleable iron and the various grades of pearlitic or martensitic malleable iron are achieved through variations in heat treatment. The microstructure of a casting of any type of malleable iron is derived by controlled annealing of white iron of suitable composition. During the annealing cycle, carbon that exists in combined form, either as massive carbides or as a microconstituent in pearlite, is converted to a form of free graphite known as temper carbon. Ferritic malleable iron requires a two-stage annealing cycle. The first stage converts primary carbides to temper carbon, and the second stage converts the carbon dissolved in austenite at the first-stage annealing temperature to temper carbon and ferrite. The microstructure of ferritic malleable iron is shown in Fig. 1. A satisfactory structure consists of temper carbon in a matrix of ferrite. There should be no flake graphite and essentially no combined carbon in ferritic malleable iron. Pearlitic and martensitic malleable irons contain a controlled quantity of combined carbon, which, depending on heat treatment, may appear in

Fig. 1

Structure of annealed ferritic malleable iron showing temper carbon in ferrite. Original magnification: 200

Fig. 2

Structure of as-cast malleable white iron showing a mixture of pearlite and eutectic carbides. Original magnification: 200

886 / Cast Irons nodule count (few graphite particles per unit area or volume), mechanical properties are reduced from optimum, and second-stage annealing time is unnecessarily long because of long diffusion distances. Excessive nodule count is also undesirable, because graphite particles may become aligned in a configuration corresponding to the boundaries of the original primary cementite. In martensitic malleable iron, very high nodule counts are sometimes associated with low hardenability and nonuniform tempering. Generally, a nodule count of 80 to 150 discrete graphite particles per square millimeter (80 to 150 in 15.5 in.2 of a micrograph at 100) appears to be optimum. This produces random particle distribution, with short distances between particles. Temper carbon is formed predominantly at the interface between primary carbide and saturated austenite at the first-stage annealing temperature, with growth around the nuclei taking place by a reaction involving diffusion and carbide decomposition. Although new nuclei undoubtedly form at the interfaces during holding at the first-stage annealing temperature, nucleation and graphitization are accelerated by the presence of nuclei that are created by appropriate melting practice. High silicon and carbon contents promote nucleation and graphitization, but these elements must be restricted to certain maximum levels because of the necessity that the iron solidify white. Control of Annealing. The rate of annealing of a hard iron casting depends on chemical composition, nucleation tendency (as discussed previously), and annealing temperature. With the proper balance of boron content and graphitic materials in the charge, optimum number and distribution of graphite nuclei are developed in the early part of first-stage annealing, and growth of the temper carbon particles proceeds rapidly at any annealing temperature. An optimum iron will anneal completely through the first-stage reaction in approximately 3 h at 940  C (1720  F). Irons with lower silicon contents or less-than-optimum nodule counts may require as much as 20 h for completion of first-stage annealing. The temperature of first-stage annealing exercises considerable influence on the rate of annealing and the number of graphite particles produced. Increasing the annealing temperature accelerates the rate of decomposition of primary carbide and produces more graphite particles per unit volume. However, high first-stage annealing temperatures can result in excessive distortion of castings during annealing, which leads to straightening of the casting after heat treatment. Annealing temperatures are adjusted to provide maximum practical annealing rates and minimum distortion and are therefore controlled within the range of 900 to 970  C (1650 to 1780  F). Lower temperatures result in excessively long annealing times, while higher temperatures produce excessive distortion. After first- stage annealing, the castings are cooled as rapidly as practical to 740 to

760  C (1360 to 1400  F) in preparation for second-stage annealing. The fast cooling step requires 1 to 6 h, depending on the equipment being employed. Castings are then cooled slowly at a rate of approximately 3 to 11  C (5 to 20  F) per hour. During cooling, the carbon dissolved in the austenite is converted to graphite and deposited on the existing particles of temper carbon. This results in a fully ferritic matrix. In the production of pearlitic malleable iron, the first-stage heat treatment is identical to that used for ferritic malleable iron. However, some foundries then slowly cool the castings to approximately 870  C (1600  F). During cooling, the combined carbon content of the austenite is reduced to approximately 0.75%, and the castings are then air cooled. Air cooling is accelerated by an air blast to avoid the formation of ferrite envelopes around the temper carbon particles (bull’seye structure) and to produce a fine pearlitic matrix (Fig. 3). The castings are then tempered to specification, or they are reheated to reaustenitize at approximately 870  C (1600  F), oil quenched, and tempered to specification. Large foundries usually eliminate the reaustenitizing step and quench the castings in oil directly from the first-stage annealing furnace after stabilizing the temperature at 845 to 870  C (1550 to 1600  F). The furnace atmosphere for producing malleable iron in continuous furnaces is controlled so that the ratio of CO to CO2 is between 1:1 and 20:1. In addition, any sources of water vapor or hydrogen are eliminated; the presence of hydrogen is thought to retard annealing, and it produces excessive decarburization of casting surfaces. Proper control of the gas atmosphere is important for avoiding an undesirable surface structure. A high ratio of CO to CO2 retains a high level of combined carbon on the surface of the casting and produces a pearlitic rim, or picture frame, on a ferritic malleable iron part. A low ratio of CO to CO2 permits excessive decarburization, which forms a ferritic skin on the casting with an underlying rim of pearlite. The latter condition is produced when a significant portion of the subsurface metal is decarburized to the degree that no temper carbon nodules can be developed during first-stage annealing. When this occurs, the dissolved carbon cannot precipitate from the austenite, except as the cementite plates in pearlite. An alternative is the use of nitrogen to control the furnace atmosphere, which is common in European foundries. The atmosphere is controlled by measuring the amount of CO2, which is normally held below 0.25%. The advantage of using nitrogen is the extreme small decarburization rim of just 0.2 mm (0.008 in.). The rate of cooling after first-stage annealing is important in the formation of a uniform pearlitic matrix in the air-cooled casting, because slow rates permit partial decomposition of carbon in the immediate vicinity of the temper carbon nodules, which results in the formation of films of ferrite around the temper carbon (bull’s-eye structure). When the extent of these films becomes excessive, a carbon gradient is

developed in the matrix. Air cooling is usually done at a rate not less than approximately 80  C (150  F) per minute. Air-quenched malleable iron castings have hardnesses ranging from 269 to 321 HB, depending on casting size and cooling rate. Such castings can be tempered immediately after air cooling to obtain pearlitic malleable iron with a hardness of 241 HB or less. High-strength malleable iron castings of uniformly high quality are usually produced by liquid quenching and tempering, using any of the three procedures. The most economical procedure is direct quenching after first-stage annealing. In this procedure, castings are cooled in the furnace to the quenching temperature of 845 to 870  C (1550 to 1600  F) and held for 15 to 30 min to homogenize the matrix. The castings are then quenched in agitated oil to develop a matrix microstructure of martensite having a hardness of 415 to 601 HB. Finally, the castings are tempered at an appropriate temperature between 590 and 725  C (1100 and 1340  F) to develop the specified mechanical properties. The final microstructure consists of tempered martensite plus temper carbon, as shown in Fig. 4. In heavy sections, higher-temperature

Fig. 3

Structure of air-cooled pearlitic malleable iron. Cooled in an air blast. Original magnification:

200

Fig. 4 200

Structure of oil-quenched and tempered martensitic malleable iron. Original magnification:

Malleable Iron Castings / 887 transformation products such as fine pearlite are usually present. Some foundries produce high-strength malleable iron by an alternative procedure in which the castings are forced-air cooled after first-stage annealing, retaining approximately 0.75% C as pearlite, and then reheated to 840 to 870  C (1545 to 1600  F) for 15 to 30 min, followed by quenching and tempering as described previously for the direct-quench process. Rehardened-and-tempered malleable iron can also be produced from fully annealed ferritic malleable iron with a slight variation from the heat treatment used for arrested-annealed (airquenched) malleable. The matrix of fully annealed ferritic malleable iron is essentially carbon free but can be recarburized by heating at 840 to 870  C (1545 to 1600  F) for 1 h. In general, the combined carbon content of the matrix produced by this procedure is slightly lower than that of arrested-annealed pearlitic malleable iron, and the final tempering temperatures required for the development of specific hardnesses are lower. Rehardened malleable iron made from ferritic malleable may not be capable of meeting certain specifications. Tempering times of 2 h or more are needed for uniformity. In general, control of final hardness of the castings is precise, with process limitations approximately the same as those encountered in the heat treatment of mediumor high-carbon steels. This is particularly true when specifications require hardnesses of 241 to 321 HB where control limits of  þ0.2 mm Brinell diameter can be maintained with ease. At lower hardnesses, a wider process control limit is required because of certain unique characteristics of the pearlitic malleable iron microstructure. The relationships between tempering conditions and yield strength and hardness properties are given in Fig. 5.

liquidus arrest. Curve A is the normal liquidus arrest of a cooling curve. The solidification starts at 1285  C (2345  F) at approximately 18 to 20 s after pouring the metal into the cup. Curve B shows a little arrest up front, the (normal) liquidus arrest. It seems that the solidification starts at 1310  C (2390  F) and approximately 12 s. This is a 25  C (45  F) higher solidification temperature and approximately 6 s earlier. Because of the solidification simulation technology used today (2008), it is essential to know how solidification behaves. This information is the basis for updating simulation technology. Simulation Technologies. Many foundries use MagmaSoft (Magma Gmbtt) software for the simulation of the casting process. The recent information from the cooling curve is transferred into the solidification simulation process. This means that the solidification simulation is not based with a calculated solidification temperature out of a chemical analysis but with a real solidification temperature, determined by the cooling curve. This step made the simulation process accurate and reliable. The simulation of a casting is done in three steps. The first step is the simulation of the part itself. This illustrates the temperature range, modules, and the shrinkage or the expected porosity areas. The second step is the simulation of the mold filling (Fig. 7). This displays the effect of the mold filling on the simulation result done previously. It shows changes in temperatures, modules, and shrinkage areas. At last, the complete gating and feeding system is simulated, which leads to the most accurate result. After a positive result, the tools are built, and the prototype parts are produced and checked with destructive testing and x-ray.

Fig. 5

Hardness and minimum yield strength of pearlitic malleable iron. Relationships of tempering time and temperature to hardness and minimum yield strength are given.

Current Production Technologies Digital Solidification Analysis Technology. Already in the 1970s, the American Foundry Society reported the importance of the use of thermal analysis during solidification. Leon, Ekpoon, and Heine (Ref 2) showed that the difference in carbon equivalent as determined by the cooling curve and by chemical analysis indicated a relation to the degree of oxidation of the liquid metal. Jacobs reported a relationship between the results of the thermal analysis and the amount of foundry scrap (Ref 3). Recent research shows the same relationship. However, the deviation in the liquidus arrest of the cooling curve was not found, but the digital cooling curve shows a first-arrest up front, the so called liquidus arrest, that comes and goes under the different melting circumstances. This phenomenon is used to control the method of charging and melting. The cooling curve can also be helpful for predicting the type of solidification structure (Ref 4). Figure 6 shows the difference in the cooling curve around the

Fig. 6

Difference in cooling curve around the liquidus arrest of a white cast iron (base malleable iron) caused by melting conditions. Curve A is a typical normal cooling curve of a solid-solution alloy with a liquidus temperature of approximately 1285  C (2345  F). Curve B shows there is a transformation occurring earlier, at approximately 1310  C (2390  F), due to different melting circumstances. Source: Ref 4

888 / Cast Irons Compared to malleable iron, which is treated in terms of tons per hour, ADI is treated in low volume. The costs are also increased by the additional transport costs to and from the heat treatment company, while malleable iron is treated in-house. The total costs are close for standard ductile iron and malleable iron, according to this new concept. Table 2 shows the high costs of the old malleable iron but also the costs of ADI, which is more than double that of standard ductile. The costs for ADI are based on the present (2008) costs of nickel of $37/kg and copper of $6/kg at an addition rate of 0.75%.

Applications

Fig. 7

Simulation of mold filling of a piston dome

Table 2 Comparison of costs Process cost(a), % Item

Alloy Melting Heat treatment Sum Casting yield(d)

Ductile standard

Malleable old

Malleable new concept(b)

Ductile ADI(c)

84 16 0 100 65

103 22 48 173 40

71 14 28 113 65

136 16 68 220 65

(a) Standard dutile iron total (100%) is base. (b) Simulation and smart engineering practices as discussed in text. (c) ADI, austempered ductile iron. (d) Mass of casting(s) divided by the total mass of metal poured into the mold, expressed as a percentage. Source: Ref 4

Smart engineering is nothing more than positive and simple thinking. Smart engineering focuses on the casting. It is not thought of as part of the complete gating system. It is important to judge the value of existing knowledge when using new technology. Knowing the results of simulation provides information about what is useful and what is useless knowledge of gating and risering. In the past, engineers made dozens of calculations and also used their past experiences to develop a gating and risering system. The results of a simulation are sometimes 100% against all the existing thinking. Now, it is important for the engineer to abandon his routine and experience and follow the simulation. Do not make it difficult or look back too often. Of course, there are some basic rules that should be obeyed, such as maximum metal velocity, pouring times, and pouring temperatures. However, all of these parameters are used in the simulation process. Smart engineering can also answer the question, “How can we use the simulation technology?” In other words,

the goal is to achieve the maximum amount of castings on a pattern and to use the simulation process to fix the gating and risering problems. Effect on Production Costs. Using new technologies improves the production costs of malleable iron. Two major developments influence the cost reduction significantly: the reduction in heat treatment time and the improvement in casting yield. Table 2 gives a review of the production costs among ductile and malleable iron castings. The costs are presented as a percentage where the base is standard ductile iron at 100%. The casting yield is absorbed in the alloy costs. Alloy costs for ductile iron are more expensive than that of malleable iron. Alloy costs are the highest for austempered ductile iron (ADI), due to the current high costs of nickel. The comparison of the costs are not complete for producing a casting because the cost for coremaking, molding, and other operations are similar for both malleable and ductile iron. The cost for the heat treatment of ADI is very high.

The requirement that any iron produced for conversion to malleable iron must solidify white places definite section thickness limitations on the malleable iron industry. Thick metal sections can be produced by melting a base iron of low carbon and silicon contents or by alloying the molten iron with a carbide stabilizer. However, when carbon and silicon are maintained at low levels, difficulty is invariably encountered in annealing, and the time required to convert primary and pearlitic carbides to temper carbon becomes excessively long. High-production foundries are usually reluctant to produce castings more than approximately 40 mm (1.5 in.) thick. Some foundries, however, routinely produce castings as thick as 100 mm (4 in.). Automotive and associated applications of ferritic and pearlitic malleable irons include many essential parts in vehicle power trains, frames, suspensions, and wheels. A partial list includes differential carriers, differential cases, bearing caps, steering-gear housings, spring hangers, universal-joint yokes, automatic-transmission parts, rocker arms, disc brake calipers, and wheel hubs. Examples are shown in Fig. 8 for small parts and in Fig. 9 to 11 for typical automotive parts. Ferritic and pearlitic malleable irons are also used in the railroad industry and in agricultural equipment, chain links, ordnance material, electrical pole line hardware, hand tools, and other parts requiring section thicknesses and properties obtainable in these materials. Malleable iron castings are often selected because the material has excellent machinability in addition to significant ductility. In other applications, malleable iron is chosen because it combines castability with good toughness and machinability. Malleable iron is often chosen because of shock resistance alone. Applications are listed by ASTM specifications in Table 3. Mechanical properties are given in Table 4. Similar standards are grouped. The ASTM standards have specified metric values. A 47 (Ref 5) is A 47/A 47M with a metric grade 22010 (tensile minimum, 340 MPa; tensile yield, 220 MPa). These have not been included for clarity of the table.

Malleable Iron Castings / 889

Fig. 8

Small parts of malleable iron. (a) Nut made of pearlitic malleable iron, grade EN-GJMB-650-2, 600 g. The insert shows the core used for the as-cast internal screw thread. (b) Balancing weight made of ferritic malleable iron, grade EN-GJMB-350-10, 1500 g. The large picture shows the application, and the insert is the actual casting. Courtesy of VS GUSS AG, Germany

Fig. 11 Fig. 9

Various gears in martensitic malleable iron. Courtesy of Gieterij Doesburg, Netherlands

Fig. 10

Wide range of pistons made in pearlitic malleable iron. Courtesy of Gieterij Doesburg, Netherlands

Ferritic Malleable Iron Because ferritic malleable iron consists of only ferrite and temper carbon, the properties of ferritic malleable castings depend on the quantity, size, shape, and distribution of the

Various transmission parts made in oilquenched pearlitic malleable iron. Most of the parts are induction hardened after machining. Courtesy of Gieterij Doesburg, Netherlands

temper carbon particles and on the composition of the ferrite. Fully annealed ferritic malleable iron castings contain 2.00 to 2.70% graphitic carbon by weight, which is equivalent to approximately 6 to 8% by volume. Because the graphitic carbon contributes nothing to the strength of the castings, those with the lesser amount of graphite are somewhat stronger and more ductile than those containing the greater amount (assuming equal size and distribution of graphite particles). Elements such as silicon and manganese in solid solution in the ferritic matrix contribute to the strength and reduce the elongation of the ferrite. Therefore, by varying base metal composition, slightly different strength levels can be obtained in the fully annealed ferritic product. The mechanical properties that are most important for design purposes are tensile strength, yield strength, modulus of elasticity, fatigue strength, and impact strength. Hardness can be considered no more than an approximate indicator that the ferritizing anneal was complete, and it is seldom used for any other

purpose. The hardness of ferritic malleable iron almost always ranges from 110 to 156 HB. The tensile properties at room and lower temperatures of ferritic malleable iron (Table 5) are usually measured on unmachined test bars. Machined test bars are normally used for pearlitic malleable iron. Fracture toughness is listed as well. The fatigue limit of unnotched ferritic malleable iron is approximately 50% of the tensile strength, or from 170 to 205 MPa (25 to 30 ksi). Fig. 12 summarizes the effects of notches on fatigue strength. Notch radius generally has little effect on fatigue strength, but fatigue strength decreases with increasing notch depth. Modulus of elasticity in tension is approximately 170 GPa (25  106 psi). The modulus in compression ranges from 150 to 170 GPa (22  106 to 25  106 psi); in torsion, from 65 to 75 GPa (9.5  106 to 11  106 psi). Because brittle fractures are most likely to occur at high strain rates, at low temperatures, and with a high restraint on metal deformation, notch tests such as the Charpy V-notch test are conducted over a range of test temperatures to establish the toughness behavior and the temperature range of transition from a ductile to a brittle fracture. Figure 13 illustrates the behavior of ferritic malleable iron and several types of pearlitic malleable iron in the Charpy V-notch test. This shows that ferritic malleable iron has a higher upper-shelf energy and a lower transition temperature to a brittle fracture than pearlitic malleable iron does. Additional information on fracture toughness of malleable irons can be found later in the discussion of “Pearlitic and Martensitic Malleable Iron” in this article. Short-term tensile properties show no significant change to 370  C (700  F). Sustained-load stress-rupture data from 425 to 650  C (800 to 1200  F) are given in Fig. 14. The corrosion resistance of ferritic malleable iron is increased by the addition of copper, usually approximately 1%, in certain applications, for example, conveyor buckets, bridge castings, pipe fittings, railroad switch stands, and freight-car hardware. One important use for copper-bearing ferritic malleable iron is chain links. Ferritic malleable iron can be galvanized to provide added protection. The corrosion behavior of cast irons and use of protective coatings is found in “Corrosion of Cast Irons” in Corrosion: Materials, Volume 13B of ASM Handbook (Ref 8). Welding and Brazing. Welding of ferritic malleable iron almost always produces brittle white iron in the weld zone and the portion of the heat-affected zone (HAZ) immediately adjacent to the weld zone. During welding, temper carbon is dissolved, and upon cooling it is reprecipitated as carbide rather than graphite. In some cases, welding with a cast iron electrode may produce a brittle gray iron weld zone. The loss of ductility due to welding may not be serious in some applications. However, welding is usually not recommended unless the castings are subsequently annealed to

890 / Cast Irons convert the carbide to temper carbon and ferrite. Ferritic malleable iron can be fusion welded to steel without subsequent annealing if a completely decarburized zone as deep as the

normal HAZ is produced at the faying surface of the malleable iron part before welding. Silver brazing and tin-lead soldering can be satisfactorily used.

Table 3 Applications of malleable iron castings Mechanical properties are given in Table 4. Specification No.

Ferritic ASTM A 47

Class or grade

Microstructure

32510 35018

Temper carbon and ferrite

ASTM A 338 32510 35018 ASTM A 197 ...

Temper carbon and ferrite

Typical applications

General engineering service at normal and elevated temperatures for good machinability and excellent shock resistance Flanges, pipe fittings, and valve parts for railroad, marine, and other heavy-duty service to 345  C (650  F) Pipe fittings and valve parts for pressure service

Free of primary graphite

Pearlitic and martensitic ASTM A 220 40010 Temper carbon in necessary matrix 45008 without primary cementite or 45006 graphite 50005 60004 70003 80002 90001

General engineering service at normal and elevated temperatures. Dimensional tolerance range for castings is stipulated

Automotive ASTM A 602 M3210 Ferritic

For low-stress parts requiring good machinability: steeringgear housings, carriers, and mounting brackets Compressor crankshafts and hubs For selective hardening: planet carriers, transmission gears, and differential cases For machinability and improved response to induction hardening For high-strength parts: connecting rods and universal-joint yokes For high strength plus good wear resistance: certain gears

M4504 Ferrite and tempered pearlite(a) M5003 Ferrite and tempered pearlite(a) M5503 Tempered martensite M7002 Tempered martensite M8501 Tempered martensite (a) May be all tempered martensite for some applications

Pearlitic and Martensitic Malleable Iron Pearlitic malleable iron is produced either by controlled heat treatment of the same base white iron used to produce ferritic malleable iron or by alloying to prevent the decomposition of carbides dissolved in austenite during cooling from the first- stage annealing temperature. It can be produced by air cooling after first-stage annealing and subsequent tempering to develop specified properties. Martensitic malleable iron is produced by liquid quenching and tempering to develop specified properties (see the section “Control of Annealing” in this article). Variations in heat treatment, coupled with variations in base composition and melting practice, make it possible to obtain a wide range of properties in pearlitic or martensitic malleable iron. The mechanical properties of pearlitic and martensitic malleable iron vary in a substantially linear relationship with Brinell hardness (Fig. 15). In the low hardness ranges, below approximately 207 HB, the properties of airquenched and tempered pearlitic malleable are essentially the same as those of oil-quenched and tempered martensitic malleable. This results from the fact that attaining the low hardnesses requires considerable coarsening of the matrix carbides and partial second-stage graphitization. Either an air-quenched pearlitic structure or an oil-quenched martensitic structure can be coarsened and decarburized to meet this hardness requirement.

Table 4 Properties of malleable iron castings Microstructure and more applications are given in Table 3. Malleable casting specifications

ASTM A 47 SAE J158 GM11M ISO 5922 EN 1562:1997 GMW1 ASTM A 220 SAE J158 GM ISO 5922 EN 1562:1997 GMW1 ASTM A 220 SAE J158 GM85M ISO 5922 EN 1562:1997 GMW1 ASTM A 220 SAE J158 GM84M ISO 5922 EN 1562:1997 GMW1 ASTM A 220 SAE J158 GM88M ISO 5922 EN 1562:1997 GMW1 NA, nonapplicable

Tensile strength Class

32510 M3210 11M JMB/350-10 GJMB-350-10 MB 350-10 50005 M4504 86M JMB/500-5 GJMB-500-5 MB 550-4 60004 M5503 85M JMB/600-3 GJMB-600-3 MB 600-3 70003 M7002 84M JMB/700-2 GJMB-700-2 MB 700-2 80002 M8501 88M JMB/800-1 GJMB-800–1 NA

ksi

50 50 50 ... ... ... 70 65 69.6 ... ... ... 80 75 79.75 ... ... ... 85 90 100 ... ... ... 95 105 104.4 ... ... NA

MPa

... ... ... 350 350 350 ... ... ... 500 500 550 ... ... ... 600 600 600 ... ... ... 700 700 700 ... ... ... 800 800 NA

Yield strength ksi

MPa

Elongation, %

Hardness, Brinell

32.5 32 32.625 ... ... ... 50 45 47.85 ... ... ... 60 55 59.45 ... ... ... 70 70 79.75 ... ... ... 80 85 84.825 ... ... NA

... ... ... 200 200 200 ... ... ... 300 300 340 ... ... ... 390 390 410 ... ... ... 530 530 530 ... ... ... 600 600 NA

10 10 10 10 10 10 5 4 5 5 5 4 4 3 3 3 3 3 3 2 2 2 2 2 2 1 2 1 1 NA

156 max 156 max 156 max 150 max 150 max 149 max 179–228 163–217 163–207 165–215 165–215 179–229 197–241 187–241 197–241 195–245 195–245 195–245 217–269 229–269 241–269 240–260 240–290 241–293 241–285 269–302 269–302 270–320 270–320 NA

Matrix structure

Ferrite

Lamellar spheroidized pearlite and ferrite

Typical applications

General engineering service at normal and elevated temperatures for good machinability and excellent shock resistance. Flanges, pipe fittings, and valve parts for railroad, marine, and other heavy-duty service to 345  C (650  F) Compressor crankshafts and hubs

Lamellar spheroidized pearlite

For good machinability and improved response to induction hardening. Pistons, planet carriers, transmission gears

Tempered martensite

High-strength parts: connecting rods, universal-joint yokes

Tempered martensite

High strength plus good wear resistance: certain gears

Malleable Iron Castings / 891

Fig. 12

Effects of notch radius and notch depth on the fatigue strength of ferritic malleable iron

Fig. 13

Charpy V-notch transition curves for ferritic and pearlitic malleable irons. Source: Ref 7

At higher hardnesses, oil-quenched and tempered malleable iron has higher yield strength and elongation than air-quenched and tempered malleable because of greater uniformity of matrix structure and finer distribution of carbide particles. Oil-quenched and tempered pearlitic malleable iron is produced commercially to hardnesses as high as 321 HB, while the maximum hardness for high-production air-quenched and tempered pearlitic malleable iron is approximately 255 HB. The lower maximum hardness is applied to the air-quenched material for several reasons:
Because hardness on air quenching normally

does not exceed 321 HB and may be as low as 269 HB, attempts to temper to a hardness range above 255 HB produce nonuniform hardness and make the process control limits excessive.
Very little structural alteration occurs during the tempering heat treatment to a higher hardness, and the resulting structure is more

difficult to machine than an oil- quenched and tempered structure at the same hardness.
There is only a slight improvement in other mechanical properties with increased hardness above 255 HB. Because of these considerations, applications for air-quenched and tempered pearlitic malleable iron are usually those requiring moderate strength levels, while the higher-strength applications need the oil-quenched and tempered material. The modulus of elasticity in tension of pearlitic malleable iron is 175 to 195 GPa (25.5  106 to 28.0  106 psi). For automobile crankshafts, the modulus is important and must be determined with greater precision. The results of Charpy V-notch tests on pearlitic malleable iron are presented in Fig. 13. The fracture toughness of ferritic and pearlitic malleable irons has not been widely studied, but one researcher has estimated KIc values for these materials by using a J-integral approach

(Ref 6). Table 5 summarizes the fracture toughness values obtained for the various grades of malleable iron at various temperatures. All of the materials exhibited stable crack extension prior to fracture for 25 mm (1 in.) wide compact-tension specimens. High Yield Strength to Tensile Strength Ratio. The oil- quenched pearlitic and martensitic malleable irons, especially, have an excellent yield strength to tensile strength ratio of up to 0.8. Compared to standard ductile iron, malleable iron is superior with a much higher yield strength. At a tensile strength of 700 MPa (102 ksi), the yield strength of ductile iron is a minimum of 430 MPa (62 ksi), while malleable iron shows a yield strength of 575 MPa (83 ksi), almost 35% higher. Figure 16 shows the relationship between tensile strength and yield strength according to the ASTM standard A 220 (Ref 9) for malleable iron and the EN 1563:1997 for ductile iron. As for ductile irons, fracture toughness testing indicates that malleable irons possess considerably more toughness than is indicated by Charpy impact toughness results. Although the fracture toughness values for pearlitic grades are similar to those obtained for ferritic grades, the higher yield strengths of the pearlitic grades indicate that their critical flaw sizes, which are proportional to (KIc/sy)2, are less than those of the ferritic grades of malleable iron. Detailed information on the principles of fracture toughness and the nomenclature associated with fracture mechanics studies is available in the article “Fracture Toughness and Fracture Mechanics” in Mechanical Testing and Evaluation, Volume 8 of ASM Handbook (Ref 10). The compressive strength of pearlitic and martensitic malleable irons is plotted in Fig. 17. Torsional strength values are presented in Table 6. Sustained-load stress-rupture data for eight different grades of pearlitic malleable iron are shown in Fig. 18. Results of high-temperature Charpy V-notch tests showing the effect of hardness on impact energy are given in Fig. 19. The fatigue limits of pearlitic and martensitic malleable irons are approximately 40 to 50% of tensile strength. Oil-quenched and tempered martensitic iron usually has a higher fatigue ratio than pearlitic iron made by the arrestedanneal method. Wear Resistance. Because of its structure and hardness, pearlitic and martensitic malleable irons have excellent wear resistance. In some moving parts where bushings are normally inserted at pivot points, heat treated malleable iron has proved to be so wear resistant that the bushings have been eliminated. One example of this is the rocker arm for an overhead-valve automotive engine. Welding and Brazing. Welding of pearlitic or martensitic malleable iron is difficult because the high temperatures used can cause the formation of a brittle layer of graphite-free white iron. Pearlitic and martensitic malleable

892 / Cast Irons and clutch hubs. An example of a flame-hardened pearlitic malleable iron part is a pinion spacer used to support the cup of a roller bearing. To preclude service failures, the ends of the pinion spacer are flame hardened to file hardness to a depth of approximately 2.3 mm (0.09). Malleable iron can be carburized, carbonitrided, or nitrided to produce a surface with improved wear resistance. In addition, heat treatments such as austempering have been used in specialized applications.

REFERENCES

Composition, % Group

Grade

C

Si

Mn

P

S

Cr

A-1 B-1 E-1 G-1

35018 32510 35018 35018

2.21 2.50 2.16 2.29

1.14 1.32 1.17 1.01

0.35 0.43 0.38 0.38

0.161 0.024 0.137 0.11

0.081 0.159 0.095 0.086

... 0.029 0.017 ...

Fig. 14

Stress-rupture plot for various grades of ferritic malleable iron. The solid lines are curves determined by the method of least squares from the existing data and are least squares fit to data. The dashed lines define the 90% symmetrical tolerance interval. The lower dashed curve defines time and load for 95% survivors, and the upper dashed curve is the boundary for 5% survivors. Normal distribution is assumed.

iron can be successfully welded if the surface to be welded has been heavily decarburized. Pearlitic or martensitic malleable iron can be brazed by various commercial processes. One application is the induction silver brazing of a pearlitic malleable casting and a steel shaft to form a planetary output shaft for an automotive transmission. In another automotive application, two steel shafts are induction copper brazed to a pearlitic malleable iron shifter shaft plate. Selective Surface Hardening. Pearlitic malleable iron can be surface hardened by either induction heating and quenching or flame heating and quenching to develop high hardness at the heat-affected surface. Considerable research has been done to determine the surface-hardening characteristics of pearlitic malleable and its capability of developing high hardness over relatively narrow surface bands. Generally, little

difficulty is encountered in obtaining hardnesses in the range of 55 to 60 HRC, with the depth of penetration being controlled by the rate of heating and by the temperature at the surface of the part being hardened (Fig. 20). The maximum hardness obtainable in the matrix of a properly hardened pearlitic malleable part is 67 HRC. However, conventional hardness measurements made on castings show less than 67 HRC because of the presence of the graphite particles, which are averaged into the hardness. Generally, a casting with a matrix microhardness of 67 HRC will have approximately 62 HRC average hardness, as measured with the standard Rockwell tester. Similarly, a Rockwell or Brinell hardness test on softer structures will show less than matrix microhardness because of the presence of graphite. Two examples of automobile production parts hardened by induction heating are rocker arms

1. Malleable Iron Castings, Malleable Founders’ Society, The Ann Arbor Press Inc., Ann Arbor, MI, 1960 2. C. Leon, U. Ekpoon, and R.W. Heine, Relationship of Casting Defects to Solidification of Malleable Base White Cast Iron, AFS Trans., Vol 81–74, p 323–344 3. F.W. Jacobs, Practical Application of Liquidus Control for Malleable Iron Melting, AFS Trans., Vol 81–56, p 261–276 4. C.J. van Ettinger and W. Baumgart, “Thermal Analysis, A Unique Fingerprint of a Melt,” Paper 78, 66th World Foundry Congress (Istanbul, Turkey), Sept 6–9, 2004, p 743–757 5. “Standard Specification for Malleable Iron Castings,” A 47/A 47M, Annual Book of ASTM Standards, ASTM International 6. W.L. Bradley, Fracture Toughness Studies of Gray, Malleable and Ductile Cast Iron, Trans. AFS, Vol 89, 1981, p 837–848 7. C.F. Walton and T.J. Opar, Ed., Iron Castings Handbook, Iron Castings Society, 1981, p 297–321 8. T.C. Spence, Corrosion of Cast Irons, Corrosion. Materials, Vol 13B, ASM Handbook, ASM International, 2005, p 43–50 9. “Standard Specification for Pearlitic Malleable Iron Castings,” A 220, Annual Book of ASTM Standards, ASTM International 10. Fracture Toughness and Fracture Mechanics, Mechanical Testing and Evaluation, Vol 8, ASM Handbook, ASM International, 2000, p 563–575

Malleable Iron Castings / 893

Fig. 15

Relationships of tensile properties to Brinell hardness for pearlitic malleable irons from two foundries. The mechanical properties of these irons vary substantially in a linear relationship with Brinell hardness, and in the low hardness ranges (below approximately 207 HB), the properties of air-quenched and tempered material are essentially the same as those produced by oil quenching and tempering.

Table 5

Fracture toughness of malleable irons Test temperature

Malleable iron grade

Ferritic M3210 Pearlitic M4504 (normalized)

M5503 (quenched and tempered)

M7002 (quenched and tempered)

Source: Ref 6

Yield strength

KIc

pffiffiffiffi ksi in.

F

MPa

ksi

pffiffiffiffi MPa m

24 19 59

75 3 74

230 240 250

33 35 36

44 41 44

40 38 40

24 19 57 24 19 58 24 19 58

75 2 70 75 3 73 75 3 72

360 380 390 410 430 440 520 550 570

52 55 57 60 64 66 75 80 83

55 48 30 45 51 30 54 39 39

50 44 27 41 47 27 49 35 36



C



Fig. 17

Fig. 16

Table 6

Compressive strength of martensitic malleable irons

pearlitic

and

Difference in yield strength and ultimate tensile strength ratio (slope of curve) of malleable iron (A 47/A 220/Armasteel GM) and ductile iron (Lineair ductile EN 1563)

Torsion test values of malleable irons Modulus of rupture

Yield strength

Modulus of elasticity

Material(a)

MPa

ksi

MPa

ksi

Ultimate twist, degree

GPa

106 psi

Ferritic 45010(b) 60003(b) 80002(b) High strength(c)

375 570 665 715 850

54 83 97 104 123

130 215 320 410 490

19 31 46 59 71

540 350 280 140 70

65.15 66.75 70.3 69.22 67.5

9.45 9.68 10.20 10.04 9.79

(a) Specimens 20 mm (0.750 in.) in diameter. (b) Pearlitic malleable iron. (c) High-strength malleable iron with yield strength of approximately 620 MPa (90 ksi), tensile strength of 830 MPa (120 ksi), and 2% elongation in 50 mm (2 in.)

Composition, % Material

C

Pearlitic (low carbon-high phosphorus) Group E-2 2.27 Group G-2 2.29 Pearlitic (high carbon-low phosphorus) Group C-2 2.65 Group W-1 2.45 Alloyed pearlitic (low carbon-high phosphorus) Group E-3 2.21 Group L-1 2.16 Group L-2 2.16 Group L-3 2.32

Fig. 18

Si

Mn

S

P

Cr

Others

1.15 1.01

0.89 0.75

0.098 0.086

0.135 0.11

0.019 ...

... ...

1.35 1.38

0.41 0.41

0.15 0.12

... 0.04

0.018 0.032

0.0020 B ...

1.13 1.18 1.18 1.14

0.88 0.72 0.80 0.82

0.110 0.120 0.123 0.117

0.122 0.128 0.128 0.128

0.021 ... ... ...

0.47Mo, 1.03Cu 0.34Mo, 0.83 Ni 0.40Mo, 0.62 Ni 0.38Mo, 0.65 Ni

Stress-rupture plot for (a) pearlitic malleable iron and (b) alloyed pearlitic malleable iron. The solid lines are curves determined by the method of least squares from the existing data. The dashed lines define the 90% symmetrical tolerance interval. The lower dashed curve defines time and load for 95% survivors, and the upper dashed curve is the boundary for 5% survivors. Normal distribution is assumed.

Malleable Iron Castings / 895

Fig. 19

Charpy V-notch impact energy of one heat of air-quenched and tempered pearlitic malleable iron

Fig. 20

Hardness versus depth for surface-hardened pearlitic malleable irons. Curves labeled “Matrix” show hardness of the matrix, converted from microhardness tests. O, oil-quenched and tempered to 207 HB before surface hardening; A, air-cooled and tempered to 207 HB before surface hardening

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 896-903 DOI: 10.1361/asmhba0005327

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

White Iron and High-Alloyed Iron Castings Richard B. Gundlach, Stork Climax Research Services

HIGH-ALLOYED WHITE CAST IRONS are an important group of materials whose production must be considered separately from that of ordinary types of cast irons. In these cast iron alloys, the alloy content is well above 4%, and consequently they cannot be produced by ladle additions to irons of otherwise standard compositions. They are usually produced in foundries specially equipped to produce highly alloyed irons. These iron alloys are most often melted in electric furnaces, specifically electric arc furnaces and induction furnaces, in which the precise control of composition and temperature can be achieved. The foundries usually have the equipment needed to handle the heat treatment and other thermal processing unique to the production of these alloys. The high-alloyed white irons are primarily used for abrasion-resistant applications and are readily cast into the parts needed in machinery for crushing, grinding, and handling of abrasive materials. The chromium content of high-alloyed white irons also enhances their corrosion-resistant properties. The large volume fraction of primary and/or eutectic carbides in their microstructures provides the high hardness needed for crushing and grinding other materials. The metallic matrix supporting the carbide phase in these irons can be adjusted by alloy content and heat treatment to develop the proper balance between the resistance to abrasion and the toughness needed to withstand repeated impact. All high-alloyed white irons contain chromium to prevent the formation of graphite upon solidification and to ensure the stability of the carbide phase. Most also contain nickel, molybdenum, copper, or combinations of these alloying elements to prevent the formation of pearlite in the microstructure. While lowalloyed white iron castings, which have an alloy content below 4%, develop hardnesses in the range of 350 to 550 HB, the high-alloyed irons range in hardness from 450 to 800 HB. In addition, several grades contain alloy eutectic carbides (M7C3 chromium carbides), which are substantially harder than the M3C iron carbides


Nickel-chromium white irons, which are low-

in low-alloyed irons. For many applications, the increased abrasion resistance of the more expensive high-alloyed white irons adds significantly to wear life, enabling them to provide the most cost-effective performance. Alloy Grades. Specification ASTM A 532/A 532M, covers the composition and hardness of the abrasion-resistant white iron grades (Table 1). Many castings are ordered according to these specifications. However, a large number of castings are produced with composition modifications for specific applications. It is most desirable that the designer, metallurgist, and foundryman work together to specify the composition, heat treatment, and foundry practice to develop the most suitable alloy and casting design for a specific application. The high-alloyed white cast irons fall into two major groups:

chromium alloys containing 3 to 5% Ni and 1 to 4% Cr, with one alloy modification that contains 7 to 11% Cr. The nickel-chromium irons are also commonly identified as Ni-Hard types 1 to 4.
The chromium-molybdenum irons containing 11 to 23% Cr, with up to 3% Mo and often additionally alloyed with nickel or copper A third group comprises the 25 or 28% Cr white irons, which may contain other alloying additions of molybdenum and/or nickel up to 1.5%.

Nickel-Chromium White Irons The oldest group of high-alloyed irons of industrial importance, the nickel-chromium white irons, or Ni-Hard irons, have been

Table 1 Composition and mechanical requirements of abrasion-resistant cast irons in accordance with ASTM A 532/A 532M Composition(a),% Class

Type

I I I I II II II III

A B C D A B D A

Designation

Ni-Cr-HC Ni-Cr-LC Ni-Cr-GB Ni-Hi Cr 12% Cr 15% Cr-Mo 20% Cr-Mo 25% Cr

C

Si

Ni

Cr

2.8–3.6 2.4–3.0 2.5–3.7 2.5–3.6 2.0–3.3 2.0–3.3 2.0–3.3 2.0–3.3

0.8 max 0.8 max 0.8 max 2.0 max 1.5 max 1.5 max 1.0–2.2 1.5 max

3.3–5.0 3.3–5.0 4.0 max 4.5–7.0 2.5 max 2.5 max 2.5 max 2.5 max

1.4–4.0 1.4–4.0 1.0–2.5 7.0–11.0 11.0–14.0 14.0–18.0 18.0–23.0 23.0–30.0

Mo

1.0 1.0 1.0 1.5 3.0 3.0 3.0 3.0

max(b) max(b) max(b) max(c) max(d) max(d) max(d) max(d)

Mechanical requirements(e) Hardness, HB Class

I I I I II II II III

Type

A B C D A B D A

Designation

Ni-Cr-HC Ni-Cr-LC Ni-Cr-GB Ni-Hi Cr 12% Cr 15% Cr-Mo 20% Cr-Mo 25% Cr

Typical section thickness, max

Sand cast, min

Chill cast, min

Hardened, min

Softened, max

in

mm

550 550 550 550 550 450 450 450

600 600 600 550 550 ... ... ...

600 600 600 600 600 600 600 600

... ... 400 ... 400 400 400 400

8 8 3 diam ball 12 1 diam ball 4 8 8

200 200 75 diam ball 300 25 diam ball 100 200 200

(a) Maximum: 2.0% Mn (b) Maximum: 0.30% P, 0.15% S. (c) Maximum: 0.10% P, 0.15% S. (d) Maximum: 0.10% P, 0.06% S, 1.2% Cu. (e) Other hardness levels and methods given; see standard. Typical section thickness is not part of standard.

White Iron and High-Alloyed Iron Castings / 897 produced for more than 70 years and are very cost- effective materials for crushing and grinding. In these martensitic white irons, nickel is the primary alloying element because, at levels of 3 to 5%, it is effective in suppressing the transformation of the austenite matrix to pearlite, thus ensuring that a hard, martensitic structure (usually containing significant amounts of retained austenite) will develop upon cooling in the mold. Chromium is included in these alloys, at levels from 1.4 to 4%, to ensure that the irons solidify carbidic, that is, to counteract the graphitizing effect of nickel. A typical microstructure is shown in Fig. 1. The optimum composition of a nickel-chromium white iron alloy depends on the properties required for the service conditions and the dimensions and weight of the casting. Abrasion resistance is generally a function of the bulk hardness and the volume of carbide in the microstructure. When abrasion resistance is the principal requirement and resistance to impact loading is secondary, alloys having high carbon contents, ASTM A 532/A 532M class I type A (Ni-Hard 1), are recommended. When conditions of repeated impact are anticipated, the lower-carbon alloys, class I type B (Ni-Hard 2), are recommended because they have less carbide and therefore greater toughness. A special grade, class I type C, has been developed for producing grinding balls and slugs. Here, the nickel-chromium alloy composition has been adapted for chill casting and specialized sand casting processes. The class I type D (Ni-Hard 4) alloy is a modified nickel-chromium iron that contains higher levels of chromium, ranging from 7 to 11%, and increased levels of nickel, ranging from 5 to 7%. Whereas the eutectic carbide phase in the lower-alloyed nickel-chromium irons is M3C (iron carbide), which forms as a continuous network in these irons, the higher chromium in the type D alloy promotes M7C3 chromium carbides, which form a relatively discontinuous eutectic carbide distribution (Fig. 2). This modification in the eutectic carbide pattern provides an appreciable improvement in resistance to fracture by impact. The higher alloy content of this iron grade also results in improved corrosion resistance, which has proved to be useful in the handling of corrosive slurries.

Melting Practice A principal advantage of the nickel-chromium irons is that they can be melted in a cupola; however, better control of composition and temperature are achieved with electric furnace melting. Consequently, electric arc furnace melting and induction furnace melting are most common. The nickel-chromium irons are readily made in either acid-, neutral-, or basic-lined electric furnaces. They are normally dead melted, and there is usually no reason to use oxygen lancing except as a means of

slightly reducing carbon content. Acid linings are generally more economical than basic linings. The type of lining affects silicon and chromium losses. Very little adjustment to slag composition is necessary in acid melting, which is used for the majority of current production. Normal charge materials are various kinds of steel scrap, foundry returns, or returns of similar alloy from service. For foundries that lack the facilities for rapid melt analysis, it may be necessary to select steel scrap rather carefully to ensure that the residual levels of alloying elements, such as manganese and copper, which have a potent effect on austenite retention, are consistent and under control. Chromium is obtained in the form of highcarbon ferrochrome and is generally added near the end of the heat to avoid excessive oxidation losses. Carbon is obtained from electrode graphite, petroleum coke, and other sources. Pig iron is also used to carburize the melt and can be an additional source of silicon. Silicon and manganese are added as ferroalloys. Molybdenum is usually added as ferromolybdenum; however, with arc furnaces, molybdenum oxide briquettes may be used. Sulfur is limited to 0.06%, and phosphorus is kept to 0.10%, as specified by ASTM A 532/A 532M. High superheating temperatures have not been necessary when melting in induction furnaces, which have good stirring action, and a final temperature of 1480  C (2700  F) is usually adequate for thick-section castings. Final temperatures of up to 1565  C (2850  F) are commonly used in arc furnace practice to ensure homogeneity of the bath composition and to accelerate the solution of carbon, added after meltdown, and late alloy additions. No particular problems have been associated with high superheat temperatures, but it is more difficult to predict melting losses. As in steel melting, deoxidation of the bath with aluminum was common, but it is no longer practiced by most producers, with no adverse effects on the soundness or mechanical properties of the casting. A late addition/inoculation of foundry-grade ferro-silicon is often made to improve toughness.

Composition Control Carbon is varied according to the properties needed for the intended service. Carbon contents in the range of 3.2 to 3.6% are prescribed when maximum abrasion resistance is desired. When impact loading is expected, carbon content should be held in the range of 2.7 to 3.2%. Nickel content increases with section size or cooling time of the casting to inhibit pearlitic transformation. For castings of 38 to 50 mm (1.5 to 2 in.) thick, 3.4 to 4.2% Ni is sufficient to suppress pearlite formation upon mold cooling. Heavier sections may require nickel levels up to 5.5% to avoid the formation of pearlite. It is important to limit nickel content to the level needed for control of pearlite; excess

Fig. 1

Typical microstructure of class I type A nickel-chromium white cast iron. Original magnification: 340

Fig. 2

Typical microstructure of class I type D nickel-chromium white cast iron. Original magnification: 340

nickel increases the amount of retained austenite and lowers hardness. Silicon is needed for two reasons. A minimum amount of silicon is necessary to improve fluidity of the melt and to produce a fluid slag, but of equal importance is its effect on as-cast hardness. Increased levels of silicon, in the range of 1 to 1.5%, have been found to increase the amount of martensite and the resulting hardness. Late additions of ferrosilicon (0.2% as 75% Si-grade ferrosilicon) have been reported to increase toughness. It is important to note that higher silicon contents can promote pearlite and may increase the nickel requirement. Chromium, primarily added to offset the graphitizing effects of nickel and silicon in

898 / Cast Irons types A, B, and C alloys, ranges from 1.4 to 4.0%. Chromium content must increase with increasing section size. In type D alloy, chromium levels range from 7 to 11% (typically 9%) for the purpose of producing eutectic carbides of the M7C3 chromium carbide type, which are harder and less deleterious to toughness. Manganese is typically held to a maximum of 0.8% even though 2.0% maximum is allowed according to ASTM A 532/A 532M specification. While it provides increased hardenability to avoid pearlite formation, it is a more potent austenite stabilizer than nickel and promotes increased amounts of retained austenite and lower as-cast hardness. For this reason, higher manganese levels are undesirable. When considering the nickel content required to avoid pearlite in a given casting, the level of manganese present should be a factor. Copper increases both hardenability and the retention of austenite and therefore must be controlled for the same reason that manganese must be limited. Copper should be treated as a nickel substitute and, when properly included in the calculation of the amount of nickel required to inhibit pearlite, it reduces the nickel requirement. Molybdenum is a potent hardenability agent in these alloys and is used in heavy-section castings to augment hardenability and inhibit pearlite.

Pouring Practices High pouring temperatures aggravate shrinkage under feeder heads and other hot spots and can lead to microshrinkage and coarse dendritic structures. Careful control of pouring temperature is particularly important in the consistent production of thick-section castings. Low pouring temperatures are necessary not only to avoid shrinkage defects but also to avoid problems such as metal penetration and burnedon sand. Low pouring temperatures are also effective in controlling dendrite size and the coarseness of the eutectic carbide structure. The eutectic temperature for the various nickelchromium iron grades is approximately 1200  C (2190  F), and solidification generally begins at temperatures of 1200 to 1280  C (2190 to 2340  F), depending on composition. When selecting a pouring temperature, common rules apply. Pouring temperatures are seldom lower than 100  C (180  F) above the liquidus temperature. Higher pouring temperatures may be necessary when pouring smaller castings. Of course, casting configurations must be considered when selecting optimum pouring temperatures.

Molds, Patterns, and Casting Design These alloys may be sand cast or chill cast in permanent molds. The chill cast microstructures develop higher hardness, strength, and impact toughness over the sand cast structures, because of the finer carbides. It is recommended that,

whenever practical, the wearing faces of the casting be cast against a chill to improve abrasion resistance and toughness. Sand molds may be made either of green sand or of dry sand, oil sand, or steel casting sand. Air-setting and thermal-setting resin sands are also commonly used. Molds must be rigid to minimize shrinkage defects. White irons are subject to hot tearing, unless suitable precautions are taken. Occasionally when castings are extracted from the mold, they are found to be cracked. Stresses large enough to cause fracture can arise when either the mold or the cores are excessively strong and there is inadequate mold or core breakdown to allow normal thermal contraction with cooling. Large cores, such as those used in pump volutes, and even small cores used to form bolt holes, can cause cracking of this type. Shrinkage allowance is 13 to 21 mm/m (5/32 to ¼ in./ft); a commonly used figure is 16 mm/m (3/16 in./ft). The actual pattern allowance depends on the choice of alloy and the geometry of the casting. Patterns should employ generous radii and avoid sharp section changes to avoid initiating cracking upon solidification and subsequent cooling. These alloys exhibit relatively high liquid-tosolid shrinkage (5%); therefore, large gates and risers are required for feeding. Special consideration should be given to the design of end gates and the positioning of risers so that they can be readily removed. Risers cannot be removed by burning and are often notched or necked down to facilitate removal by impact. They are most easily removed after the castings have cooled to room temperature, that is, when martensite is present. Martensitic transformation begins at temperatures below 230  C (450  F).

properly made, they have a martensitic matrix structure, as-cast. Tempering is performed between 205 and 260  C (400 and 500  F) for at least 4 h. This tempers the martensite, relieves some of the transformation stresses, and increases the strength and impact toughness by 50 to 80%. Some additional martensite may form upon cooling from the tempering temperature. This heat treatment does not reduce hardness or abrasion resistance. High-Temperature Heat Treatment. In the past, hardening of the class I type D, Ni-Hard 4, was performed by supercritical heat treatment when as-cast hardness was insufficient. An austenitizing heat treatment usually comprised heating the casting at temperatures between 750 and 790  C (1380 and 1450  F) with a soak time of 8 h. Air or furnace cooling, not over 30  C/h (50  F/h), was conducted, followed by a tempering/stress-relief heat treatment. Refrigeration treatment is the more commonly practiced remedy for low hardness today (2008). To achieve a hardness of 550 HB, it is necessary for the as-cast austenite-martensite microstructure to have at least 60% martensite present. When martensite content is increased to 80 to 90%, however, hardness values exceed 650 HB. To reduce the amount of retained austenite, that is, form more martensite, deep freeze treatments are commonly applied. Refrigeration to temperatures between 70 and 185  C (90 and 300  F) for ½ to 1 h will usually raise the hardness level 100 HB units. A tempering/ stress-relief heat treatment usually follows. The typical refrigerated nickel-chromium iron microstructure is shown in Fig. 3. In the heat treatment of any white cast iron, care must be taken to avoid cracking by thermal shock; the castings must never be placed in a hot furnace or otherwise subjected to rapid

Shakeout Practices The shakeout practice is a critical step in the successful production of nickel-chromium iron castings. High residual stresses and cracking can result from extracting castings from a mold at too high a temperature. Cooling all the way to room temperature in the mold is desirable because, as stated previously, transformation to martensite occurs below 230  C (450  F) during the last stages of cooling. This precaution is mandatory in heavy-section castings; the volume increase that occurs with martensite formation results in the development of high transformation stresses in castings having steep temperature gradients due to rapid cooling. Thin-section castings and castings having simple shapes are less susceptible to stresses and are often removed from the mold once the castings have reached black heat. With these alloys, slow cooling favors higher as-cast hardness.

Heat Treatment All nickel-chromium white iron castings are given a stress-relief heat treatment because,

Fig. 3

Microstructure of class I type D nickelchromium white cast iron after refrigeration. Original magnification: 340

White Iron and High-Alloyed Iron Castings / 899 heating or cooling. The risk of cracking increases with the complexity of the casting shape and section thickness.

Machining Alloyed white irons are generally ground to finish size; care must be taken to avoid cracking by overheating. Single-point turning and boring can be done using heavy tooling, rigid setups, sufficient power, and carbide- or boride-tipped cutters or special ceramic inserts. Electrical discharge machining and electrochemical machining are also feasible.

Applications Because of their low cost, the martensitic nickel-chromium white irons are consumed in large tonnages in mining operations as ball mill liners and grinding balls. Class I type A castings are used in applications requiring maximum abrasion resistance, such as ash pipes, slurry pumps, roll heads, muller tires, augers, coke-crusher segments, classifier shoes, brick molds, pipe elbows carrying abrasive slurries, and grizzly disks. Type B is recommended for applications requiring more strength and exerting moderate impact, such as crusher plates, crusher concaves, and pulverizer pegs. Class I type D, Ni-Hard type 4, has a higher level of strength and toughness and is therefore used for the more severe applications that justify its added alloy costs. It is commonly used for pump volutes handling abrasive slurries and coal pulverizer table segments and tires. The class I type C alloy (Ni-Hard 3) is specifically designed for the production of grinding balls. This grade is both sand cast and chill cast. Chill casting has the advantage of lower alloy cost, and, more important, provides a 15 to 30% improvement in life. All grinding balls require tempering for 8 h at 260 to 315  C (500 to 600  F) to develop adequate impact toughness.

Special Nickel-Chromium White Iron Alloys Certain proprietary grades of type A alloy have been developed by the rolling mill industry. The compositions of these alloys have been modified to produce mottled structures, containing some graphite. The graphite inclusions are reported to improve resistance to thermal cracking. These indefinite chill rolls are cast in thickwall gray iron chiller molds in roll diameters of up to 1015 mm (40 in.) or more. The silicon-tochromium ratios and inoculation with ferro-silicon are carefully controlled to control the amount and distribution of the graphite particles. The rolls can be double poured with a gray iron core. With molybdenum modification, the matrix of the chill cast shell becomes martensitic. Some roll alloys are designed to be heat treated, that is, by a modified normalizing heat treatment, to obtain a bainitic microstructure.

High-Chromium White Irons The high-chromium white irons have excellent abrasion resistance and are used effectively in slurry pumps, brick molds, coal-grinding mills, shot-blasting equipment, and components for quarrying, hard-rock mining, and milling. In some applications, they must also be able to withstand heavy impact loading. These alloyed white irons are recognized as providing the best combination of toughness and abrasion resistance attainable among the white cast irons. In the high-chromium irons, as with most abrasion-resistant materials, there is a trade-off between wear resistance and toughness. By varying composition and heat treatment, these properties can be adjusted to meet the needs of most abrasive applications. As a class of alloyed irons, the high-chromium irons are distinguished by the hard, relatively discontinuous M7C3 eutectic carbides present in the microstructure, as opposed to the softer, more continuous M3C eutectic carbides present in the alloyed irons containing less chromium. These alloys are usually produced as hypoeutectic compositions.

improved by increasing the amount of carbide (increasing the carbon content), while toughness is improved by increasing the proportion of metallic matrix (reducing the carbon content). The influence of carbon content on the shape and distribution of the carbide phase in these alloys is illustrated in the micrographs of Fig. 4. Large hexagonal carbide rods occur when carbon contents exceed the eutectic carbon content (Fig. 4c). These primary chromium carbides, which precipitate from the melt before eutectic solidification, are quite deleterious to impact toughness and should be avoided in castings subjected to any impact in service. The eutectic carbon content changes inversely with the chromium content in these alloys. The relationship between eutectic carbon content and chromium content is shown in Fig. 5. To avoid primary M7C3 chromium carbides in high-chromium irons, the composition should follow the formula: %C þ %Cr=15  4:5

(Eq 1)

Classes of High-Chromium Irons Specification ASTM A 532/A 532M covers the compositions and hardnesses of two general classes of the high-chromium irons (Table 1). The chromium-molybdenum irons (class II of ASTM A 532/A 532M) contain 11 to 23% Cr and up to 3.0% Mo and can be supplied either as-cast with an austenitic or austenitic-martensitic matrix, or heat treated with a martensitic matrix microstructure for maximum abrasion resistance and toughness. They are usually considered the hardest of all grades of white cast irons. Compared to the lower-alloy nickel-chromium white irons, the eutectic carbides are harder and can be heat treated to achieve castings of higher hardness. Molybdenum, as well as nickel and copper when needed, is added to prevent pearlite and to ensure maximum hardness. The high-chromium irons (class III of ASTM A 532/A 532M) represent the oldest grade of high-chromium irons, with the earliest patents dating back to 1917. These general-purpose irons, also called 25% Cr and 28% Cr irons, contain 23 to 28% Cr with up to 3.0% Mo. To prevent pearlite and attain maximum hardness, molybdenum is added in all but the lightest-cast sections. Alloying with nickel and copper up to 1% is also practiced. Although the maximum attainable hardness is not as high as in the class II chromium-molybdenum white irons, these alloys are selected when resistance to corrosion is also desired.

Microstructures Carbide. The carbides in high-chromium irons are very hard and wear resistant but are also brittle. In general, wear resistance is

Fig. 4

Microstructures of high-chromium white iron compositions. (a) Low carbon (hypoeutectic). (b) Eutectic. (c) High-carbon (hypereutectic). Original magnification: all 75. Courtesy of Climax Molybdenum Company

900 / Cast Irons

Fig. 5

Relationship between the chromium and carbon contents and the eutectic composition in high-chromium irons

Fig. 6

High-chromium iron with an as-cast austenitic matrix microstructure. Original microstructure: 500. Courtesy of Climax Molybdenum Company

Table 2 Minimum alloy content to avoid pearlite in mold-cooled castings for indicated effective section size (plate thickness or radius of rounds) ASTM A 532/A 532M class

Plate thickness or radius of rounds Cr(a), %

IIB

14–18

IID

18–23

IIIA

23–28

C(a), %

2.0 3.5 2.0 3.2 2.0 3.0

25 mm (1 in.)

1.0 2.0 0.5 1.5 ... 1.0

Mo Mo Mo Mo Mo

50 mm (2 in.)

1.5 2.5 1.0 2.0 0.5 1.5

Mo Mo Mo Mo Mo Mo

100 mm (4 in.)

1.5 2.5 1.0 2.0 1.0 1.5

Mo Mo Mo Mo Mo Mo

+ + + +

1.0 1.0 1.0 1.0

(Ni (Ni (Ni (Ni

+ + + +

Cu) Cu) Cu) Cu)

+ 1.0 (Ni + Cu)

(a) In base irons containing 0.6% Si and 0.8% Mn

As-Cast Austenitic. Solidification in the hypoeutectic alloys occurs by the formation of austenite dendrites followed by the eutectic formation of austenite and M7C3 chromium carbides. Under equilibrium conditions, additional chromium carbide precipitates from the austenite matrix upon cooling from the eutectic to the critical temperature, that is, approximately 760  C (1400  F), and transformation to ferrite and carbide occurs with subsequent cooling. However, when cooling under nonequilibrium conditions such as those encountered in most commercial castings, the austenite becomes supersaturated in carbon and chromium. Because of elevated carbon and chromium contents, a metastable austenitic matrix microstructure normally develops, provided pearlitic and bainitic transformations have been inhibited (Fig. 6). With sufficient alloying with molybdenum, manganese, nickel, and copper, pearlitic transformation can be avoided in virtually any cast section. As-Cast Martensitic Martensitic structures can be obtained as-cast in heavy-section castings that cool slowly in the mold. With slow cooling rates, austenite stabilization is incomplete, and partial transformation to martensite occurs. However, in these castings, martensite is mixed

with large amounts of retained austenite, and therefore, hardness levels are lower than can be achieved in heat treated martensitic castings. These castings must contain sufficient alloy to suppress pearlite upon cooling. Some compositions (higher silicon) have been developed to assist martensite formation in refrigeration treatments. Subcritical heat treatment has also been used to reduce austenite content and, at the same time, increase hardness and toughness. Heat Treated Martensitic. To obtain maximum hardness and abrasion resistance, martensitic matrix structures must be produced by full heat treatment (Fig. 7). The casting must contain sufficient alloy to avoid pearlite formation upon cooling from the heat treatment temperature.

Selecting Compositions to Obtain Desired Structures Many complex sections, such as slurry pump components, are often used in the as-cast austenitic/martensitic condition to avoid the possibility of cracking and distortion when heat treated. To prevent pearlite in mold cooling,

Fig. 7

High-chromium iron with a heat-treatedmartensitic matrix microstructure. Original mag-nification: 500. Courtesy of Climax Molybdenum Company

alloying additions are usually required. As the carbon content is increased, more chromium is consumed, forming additional carbide, and therefore, larger alloying additions are required. Table 2 presents a guide for appropriate alloying to prevent pearlite in the various classes of as-cast irons. Optimum performance is usually achieved with heat treated martensitic structures. Again, alloying must be sufficient to ensure a pearlite-free microstructure, with heat treatment. Of necessity, the heat treatment, discussed later in detail (see the section “Quenching” in this article), requires an air quench from the austenitizing temperature. Faster cooling rates should not be used, because the casting can develop cracks due to high thermal and/or transformation stresses. Thus, the alloy must have sufficient hardenability to allow air hardening. However, overalloying with manganese, nickel, and copper promotes retained austenite, which detracts from resistance to abrasion and spalling. Therefore, it is best to obtain adequate hardenability primarily with molybdenum. Table 3 is offered as a guide to alloying for air quenching heat treated castings of various sections.

Melting Practice The high level of carbon pickup with the high chromium contents in these alloys precludes the use of cupola melting. Consequently, electric arc and induction furnaces are normally used. The high-chromium irons are readily made in acid-, neutral-, or basic-lined electric furnaces. They are normally dead melted, and there is usually no reason to use oxygen lancing except as a means of slightly reducing carbon content. Despite rapid refractory wear due to chromium-bearing slag, acid linings are generally more economical than basic linings. Very

White Iron and High-Alloyed Iron Castings / 901 Table 3 Minimum alloy content to avoid pearlite in heat treatment for indicated effective section size (plate thickness or radius of rounds) ASTM A 532/A 532M class

Plate thickness or radius of rounds Cr(a), %

IIB

14–18

IID

18–23

IIIA

23–28

C(a), %

2.0 3.5 2.0 3.2 2.0 3.0

50 mm (2 in.)

1.5 3.0 1.0 1.5 0.5 1.5

Mo Mo Mo Mo Mo Mo

125 mm (5 in.)

1.5 2.0 2.0 2.0 1.5 1.5

Mo Mo Mo Mo Mo Mo

+ 0.5 (Ni + Cu) + 1.0 (Ni + Cu) + 0.7 (Ni + Cu) + 0.6 (Ni + Cu)

150–255 mm (6–10 in.)

2.0 2.5 2.0 2.0 1.5 1.5

Mo Mo Mo Mo Mo Mo

+ + + + + +

1.0 1.2 0.5 1.2 0.5 1.2

(Ni (Ni (Ni (Ni (Ni (Ni

+ + + + + +

Cu) Cu)(b) Cu) Cu)(b) Cu) Cu)(b)

(a) In base irons containing 0.6% Si and 0.8% Mn. (b) Nickel and copper promote retained austenite and should be restricted to combined levels of 1.2% maximum; manganese behaves similarly and should be restricted to 1.0% maximum.

little adjustment to slag composition is necessary in acid melting, which is used for the majority of current production. Viscous or nonfluid slags should be avoided. Normal charge materials are various kinds of steel scrap, foundry returns, or returns of similar alloy from service. For foundries that lack the facilities for rapid melt analysis, it may be necessary to select steel scrap rather carefully to ensure that the total content of alloying elements that have a potent effect on hardenability and austenite retention is consistent and under control. While some furnace operators prefer to add a portion of high-carbon ferrochrome to the charge, it is generally added near the end of the heat to avoid excessive oxidation losses. Carbon is obtained from electrode graphite, petroleum coke, and other sources. If pig iron is used to carburize the melt, it should have a low silicon content. The ideal silicon content is 0.6%. Less than 0.4% Si in the bath can cause difficulties with viscous slag, while higher silicon contents can promote pearlite. Selection of an incorrect grade of ferrochromium with high silicon is a common cause for an excessively high silicon content. Manganese can range up to 2.0%. Manganese on the high side of this range can erode acid furnace linings. Therefore, manganese content should be limited to 1% in meltdown; the remainder can be added to the ladle as crushed ferroalloy. Molybdenum is usually added as ferromolybdenum; however, with arc furnaces, molybdenum oxide briquettes may be used. Sulfur is limited to 0.06%, and phosphorus is kept to 0.10%. High superheating temperatures have not been necessary when melting in induction furnaces, which have good stirring action, and a final temperature of 1480  C (2700  F) is usually adequate for thick-section castings. Final temperatures of up to 1565  C (2850  F) are commonly used in arc furnace practice to ensure homogeneity of the bath composition and to accelerate the solution of carbon, added after meltdown, and late alloy additions. No particular problems have been associated with high superheat temperatures, but it is more difficult to predict melting losses. As in steel melting, deoxidation of the bath with aluminum was common in the past, but this practice has been discontinued by most producers, with no adverse effects on the

soundness or mechanical properties of the casting. Titanium is sometimes added to limit dendrite size. There have been reports that bath additions of aluminum and titanium aggravate feeding problems. For heats in which hardenability is marginal, there is evidence that these elements tend to promote pearlite with mold cooling or in heat treated castings.

Pouring Practices High pouring temperatures aggravate shrinkage under feeder heads and other hot spots and can lead to microshrinkage and coarse dendritic structures. Careful control of pouring temperature is particularly important in the production of thick-section castings. Low pouring temperatures are necessary to avoid shrinkage defects and prevent problems such as metal penetration and burned-on sand. Low pouring temperatures are also effective in controlling dendrite size and the coarseness of the eutectic carbide structure. The eutectic temperature for the various high-chromium iron grades ranges from 1230 to 1270  C (2245 to 2315  F), and solidification generally begins at temperatures up to 1350  C (2460  F), depending on composition. Pouring temperatures are seldom lower than 100  C (180  F) above the liquidus temperature. Castings thicker than 102 mm (4 in.) are generally poured between 1345 and 1400  C (2450 and 2550  F). Higher pouring temperatures may be necessary for smaller castings. Casting configurations also must be considered when selecting optimum pouring temperatures. In the ladle or during pouring, the metal appears cold and sluggish because an oxide film readily forms on the surface; but the metal is actually quite fluid and may be poured into intricate shapes. This rather viscous surface oxide is less liable to cause surface defects on castings that are poured quickly. Even when the metal is poured on the cold side, flasks have to be clamped, or weighted, to prevent the metal from running out at the parting line.

Molds, Patterns, and Casting Design Molds must be rigid to minimize shrinkage defects. Molds may be made of either green sand or of dry sand, oil sand, or steel casting sand. Air-

setting and thermal-setting resin sands are also commonly used. Because white irons are subject to hot tearing, cores should break down readily. Unless suitable precautions are taken, the high-chromium irons are somewhat more prone to crack in the foundry than are white irons of lower alloy content. Castings are occasionally found to be cracked when extracted from the mold. Stresses large enough in magnitude to cause fracture can arise when either the mold or cores are excessively strong and there is inadequate mold or core breakdown to allow normal thermal contractions upon cooling. Large cores such as those used in pump volutes, and even small cores used to form bolt holes, can cause cracking of this type. Shrinkage allowance is 18 to 21 mm/m (7/32 to ¼ in./ft). The actual pattern allowance depends on the choice of subsequent heat treatment procedures, if any, and the geometry of the casting. Patterns should employ generous radii and avoid sharp section changes to avoid initiating cracking on solidification and subsequent cooling. These alloys exhibit relatively high liquid-tosolid shrinkage; therefore, large gates and risers are required for feeding. Special consideration should be given to the design of end gates and to the positioning of risers so that they can be readily removed. Risers are often notched or necked down to facilitate removal by impact.

Shakeout Practices The shakeout practice is probably the most critical step in the successful production of high-chromium iron castings. A frequent cause of high residual stresses and of cracking is the common practice of extracting castings from the mold at too high a temperature. Cooling all the way to room temperature in the mold is desirable and can be a requirement to avoid cracking, especially if martensite forms during the last stages of cooling. This precaution is mandatory in heavy-section castings to be used in the as-cast condition where the desired, mold-cooled structure is a mixture of austenite and martensite. For irons that are heat treated, the desired mold-cooled structure is often pearlite. This softer structure facilitates removal of gates and risers, minimizes the transformational and thermal stresses that cause cracking, and shortens the response to heat treatment. Careful design of alloy composition ensures the development of a substantially pearlitic structure in the casting after mold cooling, yet provides enough hardenability to prevent pearlite formation during subsequent heat treatment. Heavy-section castings made pearlitic by such an alloy content can often be removed from the mold once the castings have reached black heat.

Heat Treatment Toughness and abrasion resistance are improved by heat treatment to a martensitic

902 / Cast Irons

Fig. 8

Heat treatment schedule high-chromium irons

for

hardening

Fig. 11

Influence of chromium on the optimum hardening temperature in high-chromium

white iron

Fig. 9

Influence of austenitizing temperature on hardness (H) and retained austenite (g) in high-chromium irons

Fig. 10

Microstructure of heat treated martensitic high-chromium iron illustrating fine secondary M7C3 carbides. Original magnification: 680

microstructure. Figure 8, which illustrates the heat treatment process, shows the importance of slow heating to 650  C (1200  F) to avoid cracking. For complex shapes, a maximum rate of 30  C/h (50  F/h) is recommended. Simple

shapes and fully pearlitic castings can be heated at faster rates. The heating rate can be accelerated above red heat. Austenitization. There is an optimum austenitizing temperature for achieving maximum hardness (Fig. 9). The austenitizing temperature determines the amount of carbon that remains in solution in the austenite matrix. Too high a temperature increases the stability of the austenite, while the higher retained austenite content reduces hardness. Low temperatures result in low-carbon martensite, reducing both hardness and abrasion resistance. Because of this sensitivity to temperature, furnaces that can produce accurate and uniform temperatures are most desirable. The successful heat treatment produces austenite destabilization by precipitation of fine, secondary M7C3 carbides within the austenite matrix (Fig. 10). The optimum austenitizing temperature varies with composition (Fig. 11). Class II irons containing 12 to 20% Cr are austenitized in the temperature range of 955 to 1010  C (1750 to 1850  F). Class III irons containing 23 to 28% Cr are austenitized in the temperature range of 1010 to 1095  C (1850 to 2000  F). Heavy sections usually require temperatures on the upper end of the range. Castings should be held at temperature long enough to accomplish equilibrium dissolution of chromium carbides to ensure a proper hardening response. A minimum of 4 h at temperature is necessary. For heavy sections, the rule of 1 h/in. of section thickness is usually adequate. For castings that are fully pearlitic before heat treatment, the holding time at temperature can be reduced. Quenching. Air quenching (vigorous fan cooling) the castings from the austenitizing temperature to below the pearlite temperature range, that is, to between 550 and 600  C (1020 and 1110  F), is highly recommended. The subsequent cooling rate should be substantially reduced to minimize stresses; still-air or even

furnace cooling to ambient temperature is common. Complex and heavy-section castings are often reinserted in the furnace, which is at 550 to 600  C (1020 to 1110  F), and allowed sufficient time to reach a uniform temperature within the casting. After the temperature is equalized, the castings are either furnace or still-air cooled to ambient temperature. Tempering. Castings can be put into service in the hardened (as-cooled) condition without further tempering or subcritical heat treatments; however, tempering in the range of 205 to 230 C (400 to 450  F) for 2 to 4 h is recommended to restore some toughness in the martensitic matrix and to further relieve residual stresses. The microstructure after hardening always contains retained austenite in the range of 10 to 30%. Some retained austenite is transformed following tempering at low temperatures, but if spalling is a problem, highertemperature tempering can be used to further reduce austenite content. Subcritical Heat Treatment. Subcritical heat treatment (tempering) is sometimes performed, particularly in large heat treated martensitic castings, to reduce retained austenite content and increase resistance to spalling. The tempering parameters necessary to eliminate retained austenite are very sensitive to time and temperature and depend on the casting composition and prior thermal history. Typical tempering temperatures range from 480 to 540  C (900 to 1000  F), and times range from 8 to 12 h. Excess time or temperature results in softening and a drastic reduction in abrasion resistance. Insufficient tempering results in incomplete elimination of austenite. The amount of retained austenite cannot be determined metallographically; those experienced with this heat treatment practice have developed techniques using specialized magnetic instruments to determine the level of retained austenite after tempering. Annealing Castings can be annealed to make them more machinable, either by sub-critical annealing or by a full anneal. Sub-critical annealing is accomplished by pearlitizing by means of soaking in the narrow range between 695 and 705  C (1280 and 1300  F) for 4 to 12 h, which produces hardnesses ranging from 400 to 450 HB. Lower hardnesses often can be achieved with full annealing, in which castings are heated in the range of 955 to 1010  C (1750 to 1850  F), followed by slow cooling to 760  C (1400  F), and holding at this temperature for 10 to 50 h, depending on composition. Annealing affects neither the primary carbides nor the potential for subsequent hardening; guidelines for hardening as-cast castings also apply to annealed castings.

Machining These alloyed white irons are generally ground to finish size; care must be exercised to avoid cracking by overheating. Single-point turning and boring can be done using heavy

White Iron and High-Alloyed Iron Castings / 903 tooling, rigid setups, sufficient power, and carbide- or boride-tipped cutters or special ceramic inserts. Electrical discharge machining and electrochemical machining are also feasible.

Applications The high-chromium white irons are superior in abrasion resistance and are used effectively in impellers and volutes in slurry pumps, classifier wear shoes, brick molds, impeller blades and liners for shot blasting equipment, and refiner disks in pulp refiners. In many applications, they withstand heavy impact loading, such as from impact hammers, roller segments and ring segments in coal-grinding mills, feedend lifter bars and mill liners in ball mills for hard-rock mining, pulverizer rolls, and rolling mill rolls.

Special High-Chromium Iron Alloys Irons for Corrosion Resistance. Alloys with improved resistance to corrosion, for applications such as pumps for handling fly ash, are produced with high chromium content (26 to 28% Cr) and low carbon content (1.6 to 2.0% C). These high-chromium, low-carbon irons provide the maximum chromium content in the matrix. The addition of 2% Mo is recommended for improving resistance to chloridecontaining environments. For this application, fully austenitic matrix structures provide the best resistance to corrosion, but some reduction

in abrasion resistance must be expected. Castings are normally supplied in the as-cast condition. Irons for High-Temperature Service. Because of castability and cost, high-chromium white iron castings can often be used for complex and intricate parts in high-temperature applications at considerable savings compared to stainless steel. These cast iron grades are alloyed with 12 to 39% Cr and are used at temperatures up to 1040  C (1900  F) for scaling resistance. Chromium causes the formation of an adherent, complex, chromium-rich oxide film at high temperatures. The high-chromium irons designated for use at elevated temperatures fall into one of three categories, depending on the matrix structure:
Martensitic irons alloyed with 12 to 28% Cr
Ferritic irons alloyed with 30 to 34% Cr
Austenitic irons, which, in addition to con-

taining 15 to 30% Cr, also contain 10 to 15% Ni to stabilize the austenite phase The carbon contents of these alloys range from 1 to 2%. The choice of an exact composition is critical to the prevention of s-phase formation at intermediate temperatures and, at the same time, avoids the ferrite-to-austenite transformation during thermal cycling, which leads to distortion and cracking. Typical applications include recuperator tubes, breaker bars and trays in sinter furnaces, grates, burner nozzles and other furnace parts, glass bottle molds, and valve seats for combustion engines.

ACKNOWLEDGMENTS The author wishes to thank Warren Spear of the Nickel Development Institute and Ralph Nelson of Thomas Foundries Inc. for helpful discussions. SELECTED REFERENCES
“Abrasion-Resistant Cast Iron. Classifica










tion,” BS ISO 21988:2006, British Standards Institute (ISO Standard), 2007 Abrasion-Resistant Cast Iron Handbook, American Foundry Society, Schaumberg, IL, 2000 Abrasion Resistant Castings for Handling Coal: Properties and Uses of the Ni-Hard Irons (1978), Nickel Institute, www.nickelinstitute.org, accessed June 2008 W. Fairhurst and K. Rohrig, Abrasion-Resistant High-Chromium White Cast Irons, Foundry Trade J., May 1974 “Founding. Abrasion-Resistant Cast Irons,” BS EN 12513:2000, British Standards Institute (EN Standard), 2005 F. Maratray, Choice of Appropriate Compositions for Chromium-Molybdenum White Irons, Trans. AFS, Vol 79, 1971, p 121–124 Ni-Hard Material Data and Applications, Reference Book, Series 11017, Nickel Development Institute, 1996 “Standard Specification for Abrasion-Resistant Cast Irons,” A 532/A 532M, Annual Book of ASTM Standards, Vol 01.02, ASTM International, 2005

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 904-908 DOI: 10.1361/asmhba0005328

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

High-Alloy Graphitic Irons Richard B. Gundlach, Stork Climax Research Services

HIGH-ALLOY GRAPHITIC IRONS have found special use primarily in applications requiring corrosion resistance or strength and oxidation resistance in high-temperature service. They are commonly produced in both flake graphite and nodular graphite versions. Those alloys used in applications requiring corrosion resistance are the nickel-alloyed (13 to 36% Ni) gray and ductile irons (also called Ni-Resist irons) and the high-silicon (14.5% Si) gray irons. The alloyed irons produced for high-temperature service are the nickel-alloyed gray and ductile irons, the high-silicon (4 to 6% Si) gray and ductile irons, and the aluminum-alloyed gray and ductile irons. Two groups of aluminum-alloyed irons are recognized: the 1 to 7% Al irons and the 18 to 25% Al irons. Neither the 4 to 6% Si irons nor the aluminum-alloyed irons are covered by ASTM International standards. Although the oxidation resistance of the aluminum-alloyed irons is exceptional, problems in melting and casting the alloys are considerable, and commercial production of the aluminum-alloyed irons is uncommon. In general, the molten-metal processing of high-alloy graphitic irons follows that of conventional gray and ductile iron production. The same attention to chill control and inoculation practice is necessary, and the conditions for successful magnesium treatment for producing nodular graphite are similar or identical to those for conventional gray and ductile irons. The high alloy contents influence the eutectic carbon content; therefore, carbon levels in these alloys are lower. In many cases, hypereutectic compositions (high carbon contents) in the ductile irons are avoided because of a greater tendency toward the formation of graphite flotation and degenerate graphite forms in sections larger than 25 mm (1 in.). The higher alloy contents affect the constitution of the irons, creating conditions that favor the formation of third phases and/or secondary eutectics during solidification. Therefore, many of the alloys commonly contain interdendritic carbides or silicocarbides in the as-cast

structure. These constituents remain after heat treatment and are an accepted part of the microstructure.

(500
F). They have been successfully used for furnace and stoker parts, burner nozzles, and heat treatment trays.

High-Silicon Irons for High-Temperature Service

High-Silicon Ductile Irons

Graphitic irons alloyed with 4 to 6% Si provide good service and low cost in many elevated-temperature applications. These irons, whether gray or ductile, provide good oxidation resistance and stable ferritic matrix structures that will not go through a phase change at temperatures up to 900
C (1650
F). The elevated silicon contents of these otherwise normal cast iron alloys reduce the rate of oxidation at elevated temperatures because silicon promotes the formation of a dense, adherent oxide at the surface, which consists of iron silicate rather than iron oxide. This oxide layer is much more resistant to oxygen penetration, and its effectiveness improves with increasing silicon content. Detailed information on the corrosion-resistant properties of high-silicon cast irons is provided in the article “Corrosion of Cast Irons” in Corrosion: Materials, Volume 13B of ASM Handbook (Ref 1).

High-Silicon Gray Irons The high-silicon gray irons were developed in the 1930s by the British Cast Iron Research Association and are commonly called Silal. In Silal, the advantages of a high critical (A1) temperature, a stable ferritic matrix, and a fine, undercooled type D graphite structure are combined to provide good growth and oxidation resistance. Oxidation resistance is further improved with additions of chromium, which in these grades can approach levels of 2% Cr. An austenitic grade called Nicrosilal was also developed, but the Ni-Resist irons have replaced this alloy. Applications. Although quite brittle at room temperature, the high-silicon gray irons are reasonably tough at temperatures above 260
C

The advent of ductile iron led to the development of high-silicon ductile irons, which currently constitute the greatest tonnage of these types of iron being produced. Converting the eutectic flake graphite network into isolated graphite nodules further improves resistance to oxidation and growth. The higher strength and ductility of the ductile iron version of these alloys qualify them for more rigorous service. The high-silicon ductile iron alloys are designed to extend the upper end of the range of service temperatures for ferritic ductile irons. These irons are used to temperatures of 900
C (1650
F). Raising the silicon content to 4% raises the A1 temperature to 815
C (1500
F), and at 5% Si the A1 is above 870
C (1600
F). The mechanical properties of these alloyed irons at the lower end of the range (4 to 4.5% Si) are similar to those of standard ferritic ductile irons. At 5 to 6% Si, oxidation resistance is improved and critical temperature increased, but the iron can be very brittle at room temperature. At higher silicon levels, the impact transition temperature rises well above room temperature, and the upper shelf energy is dramatically reduced. Ductility is restored when temperatures exceed 425
C (800
F). SAE standard J2582 (Ref 2) defines the three grades of silicon-molybdenum ductile iron most commonly used in automotive applications (Table 1). The three grades are differentiated by molybdenum concentration. For most applications, alloying with 0.5 to 1% Mo provides adequate elevated-temperature strength and creep resistance (Fig. 1). Higher molybdenum additions are used when maximum elevated-temperature strength is needed. High molybdenum additions (>1%) tend to generate interdendritic carbides of the Mo2C type, which persist even through annealing, and tend to reduce toughness

High-Alloy Graphitic Irons / 905 Table 1

Grades of high-silicon-molybdenum ductile iron (SAE J2582) Composition, wt%

Grade

1 2 3

C

Si

Mn

Mo

S

P

Mg

3.3–3.8 3.3–3.8 3.3–3.8

3.5–4.5 3.5–4.5 3.5–4.5

0.1–0.5 0.1–0.5 0.1–0.5

0.5 max 0.51–0.70 0.71–1.00

0.035 max 0.035 max 0.035 max

0.05 max 0.05 max 0.05 max

0.025–0.060 0.025–0.060 0.025–0.060

Source: Ref 2

Fig. 2

Micrograph of a 4Si-Mo ductile iron showing nodular graphite structure. Original magnification: 400

Fig. 1

Influence of molybdenum on (a) tensile properties and (b) creep resistance of 4% Si ductile iron at 705
C (1300
F)

and ductility at room temperature. Figure 2 shows the typical microstructure obtained in high-silicon-molybdenum ductile iron. Silicon lowers the eutectic carbon content, which must be controlled to avoid graphite flotation. For 4% Si irons, carbon content should range from 3.2 to 3.5% C, depending on section size, and at 5% Si it should be approximately 2.9% C. The unique properties of each element and its contribution to the total alloy system of

a nodular graphite cast iron are described in the article “Ductile Iron Castings” in this Volume. Melting Practice. For the high-silicon ductile irons, standard ductile iron melting practices apply. Cupola melting is acceptable, but these irons are commonly electric melted. Acid, neutral, or basic linings are used. Conventional ductile iron charge materials are also used; however, care should be taken to minimize chromium, manganese, and phosphorus. Manganese content should not exceed 0.5%, and it is preferable to keep it below 0.3%. Chromium should be no more than 0.1% and preferably below 0.06%. These elements promote as-cast embrittlement due to carbides and pearlite, which form networks in the interdendritic regions. These microstructural constituents are difficult to break down through subsequent heat treatment, and they degrade toughness and machinability. They are a particular problem in turbocharger applications, in which minimum ductility requirements must be met. Consequently, a larger proportion of pig iron in the charge is common. Silicon additions can be made in the ladle, but it is highly recommended that all silicon be added to the furnace, except for that required for magnesium treatment and postinoculation.

The same is true for molybdenum additions; either ferromolybdenum or molybdic oxide briquettes can be used. Conventional magnesium treatment techniques are used, and the same rules for balancing sulfur and setting residual magnesium levels apply to the high-silicon irons. These irons are more susceptible to the formation of intercellular carbides when magnesium levels exceed those needed to balance sulfur and to nodularize the graphite. Although these irons respond like standard ductile irons to most inoculants, foundry-grade ferrosilicon, containing 75 to 85% Si, is most often used for postinoculation. Pouring Practice. Due to increased silicon contents, higher pouring temperatures (>1425
C, or 2600
F) are required to develop a clean melt surface in the ladle. Therefore, pouring temperatures somewhat higher than those for conventional ductile irons should be used to minimize dross and slag defects. Molds, Patterns, and Casting Design. These alloys are normally sand cast; sand molds can be made of green sand, shell mold, and air-set chemically bonded sands suitable for gray and ductile irons. As with any graphitic iron, high mold rigidity is necessary to minimize mold wall movement and shrinkage porosity. Softer molds will produce bigger castings, but in general, little or no allowance for shrinkage is required in the pattern. As with conventional ductile irons, the highly reactive magnesium dissolved in these irons, along with the high levels of silicon, gives rise to the formation of dross and inclusions due to the reaction with oxygen in the air during pouring and filling of the mold. Consequently, clean ladles with teapot designs or skimmers are recommended. Furthermore, measures must be taken in designing the runners and gating system to minimize turbulence and to trap dross before it enters the mold cavity. Shakeout Practice. As mentioned previously, room-temperature impact resistance is low; therefore, riser and gate removal is somewhat easier with these alloys than with standard ductile iron grades. These irons are quite ductile at elevated temperatures, and they should be allowed to cool before cleaning. They do exhibit a ductility trough in the temperature range of 315 to 425
C (600 to 800
F), and riser and gate removal is aided when performed upon cooling through this temperature range. Heat Treatment. The high-silicon ductile irons are predominantly ferritic as-cast, but the presence of carbide-stabilizing elements will result in a certain amount of pearlite and often intercellular carbides. These alloys are inherently more brittle than standard grades of ductile iron and may have higher levels of internal stress due to lower thermal conductivity and higher elevated-temperature strength. These factors should be taken into account when selecting heat treatment requirements. High-temperature heat treatment is often advised to anneal any pearlite and to stabilize the casting against growth in service. A normal

906 / Cast Irons graphitizing (full) anneal in the austenitic temperature range is recommended when undesirable amounts of carbide are present. For the 4 to 5% Si irons, this will require heating to at least 900
C (1650
F) for several hours, followed by slow cooling to below 705
C (1300
F). At higher silicon contents (>5%), in which carbides readily break down, and in castings that are relatively carbide free, subcritical annealing in the temperature range of 720 to 790
C (1325 to 1450
F) for 4 h is effective in ferritizing the matrix. Compared to full annealing, the subcritically annealed material will have somewhat higher strength, but ductility and toughness will be somewhat reduced. Applications. The high-silicon and siliconmolybdenum ductile irons are currently produced as manifolds and turbocharger housings for trucks and some automobiles. They are also used in heat treating racks.

Austenitic Nickel-Alloyed Gray and Ductile Irons The nickel-alloyed austenitic irons are produced in both gray and ductile cast iron versions for high-temperature service. Austenitic gray irons date back to the 1930s, when they were specialized materials of minor importance. After the invention of ductile iron, austenitic grades of ductile iron were also developed. These nickel-alloyed austenitic irons have been used in applications requiring corrosion resistance, wear resistance, and high-temperature stability and strength (Ref 1). Additional advantages include low thermal expansion coefficients,

nonmagnetic properties, and cast iron materials having good toughness at low temperatures. When compared with corrosion and heat-resistant steels, nickel-alloyed irons have excellent castability and machinability. Applications of Nickel-Alloyed Irons. The nickel-alloyed (Ni-Resist) irons have found wide application in chemical process-related equipment such as compressors and blowers; condenser parts; phosphate furnace parts; pipe, valves, and fittings; pots and retorts; and pump casings and impellers. Similarly, in foodhandling equipment, the various alloy components include bottling and brewing equipment, canning machinery, distillery equipment, feed screws, metal grinders, and salt filters. In hightemperature applications, nickel-alloyed irons are used as cylinder liners, exhaust manifolds, valve guides, gas turbine housings, turbocharger housings, nozzle rings, water pump bodies, and piston rings in aluminum pistons. Figure 3 shows the typical microstructure obtained in Ni-Resist irons.

Austenitic Gray Irons Specification ASTM A 436 (Ref 3) defines eight types of austenitic gray iron alloys, four of which are designed to be used in elevatedtemperature applications and four in applications requiring corrosion resistance (Table 2). Nickel additions produce a stable austenitic microstructure with good corrosion resistance and strength at elevated temperatures. The nickel-alloyed irons are also alloyed with chromium and silicon for wear resistance and

oxidation resistance at elevated temperatures. Types 1 and 1b, which are designed for corrosion-resistant applications, are alloyed with 13.5 to 17.5% Ni and 6.5% Cu. Type 1 alloys are used to produce ring carriers used in conjunction with aluminum pistons in diesel engines. Types 2b, 3, and 5, which are principally used for high-temperature service, contain 18 to 36% Ni, 1 to 2.8% Si, and 0 to 6% Cr. With the development of ductile iron, most high-temperature applications shifted to similar Ni-Resist ductile iron grades. Type 4 alloys are alloyed with 29 to 32% Ni, 5 to 6% Si, and 4.5 to 5.5% Cr and are recommended for their stain-resistant properties.

Austenitic Ductile Irons Specification ASTM A 439 (Ref 4) defines the group of austenitic ductile irons (Table 3). The austenitic ductile iron alloys have similar compositions to the austenitic gray iron alloys but have been treated with magnesium to produce nodular graphite structures. Ductile iron alloys are available in every gray iron alloy type but type 1; this is because the high copper content of type 1 is not compatible with the production of nodular graphite iron. The ductile iron alloys have high strength and ductility, combined with the same desirable properties of the gray iron alloys. They provide frictional wear resistance, corrosion resistance, strength and oxidation resistance at high temperatures, nonmagnetic characteristics, and, in some alloys, low thermal expansivity at ambient temperatures.

Table 2 Compositions of austenitic gray iron alloys Composition, % Alloy

Type Type Type Type Type Type Type Type

C

1 1b 2 2b 3 4 5 6

3.00 3.00 3.00 3.00 2.60 2.60 2.40 3.00

max max max max max max max max

Si

Mn

Ni

Cu

Cr

1.00–2.80 1.00–2.80 1.00–2.80 1.00–2.80 1.00–2.00 5.00–6.00 1.00–2.00 1.50–2.50

0.5–1.5 0.5–1.5 0.5–1.5 0.5–1.5 0.5–1.5 0.5–1.5 0.5–1.5 0.5–1.5

13.50–17.50 13.50–17.50 18.00–22.00 18.00–22.00 28.00–32.00 29.00–32.00 34.00–36.00 18.00–22.00

5.50–7.50 5.50–7.50 0.50 max 0.50 max 0.50 max 0.50 max 0.50 max 3.50–5.50

1.5–2.5 2.50–3.50 1.5–2.5 3.00–6.00(a) 2.50–3.50 4.50–5.50 0.10 max 1.00–2.00

S

0.12 0.12 0.12 0.12 0.12 0.12 0.12 0.12

Mo

max max max max max max max max

... ... ... ... ... ... ... 1.00 max

(a) Where some machining is required, the 3.00–4.00% Cr range is recommended. Source: Ref 3

Table 3 Compositions of austenitic nodular irons Composition, %

Fig. 3

Micrograph of a D-5S Ni-Resist ductile iron showing nodular graphite structure. Original magnification: 400

Alloy type

C

Si

Mn

D-2(a) D-2B D-2C D-2M(b) D-3(a) D3-A D-4 D-5 D-5B D-5S

3.00 max 3.00 max 2.90 max 2.2–2.7 2.60 max 2.60 max 2.60 max 2.40 max 2.40 max 2.30 max

1.50–3.00 1.50–3.00 1.00–3.00 1.5–2.50 1.00–2.80 1.00–2.80 5.00–6.00 1.00–2.80 1.00–2.80 4.90–5.50

0.70–1.25 0.70–1.25 1.80–2.40 3.75–4.5 1.00 max(c) 1.00 max(c) 1.00 max(c) 1.00 max(c) 1.00 max(c) 1.00 max(c)

P

0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08

max max max max max max max max max max

Ni

Cr

18.00–22.00 18.00–22.00 21.00–24.00 21.0–24.0 28.00–32.00 28.00–32.00 28.00–32.00 34.00–36.00 34.00–36.00 34.00–37.00

1.75–2.75 2.75–4.00 0.50 max 0.20 max(c) 2.50–3.50 1.00–1.50 4.50–5.50 0.10 max 2.00–3.00 1.75–2.25

(a) Additions of 0.7–1.0% Mo will increase the mechanical properties above 425
C (800
F). (b) D-2M source: Ref 5; all others: Ref 4. (c) Not intentionally added

High-Alloy Graphitic Irons / 907 Austenitic ductile irons suitable for low-temperature service are specified by ASTM A 571/ A 571M (Ref 5). These are designated D-2M, with two strength classes in customary units and two classes in metric units. They are intended for pressure-containing parts in cryogenic applications. Melting Practice. In the past, these iron alloys were generally cupola melted, but today (2008), melting is almost exclusively done in electric furnaces. Choice of furnace linings is usually based on other alloys being melted in the shop; acid, neutral, or basic linings are used. Selection of charge materials is more critical in melting the higher-nickel alloys of the ductile iron versions, because they are more sensitive to the tramp elements that affect graphite structure. The high nickel content causes the materials to be more prone to hydrogen gas defects; therefore, charge materials should be thoroughly dry, and melting times should be short. The molten iron should only be superheated to the temperature necessary to treat and pour, and for as short a time as possible. Pouring is generally done above 1400
C (2550
F) to keep the molten iron surface clean and free of oxide. Magnesium treatment of the ductile iron alloys is normally accomplished with nickelmagnesium alloys. The nickel-magnesium treatment alloy is often added in the furnace; there is no concern for pyrotechnics. The same rules for balancing sulfur and setting residual magnesium levels apply to these nickel-alloyed irons. These irons are more susceptible to the formation of intercellular carbides when magnesium levels exceed those needed to balance sulfur and to nodularize the graphite. Although these irons respond like standard ductile irons to most inoculants, foundry-grade ferrosilicon, containing 75 to 85% Si, is most often used for postinoculation. Postinoculation is most often performed when tapping the furnace. Stream inoculation, where feasible, is also recommended for improved machinability. Carbon content for the type D-5S alloy must be monitored carefully, because section sensitivity is high. Although the ASTM specification allows up to 2.4% C, sections over 25 mm (1 in.) are susceptible to exploded and chunky graphite formation. Carbon levels of 1.6 to 1.8% are recommended for heavy-section castings. Molding and Casting Design. Conventional ferrous molding sands are used, including green sand, shell mold, and chemically bonded sands. The same precautions taken for high-strength irons apply to these alloys. Solidification should progress from thin to thick sections without interruption. Abrupt changes in section thickness should be avoided. Riser location should allow convenient access for riser removal. The shrinkage allowance is generally 21 mm/m ð1=4 in:=ftÞ, or 2.1%. Shakeout. These alloys can develop large thermal stresses upon cooling because of a

relatively low thermal conductivity, combined with high elevated-temperature strength and high thermal expansion rates. Consequently, care should be taken in shaking out too hot, and mold cooling is recommended for intricately shaped castings and castings of widely varying section thickness. Heat treatment of nickel-alloyed ductile irons serves to strengthen the casting and to stabilize the microstructure of the casting for increased durability. Stress-relief heat treatments are typically conducted at temperatures between 620 and 675
C (1150 and 1250
F) to remove residual casting stresses. Mold cooling to 315
C (600
F) is a satisfactory alternative to furnace stress relief. Annealing of some castings may be necessary to reduce hardness. Annealing is performed at 955 to 1040
C (1750 to 1900
F) for 30 min to 5 h, and this treatment will normally break down some of the carbides and spheroidize the rest. Heat treatment for stability of the microstructure for service at temperatures of 480
C (900
F) and above is performed by heating at 760
C (1400
F) for a minimum of 4 h and furnace cooling to 540
C (1000
F), followed by air cooling. An alternative treatment is to heat at 900
C (1650
F) for 2 h and furnace cool to 540
C (1000
F). These treatments stabilize the microstructure and minimize growth and warpage in service. The treatments are designed to reduce carbon levels in the matrix, and some growth and distortion often accompany heat treatment. Type 1 alloys are not generally amenable to this stabilizing treatment. Dimensional stability, when truly critical, can be ensured by heat treating at 870
C (1600
F) or higher for 2 h plus an additional hour per 25 mm (1 in.) of cross section, furnace cooling not faster than 55
C/h (100
F/h) to 540
C (1000
F), and then holding for 1 h per 25 mm (1 in.) of cross section and cooling uniformly. After rough machining, the casting should be reheated to 455 to 480
C (850 to 900
F) and held 1 h per 25 mm (1 in.) of cross section. Refrigeration and reaustenitization heat treatments are applied to type D-2 alloys to increase yield strength. Solution heat treating at 925
C (1700
F), refrigerating at 195
C (320
F), and then reheating between 650 and 760
C (1200 and 1400
F) will increase yield strength considerably without materially affecting magnetic properties or corrosion resistance in seawater or dilute sulfuric acid.

Detailed information on the heat treatment of ductile iron is available in the article “Ductile Iron Castings” in this Volume.

Aluminum-Alloyed Irons The aluminum-alloyed irons consist of two groups of gray and ductile irons. The lowalloyed group contains 1 to 7% Al, and the aluminum essentially replaces silicon as the graphitizing element in these alloys. The highalloyed group contains 18 to 22% Al. Irons alloyed with aluminum in between these two ranges will be white iron as-cast and have no commercial importance. The aluminum greatly enhances oxidation resistance at elevated temperatures and also strongly stabilizes the ferrite phase to very high temperatures—up to and beyond 980
C (1800
F). Like the silicon-alloyed irons, the aluminum irons form a tight, adherent oxide on the surface of the casting that is very resistant to further oxygen penetration. Unfortunately, the aluminum-alloyed irons are very difficult to cast without dross inclusions and laps (cold shuts). The aluminum in the iron is very reactive at the temperatures of the molten iron, and contact with air and moisture must be negligible. Care must be taken not to draw the oxide skin, which forms during pouring, into the mold in order to avoid dross inclusions. Methods for overcoming these problems in commercial practice are under development. At present, there is no ASTM standard covering the chemistry and expected properties of these alloys, and commercial production is very limited. In the past, the 1.5 to 2.0% Al irons have been used in the production of truck exhaust manifolds.

High-Silicon Irons for Corrosion Resistance Irons with high silicon contents (14.5% Si) constitute a unique corrosion-resistant ferritic cast iron group. These alloys are widely used in the chemical industry for processing and for transporting highly corrosive liquids. They are particularly suited to handling sulfuric and nitric acids (Ref 6). The three most common high-silicon iron alloys are covered in ASTM A 518/A 518M (Ref 7). These alloys (Table 4) contain 14.2 to 14.75% Si and 0.7 to 1.15% C.

Table 4 Compositions of high-silicon iron alloys Composition, % Alloy

Grade 1 Grade 2 Grade 3 Source: Ref 7

C

Mn

Si

Cr

Mo

Cu

0.65–1.10 0.75–1.15 0.70–1.10

1.50 max 1.50 max 1.50 max

14.20–14.75 14.20–14.75 14.20–14.75

0.50 max 3.25–5.00 3.25–5.00

0.50 max 0.40–0.60 0.20 max

0.50 max 0.50 max 0.50 max

908 / Cast Irons Grades 2 and 3 are also alloyed with 3.25 to 5% Cr, and grade 2 contains 0.4 to 0.6% Mo. Other compositions are also commercially produced with up to 17% Si. Melting Practice. Induction melting is the preferred method for these alloys. Induction melting permits the very tight control of chemistry needed in melting these materials in order to minimize scrap losses due to cracking. The alloys have a very low tolerance for hydrogen and nitrogen; thus, charge materials must be carefully controlled to minimize the levels of these gases. Vacuum treatment can be used to increase strength and density. The melting point of the eutectic 14.3% Si cast iron is approximately 1180
C (2160
F), and the alloy is generally poured at approximately 1345
C (2450
F). Because of the very brittle nature of high-silicon cast iron, castings are usually shaken out after cooling to ambient temperature. However, some casting geometries demand hot shakeout while the castings are still red hot, so that the castings can be immediately stress relieved and furnace cooled to prevent cracking. Molding and Casting Design. These alloys are routinely cast in sand molds, investment molds, and permanent steel molds. Cores must have good collapsibility to prevent fracture during solidification. Flash must be minimized because it will be chilled iron and can readily become an ideal nucleation site for cracks that propagate into the casting. Casting design for these high-silicon irons requires special considerations. To avoid cracking, sharp corners and abrupt changes in section size must be avoided. Casting designs should have tapered ingates to permit easy removal of gates and risers by impact and to minimize the amount of grinding required. Unless vacuum

degassing is employed, conventional risers are not generally used or desired. High-silicon cast irons are generally cast in sections ranging from 4.8 to 38 mm (3=16 to 1.5 in.). Thin sections fill well because of the excellent fluidity of these cast irons, but tend to chill and become extremely brittle white iron. Sections over 38 mm (1.5 in.) are prone to segregation and porosity, which reduce strength and corrosion resistance. Castings are generally designed with a patternmaker’s shrink rule of 18 mm/m (7=32 in./ft), or 1.8%. Heat Treatment. High-silicon irons can be stress-relieved by heating in the range of 870 to 900
C (1600 to 1650
F), followed by slow cooling to ambient temperatures to minimize the likelihood of cracking. Heat treatments have no significant effect on corrosion resistance. Machinability. High-silicon cast irons have hardnesses of approximately 500 HB and are normally considered machinable only by grinding. Machinability can be improved with higher additions of carbon and/or phosphorus, but only at the expense of other properties, such as strength and corrosion resistance. Applications. High-silicon irons are extensively used in equipment for the production of sulfuric and nitric acids, for sewage disposal and water treatment, for handling mineral acids in petroleum refining, and in the manufacture of fertilizer, textiles, and explosives. Specific components include pump rotors, agitators, crucibles, and pipe fittings in chemical laboratories.

REFERENCES 1. T.C. Spence, Corrosion of Cast Irons, Corrosion: Materials, Vol 13B, ASM Handbook, ASM International, 2005, p 43–50 2. “Automotive Ductile Iron Castings for High Temperature Applications,” J2582, Vol 1, SAE, Dec 2001, p 6.17 3. “Standard Specification for Austenitic Gray Iron Castings,” A 436, Annual Book of ASTM Standards, Vol 01.02, ASTM International, 2005 4. “Standard Specification for Austenitic Ductile Iron Castings,” A 439, Annual Book of ASTM Standards, Vol 01.02, ASTM International, 2005 5. “Standard Specification for Austenitic Ductile Iron Castings for Pressure-Containing Parts Suitable for Low-Temperature Service,” A 571/A 571M, Annual Book of ASTM Standards, Vol 01.02, ASTM International, 2005 6. S.K. Brubaker, Corrosion by Sulfuric Acid, Corrosion: Environments and Industries, Vol 13C, ASM Handbook, ASM International, 2006, p 659–667 7. “Standard Specification for Corrosion-Resistant High-Silicon Iron Castings,” A 518/A 518M, Annual Book of ASTM Standards, Vol 01.02, ASTM International, 2005 SELECTED REFERENCES  “High Temperature Materials for Exhaust

ACKNOWLEDGMENTS The author wishes to thank Warren Spear of The Nickel Development Institute and Don Stickle of the Duriron Company for their valuable assistance.

Manifolds,” J2515, Aug 1999, SAE Ferrous Materials Standards Manual, 2004 ed., HS30, SAE, p 365–374  “Properties and Applications of Ni-Resist and Ductile Ni-Resist Alloys,” Nickel Development Institute, Toronto, Ontario, Canada, 1998

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 911-917 DOI: 10.1361/asmhba0005295

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Steel Ingot Casting Donald J. Hurtuk

WHEN A HEAT OF STEEL is melted and refined, it is necessary to solidify it into useful forms for further processing or final use. Although most steel is now solidified in continuous casting operations, the original method for casting steel involved solidifying into ingots. This is still the preferred method for certain specialty, tool, forging, and remelted steels. The molten steel is transferred from the steelmaking vessel to a refractory-lined ladle, from which it is teemed into containers, called ingot molds, for solidification. Some of the methods, equipment, and theory for pouring and solidifying ingots are presented in this article.

Pouring Method

through a nozzle at the bottom of the ladle, but lip pouring by tilting the ladle is also done in small operations. The flow of steel from the bottom of a ladle is controlled by a stopper rod or, more commonly, a slide gate nozzle. Lip pouring from the ladle is controlled by the skill of the crane operator. The pouring rate is quite fast in top pouring, achieving a rate of rise of the steel in the mold in the range of 125 to 255 cm/min (50 to 100 in./ min). The steel stream from the ladle to the mold will oxidize during pouring, and considerable turbulence is generated in the mold. This method is damaging to the ingot mold, and mold iron consumption rates can be quite high. This method also produces a lower-quality ingot because of the oxidation and turbulence. However, toppoured steel can be poured quite rapidly. When

ingot molds are placed on a string of railroad cars, called a drag, this can be a very efficient way of pouring steel. Bottom pouring (or uphill teeming) involves teeming through a runner system that fills the ingot molds from the bottom until they are full, as shown in Fig. 2. This method generates much less turbulence than top pouring and produces better surface and internal quality in the ingots. It is also less damaging to ingot molds but does require additional equipment to direct the steel into the molds. A fountain or trumpet is required to funnel the steel from the ladle to a sprue plate or stool that has separate channels, or runners, to direct the steel to the molds. The trumpet and runners are refractory lined, and refractory ingates or outlets direct the steel into the ingot mold cavity. Bottom pouring has the

It has long been recognized that the way in which steel is poured to produce ingots is important. Of critical importance is the way in which steel fills the mold. Two basic types of pouring methods have been used for transferring steel from the ladle to the ingot mold. They are top pouring and bottom pouring. Top pouring (or direct teeming) involves teeming from the ladle directly into the top opening of an ingot mold until the mold is full, as shown in Fig. 1. Pouring is normally done

Fig. 1

Method for top pouring, or direct teeming, of steel ingots. Source: Ref 1

Fig. 2

Schematic arrangement and nomenclature for bottom pouring steel ingots. Source: Ref 2

912 / Steel Castings added advantage of filling multiple ingot molds at one time, as shown in Fig. 3. A flux is used in bottom pouring for several reasons. These include:
Prevent oxidation of the steel meniscus
Insulate the rising metal surface
Lubricate the flow of the steel meniscus

against the mold wall


Trap and absorb nonmetallic inclusions (Ref 1)

The flux is normally suspended in bags from the top of the mold and is designed to spread over the surface and begin melting as the molten steel fills the mold. The melted flux fills the gap between the steel and the mold, which improves thermal conductivity and lubricates the surface to generate excellent surface quality. The melting flux at the top helps remove inclusions from the steel to improve steel cleanliness. This is depicted in the drawings in Fig. 4. The application rate of bottom-poured fluxes is generally in the range of 1.0 to 2.3 kg/ton (2 to 5 lb/ton). Removal of nonmetallic inclusions can occur when the inclusions reach the flux/steel interface and become absorbed and removed from the steel. This absorption is controlled by surface tension according to the following relationship: sm  ss  si 0 > 0 where sm is the surface tension of the steel, ss is the surface tension of the slag, and si is the surface tension of the slag/steel interface. If the surface tension of the inclusions is similar to or greater than that of the steel, the inclusions will be absorbed into the slag (Ref 1). With good steelmaking and steel pouring practices, bottom pouring has been able to achieve steel cleanliness levels rivaling remelt quality. The grade of steel, pouring rate (or rate of rise in the mold), and flux properties must be carefully matched to achieve the best results. The pouring rate is much slower than in top pouring. Sometimes, the rate of rise of the steel in the mold must be kept as low as 50 to 75 mm/min (2 to 3 in./min) for particularly crack-sensitive grades. Normal bottom-pouring rates are in the 200 to 300 mm/min (8 to 12 in./min) range.

include mold designs that are open top, bottle top, open bottom, closed bottom, and plug bottom. Electrode ingots are tall, round ingots intended for remelting in electroslag remelting and vacuum arc remelting operations. These steels are very high quality and are applied to some of the most critical applications, such as aerospace and medical. Tool steel ingots are usually small versions of BEU designs for high-quality specialty applications. Ingots are often shaped on the top or bottom to maximize yields by optimizing deformation during rolling. These designs are incorporated into the design of the ingot molds and stools that are used to contain the molten steel during casting. Traditionally, ingots were categorized by the amount of gases released during solidification. The main categories are killed, semikilled, capped, and rimmed; examples are shown in Fig. 6. In this figure, ingot 1 is an example of

Fig. 3

Schematic for bottom pouring multiple ingots. Source: Ref 1

Fig. 4

Drawings depicting the role of bottom-poured flux in bottom pouring. Source: Ref 1

Fig. 5

Cross sections of various types of big-end-down and big-end-up mold designs. Source: Ref 2

Ingot Design The design of the ingot is dictated by the application and type of steel involved. A big-end-up (BEU) design is used to promote ingot soundness and ingot quality by establishing a solidification profile that promotes moving segregation and shrinkage out of the ingot to the hot top, or head, of the ingot. Some of the highest-quality steels are cast in this design. A big-end-down (BED) design is used to produce solidification structures that can be reduced in rolling for plate and sheet applications. With this design, centerline quality is not critical, since it will undergo significant reduction prior to achieving final size. The BEU and BED mold designs are shown in Fig. 5. These

Steel Ingot Casting / 913 a fully killed ingot where no gas is evolved during solidification. Ingots 2 to 4 are examples of semikilled ingots with increasing amounts of gas evolution. Ingot 5 is an example of a capped ingot where gas evolution is sufficient to alter solidification structures at the surface in the upper portions of the ingot. Ingots 6 to 8 are rimmed ingots with increasing amounts of gas evolution. Many of these types of ingots have now been replaced by continuous casting, where yield and quality are considerably higher. Fully killed steel ingots now represent the bulk of steel ingots cast in the steel industry.

Hot Tops Hot tops (or heads) are used on steel ingots to control the heat flow at the top of the ingot during solidification. This is done to improve the quality and yield of the ingot by providing a pool of liquid steel to feed the shrinkage as the ingot solidifies and to provide a region for segregates and nonmetallic inclusions to gather. During ingot solidification, heat flows from the hot top region, or head, in three ways:
Up to the atmosphere
Down to the ingot
Out through the sides

Hot top systems are designed to minimize the first two avenues and maximize the third. The more heat flow that can be directed down to the ingot and out through the mold walls, the better the resulting ingot quality and yield. Hot tops consist of sideboards on the sidewalls of the very top of the ingot and a topping compound or board on top of the ingot. Sideboards can be pure insulating or exothermic and are attached to the ingot mold walls or to a separate casting that sits on top of the mold. Topping compounds are generally powders placed on top of the ingots immediately after pouring. Sometimes, the topping compound can be in a preformed board much like the sideboards. Sometimes, hot tops are not used, which will change the way the ingot solidifies. Examples of hot-topped and non-hot-topped ingots are shown in Fig. 7. It is evident that the use of

hot tops confines the shrinkage that occurs during solidification to the very top of the ingot. This helps improve quality and yield. When high-temperature molten steel comes in contact with exothermic hot top materials, the following reaction occurs: FeO þ Al ! Fe þAl2 O3 þ Heat The generation of heat locally retards solidification of the steel, and the resulting properties of the exothermic materials following the exothermic reaction are highly insulating. This keeps the top of the ingot liquid while the body of the ingot solidifies. If the hot top is properly designed, it will keep the top of the ingot liquid until all of the body of the ingot has solidified. This will allow all shrinkage in the ingot to be fed from the hot top, so little or no shrinkage forms in the ingot itself. It also allows segregation and impurities to be pushed out of the ingot and into the head. By balancing the thermal properties of the sideboards and compound, the solidification of the ingot can be controlled to optimize quality and yield. Pure insulating hot top sideboards and compound produce similar effects to exothermic hot tops. However, without the heat generated by the exothermic reaction, insulating hot tops are less effective and are rarely used anymore, except in very large ingots. Insulating hot tops can be sufficient in very large ingots, where solidification times are long and heat retention in the ingot is extended. The hot top sideboards are generally secured in the mold or hot top casting prior to teeming. They begin to function as soon as teeming is complete. However, the topping compound must be added immediately after teeming is complete. It is critical that this be done as quickly as possible to avoid the start of solidification in the head of the ingot. The same is true if the topping compound is in a preformed board. If the topping compound is added immediately after teeming, solidification in the head will be retarded, and the ingot will solidify correctly.

Ingot Molds and Stools Ingot molds are containers into which molten steel is poured to solidify into workable ingots. Stools are plates on which the ingot molds are

sometimes placed. Ingot molds must be able to withstand the high temperature of liquid steel yet be able to conduct and radiate away the intense heat so the steel can solidify. For this reason, molds and stools are generally made of some form of cast iron that has the properties to satisfy these requirements. The most common ingot mold material is gray iron, comprised of large graphite flakes in a matrix of various portions of ferrite and pearlite. The graphite flakes provide excellent thermal conductivity to remove the intense heat of the liquid steel. Ferrite provides thermal shock resistance to avoid catastrophic failure when the steel is teemed into the mold. Pearlite gives the iron erosion resistance from the damaging effects of the hot steel stream. Compacted graphite iron and nodular or ductile iron have also been successfully used in ingot molds. The modified shape of the graphite in these types of irons reduces the thermal conductivity, but the erosion resistance and toughness of the iron increase significantly. Best results with these irons have been obtained when the ingot mold is of a symmetrical design, such as round or square and not too tall. Other designs tend to warp over time and cause significant operating problems. When these irons are used correctly, the life of ingot molds is generally extended over gray iron molds. These irons have not been as successfully used in stool applications but have had limited success in specialty applications. Ingot molds are recycled for use, and the cycle time between uses is a critical factor in the casting process. The cycle time must be appropriate so that the initial mold temperature at teeming is within an acceptable range. Molds that are too hot or too cold can cause serious operating and quality problems. An inventory of ingot molds must be sufficient to allow proper cycling between uses. The same is true for stools. The design of ingot molds is critical for ingot quality and mold life. The weight of the mold is generally 1 to 1.4 times the weight of the ingot, depending on the design. The BEU and closedbottom molds are at the higher end of this range, while BED and open-bottom molds are

Fig. 7

Fig. 6

Series of structures in various types of ingots. See the text for explanation. Source: Ref 2

Depiction of hot-topped and non-hot-topped killed steel ingots. 1, big-end-up, hot topped; 2, big-end-down, hot topped; 3, big-end-up, not hot topped; 4, big-end-down, not hot topped. Source: Ref 2

914 / Steel Castings at the lower end of this range. To attain uniform heat flow during solidification of the ingot, the mold midface is usually thicker than the corners. A normal ratio of corner-to-midface thickness is 0.87 to 0.92, while lower ratios are necessary when using modified flake irons, such as compacted graphite or ductile iron. The corner-to-midface ratio can have a significant impact on solidification structures, as shown in Fig. 8. When the mold is designed properly, heat flow during ingot solidification is uniform, resulting in uniform ingot structures. Considerable work has been performed to optimize mold design relative to ingot solidification. Figure 9 shows designs by A.P. Banks in a 1967 British patent. Factors such as grade

of steel, pouring temperature, and type of mold iron must be considered to fully optimize mold design. Some molds have corrugated inside surfaces to improve surface quality and crack resistance in the ingot. However, these usually shorten the life of ingot molds, since the greater surface area is prone to deteriorate faster over time than smooth-wall molds. Hot top castings that create the head in ingot casting are usually made of gray iron. Modified graphite irons have had some success in this application as well. Sprue plates are bottom plates that contain the runners in bottom pouring that feed steel to the ingot molds. Trumpets are refractory-lined castings that funnel the steel from the ladle to the sprue plate and ingot molds. Sprue plates and trumpets are usually made of gray iron and experience long life.

Ingot Solidification Ingot solidification takes place as heat is removed from the steel through the ingot mold to the surrounding atmosphere. Ingot solidification patterns follow the configuration of the inside cavity of the ingot mold and are further influenced by the hot top. Typical solidification of an 81 cm (32 in.) square BEU hot-topped ingot is shown in Fig. 10. The rate of heat removal controls the resulting solidification structures that form in the ingot. Normal structures usually include a chill zone, a columnar dendritic zone, and an equiaxed zone, as shown in Fig. 11. The chill zone forms rapidly as the hot liquid steel contacts the relatively cold mold wall, creating a

Fig. 8

Influence of mold corner-to-midface ratio on ingot solidification. Structures in ductile iron molds. Source: Ref 1

Fig. 9

Mold designs to optimize heat flow. Source: Ref 3

very high thermal gradient at the ingot surface. This zone of planar solidification develops just before an air gap forms between the steel and the mold wall as the solidified steel shrinks away from the mold wall. The structures quickly transition to columnar dendritic structures as the thermal gradient drops and stabilizes somewhat. This still-high thermal gradient is the driving force for the columnar dendritic structures, according to the principle of constitutional supercooling, as depicted in Fig. 12. Segregates ahead of the advancing interface in the region of high thermal gradient keep the liquid steel supercooled, which promotes the columnar growth. As the thermal gradient decreases as solidification progresses, this driving force disappears, and the structures become equiaxed toward the center of the ingot. The structures in continuous casting are identical to those in ingot casting, as shown in the macroetched cross section in Fig. 13. Considerable work has been done in recent years to study continuously cast solidification structures. Most of what has been learned is directly transferable to statically cast ingots. Extensive columnar structures promote segregation and bridging along the centerline. It is generally preferred to control these structures to minimize centerline segregation and shrinkage. The temperature of the liquid steel has an effect on these structures. It is known that superheat directly affects the length of the columnar dendrites, as shown in Fig. 14. Reduced superheat can be used as a way to control columnar structures. Steel chemistry also affects columnar structures. Carbon content is the most notable element and influences

Steel Ingot Casting / 915

Time after pour (minutes)

66 20

100 60

140

120 80 40

160

140

20

60 100

40 80 120

72

60

54

48

42

36

120

Height (inches)

140

30

24

100

18

80

12

Fig. 12

Constitutional supercooling in alloy solidification. Source: Ref 5

60 40 20

6

0

Fig. 10

Solidification patterns in an 81 cm (32 in.) square bottom-end-up hot-topped ingot of killed steel. Source: Ref 2

Fig. 14

Fig. 13

Macroetched cross section of a 22.2 by 25 cm (8.75 by 10 in.) continuously cast bloom in the as-cast condition. Source: Ref 6

Fig. 11 Ref 4

Sketch of ingot structures showing chill zone, columnar zone, and equiaxed zone. Source:

columnar dendrites according to the carbon content, as shown in Fig. 15. This effect is controlled by the peritectic reaction that steel undergoes upon solidification, which is: Lþd!g where L is the liquid, d is delta iron, and g is austenite. The high-temperature portion of the iron-carbon equilibrium phase diagram, where this reaction occurs, is shown in Fig. 16. This

Influence of superheat on columnar structures. Source: Ref 6

reaction is important because significant shrinkage occurs as the body-centered cubic crystal structure of d-ferrite transforms to the more dense face-centered cubic crystal structure of austenite. This is the driving force behind much of the shrinkage that occurs during solidification and the formation of the air gap between the mold and ingot. The shrinkage is largest at approximately 0.10% C, at the onset of the peritectic reaction. Columnar structures are minimized, but surface quality suffers at this carbon content. Shrinkage is least at approximately 0.60% C, beyond the limits of the peritectic reaction. Columnar structures are maximized

916 / Steel Castings

Fig. 15

Plot showing influence of carbon content on columnar structures. Source: Ref 6

at this carbon content, since better contact between the ingot and mold is maintained. This keeps the thermal gradient high and provides the driving force for columnar structures. Surface quality is usually better at this carbon content. Beyond 0.60% C, columnar structures are reduced, but, of course, there is more carbon to segregate. Centerline carbon segregation can be quite troublesome in steels above 1.0% C. Nickel content also influences columnar dendrites, much as carbon does, as seen in Fig. 17. This effect is also controlled by a peritectic reaction, and the limits of the reaction in ironnickel at 4.0 and 5.4% Ni represent the concentrations of minimized and maximized columnar

structures, respectively, as shown in Fig. 18. The mechanism involved is the same as in the case of carbon. Most other solute elements have the effect of refining solidification structures by promoting equiaxed structures with increasing concentration. Columnar dendritic solidification structures are depicted in Fig. 19. These dendrites have primary and secondary arms. New primary arms can form by branching from secondary arms. Sometimes, tertiary arms form on secondary arms. All of this generates considerable segregation during solidification and creates a structure that can easily entrap inclusions. Columnar structures will increase centerline segregation and can even bridge at the centerline to create shrinkage. Equiaxed structures reduce these problems. Examples are shown in Fig. 20. It is generally considered best to produce as many equiaxed structures as possible to maximize ingot quality.

Ingot Stripping

Fig. 16

High-temperature portion of the iron-carbon equilibrium phase diagram. Source: Ref 7

Once ingot solidification is complete, and in some cases even earlier, the ingot is removed from the mold by a process called stripping. For BED designs, the ingot mold is usually lifted from the ingot by an overhead crane. For BEU designs, the ingot can be extracted from the ingot mold by an overhead crane. Sometimes, it is necessary to invert the ingot mold and dump the ingot out. Ingots can be stripped before solidification is complete to shorten the cycle time for mold reuse. This is normally done with BED ingots, where center quality is less important. Movement of an ingot with liquid centers will certainly disrupt the solidification front and create segregation and possibly shrinkage problems. However, if the ingot will undergo significant reduction into certain flat rolled products, this will not adversely affect the quality of the final product. Ingots are then generally charged into a soaking pit for reheating before further processing by rolling. Ingots for remelt or forging are slow cooled until they reach ambient temperature and can then be further processed.

Solidification Simulation

Fig. 17

Plot showing influence of nickel content on columnar structures. Source: Ref 6

Ingot producers have found value in simulating the solidification of ingots by various available simulation packages when considering new ingots or designs. Some simulations have remarkable accuracy in predicting solidification patterns and times. It has been found that seemingly small changes to hot top design, ingot design, pouring temperature, and so on can have disastrous effects on ingot quality and yield. A consideration of these possible effects by simulation before application in the plant can avoid many problems and expense.

Steel Ingot Casting / 917

Fig. 19

Columnar dendritic solidification in alloys. Source: Ref 5

REFERENCES

Fig. 18

Iron-nickel equilibrium phase diagram. Source: Ref 8

Fig. 20

Schematic illustrations of columnar and equiaxed ingot structures. Source: Ref 5

1. R.L. Harvey and A.P. Banks, “Bottom Pouring of Steel Ingots,” Foseco Inc. Seminar (Cleveland, OH), 1983 2. H.E. McGannon, Ed., The Making, Shaping and Treating of Steel, 8th ed., United States Steel Corporation, Pittsburgh, PA, 1964 3. A.P. Banks, British Patent 1,086,946, 1967 4. T.F. Bower and M.C. Flemings, Trans. AIME, Vol 239, 1967, p 1620 5. M.C. Flemings, Solidification Processing, McGraw-Hill, Inc., 1974 6. D.J. Hurtuk, “Factors Influencing Steel Solidification Structures,” Ph.D. thesis, Case Western Reserve University, Cleveland, OH, 1981 7. J. Chipman, Thermodynamics and Phase Diagram of the Fe-C System, Met. Trans., Vol 3, 1972, p 55–64 8. J.I. Goldstein and R.E. Ogilvie, A Re-Evaluation of the Iron Rich Portion of the Fe-Ni System, Trans. Met. Soc. AIME, Vol 233, 1965, p 2083

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 918-925 DOI: 10.1361/asmhba0005296

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Steel Continuous Casting Robert D. Pehlke, University of Michigan

THE ADVANTAGES OF CONTINUOUS CASTING in primary metals production have been recognized for more than a century. In recent decades, a dramatic growth of this processing technology has been realized in ferrous as well as nonferrous metal production. The principal advantages of continuous casting are a substantial increase in yield, a more uniform product, energy savings, and higher manpower productivity. These advantages and the ease of integration into metals production systems have led to the wide application of continuous casting processes.

Historical Aspects of Continuous Casting of Steel One of the earliest references to continuous casting is a patent granted in 1840 to George Sellers, who had developed a machine for continuously casting lead pipe (Ref 1). There is some indication that this process had been underway before Sellers’s patent, which was directed toward improvement of this continuous casting process. The first work on continuous casting of steel was by Sir Henry Bessemer, who patented a process for “manufacture of continuous sheets of iron and steel” in 1846 and made plant trials on continuous casting of steel in the 1890s (Ref 2). Although continuous casting had its start before the beginning of the 20th century, it was not until the mid-1930s in Germany that commercial production of continuously cast brass billets was introduced. Sigfried Junghans, an active inventor of casting technology, provided many improvements in the process, in particular the introduction of the oscillatingmold system to prevent the casting from sticking to the mold. Further development of the process for the casting of nonferrous metals continued, including the installation of processing units in North America. Mold lubrication in the form of oil, or, more recently, lowmelting slag powders, was introduced. Taper of the mold to compensate for metal shrinkage on solidification provided improved heat transfer and, more importantly, fewer cracks. In 1935, a plant with casting rolls for continuous production of brass plates was operated

at Scovill Manufacturing in the United States, and the Vereinigte Leichtmetallwerke in 1936 started a semicontinuous casting machine for aluminum alloys. Immediately after World War II, commercial development of continuous casting of steel began in earnest, with pilot plants at Babcock and Wilcox Company (United States), Low Moor (Great Britain), Amagasaki (Japan), Eisenwerk Breitenfeld (Austria), BISRA (Great Britain), and Allegheny Ludlum Corporation (United States). These were followed by production plants for casting billets in the West and stainless slabs in the Soviet Union and Canada (the latter at Atlas Steels) (Ref 3). The Schneckenburger and Kung patent on the curved strand was filed in Switzerland in 1956, and production was commercialized with a billet machine at Von Moosche Eisenwerke (Switzerland) in 1963. In 1961, at Dillingen Steelworks (West Germany), the first verticaltype large slab machine with bending of the strand to horizontal discharge was started up. In 1964, Shelton Iron and Steel (Great Britain) was the first new steelworks to turn out its entire production by continuous casting, consisting of four machines with eleven strands for medium to very large bloom sizes, and operating in connection with Kaldo converters. That same year, the first Concast S-type curved-mold machine for large slabs was started up at Dillingen Steelworks (West Germany). The height of this type of machine was less than 50% of the corresponding height of a vertical type of machine. In the same year, a bow-type slab caster was presented by Mannesmann (West Germany). In 1968, McLouth Steel (United States) started up four curved-strand slab casting machines immediately after the first four-strand low-head caster for large slabs in the West was started up at the Weirton Steel Division of National Steel (United States). Subsequently, National pioneered the casting of slabs for tinplate applications (Ref 3). Continuous casting of steel is entering a new era of development, not only with respect to its increasing application in the production process but also in its own evolution as a process and its interaction with other processes in steel manufacture. Continuous casting output has shown an accelerating growth curve. More than 80%

of current world steel production is continuously cast, and continuous casting in Japan exceeds 90%. The advantages of the process, along with its developments and current challenges for improvement, are outlined in the following sections.

General Description of the Process The purpose of continuous casting is to bypass conventional ingot casting and to cast to a form that is directly rollable on finishing mills. The use of this process has resulted in improvement in yield, surface condition, and internal quality of product when compared to the previous ingot-made material. Continuous casting involves the following sequence of operations:  Delivery of liquid metal to the casting strand  Flow of metal through a distributor (tundish)

into the casting mold

 Formation of the cast section in a water-

cooled copper mold

 Continuous withdrawal of the casting from

the mold

 Further heat removal to solidify the liquid

core of the casting by water spraying beyond the mold  Cutting to length and removing the cast sections The main components of continuous casting machines are presented in Fig. 1. Molten steel in a ladle is delivered to a reservoir above the continuous casting machine, called a tundish. The flow of steel from the tundish into one or more open-ended, water-cooled copper molds is controlled by a stopper rod-nozzle or a slide gate valve arrangement. To initiate a cast, a starter, or dummy bar, is inserted into the mold and sealed so that the initial flow of steel is contained in the mold and a solid skin is formed. After the mold has been filled to the desired height, the dummy bar is gradually withdrawn at the same rate that molten steel is added to the mold. The initial liquid steel freezes onto a suitable attachment of the dummy bar so that the cast strand can be

Steel Continuous Casting / 919

Fig. 1

Main components of a curved continuous casting strand. Source: Ref 4

Fig. 2

Principal types of continuous casting. V, vertical; VB, vertical with bending; VPB, vertical with progressive bending; CAS, circular arc with straight mold; CAC, circular arc with curved mold; PBC, progressive bending with curved mold; H, horizontal. Source: Ref 5

withdrawn down through the machine. Solidification of a shell begins immediately at the surface of the copper mold. The length of the mold and the casting speed are such that the shell thickness is capable of withstanding the pressures of the molten metal core upon exiting from the copper mold. To prevent sticking of the frozen shell to the copper mold, the mold is normally oscillated during the casting operation, and a lubricant is added to the mold. The steel strand is mechanically supported by rolls below the mold, where secondary cooling is achieved by spraying cooling water onto the strand surface to complete the solidification process. After the strand has fully solidified, it is sectioned into desired lengths by a cutoff torch or shear. This final portion of the continuous casting machine also has provision for disengagement and storage of the dummy bar. Several arrangements are now in commercial use for the continuous casting of steel. The types of continuous casting machines in use include vertical, vertical with bending, curved or S-strand with either straight or curved mold, curved strand with continuous bending, and horizontal. Examples of the principal types of

machines currently producing slabs are shown in Fig. 2. Most of the original continuous casting machines for steel were vertical machines. Vertical machines with bending and curved strand machines, although more complicated in their construction, were developed to minimize the height of the machine and allow installation in existing plants without modification of crane height. Four basic caster designs for slabs are shown in Fig. 3, with an indication of the required installation height and the corresponding solidification distance or metallurgical length (mL).

Plant Layout The design and layout of a steelmaking facility often focuses initially on continuous casting. The optimal plant layout varies markedly from one installation to another. One major factor in the configuration is whether or not the steelmaking complex is being constructed on a greenfield site or being added (“shoehorned”) into an existing works. Many of the major

integrated steel works in Japan were constructed as greenfield installations during the period from 1960 to 1975. Most of the minimills constructed throughout the world, and in particular in the United States, were also built on greenfield sites. In building a greenfield site, the plant layout should incorporate two major features: a smooth and well-organized arrangement for materials handling and flow, and the capacity for future expansion. Generally, these plants are designed for 100% continuous casting, and no ingot facilities are included (Ref 7). Nearly all of the recently built minimills incorporate one or more electric furnaces and provide for 100% continuous casting of billets. Chaparral Steel Company at Midlothian, TX, is an excellent example (Ref 8). A profile of a proposed large-scale electric furnace billet casting plant is shown in Fig. 4. A twin-strand slab caster was shoehorned into the Number 4 basic oxygen furnace shop at the previous Inland Steel Company. The addition of this caster substantially increased the output of this facility, where ingot casting was the rate-limiting step in production. The arrangement of this installation, showing the caster and ingot facilities, is presented in Fig. 5. An important characteristic of the plant layout, and in particular of the materials handling facilities, is the concept that the continuous casting machine cannot wait. This design and operating philosophy has had a dramatic impact on steelmaking operations, which have now become a synergism to the continuous casting facilities. Because of this shift in priorities, marked improvements in productivity have been developed for continuous casting, as outlined subsequently. Dramatic increases in energy costs, as well as the desire for higher productivity, led to the development of the “hot connection.” Substantial energy savings can be achieved by directly charging the hot continuously cast slab or billet to the reheating furnaces of the rolling mill. The latest installations have included direct in-line hot rolling of the cast product.

Near-Net Shape Casting A current trend in the steel industry is to develop processes for casting steel close to final product shape, that is, near-net shape casting. A conventional caster produces slabs 150 to 350 mm (6 to 14 in.) thick, which are used to produce hot rolled strip 2.5 to 25 mm (0.10 to 1.0 in.) thick. The amount of energy to reduce the thick slabs to 50 mm (2 in.) thick represents a significant part of the total production cost. Consequently, processes for production of thinner primary thicknesses are actively being pursued. It is anticipated that following thin-slab production will come processes for strip (5 to 20 mm, or 0.2 to 0.8 in.) and thin strip (<5 mm, or 0.2 in., thick). A strip

920 / Steel Castings

Fig. 3

Four basic designs for continuous casting machines. ML, metallurgical length. Source: Ref 6

casting plant has been built in Indiana as a joint venture of Nucor Steel Corporation and Broken Hill Proprietary, Ltd. of Australia. The thin-slab casting process was developed by SMS Concast of Germany (Ref 11). This process was first commercialized by Nucor Steel Corporation at their Crawfordsville, IN, plant (Ref 12). The process uses a convex or funnel-shaped mold to cast slabs 50 mm (2 in.) thick, as illustrated in Fig. 6. The overall plant height for thin-slab casting is 4 to 5 times smaller than for conventional slab casters, because the thin slab solidifies faster and can be bent over a radius of only 2 to 3 m (6 to 10 ft), as compared to 10 to 12 m (33 to 39 ft) for a conventional slab caster. Experience in production has shown that mold life for the thin-slab caster is shorter than for conventional slab casters. This occurs because of the hotter mold temperatures at the marked change in slab contour during the casting process within the mold of the thin-slab caster. The near-net shape concept has also been applied to continuously cast steel long products. In the late 1960s, Concast installed a “dog bone”-shaped cross-sectional casting machine at Algoma Steel Corporation in Sault Ste. Marie, Ontario, Canada, for producing a primary product for rolling into small-tomedium-section I- or H-beams. More recently, Nucor Steel Corporation has installed similar casting machines using steel from an electric are furnace to produce beam products ranging up to 1145 mm (45 in.) in the web.

Tundish Metallurgy

Fig. 4

Proposed melt shop capable of producing 1.2 million Mg (1.32 million tons) of billets annually. Source: Ref 9

Fig. 5

Plant layout at the previous Inland Steel Company. Source: Ref 10

A number of methods for distributing liquid steel to the mold or to several molds of the casting machine have been investigated. Use of a tundish with appropriate flow control has been found to be a superior method for the production of quality steel. Considerable effort has been directed toward improvement of refractories and development of methods for preventing nozzle blockage. Geometric arrangements in tundishes, including the use of pouring pads, dams, and weirs, have provided suitable fluid flow characteristics to maximize the separation of nonmetallic inclusions, resulting in improved quality. The optimization of these configurations has been greatly advanced through the use of modeling of fluid flows in various tundish arrangements, including both physical (water) modeling and computer modeling. A cross section of a typical tundish design for a twin-strand slab caster showing these features is presented in Fig. 7. Another factor in separation of nonmetallic inclusions in the tundish is tundish size, because of its effect on residence time. Figure 8 indicates the trend in tundish size for major slab casting installations in Nippon Steel Corporation over the decade of the 1970s. The tundish is supported with either one of two tundish cars or a two-position turret.

Steel Continuous Casting / 921

Fig. 6

Fig. 7

Typical tundish design for a twin-strand slab caster. Source: Ref 4

Fig. 9

Pouring shrouds from ladle to tundish to mold. Source: Ref 15

Schematic drawing of the convex mold for thin-slab casting. Source: Ref 13

    

Fig. 8

Increases in capacity of tundishes for a two-strand slab caster. Source: Ref 14

Frequently, the double turret provides for changing of the tundish while continuing the casting process, which allows extended sequence casting. The changing of the tundish not only occurs at an appropriate time but also allows a change in steel composition with a minimal length of composition transition in the cast product. Tundish linings have been the subject of a considerable amount of patent effort. The lining of the steel structure includes a permanent lining plus a working lining that resists the erosion of the molten steel and tundish slag. Another

lining arrangement used a permanent lining and performed refractory boards against the liquid steel, supported by a semipermanent or back-up lining. As tundish technology has advanced, several functions have been assigned to the tundish, including:  Act as a transfer and distributing vessel

between ladle and mold(s)

 Perform as a flow buffer for control of cast-

ing speed

 Serve as a reservoir for liquid steel during

ladle changes

Become a continuous steel reactor/refiner Promote inclusion removal Act as a vessel for final alloying of the steel Thermally insulate the liquid steel Protect the liquid steel from atmospheric reoxidation

Pouring Stream Protection Reoxidation of the molten steel is to be avoided. The use of refractory shrouds between the ladle and the tundish and the tundish and the mold has been adopted for slab casting, as illustrated in Fig. 9 (Ref 15). One of the difficulties encountered in continuous casting of small sections has been the protection of the pouring stream from tundish to mold because of the inapplicability of a pouring tube and the mechanical difficulty caused by the oscillation of the mold relative to the fixed tundish. In one instance, this was overcome by the use of a flexible bellows, which was successfully applied to the continuous casting of

922 / Steel Castings

Fig. 11

Oscillating-mold systems. (a) Long lever arm. (b) Short lever arm. (c) Four arm.  indicates bearing centers. Source: Ref 4

Fig. 10

Cross-sectional views of typical continuous casting molds for (a) billets, (b) blooms, and (c) slabs. Source: Ref 4

special-product quality steel bars. More recently, further development has been made in the application of gaseous shrouds, where the steel pouring stream is surrounded by a cylinder filled with a gas stream. Also, the use of ceramic shrouds to protect pouring streams on billet machines has been advanced by materials that allow thinner cylindrical walls for the shrouds. Also, protection of the surface of the molten steel pool in the tundish has been achieved through the use of suitable synthetic oxide slags, insert gas blanketing, and a refractory cover to seal the tundish. The metal level in the tundish, as well as the fluid flow pattern, is important in avoiding the ingestion of the slag layer into the metal stream flowing downward into the mold.

Molds for Continuous Casting of Steel The flow-through water-cooled copper mold is the key element of the casting machine. Special attention has been given to problems associated with the design and material requirements for molds. A number of different designs have been used, including thin-wall tube-type molds, solid molds, and molds made from plate. Plate molds were found to provide excellent mold life and to avoid the necessity for

manufacture of molds from solid copper blocks. Steel and brass, as well as copper, have been used for molds, but the most outstanding material is nearly pure copper with small additions of elements that promote precipitation hardening or raise the recrystallization temperature, because both effects provide longer mold life. Mold coatings are applied to extend service life. A chromium coating is generally used, often with an intermediate layer of nickel for improved coherence. Cross-sectional views are shown of typical billet, bloom, and slab molds for continuous casting in Fig. 10. The most suitable length for a continuous casting mold has been found to be 510 to 915 mm (20 to 36 in.), a range that seems to remain constant regardless of section size. This surprising result can be explained by the higher rates of heat removal achieved with smaller sections and higher casting rates. Also, a thinner skin can be permitted for smaller sections exiting from the mold than for larger sections, because bulging of the solidifying shell will be less severe. At higher casting rates, the use of an increased taper in the mold is necessary to maintain high heat removal rates, particularly for the narrow faces of slab molds. Slab molds with multitapers on the narrow faces have been used with considerable success. Based on the temperature difference between the mold surface and the surface of the solidifying steel shell, which decreases with lower

position in the mold, the taper of the narrow faces should be greater near the meniscus and decrease lower in the mold. This continuous decrease in taper has been approximated by multistraight tapers. A limitation, as with single-taper molds, is that the optimal taper or mold profile changes with casting speed and steel composition. Oscillating molds, as used by Junghans, have been adopted almost universally, although fixed molds can be successfully used with efficient lubricating systems. The oscillation is usually sinusoidal, a motion that can be achieved easily with simple mechanical arrangements. Examples of three of the most common types of oscillating-mold systems are shown in Fig. 11. A fairly short stroke and a high frequency are used to provide a short period of “negative strip” during each oscillation, in which the mean downward velocity of the mold movement is greater than the speed of withdrawal of the casting strand in the casting direction. Oscillating frequencies are being increased from 50 to 60 cycles per minute (cpm) up to 250 to 300 cpm, with the benefits of shallower oscillation marks, less cracking, and reduced conditioning requirements. Molten metal in the slab mold is normally covered with a layer of mold powder to protect the metal from reoxidation and to absorb inclusions. The powder has a low melting point and flows over the liquid steel to provide mold lubrication and to help control heat transfer. Rapeseed oil, which has since been replaced

Steel Continuous Casting / 923

Fig. 13

Comparison of spray systems used in continuous casting. (a) Conventional spray. (b) Air-water mist spray. Source: Ref 17

Fig. 12

Four types of shell supporting systems used in continuous casting. (a) Roll. (b) Cooling grid. (c) Cooling plate. (d) Walking bar. Source: Ref 16

Table 1

Record of sequence casting in Japan Weight of steel per cast

Product

Company

Continuous casting plant

Heats per cast

Slab Slab Slab Slab Bloom Bloom Bloom Bloom Billet Billet Billet

Nippon Kokan Nippon Kokan Kawasaki Steel Nippon Steel Sumitomo Metals Daido Steel Sumitomo Metals Nippon Steel Osaka Steel Kokko Steel Funabashi Steel

Keihin Fukuyama Mizushima No. 5 Yamata No. 2 Wakayama No. 3 Chita Kokura Kamishi Okajima Main Plant Main Plant

270 244 204 186 1129 320 281 202 214 174 99

Mg

22,523 54,360 51,000 30,199 160,110 19,091 17,648 17,857 17,455 5,063 5,535

tons

24,775 59,796 56,100 33,219 176,124 21,000 19,413 19,643 19,200 5,569 6,089

Time of achievement

Sept 1974 Aug 1981 Nov 1980 July 1974 July 1982 Feb 1984 April 1982 Jan 1979 July 1978 Aug 1978 Oct 1975

Source: Ref 14

by synthetic oils, has typically been used to prevent sticking to the mold in billet casting. Metal flow rates are matched with slab casting speeds, using a stopper rod in the tundish, a slide gate, or a metering nozzle just above the shroud to control the delivery rate. Billets are normally cast with fixed metering nozzles, and the strand speed is adjusted to any changes in steel flow rate. Support of the thin steel shell exiting the mold is required, particularly for slab casters. Several systems have been used (Fig. 12), all of which provide intensified direct water cooling (Ref 16). Water spraying (secondary cooling) is critical to the process in that maximum cooling should be accomplished, but overcooling and large rapid temperature increases and decreases must be avoided. The amount of heat removed by water sprays depends on the volume of the water, its temperature, and, in particular, the method of delivery, including spray pressure. Pressure is important in that the spray should be sufficiently intense to penetrate the blanket

of steam on the surface of the solidifying strand. The thermal conductivity of steel is relatively low; consequently, as the surface temperature decreases and the shell thickness increases, cooling water has less influence on the solidification characteristics. The interrelationship between support rolls and the spray water and its delivery characteristics is quite important, particularly for the casting of wide slabs. Mixed spraying of air and water develops a unique secondary cooling system with a uniform water droplet size. Air-mist cooling is illustrated in Fig. 13, along with conventional water spray cooling, as applied to slab casting. Although more costly, the air-mist spray system has a more uniform spray pattern and intensity that offers excellent spray cooling control. Decreases in transverse and longitudinal cracking with air-mist cooling have been reported (Ref 18, 19). The straightening operation on curved strand casters has required special design and operating control. In general, temperatures at

Fig. 14

Dummy bar arrangement. (a) Top feeding. (b) Bottom feeding. Source: Ref 20

or above 900 to 1050  C (1650 to 1920  F) have been recommended to avoid conditions under which certain grades of steel have limited ductility and are susceptible to cracking. Multipoint and four-point straighteners have reduced imposed strains, and compression casting systems have reduced tensile stresses. Uniform temperature of the strand, including corners, which tend to cool more quickly, is required. The dummy bar, which is used to stopper the mold for the initiation of a cast, can be inserted from above or below, depending on the individual installation. Some arrangements are shown in Fig. 14. The top-feeding arrangement, which “chases” the last metal cast through the machine, offers a productivity advantage.

Productivity Improvements While continual increases in casting speeds over the years have led to improvement in casting machine productivity, the most dramatic factor has been sequence casting, that is, continuous-continuous casting. Perfection of this development has required extraordinary achievement in the design of ladle and tundish handling systems, and in design and maintenance considerations for long-term operation with processing times that extend for several days. A summary of sequence casting records in Japan has been presented by T. Harabuchi

924 / Steel Castings (Ref 14) and is reported in Table 1. In comparison, the Jan 1983 casting record at Great Lakes Steel Corporation in the Detroit district involved a large slab caster (240 mm by 2.5 m, or 9.5 by 99 in.), which cast 402 ladles and 83, 160 MT (91,480 tons) of steel continuously. In addition to the bulk steel handling requirements of sequence casting, the ability to change nozzles, shrouds, and tundishes at frequent intervals is also required. The string of casts at Great Lakes Steel Corporation involved 35 tundish changes and 177 shroud changes over a period of 13 days. This record was exceeded shortly thereafter at the Gary Works of United States Steel International, Inc. Records of this type are difficult to obtain; however, the early dates and substantial accomplishments of these examples speak to the impressive advances that have been achieved in this technology. While it has been shown that sequence casting extending beyond five or six heats does not dramatically increase productivity, provided a reasonable turnaround time and interactive scheduling with the steelmaking facilities exist, the capability for long-term sequence casting represents the opportunity for increased productivity with high quality at an essentially steady-state operation of the caster. An important characteristic of a casting operation with regard to productivity is the percentage of total clock time that steel is being processed in the machine. These percentages in high-productivity casters can exceed 90% for frequently used casting operations. The turnaround time required for the dummy bar to restart a cast is one of the factors in producing steel-in-mold results. However, under ideal conditions, scheduled maintenance will be a major factor, and other items, such as problems in steelmaking or mechanical difficulty on the strand, will be minimized. In the past and in many present installations, interruption of continuous casting production is required in order to make a width change. Several developments that avoid this requirement are now in operation. In one such arrangement, used at Great Lakes Steel Corporation, a very wide slab was cast and then slit into two or three optimal widths. Another approach is taken at Oita Works of Nippon Steel Corporation, where a sizing mill adjusts the slabs to various desired widths. Another, more versatile system involves the use of a variable-width mold in which the taper and width are adjusted continuously during the casting process. These techniques have permitted the adoption of sequence casting with a minimum mold inventory based on width. Another barrier to increasing productivity of continuous casting has been the accommodation of ladle-to-ladle composition changes with adoption of sequence casting. Under ideal circumstances, as when a plant produces a narrow range of compositions, often with overlapping compositional requirements, compositional changes can be slowly stepped through each grade to accommodate the desired sequence of

heats. However, when substantial variations in composition must be accommodated, physical barriers in the form of steel plates have been used to provide isolation of the grade changes in the strand. In this way, the transition zone can be minimized, and compositional changes can be accommodated without substantial losses in yield or quality. A study involving computer simulation and physical (water) modeling of a full-scale twin-strand slab casting tundish, both supported by in-plant sampling, has shown that the transition zone can be markedly reduced by minimizing the residual steel in the tundish when a grade change is made. This effect can be enhanced by optimizing dam and weir configurations in the tundish (Ref 21, 22).

Quality Improvements The operation of a continuous casting strand to produce good-quality steel uniformly and reliably on a routine basis can be the most valuable asset of the process. Development work continues in an effort to improve control and production reliability, particularly with regard to avoidance of inclusions and cracking for internal and surface quality. Internal cracking is less important for those products (for example, sheet) that have large reduction ratios. Radial cracks, center looseness or centerline cracking, minor amounts of gas evolution, and other internal defects are not deleterious in heavily rolled products. In the manufacture of heavy plates, however, these conditions can represent serious product defects. For aluminum-killed steel, subsurface inclusions are usually in the form of aluminum oxide. In some cases, surface scarfing can be effective in removing these inclusions, which could provide a surface defect in a final rolled product. Unfortunately, this surface conditioning results in substantial yield losses. It has been reported (Ref 23) that use of multiport shrouded nozzles can produce a fluid flow action in slab molds, which brings inclusions in contact with the molten mold powder covering to flux and dissolve oxide particles. This provides a clean subsurface zone that requires no scarfing. However, this method leads to the entrapment of complex nonmetallic inclusions containing mold powder. Clean steel practices in combination with nozzle configurations optimized for specific casting conditions are necessary to minimize nonmetallic inclusions. As noted previously, the control of fluid flow and promotion of separation in ladle, tundish, and mold can provide a substantial reduction in large inclusions. Depending on steel grade, cracking can result from intensive cooling or deformation in the casting strand. Surface cracking can be minimized by proper control of lubrication and cooling within the mold, and cooling and alignment in the upper spray zone. Other critical factors in cracking are control of spray cooling to avoid surface temperature rebound (resulting

in midway cracks) or nonuniformity of cooling across or along the strand. Control of temperature distribution at the straightener and proper roll gap settings for slabs and blooms have also been noted as being very important. Electromagnetic stirring (EMS) of solidifying steel casting strands is often used to extend the equiaxed crystal zone and to minimize macrosegregation and internal porosity. Electromagnetic stirring can be applied in the mold, below the mold, or along the strand, and the type of stirrer can be rotary, linear, or hellicoidal (Ref 24). The application of EMS in the production of high-quality steels has proven to be particularly successful. Electromagnetic stirring applied immediately below the mold on a billet caster to produce special-quality bars and low in the strand on a slab caster to produce high-quality heavy plate are two particularly notable successful applications.

Horizontal Continuous Casting Horizontal casting systems have been explored for many years and have been successfully applied to steels. Oldsmobile Division of General Motors Corporation started casting 95 mm (3.75 in.) diameter steel bars in a horizontal mold in 1969 (Ref 25, 26), with uninterrupted casts of up to 24 h. This is believed to be the longest sustained continuous casting operation for steel up to that time. Voest-Alpine is developing a horizontal continuous casting machine for steels, and the Nippon Kokan (NKK) Steel Company of Japan has been producing horizontally cast steel sections up to 210 mm (8 in.) square (Ref 27). These systems depend on a refractory nozzle or break ring at the entrance end of a stationary water-cooled metallic mold, and an intermittent motion for translating the bar, each forward stroke being followed by a discrete rest or dwell. A cross section of the tundish and mold arrangement of the NKK machine is shown in Fig. 15. Casting of larger sections is being explored throughout the world.

Future Developments The emphasis on the further development of continuous casting will focus on control systems and automation, with the objective of maintaining high quality and high productivity. Accomplishing this will include monitoring metal quality and ensuring that all aspects of the process are under proper control. Various operating parameters, such as mold, slag, or flux levels in ladle and tundish, also will be monitored directly. Sensors will be developed to automatically control the process for proper cooling in the mold, first zone, and secondary cooling systems. Hot inspection techniques will be developed that will provide a direct measure of product quality as the product leaves the casting machine and moves to the hot rolling process.

Steel Continuous Casting / 925

12. 13.

14. 15. 16. 17. 18. 19. 20.

Fig. 15

Tundish and mold arrangement for horizontal caster. Source: Ref 27

Also, current worldwide efforts on steel processing are in the area of the belt casting of thin slabs, the roll casting of thin strip, and the electromagnetic casting of steel. ACKNOWLEDGMENTS The author gratefully acknowledges the contributions to this article made by C.R. Jackson and N.T. Mills of Inland Steel Company; the late R. Lincoln of Chaparral Steel Company; K. Schwaha of Voest-Alpine A.G.; and T. Harabuchi of Nippon Steel Corporation. REFERENCES 1. G. Sellers, U.S. Patent 1,908, 1840 2. H. Bessemer, “On the Manufacture of Continuous Sheets of Malleable Iron and Steel Direct from Fluid Metal,” Paper presented at the Iron and Steel Institute Meeting, Oct 1891; also J. Met., Vol 17 (No. 11), 1965, p 1189–1191 3. Continental Iron and Steel Trade Report, The Hague, Aug 1970 4. W.R. Irving, Continuous Casting of Steel, Woodhead Publishing Ltd., Cambridge, England, 1993

5. T. Ohnishi, in Proceedings of the Nishiyama Memorial Lecture, 80–9, p 45 6. R.W. Joseph and N.T. Mills, A Look Inside Strand Cast Slabs, Association of Iron and Steel Engineers, May 1975 7. T. Haribuchi et al., Technical Report 294, Nippon Steel, Jan 1978 8. R. Lincoln, Melt Shop Reprofit at Chaparral/Goal 450,000 Tons of Prime Billets, Proceedings of the Electric Furnace Conference, Vol 36, American Institute of Mining, Metallurgical, and Petroleum Engineers, 1978, p 168–171 9. F.M. Wheeler and A.G.W. Lamont, Current Trends in Electric Meltshop Design, Proceedings of the Electric Furnace Conference, Vol 36, American Institute of Mining, Metallurgical, and Petroleum Engineers, 1978, p 139–147 10. C.R. Jackson and L.R. Schell, Fifteen Years of Looking—One Year of Operating, The Start-Up and First Year’s Operation of Inland’s No. 1 Slab Caster, Proceedings of the Open Hearth Conference, Vol 57, American Institute of Mining, Metallurgical, and Petroleum Engineers, 1974, p 55–66 11. M. Kolakowski and H. Streubel, Mold for Continuous Casting of Steel Strip, U.S.

21.

22.

23. 24.

25.

26. 27.

Patent 4,635,702, SMS Schloemann Siemag, Republic of Germany, Jan 13, 1987 R. Preston, American Steel, Prentice Hall Press, 1991 T.G. O’Connor and J. Dantzig, Modeling the Thin-Slab Continuous-Casting Mold, Metall. Trans. B, Vol 25, June 1994, p 443–457 T. Harabuchi, Nippon Steel Corporation, private communication, May 1984 R.D. Pehlke, Reoxidation of Liquid Steel, Radex Rundsch., Vol 1/2, Jan 1981, p 349–367 C.R. Jackson, Inland Steel Company, private communication, May 1983 Report 77, Steelmaking Committee of ISIJ, Nippon Steel M. Tokuda et al., Iron Steel Inst. Jpn., Vol 83, p 919 Y. Kitano et al., Iron Steel Inst. Jpn., Vol 84, p 179 K. Schwaha, Voest-Alpine, private communication, May 1984 H. Chen and R.D. Pehlke, Minimization of Transitional Slabs Based on Tundish Flow Controls, ISS-AIME Steelmaking Conference Proceedings, 1994, p 695–702 H. Chen and R.D. Pehlke, Mathematical Modeling of Tundish Operation and Flow Control to Reduce Transition Slabs, Metall. Trans. B, Vol 27 (No. 5), 1996, p. 745–756 R. Clark, Continuous Casting of Steel, Institute for Iron and Steel Studies, 1970, p 7 The Application of Electromagnetic Stirring (EMS) in the Continuous Casting of Steel, Continuous Casting, Vol 3, Iron and Steel Society of AIME, 1984 F.J. Webberre, R.G. Williams, and R. McNitt, “Steel Scrap Reclamation Using Horizontal Strand Casting,” Research Publication GMR-11, General Motors Corporation, Oct 1971 W.G. Patton, GM Casts In-Plant Scrap into In-Plant Steel, Iron Age, Dec 1971, p 53–55 F.G. Rammerstorfer et al., “Model Investigations on Horizontal Continuous Casting,” Paper presented at the Voest-Alpine Continuous Casting Conference, 1984

SELECTED REFERENCE  Continuous Casting, Vol 1–9, The Iron and

Steel Society (AIST)

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 926-948 DOI: 10.1361/asmhba0005299

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Shape Casting of Steel CAST STEEL production worldwide currently is approximately 9
106 metric tons per year, which is approximately 10% of the total worldwide casting tonnage of 85
106 metric tons (Ref 1). Sand mold and permanent mold casting are the major methods for shape casting of steels, with production closely split among green sand, chemically bonded sand, and permanent mold processes:  Green sand molding, approximately 32% of

cast steel tonnage

 Chemically bonded sand molding (princi-

pally shell molding and no-bake), approximately 30% of cast steel tonnage  Permanent molding (32%), typically to produce carbon steel railroad wheels in graphite molds Small amounts of cast steel are also produced by investment casting (3%), centrifugal casting (3%), and lost foam (0.4%). This article describes steel solidification characteristics, melting practices, melt treatment, and feeding of the molten steel into the mold. The principal focus is on sand casting, although permanent mold casting of steel is briefly described. Process design and castings of thin sections are detailed.

Castability of Steel Castability of an alloy is influenced by four parameters:  Liquid metal fluid life (typically compared

 Solidification shrinkage  Slag/dross formation tendency  Pouring temperature

These parameters may be interrelated in various ways. For example, fluidity (or metal fluid life) is influenced by inclusions and pouring temperature. This section focuses on the physical properties of fluidity and shrinkage. Colloquially, good castability refers to the ease with which an alloy responds to ordinary foundry practice without requiring special techniques for gating, risering, melting, sand conditioning, or any of the other factors involved in making good castings. Compared to other cast metals (Table 1), the steel castability may be described as “fair” relative to cast irons and aluminum (aluminum-silicon) alloys. The feasibility of a given casting geometry depends on the four parameters of castability. Techniques are also described in the section “Thin-Wall Steel Sand Castings” in this article. Fluidity is the ability of a molten alloy to completely fill a mold cavity in every detail. High fluidity often ensures good castability, but it is not the only parameter for castability. Fluidity is affected by various factors such as alloying, inclusions, and pouring temperature. Various methods are used to test fluidity, but the most common is comparing the length of a cast spiral. Fluidity of various steel compositions is compared in Fig. 1 in terms of spiral length. Fluidity increases with temperature, but the rate of increase becomes smaller as the superheating temperature becomes higher. The physical properties of metal that affect fluidity include:

        

Mold materials and surface characteristics Alloy composition Surface tension and surface films Gas content and suspended inclusions (which may alter fluidity directly or indirectly by effects of surface tension) Degree of superheat Rate of pouring Rate of heat transfer to the surroundings Heat of fusion and solidification Viscosity

Viscosity is a factor, but fluidity should not be confused with viscosity. Viscosity defines the internal friction of a liquid, and molten metals have a relatively low kinematic viscosity (viscosity divided by density). The kinematic viscosity of metals is much lower than that of water, and so the effect on fluidity is less significant. In terms of alloy composition, extensive studies concerning the fluidity of steels were performed in the early part of the 20th century, and results were summarized by Charles Briggs in various publications (Ref 2, 3). The following is an adapted summary from Ref 3. “Carbon has a slight effect on fluidity of steel, as shown in Fig. 1. Silicon has a marked effect on fluidity, as shown in Fig. 2. A notable change toward greater fluidity is found in the medium-carbon steels when silicon is increased from 0.25 to 0.45%. Presumably, this change is not an effect of silicon addition as such but rather of the state of deoxidation of the steel. Nickel improves the fluidity of steel

as fluidity from spiral tests) Table 1 General comparison of castability, weldability, and machinability of cast metals Casting

Steel Nodular iron Gray iron Malleable Aluminum base

Weldability

Excellent Difficult

Difficult Difficult Highly susceptible to cracking Copper base Careful control needed Magnesium Readily welded, base preheat needed Nickel base Fair Zinc base Difficult

Castability

Machinability

Fair Good

Fair Good

Excellent Good Excellent

Good Good Good to excellent

Fair to good Good to excellent Fair Excellent

Fair to good Excellent Fair Excellent

Fig. 1

Influence of carbon on the fluidity of steels with average manganese and silicon contents. Source: Ref 3

Fig. 2

Influence of silicon on the fluidity of steels having commercial compositions

Shape Casting of Steel / 927 progressively up to 3.25%, as shown in Fig. 3. Higher concentrations, as great as 5% Ni, result in a decrease in fluidity. An increase of manganese to 2% has virtually no greater effect on the fluidity of the alloy than that found in a regular-grade carbon steel of approximately 0.75% Mn. Steels containing more maganese (approximately 2 to 14%) will exhibit improved fluidity at low temperatures (1455 to 1540  C, or 2650 to 2800  F) as the manganese content increases, whereas at temperatures of 1595  C (2900  F) and above, the manganese-rich steels are inferior to plain carbon steels in fluidity.” “Additions of copper to plain carbon steel increase fluidity progressively as the concentration increases to approximately 4% Cu. Substantial additions of chromium to steel decrease the fluidity as compared with plain carbon steel. Small additions, less than 1.5% Cr, have no effect on the fluidity characteristics. Molybdenum and vanadium, when added to a normal plain carbon steel in amounts between 0.25 and 1%, cause a reduction in fluidity from that ordinarily obtained with no molybdenum or vanadium.” “Aluminum is more effective than silicon in promoting fluidity, especially at the lower temperatures and when small quantities of aluminum are used. As aluminum is added beyond 0.2%, or that amount required for deoxidation control, fluidity is progressively decreased to low values at 1.2% A1. This indicates that the amount of aluminum for maximum fluidity is somewhat critical. The threshold value depends on the state of oxidation of the bath prior to the addition of the element. In order to have good fluidity, steel must be well made and thoroughly deoxidized. Rimming steels have fluidity inferior to that of dead-killed steels. Also, dead-killed steels that are excessively high in deoxidizers, such as aluminum and silicon, have fluidity inferior to that of steels killed with just sufficient silicon and aluminum.” (Ref 3). Shrinkage. An understanding of castability also involves solidification behavior—particularly the volume contraction of the metal during cooling and solidification. The solidification of a casting can involve as many as three separate contractions as a result of cooling:



C, or 200 to 300  F, above its liquidus point) down to the point when solidification begins. This contraction is less significant than the other two regions of contraction. In particular, the region of liquid-to-solid contraction is much more significant. Solid-state contraction can also be significant, as the cooling occurs over a very wide temperature range from the solidus

down to room temperature. Solid-state contraction is a major concern in part design, because it is one of the primary causes of warped and cracked castings. Specific volume changes in various steels are plotted in Fig. 5 versus temperature. Linear contraction of cast steel bars is shown in Fig. 6 and Table 2. As described by Briggs (Ref

Fig. 3

Influence of nickel on the fluidity of cast steels. Source: Ref 3

Fig. 4

Volume change recorded on the cooling of a 0.35% C cast steel. Source: Ref 2

 Liquid-liquid contraction  Liquid-solid contraction  Solid-solid contraction

As a rough guide, molten steel contracts 0.9% per 55  C (100  F) as it cools from the pouring temperature (usually 110 to 165  C, or 200 to 300  F, above its melting point) to the solidification temperature. It then undergoes solidification contraction of 3% during freezing, and finally the solidified metal contracts 7.2% during cooling to room temperature. Figure 4 is a typical cooling curve that plots contraction of a carbon steel. Contraction during cooling of the molten metal occurs from its pouring temperature (usually 110 to 165

928 / Steel Castings 4), a very good estimate of the influence of composition on the total contraction can be obtained by a comparison of the specific volumes of steels in the molten state and at room temperature (Fig. 7). From the diagrams, it is clear that with the exception of carbon, the additions of other elements influence the specific volume of the melt at 1600  C (2900  F) to approximately the same extent that they influence the solid steel at room temperature. At 1600  C (2900  F), the addition of 1.00% C increases the specific volume by approximately 4%, whereas 1.00% Al increases it 2%, and other alloying elements less than 1%. In general, upon alloying, elements heavier than steel decrease the specific volume and lighter elements increase it. Freezing Range. For commercial carbon steels with manganese content below 0.45 wt % and silicon below 0.35 wt%, the solidus and liquidus points are approximately 10 to 15  C (20 to 25  F) below those of the iron-carbon system. A noticeable lowering of the melting and freezing points occurs with manganese above 0.45 wt%, but there is little difference in freezing points of cast steel containing

Fig. 5

0.65% Mn and one containing 1.2% Mn. A method of estimating freezing points of nickel and nickel-chromium steels is to add alloy contents together and multiple the sum by 10 to obtain the approximate lowering of the freezing point in degrees Fahrenheit (Ref 4).

Melting Furnaces The melting furnaces used in steel foundries are essentially the same as those used for the production of steel ingots, except that most foundry melting units are smaller. Although the equipment is the same, the processes are often different. Steel ingots can be made as rimmed, semikilled, or killed steel. Only thoroughly killed steel is used for steel foundry products. The method of production of the killed steel used for castings may differ from that used for wrought products because of the fluidity requirement. However, the salient features of making steel in a foundry are the same as those for producing fully killed steel ingots. Foundries that have access to a good grade of scrap, and therefore do not need to reduce the

Volume changes in iron-carbon alloys. Source: Ref 4

phosphorus and sulfur contents of the steel to meet specifications, usually prefer to use a furnace lined with silica refractories (acid lined). Foundries that do not have a good source of low-phosphorus, low- sulfur scrap use refractories such as magnesite and dolomite for furnace lining (basic lining). The compositions of certain steels, such as austenitic manganese steels, require that they be made in a furnace with a basic lining. Direct Arc Melting. The direct arc furnace consists essentially of a metal shell lined with refractories. This lining forms a melting chamber, the hearth of which is bowl shaped. Three carbon or graphite electrodes carry the current into the furnace. Steel, either solid or molten, is the common conductor for the current flowing between the electrodes. The metal is melted by arcs from the electrodes to the metal charge—both by direct impingement of the arcs and by radiation from the roof and walls. The electrodes are controlled automatically so that an arc of proper height can be maintained. In acid electric practice, the furnace hearth is composed of silica sand or ganister rammed into place. The furnace is charged with selected scrap low in phosphorus and sulfur because the acid process cannot eliminate these elements. Approximately 40% of the charge usually consists of foundry scrap (gates and risers). The general practice is to charge small pieces first in order to form a compact mass in the furnace, thus aiding electrical conductivity. The heavy, lumpy portion of the charge is placed over the smaller pieces, followed by the lightest portion. The charge is melted down as quickly as possible. Small amounts of sand and limestone are occasionally added to the bath during the melting period to form a protective slag of proper consistency. Once melting is complete, oxygen is used as an oxidizing agent to produce excess oxygen. Iron oxide in slag reacts with silicon and manganese in the metal bath and produces oxides and silicates in the slag. After most of the silicon and manganese have been oxidized, the bath begins to boil. The boil is evidence of carbon elimination; it results from the interaction of dissolved carbon and free oxygen, giving rise to carbon monoxide bubbles. In the oxidizing stage of melting, the slag is rich in iron oxide. As carbon is eliminated from the molten iron, the iron oxide content of the slag decreases. The finishing slag contains approximately 55 to 60% SiO2, 12 to 16% MnO, 4 to 6% Al2O3, 7 to 10% CaO, and 12 to 20% FeO. The carbon content is reduced to approximately 0.20 to 0.25% during the vigorous carbon boil. The boil is stopped by the addition of carbon in the form of a carburizing iron of low sulfur and phosphorus content. Further deoxidizers (ferromanganese and ferrosilicon) are then added to the bath. Next, the metal is tapped into the ladle. A small amount of aluminum is added to the ladle as a final deoxidizer.

Shape Casting of Steel / 929 For basic electric furnace melting, the furnace lining is a basic refractory such as magnesite or dolomite. The charge is usually composed of purchased scrap steel and foundry returns. During the melting period, small quantities of lime are occasionally added to form a protective slag over the molten metal. Iron ore is added to the bath just as melting is complete. The slag is then highly oxidizing and in the correct condition to take up phosphorus from the metal. Shortly after all of the steel has melted, this first slag is taken off (if a two-slag process is to be used), and a new slag composed of lime, fluorspar, and sometimes a little sand is added. As soon as the second slag is melted, the current is reduced, and pulverized coke, carbon, or ferrosilicon, or a combination of these, is spread at intervals over the surface of the bath. This period of furnace operation is known as the refining period, and its purposes are to reduce the oxides of iron and manganese in the slag and to form a calcium carbide slag, which is essential to the removal of sulfur from the metal. The refining slag has approximately the following composition: Element

CaO SiO2 FeO CaF

Fig. 6 Table 2

Contraction of carbon steel castings with no stress on free bar except that due to friction between sand and bar

Linear contraction in freely contracting alloy cast steel bars

Type of steel and percent of alloy(a)

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

Carbon (0.35) Manganese (1.32) Nickel (3.00) Chromium (1.03) Copper (1.39) Molybdenum (0.39) Vanadium (0.25) Mn-Si (1.35)(1.15) Ni-Mn (1.46)(1.24) Mn-Mo (1.41)(0.37) Mn-V (1.41)(0.16) Ni-Cr (2.88)(0.91)

Contraction before critical range, %

Expansion within critical range, %

Contraction below critical range, %

Total contraction, %

1.47 1.74 1.72 1.46 1.49 1.65 1.40 1.54 1.70 1.68 1.50 1.95

0.10 0.26 0.20 0.14 0.14 0.04 0.09 0.14 0.19 0.02 0.13 0.04

1.03 0.88 0.87 1.02 1.00 0.72 1.02 0.96 0.86 0.66 0.96 0.36

2.40 2.38 2.40 2.34 2.35 2.33 2.32 2.35 2.37 2.32 2.33 2.27

(a) Carbon kept between 0.32 and 0.37%, manganese 0.55 and 0.80%, and silicon 0.25 and 0.40% unless otherwise specified. Source: Ref 4

Composition, %

45–55 15–20 0.50–1.5 5–15

Adjustments are made in the carbon content of the bath by the addition of a low-phosphorus pig iron. After the proper bath temperature is obtained, ferromanganese and ferrosilicon are added, and the furnace is tapped. Aluminum is generally added in the ladle as a final deoxidizer. Induction Melting. Induction furnaces used in the steel foundry range in capacity from 14 to 22,680 kg (30 to 50,000 lb), but most have capacities of 45 to 4500 kg (100 to 10,000 lb). Firebrick is placed on the bottom of the shell, and the space between that and the coil is rammed with grain refractory. The furnace chamber may be a refractory crucible or a rammed and sintered lining. General practice is to use ganister rammed around a steel shell that melts down with the first heat, leaving a sintered lining. Basic linings are often preferred; a rammed lining of magnesia grain or a clay-bonded magnesia crucible can be used. The process consists of charging the furnace with steel scrap and then passing a high-frequency current through the primary coil, thus inducing a much heavier secondary current in the charge, which heats the furnace to the desired temperature. As soon as a pool of liquid metal is formed, a pronounced stirring action takes place in the molten metal, which helps to accelerate melting. In this process, melting is rapid, and there is only a slight loss of the easily oxidized elements. If a capacity melt is required, steel scrap is continually added during

930 / Steel Castings

Fig. 7

Influence of composition on specific volume at 20  C (68  F) and 1600  C (2912  F). Source: Ref 4

the melting- down period. Once melting is complete, the desired superheat temperature is obtained, and the metal is deoxidized and tapped. Because only 10 to 15 min elapse from the time the charge is melted down until the heat is tapped, there is not sufficient time for chemical analysis. Therefore, the charge is usually carefully selected from scrap and alloys of known composition in order to produce the desired composition in the finished steel. Composition can be closely controlled in this manner. In most induction furnaces, no attempt is made to melt under a slag cover, because the stirring action of the bath makes it difficult to maintain a slag blanket on the metal. However, slag is not required, because oxidation is slight. The induction furnace is especially valuable because of its flexibility in operation, particularly in the production of small lots of alloy steel castings. Induction melting is also well adapted to the melting of low-carbon steels because no carbon is picked up from electrodes, as may occur in an electric arc furnace.

Melting Practice (Ref 5*) One major difference between melting practice in a steel foundry and in a steel mill is the higher tapping temperature used for foundry *Adapted with permission from Steel Castings Handbook, 6th ed., Steel Founders’ Society of America and ASM International, 1995

melting to attain better fluidity of the molten steel. The producer of ingots for rolling is less concerned with fluidity because mold filling is simpler in ingot molds than in the molds used for producing relatively complex shapes. Charging and refining reactions are described in this section. More details on ladle metallurgy and degassing are given in the article “Steel Melt Processing” in this Volume. Selection of Metallic Charge Material. Selection of charge materials is a major concern for steel foundries. The foundry will remelt the risers, runners, pigs, and other scrap produced from earlier heats where the analysis is appropriate. Recycled scrap from sheet steel tamping, shredded automobiles, and used and worn parts is a major source of material. Virgin materials (iron, nickel, copper, chromium, molybdenum, silicon, manganese) in various forms make up the balance of the charge. The meltdown composition must be carefully controlled to enable the proper refining actions to occur. Carbon is added in excess of the specification limits on many alloys to permit an oxygen blow. Oxidizable elements such as silicon, manganese, and chromium may need to be restricted. Tramp elements such as sulfur, phosphorus, lead, tin, zinc, arsenic, bismuth, and so on must be controlled. Induction charges must be clean and dry to avoid the pickup of hydrogen. Since relatively little refining occurs during melting, the charge chemistry is the same as the desired final chemistry. Refining Reactions. During the initial melting of the metal, oxidation of aluminum, silicon, manganese, and chromium may occur.

These oxides float to the top of the metal surface to form a slag cover. Basicity of the slag may be controlled by adding CaO in the form of lime. Boiling. The most common reaction, called boiling, occurs at temperatures above 1510  C (2750  F). Oxygen reacts with carbon, causing the production of carbon monoxide gas bubbles at the bottom of the bath. These bubbles rise through the metal, causing the top of the metal to appear to be boiling. The oxygen is often injected through a pipe or lance during an oxygen blow. This reaction is exothermic and causes the temperature of the bath to rise. It is also possible to add oxygen from other sources, such as iron oxide. Hydrogen and nitrogen are soluble in liquid iron at high levels (Fig. 8). These levels must be reduced below the solid solubility level or defects will be formed on solidification. The nitrogen and hydrogen migrate into the rising bubbles of carbon monoxide, and these gases leave the metal with the carbon monoxide. Alloys containing chromium are more difficult to blow down to low carbon levels, as can be seen from Fig. 9. It may be necessary to restrict the meltdown chromium or blow at high temperatures to achieve the proper final carbon content. Removal of Phosphorus. To remove phosphorus, a high calcium oxide content slag is used that contains large amounts of iron oxides (Fig. 10). The phosphorus and all other oxidizable elements react with this oxidizing slag and become part of the slag. If the high oxide content of the slag is not retained, the slag containing the phosphorus can revert and return the phosphorus to the metal. Care is taken to prevent the slag from reverting. The usual practice is to manually remove the slag from the molten bath, thus permanently removing the phosphorus from the furnace. A new slag, suitable for the next operation, is then charged and melted. Removal of Sulfur. The removal of sulfur from the molten-metal bath is also done under a slag with a very high content of calcium oxide and high basicity. In order to remove the sulfur, this slag must be reducing, that is, free of iron oxides. This is the opposite of the oxidizing condition required to remove phosphorus from the metal (Fig. 11). The slag is generally made reducing by the addition of silicon, carbon, aluminum, or other reducers to the slag in order to either remove or tie up metallic oxides. The sulfur from the metal then chemically combines into the reducing slag and becomes part of it, as long as the slag remains reducing. The sulfur-laden slag can then be removed from the furnace, taking the sulfur with it. If both sulfur and phosphorus are to be removed, the operator can choose to first perform either the sulfur-removal process or the phosphorus-removal process. In some cases, the sulfur is removed first, but general practice is to first remove phosphorus and then remove sulfur, tapping the sulfur-containing slag into the ladle, where it is then removed.

Shape Casting of Steel / 931

Fig. 8

Hydrogen and nitrogen solubility in iron

Recovery of Elements from Slag. Expensive and desirable elements may be unavoidably oxidized into the slag during the melting and blowing operations. If the slag containing these oxides has not been removed, it may be practical to recover them by reducing them back into the metal bath. This is achieved by deoxidizing the slag with additions of one or more reducing elements such as carbon, silicon, and/or aluminum. See Fig. 12 for the relative influence on oxygen concentration using different deoxidizers. The recovery of chromium from the slag is a notable example of this practice. During the melting and carbon-removal operations, much of the chromium content is oxidized and enters the slag. The amount of chromium in the slag is strongly influenced by the basicity of the slag and the silicon content of the bath, as seen in Fig. 13 and 14. By controlling the chemistry of the slag and metal, most of the chromium can be recovered. The reactions to remove sulfur and phosphorus and to recover chromium are carried out in basic-lined furnaces. Acid-lined furnaces carry an acid slag that precludes refining operations. For this reason, the choice of raw materials is extremely important when melting in acid-lined furnaces. All of these reactions are strongly influenced by the kinetics of the system. The ability of the slag to come into contact with the metal is influenced by the viscosity of the slag, which is determined by the chemistry and melting point of the slag. Since the slag has many different oxides, it is impossible to predict exactly the kinetic response. It should also be pointed out that are furnace design gives a bath that is very shallow over a large area. Reactions that take place in one

Fig. 10 Fig. 9

Relation among chromium, carbon, and temperature in oxygen-saturated bath. Source: Ref 5

Relationship between phosphorus distribution and %FeO at different CaO/SiO2 ratios. Source: Ref 5

932 / Steel Castings

Fig. 11

Effects of slag basicity and iron oxide on sulfur distribution ratio. Source: Ref 5

Fig. 12

Effect of deoxidizers on the solubility of oxygen in liquid iron and nickel at 1600  C (2912  F). Source: Ref 6

area, such as in front of the door, may be complete long before reactions are complete in other areas of the furnace. Stirring the bath and slag helps to speed up the reactions. Temperature Control. The control of temperature during melting and pouring is critical. All of the reactions previously mentioned are strongly influenced by temperature to the point that a change of 28  C (50  F) will greatly change the result. One of the most used controls is the immersion thermocouple, which is plunged into the metal and gives an accurate temperature reading in less than 4s. It can be used in furnaces, ladles, pouring basins, and in casting risers. Deoxidation Practice. Proper melting practice and, to a lesser degree, proper heat treatment can limit the gas content of conventionally melted steel to acceptable levels, regardless of the type of furnace equipment employed. Failure to control oxygen, hydrogen, and nitrogen contents may result in porosity or a serious decrease in ductility, or both. Gas content is largely adjusted during the oxygen boil. After the cold charge is melted and the bath is in the temperature range of 1510 to 1540  C (2750 to 2800  F), oxygen is introduced into the molten metal, usually by means of a piping arrangement. The oxygen combines with the dissolved carbon in the steel to form bubbles of carbon monoxide. As the bubbles form, dissolved hydrogen and nitrogen are caught up in the bubbles in much the same way that dissolved oxygen finds its way into bubbles of boiling water. Thus, the bubbles of gas contaminants are boiled out. The normal hydrogen content in acid-melted steel is in the range of 2 to 4 ppm. Hydrogen content resulting from basic practice is slightly higher. Austenitizing and tempering treatments serve to lower the hydrogen content further in sections of average thickness. However, these treatments may not suffice when very heavy sections are involved. The alternative is to degas the molten steel under vacuum, particularly when pouring heavy castings that will be subjected to severe dynamic loading. The lowest nitrogen contents are obtained with either acid or basic open-hearth practice. Nitrogen contents in electric arc melted steels range from 3 to 10 ppm. The primary role of deoxidation is to prevent pinholes due to carbon monoxide formation during solidification. Oxygen content should be less than 100 ppm to prevent porosity. Silicon and manganese are mild deoxidizers; they are added to stop the carbon boil and to adjust the chemistry. Manganese (>0.6%) and silicon (0.3 to 0.8%) additions are limited by other alloy effects and are normally inadequate alone to prevent pinholes. Aluminum is the most common supplemental deoxidizer used to prevent pinholes. As little as 0.01% Al will prevent pinholes. Aluminum is normally added at the tap—approximately 1 kg/Mg (2 lb/ton) (0.10% added) with a recovery of 30 to 50% for a final content of

Shape Casting of Steel / 933

Fig. 13

Fig. 14

Effect of slag basicity (V-ratio) and bath silicon on the retention of chromium in a reduced slag. Source: Ref 5

Chromium and silicon contents of liquid iron in equilibrium with silica-saturated slags

approximately 0.03 to 0.05%. This is normally supplemented at the pour ladle with additional deoxidation, which could be more aluminum or calcium, barium, silicon, manganese, rare earths, titanium, or zirconium. More than the minimum amount of deoxidizer required for preventing porosity is needed to maintain sulfide inclusion shape control, but excessive amounts can cause intergranular failures or dirty metal. The second addition at the pour ladle is normally 1 to 3 kg/Mg (2 to 7 lb/ton). The commonly used elements for deoxidation are (in order of decreasing power) zirconium, aluminum, titanium, silicon, carbon, and manganese. Argon-Oxygen Decarburization (AOD). Some foundries have recently installed AOD units to achieve some of the results that vacuum melting can produce. These units look very much like Bessemer converters with tuyeres in the lower sidewalls for the injection of argon or nitrogen and oxygen. They are processing units that must be charged with molten metal from an arc or induction furnace. Up to approximately 20% cold charge can be added to an

AOD unit; however, the cold charge is usually less than 20% and consists of solid ferroalloys. The continuous injection of gases causes a violent stirring action and intimate mixing of slag and metal, which can lower sulfur values to below 0.005%. The gas contents approach or may be even lower than those of vacuum induction melted steel. The dilution of oxygen with inert gas, argon, or nitrogen causes the carbon-oxygen reaction to go to completion in favor of the oxidation reaction of iron and the oxidizable elements, notably, chromium in stainless steel. Therefore, superior chromium recoveries from less expensive high-carbon ferrochromium are obtained compared to those of electric arc melting practices. Argon-oxygen decarburization units are used in the production of high-alloy castings, particularly of grades that are prone to defects due to high gas contents. Carbon and low-alloy steels for castings with heavy wall sections may be subject to hydrogen embrittlement and are also processed in these units with good results. Powder Injection/Wire Injection. The powder injection or wire injection of reactive metals (calcium, magnesium, or rare earths) into liquid steel for desulfurization has come to the forefront of technology in the past decade. Ultralow levels of sulfur (<0.005%) can be obtained with this technique. The basis of reactive metal injection desulfurization is the chemical combination of the reactive metal with sulfur dissolved in the liquid steel. Because the reactive metals are strong deoxidizers, it is necessary to have a low oxygen level in the liquid steel before the start of the injection process. Otherwise, the reactive metal would combine with the dissolved oxygen and not the dissolved sulfur. Aluminum is used to obtain the necessary low oxygen levels. It is also necessary to have a slag present to hold the sulfur removed by the reactive metal. If such a slag is not present, the metal sulfide

will be oxidized, and the sulfur will return to the liquid metal. Plasma Refining. Steadily increasing requirements for steel cleanliness have led producers to adopt several novel refining technologies and process routes, many of which involve increased use of the ladle as a refining vessel. Such procedures require longer holding times in the ladle, which necessitate either increased superheats in the furnace or external heating in the ladle in order to avoid early solidification. Higher superheat, in addition to requiring excessive energy expenditure, can contribute to the problem of dirty steel. The preferred solution is to supply heat to the ladle, maintaining the steel at minimal superheat during refining. This can be accomplished by the transferred arc plasma torch, with the added benefit of enhanced refining reactions that aid in the production of clean steel with low levels of residual elements. In this work, experiments have been carried out in an induction furnace equipped with a gas-stabilized graphite electrode to investigate the control of oxygen and inclusion levels and the enhancement of desulfurization afforded by the transferred arc plasma.

Pouring Practice The factors affecting pouring are the pouring temperature and pouring rate. The pouring temperature is usually approximately 110 to 165  C (200 to 300  F) above the melting (liquidus) point. Too slow of a pouring rate results in freezing. Too rapid pouring results in turbulence, which can increase oxide formation and mold erosion. Ideally, the optimum pouring time for a given steel casting would be established on the basis of the weight and shape of the casting, the temperature of the molten metal and its solidification characteristics, and the heat-transfer characteristics and thermal

934 / Steel Castings stability of the mold. Many foundries, however, are required to pour many different castings from one heat, or even from one ladle. Therefore, these foundries do not attempt to control pouring time, but they do control the speed at which molten steel enters the mold cavity. This control is achieved through the design of the gating system. Pouring analysis is based on Bernoulli’s theorem of fluid flow. Ladles. The types of ladles used for pouring steel castings include bottom-pour, teapot, and lip-pour ladles. Ladle capacities normally range from 45 kg to 36.3 Mg (100 lb to 40 tons), although ladles having much larger capacities are available. Bottom pouring is preferred for pouring large castings from large ladles because it is difficult to tip a large ladle and still control the stream of molten steel. In addition, the bottom-pour ladle delivers cleaner metal to the mold. Inclusions, pieces of ladle lining, and slag float to the top of the ladle; therefore, pouring from the bottom greatly reduces the risk of passing nonmetallic particles into the mold cavity. On the other hand, it is impractical to pour from a large bottom-pour ladle into small molds. The pressure head created by the metal remaining in the ladle delivers the molten steel too fast. The sand mold is not strong enough to withstand this heavy on-rush of molten steel. In addition, because little time is needed to fill a small mold, a large bottom-pour ladle must be opened and closed many times in order to empty it. This may cause the ladle to leak; however, special nozzles have been developed to minimize leakage. Although bottom-pour ladles could be scaled down in size for pouring smaller castings, this is unnecessary because of the almost equal ability of the teapot ladle to deliver clean metal. The teapot ladle has a ceramic wall, or baffle, that separates the bowl of the ladle from the spout. As the ladle is tipped, hot metal flows from the bottom of the ladle up the spout and over the lip. Because the metal is taken from near the bottom of the ladle, it is free of slag and pieces of eroded refractory, although it may pick up foreign materials in the spout section or at the lip. The teapot design is feasible in various sizes, generally covering the entire range of casting sizes that are below the minimum size for which the bottom-pour ladle is used. Lip-pour ladles are similar to the teapot type, except they have no baffles to hold back the slag. Thus, this type of ladle pours a dirtier steel and is seldom used to pour steel castings. Nevertheless, it is widely used as a tapping ladle (at the melting furnace) and as a transfer ladle to feed smaller ladles of the teapot type. Ladle Linings. Most ladle problems in the steel foundry are caused by deficiencies in preparing and maintaining the linings. Ladles with capacities above approximately 360 kg (800 lb) are lined with layers of firebrick covered with a thick refractory coating. When used to pour steel made by the acid process and when properly maintained, these linings will last for 60 to 75 heats. However, when pouring steel made by

the basic process, the life of the lining is only 10 to 25 heats because of the erosion of refractory by the slag. Smaller ladles are protected by monolithic linings prepared by ramming or by pouring and vibrating refractory mixtures around forms. The optimum refractory for this purpose is highpurity alumina, but because of the high cost of alumina, alternative materials such as olivine and zircon are more widely used. The service lives of these monolithic linings, which are most commonly used for small ladles, vary considerably, depending on the refractory mixture and the type of service. Thermal expansion and contraction can cause hairline cracks, into which hot metal enters. This results in erosion of the lining. Erosion not only shortens the life of the lining but also contributes to excessive dirt in the steel. Disposable fiber linings that do not require preheating are gaining acceptance. Preheating the ladle helps to prolong the life of the ladle lining. Heating the lining to a temperature close to that of molten steel serves to lessen the thermal shock to the lining when the molten steel contacts it. Common practice is to preheat ladle linings to 815 to 1095  C (1500 to 2000  F) with oil-fired or gas-fired torches. Ladles lined with high-purity alumina can be preheated to temperatures as high as 1370  C (2500  F).

Foundry Factors in Design Gating Design. An effective gating system for pouring steel and other metals into sand molds is one that fills the mold as rapidly as possible without developing pronounced turbulence. It is essential that the mold be filled rapidly, primarily because heat is radiated to the walls and top surface of the mold as the molten steel rises in the mold cavity; this heat can destroy the binder in the molding sand, causing the mold to collapse unless the metal rises rapidly to provide the necessary support. Turbulent metal flow is harmful because it breaks up the metal stream, exposing more surface area to air and forming metallic oxides that rise to the top of the mold cavity; this produces a rough surface or inclusions in the casting. In addition, turbulent flow erodes the mold material. These eroded particles also float to the top of the mold cavity. Preferred Metal Flow. According to preferred practice, the pourer directs the metal stream toward the pouring cup at the top of the mold, controlling the pouring rate to keep the cup full of molten steel throughout the pouring cycle. The opening in the bottom of the cup is directly over the sprue, or downgate, which is tapered with the large end up, thus reducing the diameter of the stream of descending metal. The taper prevents the stream from pulling away from the walls and drawing air into the gating system. The descending metal impinges on the sprue well at the bottom of the sprue, and the direction of flow changes from vertical

to horizontal, with the metal flowing along runners to gates (ingates) and then to the main body of the casting. A gating system that incorporates these features is shown in Fig. 15. The design of the gating system will largely determine the manner in which molten steel is fed into the mold, as well as the feed rate. Therefore, a system having several gates will influence the distribution of the flow between gates. A good design will have even distribution between gates, both initially and while the mold is filling. The distribution of flow during filling of the mold will affect the type of flow that occurs in the main body of the casting. A large difference in the flow between gates will create a swirl of metal in the mold about a vertical axis, in addition to that occurring about a horizontal axis. The gating system shown in Fig. 15 is an example of a finger-type parting line system, in which the fingers feed metal to the casting just above the parting line. Other major types of gating systems used in steel foundries include the bottom gate, which feeds metal to the bottom of the casting, and the step gate, which feeds metal through a number of stepped gates, one above another. In the system shown in Fig. 15, the ratio of the cross-sectional area of the choke of the sprue to all of the runners emanating from the sprue basin and to all of the gates is 1 to 4 to 4. The runner area decreases progressively by an amount equal to the area of each gate it passes. This practice ensures that once the system is filled with metal, it will remain full during the pouring cycle and will feed equally to each gate. Further, the gates are located in the cope, while the runner, which extends beyond the last gate, is located in the drag. The extension of the runner serves as a trap for the first, and usually the dirtiest, metal to enter the system. The entire runner must fill before the metal will rise to the level of the gates. Thus, each gate begins feeding at the same time. The runners and gates are curved wherever a change in direction occurs. This streamlining reduces turbulence in the metal stream and minimizes mold erosion.

Fig. 15

Gating system that allows good metal flow. The desired metal flow is obtained in this system by proportioning the cross-sectional area of the choke of the spur to all the runners emanating from the sprue and to all of the gates in accordance with a 1:4:4 ratio.

Shape Casting of Steel / 935 In contrast to the ratio of the system shown in Fig. 15 (1 to 4 to 4), if the total cross-sectional area of the gates is less than that of the runners (1 to 2 to 0.8, for example), the result is a pressurized system. With a pressurized system, the metal squirts into the mold cavity and flows turbulently over the mold bottom. This may cause roughening of the bottom surfaces. Conversely, if the total cross-sectional area of the gates is significantly greater than that of the runners (1 to 2 to 3, for example), the gating system will be incompletely filled, and flow from the gates will be uneven. This condition increases the likelihood of mold erosion. When required, other more complicated additions to gating systems are used, including whirl gates, horn gates, strainer cores, tangential gates, and slit gates. However, any addition to the gating system usually increases the cost of the casting because all gating must be removed. Mold Erosion. In addition to the contribution of gating design to a reduction in mold erosion, further reduction can be achieved by making the gating system out of tile, which is superior to green sand in resistance to erosion. However, the use of tile is generally limited to gating systems for large castings; in such cases, the quantity and speed of molten metal passing through the gating system would seriously erode green sand. The automated production of smaller molds is hampered by the addition of tile gates and runners. Thus, gating systems for these smaller castings are rammed in sand, usually with a semicircular or rectangular cross section for the gates and runners. Riser Design. An ample supply of molten metal must be available from risers (reservoirs) to compensate for the volume decrease; otherwise, shrinkage cavities form in the locations that solidify last. Formulas based on surface area, volume, and freezing time of the casting are used to determine riser size. Solidification time is a function of the size and shape by Chvorinov’s empirical relationship such that solidification time (Ts): Ts ¼ C

 2 V A

where V is volume, A is surface area, and C is an experimentally determined value that depends on mold material, thermal properties of casting metal, and pouring temperature relative to melting point. A casting with a higher volume-to-surface-area ratio solidifies more slowly than one with a lower ratio. In riser design, the solidification time of the riser must be equal to the solidification time of the casting. Carbon steel shrinks about 3% during solidification with additional contraction during cooling. These contractions can cause internal unless liquid metal reservoirs (risers) help feed liquid metal into the mold cavities until the end of the solidification process. Risers also can serves as a heat reservoir, creating a temperature gradient that induces directional solidification. Without directional solidification,

liquid metal in the casting may be cut off from the riser, resulting in the development of internal porosity. Two criteria determine whether or not a riser is adequate:  The solidification time of the riser relative to

that of the casting, and

 The feeding distance of the riser.

To be effective, a riser should continue to feed liquid metal to the casting until the casting has completely solidified. Thus, the riser must have a longer solidification time than the casting. Feeding and risering guidelines and calculations for steel casting are described in more detail in Ref 7. Because feeding from the riser depends on gravity, risers are usually located at the top of the casting. Riser forms are placed on the pattern and molded into the cope half of the mold. The riser cavity is usually open to the top of the mold, although blind risers are sometimes used. Most risers are cylindrical, with their height approximately equal to their diameter. This configuration provides a low ratio of surface area to volume, which prolongs the time the steel remains liquid. A spherical riser would constitute an optimal design, but spherical shapes are difficult to mold. The placement of a riser, in conjunction with its size, determines its effectiveness. The thicker sections of a casting act as reservoirs for feeding the thinner sections, which solidify first. Therefore, risers are placed over thick sections that cannot be fed by other areas of the casting. Feeding Distance. Castings of uniform thickness present a different problem. Studies have established the feeding distances of a riser for various rectangular shapes in both the horizontal and vertical planes, with and without an end effect. An end effect is the additional cooling provided by the sand cover of an end surface. The maximum feeding distance can be extended for a uniform section by adding a taper. The progressively thicker section solidifies in a progressively longer time, so that a favorable temperature gradient is established from the end of the section to the riser. A tapered pad of exothermic material placed in the mold along the length of the casting will also produce a favorable temperature gradient. Pouring Direction. Steel founders generally find it necessary to make their molds and cores from established natural materials and to bottom run the castings, that is, to introduce the liquid metal into the lower part of the molds. Generally, by adopting this recognized method, defects, such as sand inclusions and displaced cores, and turbulent pouring can be avoided when filling the mold. The adoption of this procedure ensures that the casting has the best possible external appearance. Because of the considerable shrinkage that occurs in steel as it solidifies, shrinkage cavities

may form in those parts of castings insufficiently fed with liquid metal as solidification progresses. Once the mold is filled, extra metal cannot be provided except by means of the feeding heads. Thus, the feeding of castings is essentially a function of the direction of solidification. It is important that the designer have a knowledge of the principal factors on which the founder bases his choice of gating and risering practice. The principal factors that determine the rigging (or gating and risering) of a casting are, in their order of importance, as follows:  Position of the massive parts  Position of the surfaces to be machined  Possibility of venting the core gases

Positioning Massive Parts. Since the feeding of a casting takes place by gravity, it is important that the most massive parts be arranged uppermost in the mold. Since these are the last to cool, they can feed the lower part of the casting with molten metal and draw on the feeding heads for the extra amount of metal needed to compensate for this supply and for their own shrinkage. This condition is all the more essential the greater the proportion of massive parts present in the casting as a whole. It is for this reason that in some castings the pouring direction is imposed on both the designer and founder alike. This is what happens with the pressure vessel shown in Fig. 16, the flange of which must be uppermost so that it is properly fed by the riser. Similarly, to obtain proper feeding of the rib, the change in the design indicated by the broken hatching is necessary in practice. Surfaces to Be Machined. Dense and sound surfaces for machining should be arranged vertically or at the bottom of the mold. The upper parts can, in effect, contain inclusions (gas bubbles or particles of slag), which have a tendency to rise to the top of the casting because of their low density compared with the metal; these impurities are discovered during machining and affect the condition of the surface obtained.

Fig. 16

Pressure vessel cast upright

936 / Steel Castings This second condition is less essential than the first; it can, however, predominate and decide the pouring direction if the casting has no major differences in thickness. For this reason, cylinders that are to be bored are usually cast vertically. Casting horizontally would give a less clean surface along the upper half-circle. The vital part of the double-cylinder casting shown in Fig. 17 is the 200 mm (8 in.) bore. The casting would be poured in the position shown, the critical section being in the upper part of the mold, where the required machining allowance will give it an extra thickness in relation to the lower cylinder. The feeding of the bottom portion of this casting would require special precautions, such as blind risers or chills. Similarly, it is always essential to avoid arranging machined parts in such a way that the metal reaches them cool, slowly, and without pressure after having traversed a considerable distance in the mold. The casting shown in Fig. 18(a) is a flanged T-piece to be machined. There are three possible molding and pouring positions, depending on which of the axes, ab, cd, ef, is vertical. Choice of Pouring Direction. First position: ab vertical (Fig. 18(b)). The flange B cannot be

fed, and the machined part of the flange A is at the top of the mold. This position need not be considered. Second position: cd vertical (Fig. 18(c)). The three flanges may be fed (separate runners), but the machined part of the flange C is at the top of the mold; better position. Third position: ef vertical (Fig. 18(d)). The three flanges may be fed (separate runners), and their machined parts are vertical; excellent position. Choice of Molding Direction. First position: ab vertical (Fig. 18(b)). The pattern has to be divided at three points (very complicated); the molding has to be done in four boxes (very expensive); assembly is difficult; cleaning is normal. A very complicated and expensive solution. Second position: cd vertical (Fig. 18(c)). The pattern has to be divided at two points (complicated); the molding has to be done in three boxes (expensive); assembly is difficult; cleaning is normal. A better solution, but still costly. Third position: ef vertical (Fig. 18(d)). The pattern can be divided advantageously in the plane ab, cd, which simplifies molding. Molding in two boxes will also be simplified, thanks to the use of the natural taper of the pattern. Assembly is easy, and cleaning is normal. This is the solution to be adopted because it is the least complicated and the cheapest. The third position provides simultaneously the pouring direction enabling a sound casting to be obtained and the cheapest molding position. It sometimes happens that the most suitable pouring and molding directions do not emerge as clearly as in this well-known example. It is often necessary to compromise between several solutions, choosing the one embodying the fewest drawbacks and with the maximum advantages.

Design Factors Considerations of Shape. When choosing shapes giving the required mechanical strength, the following principles should be observed:  From a structural standpoint, in castings sub-

ject to flexural stresses, minimize the amount of material on the neutral axis (at the position where there is no stress).

Considering the three equal sections A, B, and C of Fig. 19, under the same bending load F applied between two points at a distance L, the deflections vary as 100 to 16 to 10, the increased resistance to bending being attributable to the provision of ribs of different heights and thicknesses. A shape compatible with the requirements for correct production of the casting in a foundry involves the adoption of web-carrying alternating ribs, which may be one-third of the plate thickness.  In castings subject to torsional stresses, closed sections should be considered. Increased resistance to torsion is secured by the use of closed profiles. However, they are difficult to produce in the foundry, and it is wise to make use of them only when it is essential to obtain high torsional resistance or when the casting must be of closed section (wheels, frames). Certain precautions must be taken in design in order to minimize the risk of cracking.  Sharp angles should be avoided where stresses can become concentrated and be the cause of fractures. A notched testpiece fractures more easily than one that is not notched. The effect of notching is harmful both from mechanical and foundry viewpoints. It can be the cause of the formation of shrinkage cavities and of cracks or tears in the skin of the metal. These risks can be reduced to a minimum by appropriately shaping the junctions of walls, webs, and ribs. Selection of Wall Thicknesses. Having determined the shape of castings, the first guiding principle is to make all wall thicknesses as uniform as possible, compatible with foundry requirements. If they cannot be properly risered, localized masses of metal attributable to sudden change of thicknesses are the causes of foundry defects, notably tears and surface cracks. An increase in some local thicknesses, in the hope of securing only a slight increase in the strength, often introduces the risk of compromising the soundness of the casting and causing defects. The strength and soundness of a casting depend more on uniformity of thickness than

Fig. 17

Double cylinder cast upright, large diameter at top

Fig. 18

The triple-flanged T-piece (a) may be poured in complicated molds with poor pouring directions (b and c), or in a simple, well-directed mold (d).

Shape Casting of Steel / 937 on the local reinforcement of certain parts. For example, casting shape may be modified instead of reinforcing certain thicknesses. Thus, in the charging arm for an open-hearth furnace (Fig. 20) that had broken in service, it became necessary to modify the design in order to strengthen it. Instead of increasing the thickness of the arm, the same result was reached simply by making the internal diameter 203 instead of 163 mm (8 instead of 6.4 in.), which, with the same thickness, gave higher bending and torsional moments. The use of cast steel instead of cast iron permits reductions in section, as well as equalizing the sections by doing away with some massive parts. An important application of steel castings involves the production of complicated shapes, which cannot always be given a uniform thickness. In order to approach as closely as possible to the ideal of having a uniform section thickness, the number of different thicknesses must be reduced to a minimum. In the steam T-piece shown in Fig. 21, the design A is faulty because the nonuniformity of the thickness of the cylindrical parts interferes with the satisfactory behavior of the fitting under steam pressure. In order to correct this defect, the thickness was made uniform at 1.4 mm (0.055 in.) as shown in design B. In addition, the first design of the cylindrical body and the flanges of the fitting was defective. The outside diameter of the flanges was too great with respect to that of the cylindrical parts, making it necessary to provide ribs to brace the flanges. The ribs adopted, being much too large, were the cause of tears during casting, which gave rise to leakage during testing. In the corrected design, the flange diameter was reduced, which made it possible to do away with the ribs, and the number of different thicknesses in the whole assembly was reduced to two. It is necessary to provide for progressive and uniform variations of thickness. Having reduced to a minimum the number of different thicknesses, the transitions from one to the other should be progressive toward an area that can be risered and should avoid sharp changes of section, which always give rise to defects. Considering flat walls having different thicknesses, the progressive variations should be made according to the recommendations in Fig. 22. Thicknesses of Walls, Webs, and Ribs. The walls constitute the external contour of the casting; their thickness is governed by the strength required. The webs unite several walls, and their object is usually that of increasing the rigidity of the casting. The ribs serve particularly to strengthen certain parts such as joints between webs and walls. The webs are not usually in direct contact with the risers and consequently are less well fed with metal. It is therefore necessary to reduce their thickness and it is usual to adopt a value of four-fifths of the thickness of the walls, which are connected with the risers.

Fig. 19

Fig. 20

Under a uniform bending load F exerted midway between two supports of span L, deflections are in the ratio shown.

Charging arm for open-hearth furnace, showing original and modified designs

Fig. 21

Fig. 22

Section changes on one side

T-piece for steam-pressure service, showing defective (A) and correct (B) designs

938 / Steel Castings A lower limit to the thickness of the webs is set by the fluidity of the steel. Thus, for two vertical walls 25 mm (1 in.) thick, fed by runners and joined by a horizontal web, the latter will be given a thickness of approximately 19.8 mm (25/32 in.). Ribs are often made two-thirds the wall thickness. Consider the valve body shown in Fig. 23. If the thickness of the seating is made equal to that of the external walls, there is a risk of having tears at the points S, since solidification is slower at these points. To avoid this defect, the thickness of the seating should be reduced by one-fifth to one–third. Machining Allowances. Assuming that the minimum extra thickness called for by a manufacturer for machining is of the order of 3.0 mm (0.120 in.), Table 3 indicates the limits within which the maximums of the extra thickness for machining may vary as a fraction of the dimension D, on which the allowance is based. Tables 3 and Table 4 indicate that the extra thicknesses for machining may sometimes equal that of the walls themselves. It is difficult to form an idea of the relative importance of these extra thicknesses merely by considering the drawing of the finished casting after machining, without studying the design of the rough casting. It is for this reason that it is advisable for designers to show on separate drawings, where necessary, various sections indicating actual wall thicknesses of the casting and extra thickness for machining. In Fig. 24, the design A of a finished casting may seem correct in that it shows no excessively sharp differences in thickness. In effect, the addition of extra thicknesses for machining, provided by the patternmaker, may make the casting incorrect from the foundry viewpoint.

There will be a risk of tears occurring at the points C1 and C2. On the other hand, by showing the extra thickness on a supplementary drawing as well as on the drawing of the finished machine casting (design B), the design of the rough casting is provided so that one can deduce from it what precautions to take in order to avoid sharp differences in thickness. Minimum Thickness as a Function of Dimensions. To profit by reduction in weight often afforded by cast steel, specified wall thicknesses should be sufficient to avoid the risk of casting defects. The minimum thicknesses obtainable depend on the casting involved and the position of the particular wall with respect to the runners and the risers. They are made greater for external horizontal webs than for vertical walls. They also depend on the width of the wall, since, for a given thickness, the steel flows better in a narrow web than it does in a wide one. The minimum wall thicknesses as a function of wall length, shown in Fig. 25, are therefore only a general guideline. The normal minimum thickness shown should be adopted initially when replacing other metals by steel. The exceptional minimum thicknesses are those below which one does not usually go. However, it is possible that certain highly specialized founders can produce minimum thicknesses lower than those indicated here. Their success, however, requires special and costly precautions.

Valve body, showing defective and correct designs

Table 3 Effect of size of steel castings on machining allowances Largest dimension of casting Less than 230 230 to 915 mm mm 9 in. (9 to 36 in.)

More than 915 mm (36 in.)

3.0 (0.120)

3.0 (0.120)

3.0 (0.120)

4.0 (0.160) + 6D/ 1000

4.5 (0.180) + 5D/1000

5.0 (0.200) + 4D/1000

D = Longest dimension of the casting in inches

Fig. 24

Influence of extra machining thicknesses, illustrating defective design (sections paired at left) and improved design providing uniform thickness of the rough casting (pair at right)

Hot spots are those parts of castings that cool more slowly than the rest. Unless adequately fed by an adjacent riser, they are the points where shrinkage cavities and tears are likely to occur (Fig. 26). The risk of defects attributable to variations in thickness increases with the massiveness of the part concerned with respect to the casting as a whole. The designer should evaluate the relative importance of the massive parts caused by variations in thickness. It can be done by simple means. The inscribed circle method enables the mass effect in any particular section of a casting to be represented by the surface of the inscribed circle in this section. It is assumed that the relationship of the metal masses at two different points in the same section is equal to that of the surfaces of the corresponding inscribed circles (that is, the relationship of the squares of the radii). This method is not strictly accurate, although it provides an approximation that is more than sufficient and can be of great service to designers (Fig. 27 to 30). The application of the inscribed circle method enables an evaluation of the effect of accumulations of metal resulting particularly from the intersections of walls, webs, and ribs, and to deduce the shapes these points should have in order to reduce to a minimum the risk of the formation of objectionable hot spots. Table 4 Machine finish allowances for radii in steel castings

Minimum extra thickness, mm (in.) Maximum extra thickness, mm (in.)

Fig. 23

Avoidance of Hot Spots

Fig. 25

Minimum thickness as a function of length

Diameter, mm (in.)

Allowance on radius, mm (in.)

Outside radii on circular castings(a) Up to 460 (Up to 18) 460 to 915 (18 to 36) 915 to 1220 (36 to 48) 1220 to 1830 (48 to 72) 1830 to 2745 (72 to 108) Over 2745 (Over 108)

6.3 (¼) 7.9 (5/10) 9.5 (3/8) 13 (½) 16 (5/8) 19 (3/4)

Bore radii on holes Up to 25 (Up to 1) 50 to 178 (2 to 7) 178 to 305 (7 to 12) 305 to 508 (12 to 20)

Solid 6.3 (¼) 9.5 (3/8) 13 (½)

(a) Rings, spoked wheels, spoked gears, and other circular castings

Shape Casting of Steel / 939

Fig. 26

Defects in the principal types of web intersection. L- and T-junctions are best.

Fig. 27

Use of inscribed circles to determine effect of mass

Fig. 29

Cross section of a motor frame casting, showing defective and improved junctions

The problems of defects caused by the accumulations of metal have been investigated experimentally. It has been found that the increase in mass at the right angle of a junction by the inscribed circle method may be evaluated as a percentage with respect to the mass of the wall by taking the quotient of the squares of the radii of the circles inscribed in the junction and the wall. If the figures thus obtained

Fig. 28

Design of X-sections (top) and T-sections (bottom)

Fig. 30

Evolution of the L-design, eliminating shrinkage

for the four kinds of junction shown in Fig. 27 are compared with the corresponding defective areas determined experimentally (Fig. 26), It will be seen that the size of the defect is largely proportional to the increase in mass, which proves the efficacy of the inscribed circle method. Figure 26 also shows that the use of external chills tends to lessen the seriousness of the

defects and even to overcome them in the case of L- and T-pieces, which are the least bad. By using the graphic method (measuring the radii of the inscribed circles), the designer can easily evaluate the increase in the mass with sufficient accuracy. Fig. 27 shows the results obtained for a wall thickness of 38 mm (1½ in.) and fillets having a radius of 19 mm (3/4 in.). It will be seen that the crosses and acute

940 / Steel Castings angle branches should be avoided and that T- and L-junctions are to be recommended. To reduce the accumulation of metal at the intersections of walls, webs, and ribs, it is advisable, where possible, to modify the design as shown in the following examples. Replacement of a cross by an off-set crosspiece is illustrated in Fig. 28. If the intersections of walls and webs are examined, it will be seen from Fig. 28 that the accumulations of metal are not so great when the webs are thinner than the walls. The sharp-angled union shown in Fig. 29 on the left is defective, while that on the right is correct. The dotted union has the added advantage of facilitating molding. The fillets to be provided at junctions between walls and webs should be wide enough to avoid a notching effect, but they should not be so wide as to involve a significant increase in the mass of metal. Junctions to be adopted for the different types of walls are as follows. Junction of Two Walls (Right-Angle Intersection). In a wall, the rate of solidification of the metal is a function of the external perimeter of its section. In the case of a flat wall (one having an infinite curvature), the rates of solidification of the surfaces in contact with the sand are equal. If the radius of curvature decreases, the outer surface, having an area greater than that of the internal part, cools more rapidly, since the heat liberated by the metal during solidification can be absorbed by a greater volume of sand. With a radius of zero (Fig. 30, left), this difference attains its maximum, and the process of solidification is as follows: the outer area of sand being very large, the molten metal

solidifies rapidly on coming into contact with it; the inner area of sand being very small, the sand becomes overheated and within a short time is raised to the temperature of the molten metal itself. This results in a hot spot within the solidifying metal section, so that after solidification there is a shrinkage cavity (often complicated by cracks) at the angle. If the internal radius r is not equal to zero, the size of the shrinkage cavity decreases; it becomes smaller as the outer radius increases, that is, as the ratio of the inscribed circles becomes closer to unity. The case of a 90 angle is shown in Fig. 31 (left), indicating the preferred relationships for r and d under three different conditions. With certain L-junctions, one arm of the L is thicker than the other (Fig. 31, right), as in a flanged fitting. The radius must then be based on the thickness of the largest section and may be connected by a tangent of 15 slope to the lighter section. With a junction angle less than 90 (V-junction), it is advisable to provide for equal wall thicknesses when r is less than 38 mm (11/2 in.); recommended design relationships are shown in Fig. 32. Junction of Three Walls (T-Piece). For T-junctions, the inscribed circle method shows that the accumulation of metal increases with the radius of the fillets. It is therefore necessary to reduce this radius to a minimum, but this is limited by the drawbacks attributable to sharp angles (effects of notches, hot spots). If the thicknesses to be joined differ widely, the thin parts cool more rapidly and exert a pull on the thick parts. To avoid tears, or even fractures, it is advisable to provide tapered junctions.

Fig. 31

Dimensional relationships of L-design with exterior corners

Fig. 32

Recommended V-design relationships

Fig. 33

Recommended design relationships for simple T-junctions and tie-in members in the form of T-junctions are given in Fig. 33. The inscribed circle method shows that Yjoints are very defective. They can be replaced by a combination of a T and an L- piece, which is definitely more advantageous. This can be effected for two equal thicknesses, bearing in mind the above rules for L- and T-junctions. For unequal thicknesses, the necessary information is given in Fig. 33. Junction of Four Walls. In Fig. 26, it was shown that the action of external chills is not sufficient to prevent the formation of shrinkage cavities in such a joint (at least for thicknesses of 76 mm, or 3 in.). If soundness of the casting is to be guaranteed, this kind of junction should not be used unless its length is short and it is possible for it to be cast with direct risers. However, in a crosspiece, the central part is the least stressed, so that a casting is just as strong and much easier to produce sound if made either by inserting a central chill or by providing a central cavity. The designer should know that, when designing a crosspiece, the founder usually will be compelled to make use of an internal chill at the center part with the object of improving the soundness of the casting and, in consequence, its strength, because the accumulation of metal is such that it cannot be fed completely unless it is placed directly under a riser. Alternatively, to avoid the use of an internal chill, a central core may be used, provided it is of sufficient diameter (see the section “Dimensioning of Holes” in this article) to avoid fusing of the core. This is equivalent to replacing the crosspiece by four T-pieces having uniform thicknesses around the central cavity. It is often possible, while still retaining the same moments of inertia, to replace the cross by an L- or a Z-piece and even by a closed section in the form of an O. Figure 34 indicates the various methods of joining two webs without using X-junctions. The first two should be avoided. They can be substituted by No. 3, 4, and 5, which are better from both mechanical and foundry viewpoints. Number 6 could be replaced with advantage by No. 7, 8, and 9, which possess better mechanical properties and lessen the risk of tears. Numbers 10 and 11, which call for the use of a core, have, on

Recommended T-design (left) and T-junction joining unequal walls

Shape Casting of Steel / 941 the other hand, the advantage of offering high resistance to bending and torsional stresses. For the internal ribbing of a plate or frame, it is advantageous to replace the checkerwork consisting of four walls at an angle of 90 by junctions of three walls at 120 . Junction of Five or More Walls. Where the junction of five or more walls can be envisaged, it is advisable to place a central clearance in such a way as to avoid the meeting of more than three. The solution to this important problem of joining walls, webs, and ribs can be summarized by the two following rules:  Utilize L- and T-pieces as much as possible,

and avoid crosses or acute angles.

 Arrange for ordinary junctions if the ratio of

the thicknesses D/d is less than 1.5 and for tapered junctions if the ratio D/d is higher than 1.5. Strengthening Ribs and Recessing. The addition of ribs and the use of recessing often permit the avoidance of accumulations of metal at certain points. These two means, when judiciously adopted, help the founder economically to obtain sound castings that are free from tears. Increasing the thickness of a casting to obtain higher strength often results in the production of a massive part that may not be sound. By adding ribs rather than increasing wall thickness, the casting is strengthened without adding to possible defects. The double-flanged sleeve of Fig. 35 (top) is a common design. The flanges are often thicker than the barrel, resulting in accumulations of metal likely to cause cracks if certain preventive measures are not taken (horizontal pouring, separate feeding heads for each flange). When this sleeve must be cast upright, an obvious step is to make the lower flange of the same thickness as the barrel and to strengthen it with ribs, which by quick solidification serve to resist the shrinkage forces that may distort the flange, cause hot tears, and interfere with the mechanical strength of the flange. In order to guard against distortion of the flanges or the formation of tears, the founder is often compelled to make use of strengthening ribs that must be removed subsequently, thus adding to the cost of the casting. It would be more economical for the designer to provide strengthening ribs having the double role of replacing objectionable increased thicknesses, and strengthening the profile in its resistance to contraction forces. In the example examined previously, the addition of ribs 19 mm (3/4 in.) thick will enable the thickness of the lower flange to be reduced from 38 to 25 mm (1 1/2 to 1 in.). Recessing enables massive parts to be lightened and, at the same time, lessens the risks of defects resulting from the formation of hot spots in such parts. However, the recessing should be done with great care. The placing of small cores complicates molding; they are likely to shift and, because of their low density

Fig. 34

Methods for joining two walls by means of open or closed sections

Fig. 35

Use of brackets in steel casting design (top). Lightening of a casting that is too massive by means of ribs and recessing (bottom)

compared with that of the molten metal, they tend to become entrapped in the casting under the effect of ferrostatic pressure. Moreover, small cores can fuse and be very difficult to remove. The use of recesses, although sometimes essential, may entail a considerable increase in the cost of the casting. The following examples show the possibilities and the limits governing their use. To lighten the casting (Fig. 35, bottom), the massive body has been recessed and strengthened by ribs; however, to prevent a crack at the top of the rib on the left under a massive part, this rib has been recessed. Additional coring costs are more than offset by the elimination of casting defects. Where the area in

question can be fed directly by a riser, recessing is usually unnecessary, since it does not contribute to metal soundness and it complicates production. Hot Spots in Sand. Consideration of the L-piece showed that the overheating of thin parts of sand enclosed between two walls gave rise to the formation of hot spots. The same applies when a part of the mold is enclosed on several sides by the walls of the casting; the sand there heats up much more than in the rest of the mold, causing the formation of a hot spot and the occurrence of defects. Regions of sand, belonging to either the mold or the cores, that are not sufficiently voluminous must be avoided. In this respect, it is

942 / Steel Castings advisable to arrange for cores and mold parts to be at least twice the thickness of the metal surrounding them. For example, the fitting shown in Fig. 21, which had already been modified by making the thicknesses uniform, can undergo a second improvement involving an increase in the sand thickness at the point C. This can be done either by increasing the distance between the two vertical flanges or by displacement of the horizontal barrel. In general, recesses and holes of too-small dimensions should be avoided. Dimensioning of Holes. As-cast holes are made by inserting sand cores in the mold. The holes may be straight through or blind (Fig. 36). A core becomes fused when surrounded by a mass of metal too great in relation to its dimensions. This causes great difficulties when the casting is being cleaned. In some instances, it is almost impossible to remove a fused core from the casting. Careful choice of the diameter and length of holes is closely related to the thickness of the surrounding metal. In effect, the process of cleaning, which in foundry practice is an operation just as costly as molding, is largely dependent on the dimensional ratio between the diameter and length of the holes. Three cases serve to illustrate this principle:  If the diameter D of the hole is less than

twice the thickness e surrounding the core, its length L should be no more than equal to the diameter for a cylindrical hole (Fig. 36(a)); thus, for holes in walls and webs, the diameter of the hole D may equal the thickness of the wall L. However, for a blind hole, the length should not exceed half the diameter (Fig. 36(b)).  If the diameter D of the hole is between twice and three times the thickness e surrounding the core, its length L should not exceed three times the diameter for a cylindrical hole (Fig. 36(c)) and twice the diameter for a blind hole (Fig. 36(d)).  If the diameter D of the hole exceeds three times the thickness, the length of the core is no longer limited by fusing but by the fluidity of the metal being poured and the ability of the cores to withstand the ferrostatic forces to which they are subjected when the mold is being filled. Under these conditions, the graph in Fig. 25 should be consulted in order to ascertain the permissible length of the core as a function of the surrounding thickness.

Fig. 36

The difficulty in obtaining clean surfaces free from sand usually requires considerable reduction in the machining speeds of cast steel when compared with other metals. It is therefore seldom advisable to provide holes in castings that can be more rapidly and economically made by drilling. It is preferable to provide holes in castings only when their machined diameter is greater than 32 mm (1 ¼ in.), which corresponds to an as- cast hole of approximately 19 mm (3/4 in.). On the other hand, in order to avoid breakage of drills, it is necessary that the bosses be cast sound. The absence of cores or chills in the bosses often gives rise to shrinkage cavities. For this reason, the founder frequently adopts expedients such as internal chills made of iron rods, soft iron wire spirals, or merely the use of large-headed nails. This practice can greatly simplify the work of the founder without affecting the mechanical strength of the casting, especially if the chills used are removed during machining. The designer should leave to the founder the responsibility of choosing the chills and should prohibit their use only where, for reasons of mechanical strength, the use of internal chills in a massive part of the casting cannot be permitted. This should be clearly indicated on the drawing, and its effect on the cost should be understood. Closed Sections. Shrinkage can vary considerably from one part of a casting to another. These shrinkage variations give rise to residual stresses that may show themselves as soon as solidification takes place, when the metal has little strength. They can cause cracks and even fractures if the profile involved cannot withstand a certain amount of deformation. Two typical closed sections, a pulley and a flywheel, illustrate the problem. If the thicknesses of the spokes and rim are different, stresses are set up during cooling, and the stresses can be the cause of cracks. Thus, a pulley or a flywheel with straight spokes can develop cracks beside the hub or the rim. On the other hand, when a design with curved spokes is adopted, the risk of cracks is overcome since the stresses are accommodated by the variations in the curvature of the spokes. Generally speaking, it is best in closed sections to replace some straight elements by curved ones, to accommodate stresses set up during the transition through the critical hot tearing temperatures without affecting the strength of the whole assembly when cold. For

Dimensional relationships for cylindrical through holes and blind holes

the same reason, wheels should be given an odd number of spokes. The principle of wave construction is useful in providing for the relief of stresses in cast structures. It makes use of members slightly waved or curved. This principle has been used in the design of the wheel shown in Fig. 37 (left). A certain elasticity is imparted by the curvature of the spokes and their inclination relative to the axle. Drawing on the same principle, Fig. 37 (right) compares a rigid and an elastic design for a flywheel, the desired elasticity resulting from the use of an offset section. The susceptibility of closed sections to cracking increases with any variation in thicknesses and large accumulations of metal. In a frame heavily ribbed by webs intersecting at right angles, accumulations of metal occurring at these intersections will give rise to hot spots and the risk of tears. By reducing these accumulations of metal, the tendency to tearing is diminished. This can be attained by coring out heavy sections, provided the cores are given a diameter at least twice the thickness of surrounding metal and have uniform thicknesses of metal around them. If, in the same design, the use of cores is to be avoided, the webs can be staggered, allowing the central bar to take a slight deformation instead of breaking. This procedure also has the advantage of lessening the accumulations of metal by replacing crossings with T-junctions. Accuracy of Dimensions. The considerable contraction of steel during cooling causes the patternmaker to calculate the dimensions of his pattern by increasing their values, which may differ in different parts of the casting, in order to compensate for the contraction. However, owing to variations in contraction and the molding operations, it is impossible to be sure of obtaining castings having exactly the dimensions given in the drawing. Therefore, it is permissible that the ascast dimensions of a casting may differ from the exact dimensions laid down, provided that the plus or minus differences do not exceed certain limits or tolerances. The tolerances in dimensions allowed by standards or specifications are of the form: t ¼ ðaD þ bÞ

when D denotes the length involved, while a and b are constants.

Shape Casting of Steel / 943 The difference of a casting dimension from its exact value is obviously the sum of two terms, one being a constant arising from molding and coremaking procedures, the other being proportional to the length and corresponding to the uncertainty of the contraction (Fig. 38). Table 5 indicates the tolerances calculated on this principle for unalloyed steel castings. These tolerances apply to the parts of castings destined to remain as-cast. The system of tolerances adopted, however, can differ for various parts of the same casting. It is often found that certain as-cast parts are indicated as having a dimensional accuracy unattainable within the limits of the tolerances. It should be borne in mind that the responsibility of the founder concerning the accuracy of the dimensions of parts to be machined is limited to making this machining possible by providing the necessary extra thickness. Machining allowances are shown in Tables 3 and 4; they may vary, taking into account the tolerances, so that the maximum extra thickness equals the minimum extra thickness with the plus tolerance added. Setting Up for Machining. When castings are to be machined under mass-production conditions, the machining is carried out on jigs, which determine the position of the surface to be machined with respect to as-cast surfaces. Therefore, the as-cast surfaces coming into contact with the jigs should vary as little as possible from one casting to another; that is, they should be obtained by direct molding. For this reason it is essential to indicate jig location points to the founder, to enable him to select an appropriate method of molding. Continuity of Machining Operations. Boring or thread- cutting operations can be correctly carried out only on continuous cylindrical surfaces. To ensure continuous working of the tool, it is advisable to complete the segments of the cylindrical surfaces by grouping the castings. Thus, a divided nut can be made from a single pattern providing two castings (Fig. 39, left). If grouping is not feasible, the cylindrical shape is completed with supplementary material (Fig. 39, right). This is also true where two boring operations are to be carried out. Tool clearances are often shown on drawings to provide for machining. However, they are usually insufficient because the designer has not taken machining allowances into

account, with the result that the castings contain layers of sand practically surrounded by metal (Fig. 40). It is impossible to remove the sand, and the tools are quickly blunted. It is necessary to provide very wide tool clearances. In this connection, it is well to show on the drawings the tool clearance in full-scale transverse section, together with the machining allowance and the tolerance, and to ensure that the gaps provided satisfactorily comply with the conditions regarding core sizes (Fig. 40).

Thin-Wall Steel Castings Thin sections save weight and may thus contribute to a more favorable strength-to-weight ratio. By requiring a smaller volume of metal, thin walls (within practicable limits) may also lower casting costs, particularly when an expensive alloy is being poured. Realization of the advantages of thin walls in a casting depends largely on certain physical limitations of metal in the casting process. These limitations of metal in the casting process. These limitations are concerned with the fluidity of molten metal as it affects mold filling, with the processing required for metal soundness, with differences in alloy and in casting process, with distortion and heat treating problems (in some designs), and with the cost of foundry engineering and development. Variously published rules and suggestions regarding the minimum thicknesses to which cast sections can be designed are in conflict, because they are based on different points of view. Some tables, such as Table 6, are intended to prevent an increase of casting cost because of thinness of section. Others consider the actual limits of the process, rather than cost, but often fail to define the applicable processing conditions. Because so many aspects of design (such as the area of the thin-wall section, its proximity to a gate, and the availability of heavy sections to conduct molten metal to it) have an influence on the actual minimum wall obtainable for any particular casting, recommendations such as those in Table 6 have limited usefulness when applied to a specific part. They do, however, provide the casting designer with a starting

point, by familiarizing him with minimum thicknesses that normally can be produced efficiently. As an example, Fig. 41 shows a thin-wall austenitic stainless steel sand casting with a number of design features that helped to simplify casting problems and permit production at minimum cost. The casting is of relatively uniform section thickness, except for the heavy boss at the top. This boss was ideally placed to facilitate filling the sand mold and to avoid distortion as the steel solidified and cooled. The casting has good stability and rigidity, and its slightly curved surfaces minimize distortion. As originally designed, the walls in the lower part of the casting were 1.52 mm (0.060 in.) thick. However, because of difficulty in eliminating porosity in the walls, caused by the backpressure of mold gas, rejection rate was approximately 50%. Increasing the thickness of these walls to 1.91 mm (0.075 in.) reduced the rejection rate to approximately 25%. When the wall thickness was increased to 2.29 mm (0.090 in.), the rejection rate was further decreased to 5%. Even the thickest wall (2.29 mm, or 0.090 in.) is well below the minimum usually suggested in tables such as Table 6, which are based on economy rather than absolute process capability. This casting illustrates what can be done when the design of a casting is favorable for producibility. In all casting processes, castings that are designed so that rapid and orderly flow of metal to all parts of the mold is promoted are easiest to produce with a minimum of flaws or defects. However, the specific casting process can influence metal flow, and, because of inherent differences, each casting process exerts its particular influence on the production of thinwall castings. The relative unsuitability of green sand molds to the casting of thin-wall sections is considerably magnified in the casting of either alloy steels or superalloys. For this reason, most aircraft castings made of these materials are produced in shell, ceramic, or baked sand molds. An analytical approach to the design of thinwall steel castings to be produced in green sand molds necessitates an understanding of four major production problems involved: mold

Table 5 Dimensional tolerances for steel castings not to be machined Dimension to which tolerance applies

Average Concise Minimum

Fig. 37

Wave construction, showing rigid and elastic design

Fig. 38

Variations in tolerance, shown as a junction of dimensions

Under 305 mm (12 in.)(a)

305 to 914 mm (12 to 36 in.)

914 to 3048 mm (36 to 120 in.)

0.060 + 0.006D 0.040 + 0.005D 0.030 + 0.004D

0.060 + 0.006D 0.050 + 0.005D 0.040 + 0.004D

0.080 + 0.006D 0.070 + 0.005D 0.060 + 0.004D

(a) Tolerance, 1.6 mm (1/16 in.) min. D is the longest dimension of the casting in inches.

944 / Steel Castings filling, distortion, soundness, and the influence of design on heat treating problems. Mold-filling problems become increasingly acute as wall thicknesses are decreased to less than 4.75 mm (3/16 in.). Above this thickness, mold filling is seldom critical. When wall thickness is decreased to less than 1.6 mm (1/16 in.) adequate mold filling becomes almost impossible, except for extremely short distances, and misruns result because of the freezing of metal before the mold is filled. To achieve minimum weight, the designer will often incorporate minimum section thicknesses within a casting, in all locations where high strength is not a requirement. For instance, in the turbine bearing housing shown in Fig. 42 (a green sand casting made of alloy steel), eight posts, each incorporating a stiffening rib, connect the top and bottom portions. It was calculated that a rib thickness of 2.54 mm (0.100 in.) would be adequate to meet strength requirements. However, ribs of this thickness could not be filled completely; all castings produced were rejected for misruns. No practical improvement was possible in foundry processing. Consequently, the patterns were revised to increase rib thickness to 3.2 mm (0.125 in.) rejections for misruns were almost entirely eliminated. Alloy Selection. Important in the problem of mold filling are the fluidity and ability to feed of the metal specified. These properties of

castability are low in steels and superalloys, as compared with light-metal alloys. Among steels, the castability of one alloy differs from that of another. As a result, a casting that can be produced successfully in one alloy may not be castable in another. Among foundries, the comparative moldfilling abilities of metals are controversial, and relative ratings of metals in regard to this characteristic are to some extent empirical. Existing disagreement reflects the many factors that may affect the ability of metal to flow in a mold. From its experience with many alloys, one foundry lists alloys in order of decreasing ability to fill thin cavities as follows:  Cobalt-base alloys (except precipitation-

hardening alloys)

 Nickel-base alloys (except precipitation   

hardening alloys) 11 to 13% Cr steels Series 4300 alloy steels Series 300 stainless steels Series 300 stainless steels molybdenum

containing

Rigging and processing procedures vary from foundry to foundry, and foundries generally specialize in respect to the range of castings and alloys with which they prefer to work. This should be recognized by the engineer when he

designs a casting with rigid requirements that may tax the capabilities of the process best suited to produce the casting. The aft-fuselage speed-brake hinge shown in Fig. 43 was designed to be sand cast from 4340 steel and heat treated to a tensile strength of 1240 to 1380 MPa (180,000 to 200,000 psi). It was to be produced to close tolerances and rigid standards of inspection. The foundry accepted the job with the understanding that the material could be changed from 4340 steel to type 431 stainless, and that certain of the more stringent tolerances could be modified. The change in material was requested because of the foundry’s extensive experience with stainless steel. Although type 431 stainless costs considerably more than 4340, this increase was more than offset by the higher percentage of acceptable castings produced in stainless steel. Friction between molten metal and mold may seriously hinder mold filling. It becomes increasingly troublesome as the ratio of mold surface to metal volume increases, a condition that is characteristic of thin-wall castings. A high ratio promotes premature freezing in thin sections. Mold gases may create backpressures that retard the flow of metal. These gases usually originate from the burning of organic moldbonding material and from the heated air in the mold cavity. Gases are expelled through

Fig. 39

Grouping of two castings to facilitate machining (left). Providing additional metal to facilitate machining (right)

Fig. 40

Design for tool clearance

Table 6 Typical minimum section relations for three metals in five casting processes(a) Minimum section thickness, mm (in.) Casting method

Sand Permanent mold Investment Die casting Plaster mold

Aluminum

3.18 þ 0.787  2.36  þ 0.381 1.58 þ 0.254  1.58  þ 0.254 2.03 þ 0.381 

(0.125 þ 0.031)  (0.093  þ 0.015) (0.062 þ 0.010)  (0.062  þ 0.010) (0.080 þ 0.015) 

Magnesium

3.96 þ 0.787  3.18  þ 0.381 1.58 þ 0.254  2.36  þ 0.254

(0.156 þ 0.031)  (0.125  þ 0.015) (0.062 þ 0.010)  (0.093  þ 0.010) ...

(a) For compatibility with economical production. Not the thinnest producible by the processes

Steel

4.75 þ 0.787 (0.187 þ 0.031)   ... 2.36 þ 0.254 (0.093 þ 0.010)   ... ...

Fig. 41

Thin-wall sand casting produced from austenitic stainless steel. One section of the casting required two revisions in wall thickness to bring rejection rate to an acceptable level. Rejections were 50% with 1.52 mm (0.060 in.) wall, 25% with 1.91 mm (0.075 in.) and 5% with 2.29 mm (0.090 in.).

Shape Casting of Steel / 945 the mold wall in proportion to the static head of metal in the sprue and risers, depending also on the permeability of the mold material. Designs that must be oriented in the mold so as to have domelike thin sections are especially likely to create problems in venting of mold gases, which can generate backpressure and cause the metal to freeze before the mold is filled. The most obvious solution in the foundry is to use a higher pressure head, in order to force metal into the cavity more rapidly, but this may cause undesirable turbulence of the metal at or near its point of entry into the mold. Mold-Filling Problems. An unusual moldfilling problem that, because of the design of the casting, had no satisfactory solution was encountered in the production of the alloy steel part shown in Fig. 44. This casting was molded in the position shown, because, had the mold been inverted, the heavy section at the junction of the shaft and plate sections could not have been fed through the long shaft. Because of its diameter, the entrance to the saucer shape is small in proportion to the volume of metal that must pass through it. Thus, the metal broke into separate streams and trickled down to various points on the periphery of the plate section. Such separate streams oxidized rapidly and resulted in misruns, cold shuts, and “BB shot.” (The BB-shot defect refers to drops of molten metal that separate from the main body of fluid metal, freeze prematurely, and fail to fuse with the main body of metal on recontacting it.)

Fig. 42

Eight stiffening ribs in this alloy steel green sand casting were originally 2.54 mm (0.100 in.) thick; misruns made all castings produced unacceptable. Increasing thickness of these ribs to 3.18 mm (0.125 in.) almost completely eliminated rejections for misruns.

A possible solution may have been provided by the addition of a reservoir (indicated by the broken line, B, in Fig. 44) surrounding the periphery of the plate. Such a ring would have been removed subsequently by machining. Although the wall thickness of 1.52 mm (0.060 in.) would probably have caused the metal to freeze before it could flow into the reservoir, increasing the plate thickness to 3.05 mm (0.120 in.) should have permitted the flow of metal into the reservoir. Unfortunately, this approach would have been impractical; the additional metal would have had to be removed by machining to a curved surface. The cast volute shown in Fig. 45 is a common configuration that illustrates another mold-filling problem. There are three ways to gate such a volute. The first and most common method is to gate at the inside of flange A. The second is to gate at flange B. The third method is to gate at both flanges A and B. If the mold is filled at flange A, the metal will take a relatively long time to fill the thinwall section before filling flange B. Thus, if the wall of the volute is less than approximately 4.76 mm (3/16 in.) thick, the metal is likely to freeze before filling the outside flange of the casting. If the mold is filled at flange B, the same problem exists, with the added danger that metal will cascade at angle C, producing cold shuts and other defects. Finally, if the mold is filled at flanges A and B, entrapped air and mold gases in the thin-wall section may cause bubbles and porosity. Also, if the distance from

C to D exceeds approximately 150 mm (6 in.), this thin section will be very difficult to feed well enough to make it sound. The broken lines in Fig. 45 indicate a method for improving the castability of such a part by adding to the thickness of the thin sections and by tapering the walls to the center of the volute (E). Illustrating another mold-filling problem, the steel casting shown in Fig. 46 indicates how obstructions to the flow of metal as it fills the mold can result in casting defects. In this casting, metal entering the mold cavity through the heavy section (B) flowed into the thin plate area and flanged sections. In pouring, the mold was tilted, with the heavy section down, in order to prevent a trickling of metal into the plate and flanged areas. To assure metal flow to all parts of the mold, this casting was poured rapidly. However, as the wall of metal advanced to each of the three sand cores C, an eddy was formed behind each core, as indicated by the shaded areas, resulting in defects. The defects became increasingly critical as the thickness of the plate was decreased. The possible revision of core shape shown in the lower portion of Fig. 46 was not permissible. Consequently, in order to produce a sound casting, it was necessary to eliminate the coring of holes and to drill these holes. Another problem in mold filling results from the cascading of molten metal down one side of the mold for a casting designed with a ladder shape. For the original design illustrated in

Fig. 43

This close-tolerance sand casting, although designed to be cast from 4340 steel, was produced in type 431 stainless because of the foundry’s greater experience with stainless steel. Increase in material cost was more than offset by higher percentage of acceptable castings.

Fig. 45 Fig. 44

Alloy steel sand casting that was impractical to produce because its inverted saucer shape contributed to cold shuts and misruns. See text for discussion.

Thin-wall volute castings such as this present problems of mold filling because molten metal must travel a long distance to fill the thin walls. See text.

946 / Steel Castings Fig. 47, metal must enter the mold cavity at points A and rise evenly into the two outer walls. In walls less than approximately 4.75 mm (3/16 in.) thick, this is difficult to assure. If the outer walls fill unevenly, metal will flow across a cross member B and cascade down the opposite wall. As descending metal makes contact with metal rising in the wall, bubbles are likely to be trapped, and cold shuts and other defects may appear. The weight of metal in a thin-wall casting is usually insufficient to create enough pressure to force entrapped air out through the mold walls. The result is voids and porosity. Thus, this 200 mm (8 in.) long casting as originally designed, with a 1.5 mm (0.060 in.) wall thickness and five cross members, is incapable of being properly filled. The problem can be solved in at least three ways. First, the outer wall thickness can be increased to 4.75 mm (0.187 in.) or more, as in Fig. 47(a). Second, the inner cross members may be slanted, parallel to each other, as in Fig. 47(b), eliminating most of the danger of cascading. Third, the inner cross members may be slanted in opposite directions to form a truss, as in Fig. 47(c). Distortion is a major problem in the production of thin-wall steel castings. During solidification and in cooling to room temperature, distortion may be caused by variations in the cooling rate and in the contraction of sections of different thickness in the casting. In fragile castings, distortion may be caused by mold restraint. (It may also occur during heat treating). Soundness. Because of the small cross-sectional area of a thin wall, it is difficult for a riser to replace the metal lost in shrinkage. In very thin sections, feeding may be less of a problem, because the likelihood of centerline shrinkage is slight. Examples of Good Design. The chine fitting (a steel casting) shown in Fig. 48 is an excellent example of good casting design. The heaviest section is at the center, where it can be fed

readily. The legs of the casting are tapered in steps. All wall sections of uniform thickness are less than 75 mm (3 in.) long and can be fed adequately from the center, with metal soundness assured. Changes in section thickness are not abrupt, and all inside corners are provided with generous fillets to resist tearing. Radii on outside corners serve to maintain constant wall thickness. A stout tie bar crosses the center, where the danger of distortion is greatest. The walls of the casting are thick enough to resist distortion. Finally, the radii at points C are large enough so there is no cracking or tearing in this area. This casting was producible without difficulty. Another well-designed steel casting is shown in Fig. 49. This part can be fed with a riser at each end and at each side at points A; thus, the casting can be filled from the center or either end. Wall thicknesses are approximately 3.0 mm (0.12 in.) and the T-sections are easy to feed, because they are not more than 100 mm (4 in.) long nor more than 50 mm (2 in.) from the nearest riser. As originally designed, the casting was weak at points B. However, this deficiency was readily corrected by the addition of outer ribs. The revised design has good rigidity, with all areas of stress well braced and with ample corner radii and fillets. The casting did not distort in production. The wing spar casting shown in Fig. 50, although large and rather complex, is another example of good design. Basically, the design is that of a box section with a flat side approximately 6.4 mm (¼ in.) thick. One end of this steel casting is closed and varies in thickness from 4.52 to 38 mm (0.178 to 1.50 in.). Except for the round bearing support at the other end, the remainder of the casting varies in thickness from 4.75 to 6.35 mm (0.187 to 0.250 in.). Most of the heavy boss sections are located on the inside of the flat plate area. In casting,

the part was inverted so that the bosses could be fed through the plate area, thus making this the cope surface. All massive areas could be reached from the top, the two ends, or the sides. The design of this casting exemplifies the principle that areas of increased mass should be restricted to five sides of a basically cubic configuration. All corners of the casting have generous fillets and radii. The boxlike shape resists distortion in cooling from the liquid and during heat treatment. The numerous blanks attached to the horizontal ribs are intended to provide specimens for test purposes and can be removed easily without damaging the casting. For thin-wall castings, such specimens provide more reliable test results than do specimens that are separately cast.

Tolerances Shell mold casting is particularly adapted to the production of large quantities of relatively small, high-quality components. Because of reduced draft, greater dimensional accuracy of pattern, and superior mold stability, this technique usually permits closer dimensional tolerances than sand mold casting. Minimum dimensional tolerances for shell mold steel castings that are not to be machined are shown in Table 7. Recommended minimum section thicknesses for shell mold castings as a function of section length are shown in Table 8. The investment mold casting process is best suited to production of fine detail and close dimensional tolerances. Close control of dimensions in the wax, plastic, or frozen mercury pattern, smooth surface and rigidity of the ceramic investment or shell mold, and melting and drainage of the pattern all contribute to accuracy. Suggested minimum tolerances for investment castings are shown in Table 9 and Fig. 51.

Fig. 46

This thin-wall steel casting required rapid pouring to fill the mold completely. Three cores obstructed the free flow of metal, creating eddies that resulted in defects. Redesign of cores as shown, had it been otherwise acceptable, would have solved the problem of metal flow.

Fig. 47

Uneven filling of the sides of the “ladder” configuration of this casting as originally designed caused metal to flow through the cross members and cascade down either side. Three revised designs are shown.

Shape Casting of Steel / 947 The minimum thickness that can be produced by the investment mold process is considerably less than by either sand or shell mold casting, primarily because of the higher temperature of the investment mold during pouring. The pouring methods used also contribute to soundness in thin sections. Table 10 shows minimum thickness of sections of investment castings as related to length. The casting designer should carefully consider all specified tolerances. The closer the

tolerance, the higher the final cost of the casting. For greater economy, tolerances should be only as close as the application actually requires.

Size of Cores Another dimensional limitation in making steel castings is the minimum diameter of cores that can be used without core failure. In green sand molds, cores of dry sand, oil-bonded baked sand, and shell-molded sand have similar limitations. The minimum diameter core that can be used successfully in steel castings depends on the thickness of the metal section surrounding the core, the length of the core, and the special precautions and procedures used by the foundry. The thermal conditions that the core must withstand increase in severity as metal thickness increases and core diameter decreases. More heat must be dissipated in heavier sections, and the core is less able to absorb and dissipate this heat as the diameter decreases. As heat conditions become more severe, the possibilities for metal penetration and sand burn-on

increase, making the cleaning of the casting and removal of the core more difficult. The thickness of the metal section surrounding the core and the length of the core affect the bending stresses induced in the core by flotation forces. Thus, they affect the ability of the foundry to obtain the required tolerances in cored cavities. The core must be large enough so that it can be reinforced with wire or rod to withstand these stresses. As metal thickness and core length increase, the amount of rodding required increases. The minimum core diameter must also increase to accommodate the extra rodding. Fig. 52(a) shows the recommended minimum diameter of cores to be used in cylindrical or boss sections in steel castings, as a function of section thickness and core length. These curves are for cores assembled in a mold in a horizontal position and supported at the ends only. Minimum core diameters determined from these curves can be reduced by 25% if the core is to be used in a vertical position. A similar curve for cores in plate sections of steel castings is presented in Fig. 52(b). Smaller core diameters may be used, but they involve special practices, such as the use of stronger core materials, which increase the cost of the castings.

Fig. 48

Example of good design in terms of foundry producibility. The heaviest section of this cast steel chine fitting is at its center and is easily risered. From the center, each section tapers in steps to the extreme ends.

Table 7 Minimum dimensional tolerances for shell mold steel castings not to be machined Dimension

Length, mm (in.) Less than 100 (4) 100 to 200 (4 to 8) 200 to 400 (8 to 16) 400 to 800 (16 to 32) Across the parting line Draft Hole diameter Concentricity Angular tolerance

Tolerance(a)

þ0.25 (þ0.010)   þ0.375 (þ  0.015) þ0.875 (þ  0.035) þ2.5 (þ0.10)   þ0.25 (þ  0.010) 0.05 mm/mm (0.002 in./in.) þ 0.13 mm/mm (þ  0.005 in./in.) 0.05 mm/mm (0.002 in./in.) þ0 , 450 

(a) Between unmachined surfaces

Fig. 50

Box-shaped wing spar casting in which heavy sections can be easily risered from the top, the two sides, and the two ends. This steel casting demonstrates the rule that areas of increased mass should be restricted to five sides of a basically cubic configuration.

Fig. 49

Well-designed steel casting in which all sections are accessible to properly located

risers

Table 8 Recommended minimum section thickness for steel castings produced in shell molds Length, mm (in.)

Less than 100 (4) 100 to 300 (4 to 12) 300 to 450 (12 to 18) 450 to 1220 (18 to 48) 1220 to 2000 (48 to 80) Over 2000 (over 80)

Minimum section thickness, mm (in.)

5 (3/16) 6.5 (¼) 8 (5/16) 9.5 (3/8) 13 (½) 16 (5/8)

Table 9 Minimum tolerances for investment mold steel castings Dimension, mm (in.)(a)

Minimum tolerance,  þmm (þ  in.)

Dimension, mm (in.)(a)

Minimum tolerance,  þmm (þ  in.)

13 (½) 25 (1) 38 (1½) 50 (2) 63.5 (2½) 100 (4) 125 (5) 150 (6) 180 (7)

0.125 (0.005) 0.125 (0.005) 2.3 (0.009) 0.25 (0.010) 0.38 (0.015) 0.51 (0.020) 0.635 (0.025) 0.75 (0.030) 0.9 (0.035)

200 (8) 230 (9) 250 (10) 380 (15) 405 (16) 510 (20) 635 (25) 760 (30) 810 (32)

1.00 (0.040) 1.15 (0.045) 1.25 (0.050) 1.90 (0.075) 2.06 (0.080) 2.54 (0.100) 3.2 (0.125) 3.86 (0.150) 4.0 (0.160)

(a) As, for example, the dimension between centerlines of the large bore and the small bore of a connecting rod

Fig. 51

Minimum and recommended tolerances for steel investment castings on one side of parting line, as a function of the plan area of the casting

948 / Steel Castings Table 10 Minimum thickness of investment cast steel sections Length, mm (in.)

Minimum thickness, mm (in.)

Length, mm (in.)

6.35 (¼) 12.7 (½) 19 (3/4) 25.4 (1)

0.75 (0.030) 1.00 (0.040) 1.25 (0.050) 1.5 (0.060)

33 (1¼) 38 (1½) 50 (2) 63.5 (2½)

Minimum thickness, mm (in.)

1.5 1.5 1.5 1.5

(0.060) (0.060) (0.060) (0.060)

Because stronger molding materials are used in the investment casting process, thinner and longer cores may be used. It is feasible to use cores of considerable length (up to 63.5 mm, or 2½ in.) that are supported at one end only.

Casting in Graphite Molds Steel railroad-car wheels have been cast in reusable graphite molds, using pressure pouring techniques. Graphite is a suitable mold material because it does not fuse with molten steel at the high pouring temperatures required. For carwheel applications, the graphite molds (cope and drag) are machined from graphite blocks with a minimum diameter of 110 cm (43 in.). Graphite molds, because of their exceptional resistance to burn-in, ensure that the wheel casting will not stick to the mold. Graphite also exhibits enough resistance to thermal shock to endure high pouring temperatures without cracking. Its low coefficient of expansion and ability to resist distortion ensure re-producibility of successive castings made in the same mold. On the other hand, because of its relative softness and low mechanical strength, graphite is highly susceptible to damage and breakage and must be handled with care. Conventional gravity pouring from a bottom-pour ladle is not suitable for graphite molds, because of the severe erosion that results. Pressure Pouring. To avoid erosion and to control the rate of pouring in graphite molds,

Fig. 52

(a) Recommended minimum diameters for horizontal cores supported at ends only, in cylindrical or boss sections of steel castings. Minimum core diameters shown in the three curves can be reduced by 25% if the core is vertical in the mold. (b) Recommended minimum core diameter for plate sections in steel castings

a technique known as pressure pouring is used. Pressure pouring consists of forcing molten metal up through a refractory tube and into the bottom of a graphite mold cavity by means of compressed air. The height to which the metal may be raised depends on the maximum air pressure applied. The rate of pouring is controlled simultaneously by the inside diameter of the pouring tube and the rate at which air pressure increases. Advantages of pressure pouring include the ability to pump clean, slag-free metal from the bottom of the ladle, elimination of the erosion of bottom-pour stoppers and nozzles required on bottom-pour ladles, and elimination of the mold erosion that often occurs when metal falls on or against the mold surface at the start of a gravity pour. ACKNOWLEDGMENTS Portions of this article were adapted from cited references and the article “Steel Castings”

in Casting Design Handbook, American Society for Metals, 1962.

REFERENCES 1. Modern Casting, 2005 census 2. C. Briggs, The Metallurgy of Cast Steel Castings, McGraw Hill, 1946 3. C. Briggs, Fluidity of Metals, Metals Handbook, American Society for Metals, 1948 4. C. Briggs, Steel Castings Handbook, Engineer’s Edition (100th Anniversary of the Steel Casting Industry in the USA), Steel Founders’ Society of America, 1960 5. Steel Castings Handbook, 6th ed., Steel Founders’ Society of America and ASM International, 1995 6. Metall. Metalloved, No. 11, 1962, p 36–53 7. Feeding and Risering Guidelines for Steel Castings, Steel Founders’ Society of America, 2001

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 949-974 DOI: 10.1361/asmhba0005329

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Steel Castings Properties Revised by David Poweleit, Steel Founders’ Society of America

ALL STEEL is cast. Traditional integrated mills melted ore in blast furnaces, converting blast furnace iron to steel and cast steel ingots. The ingots are rolled into plate or bar and then finally rolled into the desired structural shape. Minimills melt steel scrap in electric furnaces and then continuously cast the steel into bars that are directly rolled into the structural shape. Like minimills, foundries melt steel scrap in electric furnaces but cast the steel directly to shape in molds. Many believe that cast steel is brittle because they associate it with cast iron commonly used in automotive and household goods, such as cookware, that easily cracks. However, the properties of steel are very different from iron. Steel castings can meet or exceed the ductility, toughness, or weldability of rolled steels. Designers generally think of design requirements in terms of strength, but the design is commonly constrained by modulus of elasticity, fatigue, toughness, or ductility. Increasing the strength of steel normally reduces the ductility, toughness, and weldability. It is often more cost-effective in steel casting design to use more of a lower-strength grade and increase the section size or modify the shape than to overdesign a part using a more expensive material. The design freedom makes castings an attractive way to obtain the best fabrication and material performance and the needed component stiffness and strength. Rolled sections of steel have their microstructure elongated in the direction of rolling. The strength and ductility are improved in that direction, but they are reduced across the rolling direction. The lack of a rolling direction in steel castings gives them uniform properties in all directions. Rolling steel cold can also strengthen the steel but reduces ductility and toughness. Similarly, forgings have higher mechanical properties based on the grain flow from the forging process. However, properties across the grain are degraded similar to using lumber (Fig. 1). Cast steel grades achieve their properties by alloying and heat treatment. These properties are isotropic, or the same in any direction of applied load. The biggest advantage in quality that forged or rolled shapes have over steel castings is their ability to begin with a simple optimal casting.

The ingot or bar can be easily inspected prior to rolling or forging. The use of casting processes to make uniquely designed shapes requires inspection that is correlated to the casting process, part design, and performance requirements. Often, the purchaser of steel castings uses nondestructive examination, mechanical testing, and engineering analysis to ensure the desired reliability. There are two major categories of cast steel: carbon and low-alloy, and high alloy. Carbon and low-alloy steel castings are similar to many of the standard wrought steels, such as 1020, 8630, 4140, and 4340, and provide the same great mechanical and physical properties that the wrought grades recommend. High-alloy steel castings are used for steel mechanical properties along with other attributes for severe service. High-alloy steels are broken down into three subcategories: wear resistant, corrosion resistant, and heat resistant.

Fig. 1

Carbon Steels Cast carbon steels are used when the strength and endurance of a steel is required and special properties, such as resisting heat, corrosion, and wear, are not. Carbon steels contain only carbon as the principal alloying element. Other elements are present in small quantities, including those added for deoxidation. Silicon and manganese in cast carbon steels typically range from 0.25 to approximately 0.80% Si and 0.50 to approximately 1.00% Mn. Carbon steels can be classified according to their carbon content into three broad groups:
Low-carbon steels: 0.20% C
Medium-carbon steels: 0.20 to 0.50% C
High-carbon steels: 0.50% C

Low-carbon cast steels are those that have a carbon content less than 0.20%. The elements

Influence of forging reduction on anisotropy for a 0.35% C wrought steel. Properties for a 0.35% C cast steel are shown in the graph by a star (*) for purposes of comparison.

950 / Steel Castings present in low-carbon cast steel, other than carbon, are manganese (0.50 to 1.00%), silicon (0.25 to 0.80%), and sulfur and phosphorus (up to 0.05% maximum each). Residual elements such as nickel, chromium, copper, and molybdenum will be present in small amounts. Low-carbon steel castings can be case hardened, a process by which the castings are given a hard, wear-resistant exterior and a tough, ductile core. Cast carbon steels containing less than 0.10% C are mainly produced for electrical and magnetic equipment and are normally given a full anneal heat treatment. Medium-carbon cast steels have carbon contents of 0.20 to 0.50%. This class of steels represents the bulk of steel casting production. A very large proportion of cast steel in the medium-carbon class is heat treated by normalizing, which consists of cooling the castings in air from approximately 55  C (100  F) above the upper critical temperature. A stress-relief treatment can be used to relieve stresses set up in the casting by cooling conditions or welding operations and to soften the heat-affected zone resulting from welding. These parts are used in a wide variety of ways, including applications in the railroad and other transportation industries, machinery and tools, equipment for rolling mills, mining and construction equipment, and many other miscellaneous applications. High-carbon cast steels contain more than 0.50% C. Other elements present in these grades are manganese (0.50 to 1.0%), silicon (0.30 to 0.80%), and phosphorus and sulfur (0.05% maximum each). Because of their high carbon contents, these grades are the most hardenable of the plain carbon cast steels. They are therefore used in applications that require relatively high strength levels.

Low-Alloy Steels Cast low-alloy steels are used when the strength and endurance of a steel is required. Low-alloy steels contain alloying elements, in addition to carbon, up to a total alloy content of 8%. Cast steels containing more than the following amounts of at least one of these alloying elements are considered low-alloy cast steels: Element

Manganese Silicon Nickel Copper Chromium Molybdenum Vanadium Tungsten

Amount, %

1.00 0.80 0.50 0.50 0.25 0.10 0.05 0.05

The compositions of low-alloy cast steels are characterized by carbon contents primarily under 0.45% and by small amounts of alloying elements, which are added to produce certain definite properties. Low-alloy steels are applied

when strength requirements are higher than those obtainable with carbon steels. Low-alloy steels also have better toughness and hardenability than carbon steels. Carbon-Manganese Cast Steels. Cast steels containing 1.00 to 1.75% Mn and 0.20 to 0.50% C have received considerable attention from engineers in the past because of the excellent properties that can be developed with a single, relatively inexpensive alloying element and by a single normalizing or normalizing-andtempering heat treatment. Carbon-manganese steels are also referred to as medium-manganese steels and are represented by the cast 1300 series of steels (1.60 to 1.90% Mn). Manganese-molybdenum cast steels are very similar to the medium-manganese steels, with the added characteristics of high yield strength at elevated temperatures, higher ratio of yield strength to tensile strength at room temperature, greater freedom from temper embrittlement, and greater hardenability. Therefore, these steels have replaced mediummanganese steel for certain applications. There are two general grades of manganese-molybdenum cast steels:
8000 series (1.0 to 1.35% Mn, 0.10 to

0.30% Mo)


8400 series (1.35 to 1.75% Mn, 0.25 to

0.55% Mo) For both of these alloy types, the carbon content is frequently selected between 0.20 and 0.35%, depending on the heat treatment employed and the strength characteristics desired. Manganese-Nickel-Chromium-Molybdenum Cast Steels. The cast 9500-series lowalloy steels are primarily produced for their high hardenability. Sections exceeding 127 mm (5 in.) in thickness can be quenched and tempered to obtain a fully tempered martensitic structure. The composition range employed for the 9500 series is: Element

Manganese Nickel Chromium Molybdenum

Composition, %

1.30–1.60 0.40–0.70 0.55–0.75 0.30–0.40

Nickel Cast Steels. Among the oldest cast steel alloys are those containing nickel. These steels are characterized by high tensile strength and elastic limit, good ductility, and excellent resistance to impact. The cast steels of the 2300 series contain 2.0 to 4.0% Ni, depending on the grade required. Nickel-Chromium-Molybdenum Cast Steels. The addition of molybdenum to nickelchromium steel significantly improves hardenability and makes the steel relatively immune to temper embrittlement. Nickel-chromiummolybdenum cast steel is particularly well suited to the production of large castings because of its deep-hardening properties.

In addition, the ability of these steels to retain strength at elevated temperatures extends their usefulness in many industrial applications. Chromium-Molybdenum Cast Steels. Chromium contents of approximately 1.00% or more provide a nominal improvement in elevated-temperature properties. The chromium cast steels (5100 series; 0.70 to 1.10% Cr) are not in common use in the steel casting industry. However, the chromium-molybdenum lowalloy cast steels are widely used. Copper-Bearing Cast Steels. There are several types of copper-containing steels. Selection among these various types is primarily based on either their atmospheric corrosion resistance (weathering steels) or the age-hardening characteristics that copper adds to steel. High-strength cast steels cover the tensile strength range of 1207 to 2068 MPa (175 to 300 ksi). Cast steels with these strength levels and with considerable toughness and weldability were originally developed for ordnance applications. These cast steels can be produced from any of the previously mentioned mediumalloy compositions by heat treating with liquidquenching techniques and low tempering temperatures. Cast 4300-series steels or modifications thereof are usually employed.

Wear-Resistant Steels Cast wear-resistant steels are used in wear applications such as ground-engaging equipment or scrap yard shredders. Wear-resistant alloys can be either manganese steels or heat treated low-alloy steels. The mechanism causing the wear can be categorized as frictional, abrasive, corrosive, or deformation. Austenitic manganese steels are used in many wear environments. The compositions of the austenitic manganese steels (Table 1) can be varied to achieve differing combinations of strength, ductility, wear resistance, and machinability (see ASTM A 128IA 128M for specified chemical requirements). The chromium-alloyed manganese steels are the most extensively used, following the standard ASTM A 128 grade. Medium-thickness (50 to 125 mm, or 2 to 5 in.) crusher castings generally exhibit increased wear life, which is

Table 1 Compositions of austenitic manganese steels Composition, % Grade

Standard Chromium 1% Mo 1% Mo (lean) 2% Mo High yield strength Machinable

C

Mn

Cr

Mo

1.0–1.4 12.0–14.0 . . . ... 1.0–1.4 12.0–14.0 1.5–2.5 . . . 0.8–1.3 12.0–15.0 . . . 0.8–1.2 1.1–1.4 5.0–7.0 . . . 0.8–1.2 1.0–1.5 12.0–15.0 0.4–0.7 12.0–15.0

... ...

0.3–0.6 18.0–20.0

...

Other

... ... ... ...

1.8–2.2 ... 1.8–2.2 2.0–4.0Ni, 0.5–1.0V . . . 2.0–4.0Ni, 0.2–0.4Bi

Steel Castings Properties / 951 attributable to the chromium alloying. In some installations, however, it is difficult to substantiate the degree of wear improvement, and lowered ductility can lead to premature breakage in severe service. Figure 2 shows that the tensile elongation in 150 mm (6 in.) thick chromiummanganese steels is 30 to 40% lower than that of the standard grades. Impact property comparisons (Fig. 3) also show a decided impairment. The molybdenum-alloyed manganese steels can be divided into two groups: those with 1% Mo and those with 2% Mo. The 1% Mo category is further divided into a normal alloy segment and a so-called lean alloy category. The lean alloy austenitic grade has manganese contents from 5 to 7%; this contrasts with the 12 to 14% level usually associated with austenitic manganese steels. A normal 1% Mo-alloyed grade can involve carbon levels as low as 0.80% and as high as 1.30%. Each of the extremes of the carbon range provides a distinct benefit. Heavy-section mechanical property improvement and improved resistance to mechanical property degradation as a result of elevated temperatures can lead to the use of lower-carbon materials. The higher-carbon grades can be used where improved abrasive wear resistance is needed. The lean grade of the 1% Mo steels finds limited use in crushing applications, in which lower ductility and lower impact levels are permissible. The more rapid work-hardening tendency of the material has been claimed to account for improved wear life in certain applications, such as ball mills and rod mills. The 2% Mo grades follow the more classical lines in that manganese contents tend to fall within the 12 to 15% range. Carbon content can vary from 1 to 1.5% in alloys that may

Fig. 2

Typical tensile elongations of austenitic manganese steels

either undergo a standard austenitizing heat treatment of a special two-step dispersion-hardening treatment. Before the availability of the multiple-alloyed, high-yield-strength manganese steels, the 2% Mo grades were used in the dispersion-hardened condition, in which they develop high yield strengths. The desirability of high yield strength is evident in such casting applications as large jaw plates and concave segments; this service results in a high degree of kneading of the working surface, which in turn develops extensive surface expansion. The cumulative effect of this growth can be sufficient to raise the casting from its seat or backing plate. In extreme cases, the wear castings can be physically displaced or can crack restraining structures. The 2% Mo grade can often provide a discernible improvement in overall wear life if premature failure is not encountered. However, economic considerations become involved to an extent that judgment must be exercised regarding the added cost in relation to additional wear life. The machinable grade of manganese steel listed in Table 1 plays a somewhat parallel role but in a different property area. Austenitic manganese steels exhibit poor machinability, certainly with regard to the drilling and tapping of small holes. The machinable grade, however, exhibits drilling and tapping advantages at a level somewhat higher than that of a wrought type 308 stainless steel.

Corrosion-Resistant Steels Cast corrosion-resistant steels, more commonly called cast stainless steels, are alloyed to prevent corrosion from aqueous solutions or liquid vapor normally below 315  C (600  F). They are iron alloyed with chromium, nickel, and carbon. The principal applications for these

Fig. 3

Typical Izod impact energies of austenitic manganese steels. See Fig. 2 for key.

steels are as materials of construction for chemical-processing and power-generating equipment involving corrosion service. An appropriate definition for cast stainless steels is the familiar one based on the discovery in 1910 that a minimum of 12% Cr will impart resistance to corrosion and oxidation to steel. Cast stainless steels are defined as ferrous alloys that contain a minimum of 12% Cr for corrosion resistance. Most cast stainless steels are, of course, considerably more complex compositionally than this simple definition implies. Stainless steels typically contain one or more alloying elements in addition to chromium (for example, nickel, molybdenum, copper, niobium, and nitrogen) to produce a specific microstructure, corrosion resistance, or mechanical properties for particular service requirements. Corrosion-resistant cast steels are usually classified on the basis of composition or microstructure. It should be recognized that these bases for classification are not completely independent in most cases; that is, classification by composition also often involves microstructural distinctions. Table 2 lists the compositions of the commercial cast corrosion-resistant alloys. Alloys are grouped as chromium steels, chromiumnickel steels in which chromium is the predominant alloying element, and nickel-chromium steels in which nickel is the predominant alloying element. The serviceability of cast corrosion-resistant steels depends greatly on the absence of carbon, and especially precipitated carbides, in the alloy microstructure. Therefore, cast corrosion-resistant alloys are generally low in carbon (typically <0.08% C). As shown in Table 2, high-alloy cast steels are classified on the basis of microstructure. Structures may be austenitic, ferritic, martensitic, or duplex (Ferrite and austenite); the structure of a particular grade is primarily determined by composition. Chromium, nickel, and carbon contents are particularly important in this regard. In general, straight chromium grades of high-alloy cast steel are either martensitic or ferritic, the chromium-nickel grades are either duplex or austenitic, and the nickelchromium steels are fully austenitic. Martensitic grades include alloys CA15, CA40, CA15M, and CA6NM. The CA15 alloy contains the minimum amount of chromium necessary to make it essentially rustproof. It has good resistance to atmospheric corrosion as well as to many organic media in relatively mild service. A higher-carbon modification of CA15, CA40, can be heat treated to higher strength and hardness levels. Alloy CA15M is a molybdenum-containing modification of CA15 that provides improved elevated-temperature strength. Alloy CA6NM is an Fe-Cr-NiMo alloy similar to CA15M, with higher nickel and molybdenum that improves general carrosion resistance. It was developed for improved castability, especially in heavy sections. Alloy CA6NM-B is a lower-carbon vesion of this alloy.

952 / Steel Castings Table 2 Compositions and microstructures of corrosion-resistant high-alloy cast steels Alloy

Most common end-use microstructure

Wrought alloy type(a)

Composition(b), % Cr

Ni

Mo

Si

Mn

P

S

C

0.50 0.15–1.0 0.5 ... ...

1.50 0.65 1.50 1.50 1.50

1.00 1.00 1.00 1.00 1.00

0.04 0.04 0.04 0.04 0.04

0.04 0.04 0.04 0.04 0.04

0.15 0.15 0.20–0.40 0.30 0.50

Others

Chromium steels CA15 410 CA15M ... CA40 420 CB30 431,442 CC50 446

Martensite Martensite Martensite Ferrite + carbides Ferrite + carbides

11.5–14.0 11.5–14.0 11.5–14.0 18.0–21.0 26.0–30.0

Chromium-nickel CA6NM CB7Cu CD4MCu CE30 CF3 CF8 CF20 CF3M CF8M CF12M CF8C CF16F CG8M CH20 CK20

Martensite Martensite—age hardenable Austenite in ferrite—age hardenable Ferrite in austenite Ferrite in austenite Ferrite in austenite Austenite Ferrite in austenite Ferrite in austenite Ferrite in austenite or austenite Ferrite in austenite Austenite Ferrite in austenite Austenite Austenite

11.5–14.0 15.5–17.0 25.0–26.5 26.0–30.0 17.0–21.0 18.0–21.0 18.0–21.0 17.0–21.0 17.0–21.0 18.0–21.0 18.0–21.0 18.0–21.0 18.0–21.0 22.0–26.0 23.0–27.0

3.5–4.5 3.6–4.6 4.75–6.0 8.0–11.0 8.0–12.0 8.0–11.0 8.0–11.0 9.0–13.0 9.0–13.0 9.0–12.0 9.0–12.0 9.0–12.0 9.0–13.0 12.0–15.0 19.0–22.0

0.40–1.0 ... 1.75–2.25 ... ... ... ... 2.0–3.0 2.0–3.0 2.0–3.0 ... 1.50 3.0–4.0 ... ...

1.00 1.50 1.00 2.00 2.00 2.00 2.00 1.50 1.50 2.00 2.00 2.00 1.50 2.00 1.75

1.00 1.00 1.00 1.50 1.50 1.50 1.50 1.50 1.50 1.50 1.50 1.50 1.50 1.50 1.50

0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.17 0.04 0.04 0.04

0.03 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04

0.06 0.07 0.04 0.30 0.03 0.08 0.20 0.03 0.03 0.12 0.08 0.16 0.08 0.20 0.20

... 2.3–3.3 Cu 2.75–3.25 Cu ... ... ... ... ... ... ... Nb = 8  C, 1.0 max 0.20–0.35 Se ... ... ...

Austenite

19.0–22.0

27.5–30.5

2.0–3.0

1.50

1.50

0.04

0.04

0.07

3.0–4.0 Cu

steels ... 17-4PH ... ... 304L 304 302 316L 316 ... 347 303 317 309 310

Nickel-chromium steel CN7M 320

1.00 1.00 1.00 2.00 4.00

... ... ... ... ...

(a) Wrought alloy type numbers are AISI designations for grades most closely corresponding to casting alloys. Wrought alloy type numbers are given only as a guide for determining corresponding cast and wrought grades. Buyers should use cast alloy designations when specifying castings. (b) Maximum unless range is given. All compositions contain balance of iron.

Table 3 Nominal compositions of first- and second-generation duplex stainless steels UNS designation

Common name

First-generation steels S31500 3RE60 S32404 Uranus 50 S32900 Type 329 J93370 CD4MCu

Composition(a), % Cr

Ni

Mo

Cu

N

18.5 21 26 25

4.7 7.0 4.5 5

2.7 2.5 1.5 2

... 1.5 ... 3

... ... ... ...

1.7Si ... ... ...

6 7 5 6

1.7 3 3 3

... 0.5 ... 2

0.15 0.15 0.15 0.20

... 0.3W ... ...

Second-generation steels S31200 44LN 25 S31260 DP-3 25 S31803 Alloy 2205 22 S32550 Ferralium 25 255 S32950 7-Mo 26.5 PLUS J93404 Atlas 958, 25 COR 25

Others

4.8 1.5 . . . 0.20 . . . 7

4.5 . . . 0.25 . . .

(a) All compositions contain balance of iron.

Austenitic grades include CH20, CK20, and CN7M. The CH20 and CK20 alloys are highchromium, high-carbon, wholly austenitic compositions in which the chromium exceeds the nickel content. The more highly alloyed CN7M has excellent corrosion resistance in many environments and is often used in sulfuric acid service. In the 1980s and 1990s, high-performance, more highly alloyed grades of austenitic (as well as ferritic and duplex) stainless steel were commercialized. These grades have improved corrosion resistance. The 6-Mo grades were introduced to resist localized crevice corrosion and pitting in stagnant marine environments. They are also suited for flue gas desulfurization service. Ferritic grades are designated CB30 and CC50. Alloy CB30 is practically nonhardenable

by heat treatment. As this alloy is normally made, the balance among the elements in the composition results in a wholly ferritic structure similar to wrought AISI type 442 stainless steel. Alloy CC50 has substantially more chromium than CB30 and has relatively high resistance to localized corrosion in many environments. Austenitic-ferritic alloys include CE30, CF3, CF3A, CF8, CF8A, CF20, CF3M, CF3MA, CF8M, CF8C, CF16F, and CG8M. The microstructures of these alloys usually contain 5 to 40% ferrite, depending on the particular grade and the balance among the ferritepromoting and austenite-promoting elements in the chemical composition. Duplex alloys for corrosion resistance are specified by ASTM A 890/A 890M. Grade 1A, Alloy Casting Institute alloy CD4MCu, is the base alloy. Grade 1B, CD4MCuN, is grade 1A with approximately 0.15% N added. With high levels of ferrite (approximately 40 to 50%) and low nickel, the duplex alloys have better resistance to stress-corrosion cracking (SCC) than CF3M. Alloy CD4MCu, which does not contain a specified amount of nitrogen and has relatively low molybdenum content, has only slightly better resistance to localized corrosion than CF3M. Alloy CD4MCuN exhibits better localized corrosion resistance than either CF3M or CD4MCu. The need for improvement in the weldability and corrosion resistance of these alloys resulted in the second-generation alloys, which are characterized by the addition of nitrogen as an alloying element. Improvements in stainless steel production practices (for example, electron beam refining, vacuum and argon-oxygen decarburization, and vacuum induction melting) have created a second generation of duplex stainless steels.

These steels offer excellent resistance to pitting and crevice corrosion, significantly better resistance to chloride SCC than the austenitic stainless steels, good toughness, and yield strengths two to three times higher than those of type 304 or 316 stainless steels. Second-generation duplex stainless steels are usually approximately a 50-50 blend of ferrite and austenite. The new duplex alloys combine the near immunity to chloride SCC of the ferritic grades with the toughness and ease of fabrication of the austenitics. Among the secondgeneration duplexes, alloy 2205 (CD3MN, UNS J92205) seems to have become the general-purpose stainless. Table 3 lists the nominal compositions of first- and second-generation duplex alloys. Precipitation-Hardening Grades. The alloys in this group are CB7Cu and CD4MCu. Alloy CB7Cu is a low-carbon martensitic alloy that may contain minor amounts of retained austenite or ferrite. The copper precipitates in the martensite when the alloy is heat treated to the hardened (aged) condition.

Heat-Resistant Steels Cast heat-resistant steels are alloyed to prevent creep or other failure due to elevated operating temperatures or cyclic heating, from 650 to 1315  C (1200 to 2400  F). Strength at these elevated temperatures is only one of the criteria by which these materials are selected, because applications often involve aggressive environments to which the steel must be resistant. The atmospheres most commonly encountered are air, flue gases, or process gases; such atmospheres may be either oxidizing or

Steel Castings Properties / 953 reducing and may be sulfidizing or carburizing if sulfur or carbon are present. Only heat-resistant steels exhibit the required mechanical properties and corrosion resistance over long periods of time without excessive or unpredictable degradation. In addition to long-term strength and corrosion resistance, some cast heat-resistant steels exhibit special resistance to the effects of cyclic temperatures and changes in the nature of the operating environment. A number of cast high-alloy grades have been developed and successfully used for a variety of service requirements. There are three principal categories, based on composition:
Iron-chromium alloys
Iron-chromium-nickel alloys
Iron-nickel-chromium alloys

Common Alloys

These alloy types resemble high-alloy corrosion-resistant steels except for their higher carbon contents, which impart greater strength at elevated temperature. The higher carbon content and, to a lesser extent, alloy composition ranges distinguish cast heat-resistant steel grades from their wrought counterparts. Table 4 summarizes the compositions of standard cast heat-resistant grades. Iron-chromium alloys contain 8 to 30% Cr and little or no nickel. They are ferritic in structure and exhibit low ductility at ambient temperatures. Iron-chromium alloys are primarily used where resistance to gaseous corrosion is the predominant consideration, because they possess relatively low strength at elevated temperatures. Examples of such alloys are the cast HA, HC, and HD grades listed in Table 4. Iron-chromium-nickel alloys contain more than 18% Cr and more than 8% Ni, with the chromium content always exceeding that of nickel. They exhibit an austenitic matrix, although several grades also contain some ferrite. These alloys exhibit greater strength and ductility at elevated temperatures than those in the iron-chromium group and withstand moderate thermal cycling. Examples of these alloys Table 4

are the HE, HF, HH, HI, HK, and HL grades listed in Table 4. Iron-nickel-chromium alloys contain more than 10% Cr and more than 23% Ni, with the nickel content always exceeding that of chromium. These alloys are wholly austenitic and exhibit high strength at elevated temperatures. They can withstand considerable temperature cycling and severe thermal gradients and are well suited to many reducing, as well as oxidizing, environments. Examples of Fe-Ni-Cr alloys are the HN, HP, HT, HU, HW, and HX grades listed in Table 4. Even though nickel is the major element in the HW and HX grades, these grades are ordinarily referred to as highalloy steels rather than nickel-base alloys.

WCB, similar to 1020 steel, is a medium-carbon steel suited to sand and investment casting, combining good machinability, workability, and weldability. WCB is normally specified for low-cost parts, machinery components and case-hardened parts, forming and spinning tools, chain/sprocket assemblies, and pipe and valves. There are ASTM International specifications (A 915, A 958) for steel casting alloys similar to wrought grades. SC4140 or SC4340, similar to 4140 or 4340 steel, has high hardenability, good fatigue, and abrasion and impact resistance with outstanding tensile strength when fully hardened. SC4140 or SC4340 is commonly used for machine tool parts, flanges, fittings, gears, and applications requiring high tensile strength and impact resistance. SC8620 is more easily produced by investment casting than WCB. SC8630, similar to 8630 steel, has excellent castability, good fatigue and impact resistance, high tensile strength, and is readily machined. SC8630 is used in handling equipment and power tool parts, cams, and chuck jaws and is the most widely used low-alloy cast steel. CF8M, similar to 316 stainless steel, is the most widely used corrosion-resistant alloy

Compositions of cast heat-resistant high-alloy steels Composition(b), %

Cast alloy designation

Wrought alloy type(a)

C

Mn

Si

P

S

Cr

Ni

Mo

HA HC HD HE

... 446 327 ...

0.20 0.50 0.50 0.20–0.50

0.35–0.65 1.00 1.50 2.00

1.00 2.00 2.00 2.00

0.04 0.04 0.04 0.04

0.04 0.04 0.04 0.04

8–10 26–30 26–30 26–30

... 4 4–7 8–11

0.90–1.20 0.5(c) 0.5(c) 0.5(c)

HF HH HI HK

302B 309 ... 310

0.20–0.40 0.20–0.50 0.20–0.50 0.20–0.60

2.00 2.00 2.00 2.00

2.00 2.00 2.00 2.00

0.04 0.04 0.04 0.04

0.04 0.04 0.04 0.04

18–23 24–28 26–30 24–28

8–12 11–14 14–18 18–22

0.5(c) 0.5(c) 0.5(c) 0.5(c)

HL HN HP HT HU

... ... ... 330 ...

0.20–0.60 0.20–0.50 0.35–0.75 0.35–0.75 0.35–0.75

2.00 2.00 2.00 2.00 2.00

2.00 2.00 2.50 2.50 2.50

0.04 0.04 0.04 0.04 0.04

0.04 0.04 0.04 0.04 0.04

28–32 19–23 24–28 15–19 17–21

18–22 23–27 33–37 33–37 37–41

0.5(c) 0.5(c) 0.5(c) 0.5(c) 0.5(c)

HW HX

... ...

0.35–0.75 0.35–0.75

2.00 2.00

2.50 2.50

0.04 0.04

0.04 0.04

10–14 15–19

58–62 64–68

0.5(c) 0.5(c)

(a) Wrought alloy type numbers are listed only as a guide for determining equivalent cast and wrought grades. Buyers should use cast alloy designations when specifying castings. (b) Maximum unless range is given. All compositions contain balance of iron. (c) Molybdenum not intentionally added

and is particularly useful in handling oxidizing solutions; it is resistant to most organic acids and waters. CF8M is used for chemical, paper and pulp, dairy, dye, textile, pharmaceutical, and food processes, valves, pumps, and boat fittings. Note that cast duplex steels can offer improved strength with better corrosion resistance (especially to localized corrosion) than that of the CF grades. CA15, similar to a 410 stainless steel, has an excellent resistance to atmospheric corrosion; it resists abrasive chemicals, food products oxidizing acids, pulp, sodium nitrate, and steam. CA15 is used for chemical- and food-processing equipment, paper and pulp equipment, valves, impellers, bushings, and pumps. CB7Cu-1, similar to 17-4 Ph steel, has good wear and corrosion resistance and is superior to standard chromium grades, with excellent strength and ductility. CB7Cu-1 and CB7Cu-2 (similar to 15-5) are used for gears, cams, pinions, valve plugs, seats, stems, pump and propeller shafts, hardware, aircraft parts, instruments, and a multitude of military components. These precipitation-hardening grades are suitable up to 315  C (600  F) in accordance with ASTM A 747. Tables 5 and 6 provide some specification requirements for selected steel alloys.

Carbon and Low-Alloy Mechanical Properties Carbon and low-alloy steel castings are produced to a great variety of properties, because composition and heat treatment can be selected to achieve specific combinations of properties, including hardness, strength, ductility, fatigue resistance, and toughness. Although selections can be made from a wide range of properties, it is important to recognize the interrelationships among these properties. For example, higher hardness, lower toughness, and lower ductility values are associated with higher strength values. Property trends among carbon steels are illustrated as a function of the carbon content in Fig. 4. Hardenability is the property of steel that governs the depth to which hardening occurs in a section during quenching. It should not be confused with hardness, which is the resistance to penetration as measured by Rockwell, Brinell, or other hardness tests. Hardenability is of considerable importance because it relates directly to the strength of steel and to many other mechanical properties, notably, toughness and fatigue properties. The principal method of hardening carbon and low-alloy steels consists of quenching the steel from the austenitizing temperature. Steels vary in their response to this quenching operation because the depth below the surface of a part to which the part hardens will depend on the composition of the steel and the severity of the quench. Because cooling during a quench is fastest at the surface of a part and slowest at its center, and because the hardening reaction is

954 / Steel Castings Table 5 Summary of specification requirements for various carbon steel castings Unless otherwise noted, all the grades listed in this table are restricted to a phosphorus content of 0.040% max and a sulfur content of 0.045% max. Tensile strength(a)

Yield strength(a)

MPa

MPa

Chemical composition(b), %

Minimum elongation in 50 mm (2 in.), %

Minimum reduction in area, %

... ... 22 24 24 22 22

... ... 30 35 35 30 30

0.25(c) 0.35(c) 0.25(c) 0.30(c) 0.30(c) 0.35(c) 0.25(c)

ASTM A 148: carbon steel castings for structural applications(d) 80–40 (D50400) 550 80 275 40 18

30

80–50

(D50500)

550

80

345

50

22

90–60

(D50600)

620

90

415

60

105–85

(D50850)

725

105

585

85

Class or grade

(UNS) No.

ksi

ksi

ASTM A 27: carbon steel castings for general applications N-1 (J02500) ... ... ... ... N-2 (J03500) ... ... ... ... U60-30 (J02500) 415 60 205 30 60–30 (J03000) 415 60 205 30 65–35 (J03001) 450 65 240 35 70–36 (J03501) 485 70 250 36 70–40 (J02501) 485 70 275 40

Si

Other requirements

0.75(c) 0.60(c) 0.75(c) 0.60(c) 0.70(c) 0.70(c) 1.20(c)

0.80 0.80 0.80 0.80 0.80 0.80 0.80

0.06% S, 0.05% P 0.06% S, 0.05% P 0.06% S, 0.05% P 0.06% S, 0.05% P 0.06% S, 0.05% P 0.06% S, 0.05% P 0.06% S, 0.05% P

(e)

(e)

(e)

35

(e)

(e)

(e)

20

40

(e)

(e)

(e)

17

35

(e)

(e)

(e)

0.06% S, 0.05% P Composition and heat treatment necessary to achieve specified mechanical properties 0.06% S, 0.05% P Composition and heat treatment necessary to achieve specified mechanical properties 0.06% S, 0.05% P Composition and heat treatment necessary to achieve specified mechanical properties 0.06% S, 0.05% P Composition and heat treatment necessary to achieve specified mechanical properties

SAE J435 OCT2002: see Table 6 for allay steel 0000 ... ... ... 415 ... 415 60 450 ... 450 65 585 ... 585 85 690 ... 690 100 550 ... 550 80 620 ... 620 90

castings specified ... ... 207 30 241 35 310 45 485 70 345 50 415 60

ASTM A 216: carbon steel castings suitable for WCA (J02502) 415–585 60–85 WCB (J03002) 485–655 70–95 WCC (J02503) 485–655 70–95

fusion welding and high-temperature service 205 30 24 35 0.25 250 36 22 35 0.30 275 40 22 35 0.25

Other ASTM cast steel specifications with carbon steel grades(g) A 352-LCA (J02504) 415–585 60–85 205 30 A 352-LCB (J03003) 450–620 65–90 240 35 A 356-grade 1 ... 485 70 250 36 A 757-AIQ

...

450

65

240

35

in SAE J435 ... 22 24 16 10 22 20

C

... 30 35 24 15 35 40

0.12–0.22 0.25(c) 0.30(c) 0.40–0.50 0.40–0.50 ... ...

Mn

0.50–0.90 0.75(c) 0.70(c) 0.50–0.90 0.50–0.90 ... ...

0.60 0.80 0.80 0.80 0.80 ... ...

0.70(c) 1.00(c) 1.20(c)

0.60 0.60 0.50

187 HB max 187 HB max 131–187 HB 170–229 HB 207–255 HB 163–207 HB 187–241 HB (f) (f) (f)

24 24 20

35 35 35

0.25 0.30 0.35

0.70(c) 1.00 0.70(c)

0.60 (f)(h)(i) 0.60 (f)(i)(j) 0.60 0.035% P max, 0.030% S max

24

35

0.30

1.00

0.60

(i)(j)(k)

Condition or specific application

Chemical analysis only Heat treated but not mechanically tested Mechanically tested but not heat treated Heat treated and mechanically tested Heat treated and mechanically tested Heat treated and mechanically tested Heat treated and mechanically tested

Low-carbon steel suitable for carburizing Carbon steel welding grade Carbon steel welding grade Carbon steel medium-strength grade Carbon steel medium-strength grade Medium-strength low-alloy steel Medium-strength low-alloy steel Pressure-containing parts Pressure-containing parts Pressure-containing parts Low-temperature applications Low-temperature applications Castings for valve chests, throttle valves, and other heavy-walled components for steam turbines Casting for pressure-containing applications at low temperatures

(a) Where a single value is shown, it is a minimum. (b) Where a single value is shown, it is a maximum. (c) For each reduction of 0.01% C below the maximum specified, an increase of 0.04% Mn above the maximum specified is permitted up to the maximums given in the applicable ASTM specifications. (d) Grades may also include low-alloy steels: see Table 6 for the stronger grades of ASTM A 148. (e) Unless specified by purchaser, the compositions of cast steels in ASTM A 148 are selected by the producer in order to achieve the specified mechanical properties. (f) Specified residual elements include 0.30% Cu max, 0.50% Ni max, 0.50% Cr max, 0.20% Mo max, and 0.03% V max, with the total residual elements not exceeding 1.00%. (g) These ASTM specifications also include alloy steel castings for the general type of applications listed in the Table. (h) Testing temperature of 32  C (25  F). (i) Charpy V-notch impact testing at the specified test temperature with an energy value of 18 J (13 ft  lbf) min for two specimens and an average of three. (j) Testing temperature of 46  C (50  F). (k) Specified residual elements of 0.03% V, 0.50% Cu, 0.50% Ni, 0.40% Cr, and 0.25% Mo, with total amount not exceeding 1.00%. Sulfur and phosphorus content, each 0.025% max

time-dependent, hardenability is a vital consideration in alloy selection. Figure 5 shows schematically the cooling curves for the center and surface of a hypothetical steel section superimposed on the continuous cooling transformation diagram. The curve labeled “50% martensite” intersects the pearlite area briefly; thus, half of the mass is transformed to pearlite and half transforms later to martensite. At some point closer to the surface, the cooling rate is given by the curve labeled “Critical cooling rate.” This curve represents the cooling rate that is just fast enough to avoid transformation to any type of metallurgical structure except martensite. Strength and Hardness. Depending on alloy choice and heat treatment, ultimate tensile strength levels from 414 to 1724 MPa (60 to 250 ksi) can be achieved with cast carbon and low-alloy steels. Figure 6 illustrates the tensile strength and tensile ductility values that can be expected from normalized and quenchedand-tempered cast carbon steels having a range of Brinell hardness values. The hardness and

strength values are largely determined by carbon content and the heat treatment, Fig. 4(c) for carbon steels and Fig. 7 for low-alloy steels. The effect of tempering normalized carbon steel is shown in Fig. 7. Strength and Ductility. Ductility depends greatly on the strength, or hardness, of the cast steel (Fig. 6). Actual ductility requirements vary with the strength level and the specification to which a steel is ordered. Because yield strength is a primary design criterion for structural applications, the relationships shown in Fig. 4(a) and (b) are replotted in Fig. 8 to reveal the major trends for cast carbon steels and Fig. 9 for low-alloy steels. Quenched-and-tempered steels exhibit higher ductility values for a given yield strength level than normalized, normalized-and-tempered, or annealed steels. Strength and Toughness. Several test methods are available for evaluating the toughness of steels or the resistance to sudden or brittle fracture. These include the Charpy V-notch impact test, the drop-weight test, the dynamic

tear test, and specialized procedures to determine plane-strain fracture toughness. Charpy V-notch impact energy trends at room temperature (Fig. 8, 9) reveal the distinct effect of strength and heat treatment on toughness. Higher toughness is obtained when a steel is quenched and tempered, rather than normalized and tempered. The effect of heat treatment and testing temperature on Charpy V-notch toughness is further illustrated in Fig. 10 for a carbon steel. Quenching, followed by tempering, produces superior toughness, as indicated by the shift of the impact energy transition curve to lower temperatures. Nil Ductility Transition Temperatures (NDTT) ranging from 38  C (100  F) to as low as 90  C (130  F) have been recorded in tests on normalized-and-tempered cast carbon and low-alloy steels in the yield strength range of 207 to 655 MPa (30 to 95 ksi) (Fig. 11). Comparison of the data in Fig. 11 with those of Fig. 12 shows the superior toughness values at equal strength levels that low-alloy steels

Steel Castings Properties / 955 Table 6 Material class (UNS No.)

Summary of specification requirements for various alloy steel castings with chromium contents up to 10% Tensile strength(a)

Yield strength(a)

MPa

MPa

ksi

ksi

Minimum elongation in 50 mm (2 in.), %

ASTM A 148: steel castings for structural applications(c) 115–95 795 115 655 95 135–125 930 135 860 125 150–135 1035 150 930 135 160–145 1105 160 1000 145 165–150 1140 165 1035 150 165–150L 1140 165 1035 150 210–180 1450 210 1240 180 210–180L 1450 210 1240 180 260–210 1795 260 1450 210 260–210L 1795 260 1450 210 SAE J435OCT 2002: see Table 725 725 830 830 1035 1035 1205 1205

14 9 7 6 5 5 4 4 3 3

Composition(b), %

Minimum reduction in area, %

C

Mn

Si

Cr

Ni

Mo

Other

30 22 18 12 20 20 15 15 6 6

... ... ... ... ... ... ... ... ... ...

... ... ... ... ... ... ... ... ... ...

... ... ... ... ... ... ... ... ... ...

... ... ... ... ... ... ... ... ... ...

... ... ... ... ... ... ... ... ... ...

... ... ... ... ... ... ... ... ... ...

(d) (d) (d) (d) (e) (e) (e) (e) (e) (e)

... ... ... ...

... ... ... ...

... ... ... ...

... ... ... ...

... ... ... ...

... ... ... ...

(g) (g) (g) (g)

0.50–0.80 0.50–0.80 0.40–0.70 0.50–0.80 0.40–0.70 0.50–0.80 0.40–0.70 0.35–0.65

0.60 0.60 0.60 0.60 0.60 0.30–0.60 0.75 1.00

0.35(h) 0.50–0.80 0.50–0.90 1.00–1.50 2.00–2.75 1.00–1.75 4.00–6.50 8.00–10.00

0.50(h) 0.70–1.10 0.60–1.00 0.50(h) 0.50(h) 0.50(h) 0.50(h) 0.50(h)

0.45–0.65 0.45–0.65 0.90–1.20 0.45–0.65 0.9–1.20 0.45–0.65 0.45–0.65 0.90–1.20

(h)(i) (i)(j) (i)(j) (h)(i) (h)(i) (h)(k) (h)(i) (h)(i)

0.60 0.60

1.00–1.50 1.00–1.25

... ...

0.45–0.65 0.90–1.20

(g)(l) (g)(l)

5 for the carbon steel castings specified in SAE J435(f) 105 585 85 17 35 120 655 95 14 30 150 860 125 9 22 175 1000 145 6 12

ASTM A 217: alloy steel castings for pressure-containing parts and high-temperature service WC1 (J12524) 450–620 65–90 240 35 24 35 0.25 WC4 (J12082) 485–655 70–95 275 40 20 35 0.20 WC5 (J22000) 485–655 70–95 275 40 20 35 0.20 WC6 (J12072) 485–655 70–95 275 40 20 35 0.20 WC9 (J21890) 485–655 70–95 275 40 20 35 0.18 WC11 (J11972) 550–725 80–105 345 50 18 45 0.15–0.21 C5 (J42045) 620–795 90–115 415 60 18 35 0.20 C12 (J82090) 620–795 90–115 415 60 18 35 0.20

ASTM A 389: alloy steel castings suitable for fusion welding and pressure-containing parts at high temperatures C23 (J12080) 485 70 275 40 18 35 0.20 0.30–0.80 C24 (J12092) 550 80 345 50 15 35 0.20 0.30–0.80 ASTM A 487: alloy steel castings for pressure-containing parts at high temperatures 1A 585–760 85–110 380 55 22 40 2B 620–795 90–115 450 65 22 45 1C 620 90 450 65 22 45 2A 585–760 85–110 365 53 22 35 2B 620–795 90–115 450 65 22 40 2C 620 90 450 65 22 40 4A 620–795 90–115 415 60 20 40 4B 725–895 105–130 585 85 17 35 4C 620 90 415 60 20 40 4D 690 100 515 75 17 35 4E 795 115 655 95 15 35 6A 795 115 550 80 18 30 6B 825 120 655 95 15 35 7A(p) 795 115 690 100 15 30 8A 585–760 85–110 380 55 20 35 8B 725 105 585 85 17 30 8C 690 100 515 75 17 35 9A 620 90 415 60 18 35 9B 725 105 585 85 16 35 9C 620 90 415 60 18 35 9D 690 100 515 75 17 35 10A 690 100 485 70 18 35 10B 860 125 690 100 15 35 11A 485–655 70–95 275 40 20 35 11B 725–895 105–130 585 85 17 35 12A 485–655 70–95 275 40 20 35 12B 725–895 105–130 585 85 17 35 13A 620–795 90–115 415 60 18 35 13B 725–895 105–130 585 85 17 35 14A 825–1000 120–145 655 95 14 30 16A (t) 485–655 70–95 275 40 22 35

0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.38 0.38 0.20 0.20 0.20 0.20 0.33 0.33

1.00 1.00 1.00 1.10–1.40 1.10–1.40 1.10–1.40 1.00 1.00 1.00 1.00 1.00 1.30–1.70 1.30–1.70 0.60–1.00 0.50–0.90 0.50–0.90 0.50–0.90 0.60–1.00 0.60–1.00 Composition same 0.33 0.60–1.00 0.30 0.60–1.00 0.30 0.60–1.00 0.20 0.50–0.80 0.20 0.50–0.80 0.20 0.40–0.70 0.20 0.40–0.70 0.30 0.80–1.10 0.30 0.80–1.10 0.55 0.80–1.10 0.12(u) 2.10(u)

0.80 0.35(m) 0.50(m) 0.25(m)(n) 0.5 Cu(g)(m) 0.80 0.35(m) 0.50(m) 0.25(m)(n) 0.5 Cu(g)(m) 0.80 0.35(m) 0.50(m) 0.25(m)(n) 0.5 Cu(g)(m) 0.80 0.35(h) 0.50(h) 0.10–0.30 (h)(o) 0.80 0.35(h) 0.50(h) 0.10–0.30 (h)(o) 0.80 0.35(h) 0.50(h) 0.10–0.30 (h)(o) 0.80 0.40–0.80 0.40–0.80 0.15–0.30 (j)(o) 0.80 0.40–0.80 0.40–0.80 0.15–0.30 (j)(o) 0.80 0.40–0.80 0.40–0.80 0.15–0.30 (j)(o) 0.80 0.40–0.80 0.40–0.80 0.15–0.30 (j)(o) 0.80 0.40–0.80 0.40–0.80 0.15–0.30 (j)(o) 0.80 0.40–0.80 0.40–0.80 0.30–0.40 (j)(o) 0.80 0.40–0.80 0.40–0.80 0.30–0.40 (j)(o) 0.80 0.40–0.80 0.70–1.00 0.40–0.60 (j)(o)(q) 0.80 2.00–2.75 ... 0.90–1.10 (j)(o) 0.80 2.00–2.75 ... 0.90–1.10 (j)(o) 0.80 2.00–2.75 ... 0.90–1.10 (j)(o) 0.80 0.75–1.10 0.50(h) 0.15–0.30 (h)(o) 0.80 0.75–1.10 0.50(h) 0.15–0.30 (h)(o) as 9A but with a slightly higher tempering temperature 0.80 0.75–1.10 0.50(h) 0.15–0.30 (h)(o) 0.80 0.55–0.90 1.40–2.00 0.20–0.40 (j)(o) 0.80 0.55–0.90 1.40–2.00 0.20–0.40 (j)(o) 0.60 0.50–0.80 0.70–1.10 0.45–0.65 (o)(r) 0.60 0.50–0.80 0.70–1.10 0.45–0.65 (o)(r) 0.60 0.50–0.90 0.60–1.00 0.90–1.20 (o)(r) 0.60 0.50–0.90 0.60–1.00 0.90–1.20 (o)(r) 0.60 0.40(s) 1.40–1.75 0.20–0.30 (o)(s) 0.60 0.40(s) 1.40–1.75 0.20–0.30 (o)(s) 0.60 0.40(s) 1.40–1.75 0.20–0.30 (o)(s) 0.50 0.20(r) 1.00–1.40 0.10(r) (r)(v)

(a) When a single value is shown, it is a minimum. (b) When a single value is shown, it is a maximum. (c) Unless specified by the purchaser, the compositions of cast steels in ASTM A 148 are selected by the producer and therefore may include either carbon or alloy steels: see Table 5 for the lower-grade steels specified in ASTM A 148. (d) 0.06% S (max), 0.05% P (max). (e) 0.020% S (max), 0.020% P (max). (f) Similar to the cast steel in ASTM A 148. (g) 0.045% S (max), 0.040% P (max). (h) When residual maximums are specified for copper, nickel, chromium, tungsten, and vanadium, their total content shall not exceed 1.00%. (i) 0.50% Cu (max), 0.10% W (max), 0.045% S (max), 0.04% P (max). (j) When residual maximums are specified for copper, nickel, chromium, tungsten, and vanadium, their total residual content shall not exceed 0.60%. (k) 0.35% Cu (max), 0.03% V (max), 0.015% S (max), 0.020% P (max). (l) 0.15–0.25% V. (m) The specified residuals of copper, nickel, chromium, and molybdenum (plus tungsten), shall not exceed a total content of 1.00%. (n) Includes the residual content of tungsten. (o) 0.50% Cu (max), 0.10% W (max), 0.03% V (max), 0.045% S (max), 0.04% P (max). (p) Material class 7A is a proprietary steel and has a maximum thickness of 63.5 mm (2½ in.). (q) Specified elements include 0.15–0.50% Cu, 0.03–0.10% V, and 0.002–0.006% B. (r) When residual maximum are specified for copper, nickel, chromium, tungsten, molybdenum, and vanadium, their total content shall not exceed 0.50%. (s) When residual maximums are specified for copper, nickel, chromium, tungsten, and vanadium, their total content shall not exceed 0.75%. (t) Low-carbon grade with double austenitization. (u) For each reduction of 0.01% C below the maximum, an increase of 0.04% Mn is permitted up to a maximum of 2.30%. (v) 0.20% Cu (max), 0.10% W (max), 0.02% V (max), 0.02% S (max), 0.02% P (max)

offer compared to carbon steels. When cast steels are quenched and tempered, the range of strength and of toughness is broadened. Depending on alloy selection, NDTT values of as high as 10  C (50  F) to as low as

107  C (160  F) can be obtained in the yield strength range of 345 to 1345 MPa (50 to 195 ksi) (Fig. 12). An approximate relationship exists between the Charpy V-notch impact energy-temperature

behavior and the NDTT value. The NDTT value frequently coincides with the energy transition temperature determined in Charpy V-notch tests. Applicable data for cast carbonmanganese steels are shown in Fig. 13.

956 / Steel Castings

Fig. 6

Fig. 4

Fig. 5

Properties of cast carbon steels as a function of carbon content and heat treatment. (a) Tensile strength and reduction of area. (b) Yield strength and elongation. (c) Brinell hardness. (d) Charpy V-notch impact energy

Difference in cooling rates of the surface and center of a steel casting and the resulting microstructures obtained

Fig. 7

Hardness versus tensile strength of low-alloy cast steels

Tensile properties of cast carbon steels as a function of Brinell hardness

Plane-strain fracture toughness (KIc) data for a variety of steels reflect the important strengthtoughness relationship (Table 7 for low-alloy steels). For quenched-and-tempered Ni-Cr-Mo steels, Fig. 14 indicates of ffiffiffiffi pffiffiffiffihigh KIc pvalues approximately 110 MPa mð100 ksi in:Þ, at a 0.2% offset yield strength level of 1034 MPa (150 ksi). At a yield strength level of 1655 level MPa (240 ksi), KIc values p ffiffiffiffi off to approxipffiffiffiffi mately 66 MPa mð60 ksi in:Þ. Data are also plotted in Fig. 14 for wrought plates made of comparable steel of somewhat higher carbon content. Fracture mechanics tests have the advantage over conventional toughness tests of being able to yield material property values that can be used in design equations. Strength and Fatigue. The most basic method of presenting engineering fatigue data is by means of the S-N curve, which relates the dependence of the life of the fatigue specimen in terms of the number of cycles to failure, N, to the maximum applied stress, S. Constant Amplitude Tests. The endurance ratio (endurance limit divided by the tensile strength) of cast carbon and low-alloy steels as determined in Moore rotating-beam bending fatigue tests (mean stress = 0) is generally taken to be approximately 0.40 to 0.50 for smooth bars. The data given in Table 8 indicate that this endurance ratio is largely independent of strength, alloying additions, and heat treatment. The fatigue notch sensitivity factor, q, determined in rotating-beam bending fatigue tests is related to the microstructure of the steel

Steel Castings Properties / 957

Fig. 10

Effect of various heat treatments on the Charpy V-notch impact energy of a 0.30% C steel

Table 7 Plane-strain fracture toughness of cast low-alloy steels at room temperature Yield strength, 0.2% offset

Fig. 9

Room-temperature properties of cast low-alloy steels. QT, quenched and tempered; NT, normalized and tempered

Fig. 8

Room-temperature properties of cast carbon steels after different heat treatments

Fig. 12

Nil ductility transition temperatures and yield strengths of quenched-and-tempered commercial cast steels

Alloy type

Fe-1.25Cr0.5Mo Cast 1030 Fe-0.5Cr0.5Mo0.25 V Fe-0.5C1.5Mn Fe-0.5C1Cr Fe-0.5C Cast 9535 Fe-0.35C0.6Ni0.7Cr0.4Mo Cast 4335 Cast 9536 Fe-0.3C1Ni1Cr0.3Mo Cast 4335 Cast 4335 Cast 4335 Cast 4335 Ni-Cr-Mo Cast 4340 Cast 4325 Cast 4325 Cr-Mo Cast 4340

Plane-strain fracture toughness, KIc pffiffiffiffiffiffi pffiffiffiffi MPa m ksi in:

Heat treatment(a)

MPa

ksi

SRANTSR

275

40

88

80

NT NT

303 367

44 53

127 55

116 50

NT

412

59

107

98

NT

413

60

58

53

NT NT NT

425 614 683

61 89 99

65 67 64

59 61 58

747 108 752 109 787 114

69 59 66

63 54 60

97 105 115 92 98 115 75 104 84 67

87 96 105 84 89 105 82 95 76 61

SLQT NT NT

SLQT SLQT QT QT QT QT QT QT QT QT

814 903 1090 1166 1207 1207 1263 1280 1379 1450

118 131 158 169 175 175 183 186 200 210

(a) SR, stress relieved: A. annealed; N. normalized: T. tempered; Q, quenched; SLQ, slack quenched

Fig. 13 Fig. 11

Nil ductility transition temperatures and yield strengths of normalized-and-tempered commercial cast steels

Nil ductility transition temperature values (triangles) and the scatter band of Charpy V-notch energy transition curves of several quenchedand-tempered carbon-manganese cast steels. Nominal ultimate tensile strength: 552 MPa (80 ksi)

(composition and heat treatment) and the strength. Table 9 shows that q generally increases with strength—from 0.23 for annealed carbon steel at a tensile strength of 577 MPa (83.5 ksi) to 0.68 for the higherstrength normalized-and-tempered low-alloy steels. The quenched-and-tempered steels with a martensitic structure are less notch sensitive than the normalized-and-tempered steels with a ferrite-pearlite microstructure. Similar results and trends in notch sensitivity have been reported for tests with sharper notches.

958 / Steel Castings Cast steels suffer less degradation of fatigue properties due to notches than equivalent wrought steels. When the ideal laboratory test conditions are replaced with more realistic service conditions, the cast steels exhibit much less notch sensitivity to variations in the values

of the test parameters than wrought steels. Table 10 shows that the q values for wrought steels are 1.4 to 2.3 times higher than those for cast steels. Under laboratory test conditions (uniform specimen section size, polished and honed surfaces, and so on), the endurance limit of wrought steel is higher than that of cast steel. The same fatigue characteristics as those of cast steel, however, are obtained when a notch is introduced or when standard lathe-turned surfaces are employed in the rotating-beam bending fatigue test. These effects are illustrated in Fig. 15 and 16 for carbon and Fig. 17 for low alloy. The cyclic stress-strain characteristics shown in Fig. 18 and Table 11 indicate a reduction of the strain-hardening exponent n of the normalizedand-tempered cast carbon steel (SAE 1030) from 0.3 in monotonic tension to 0.13 under cyclic strain-controlled tests. Figure 19 shows that the strain life characteristics of normalized-and-tempered cast carbon steel (SAE 1030) and wrought steel are similar. These data were obtained from strain-controlled constant-amplitude low-cycle fatigue tests (0.001 to 0.02 strain range

Fig. 14

Plane-strain fracture toughness (Klc) and strength relationships at room temperature for quenched-and-tempered Ni-Cr-Mo steels

Tensile strength

Yield strength

MPa

ksi

MPa

ksi

Reduction in area, %

Carbon steels 60 A 65 N 70 N 80 NT 85 NT 100 QT

434 469 517 565 621 724

63 68 75 82 90 205

241 262 290 331 379 517

35 38 42 48 55 75

54 48 45 40 38 41

30 28 27 23 20 19

Low-alloy steels(c) 65 NT 70 NT 80 NT 90 NT 105 NT 120 QT 150 QT 175 QT 200 QT

469 510 593 655 758 883 1089 1234 1413

68 74 86 95 110 128 158 179 205

262 303 372 441 627 772 979 1103 1172

38 44 54 64 91 112 142 160 170

55 50 46 44 48 38 30 25 21

32 28 24 20 21 16 13 11 8

Class(a)

Fatigue endurance limit MPa

ksi

Ratio of endurance limit to tensile strength

131 131 143 163 179 212

207 207 241 255 269 310

30 30 35 37 39 45

0.48 0.44 0.47 0.45 0.43 0.47

137 143 170 192 217 262 311 352 401

221 241 269 290 365 427 510 579 607

32 35 39 42 53 62 74 84 88

0.47 0.47 0.45 0.44 0.48 0.48 0.47 0.47 0.43

Elongation, %

Hardness, HB

(a) Class of steel based on tensile strength (ksi). (b) A, annealed; N, normalized; NT, normalized and tempered; QT, quenched and tempered. (c)Below 8% total alloy content

Table 9

Tensile strength

MPa

ksi

Fatigue notch sensitivity factor q(a)

576 561

83.5 81.4

0.23 0.43

Normalized and tempered 1040 cast 649 1040 wrought 620 1330 cast 669 1340 wrought 702 4135 cast 777 4140 wrought 766 4335 cast 872 4340 wrought 859 8630 cast 762 8640 wrought 748

94.2 90.0 97.0 101.8 112.7 111.1 126.5 124.6 110.5 108.5

0.29 0.50 0.28 0.65 0.45 0.81 0.68 0.97 0.53 0.85

Quenched and tempered 1330 cast 843 1340 wrought 836 4135 cast 1009 4140 wrought 1012 4335 cast 1160 4340 wrought 1161 8630 cast 948 8640 wrought 953

122.2 121.2 146.4 146.8 168.2 168.4 137.5 138.2

0.48 0.73 0.43 0.93 0.51 0.92 0.57 0.90

(a) q = (Kf  1)/(Kt  1), where Kf is the notch fatigue factor (endurance limit unnotched/endurance limit notched), and Kt is the theoretical stress-concentration factor.

MPa

Unnotched

Ratio of fatigue endurance limit to tensile strength

Notched(a)

ksi

MPa

ksi

MPa

ksi

Unnotched

Notched

Normalized and tempered 1040 648 1330 685 1330 669 4135 777 4335 872 8630 762

94.2 99.3 97 112.7 126.5 110.5

260 334 288 353 434 372

37.7 48.4 41.7 51.2 63 54

193 219 215 230 241 228

28 31.7 31.2 33.3 34.9 33.1

0.40 0.49 0.43 0.45 0.50 0.49

0.30 0.32 0.32 0.30 0.28 0.30

Quenched and tempered 1330 843 4135 1009 4335 1160 8630 948

122.2 146.4 168.2 137.5

403 423 535 447

58.5 61.3 77.6 64.9

257 280 332 266

37.3 40.6 48.2 38.6

0.48 0.42 0.46 0.47

0.31 0.28 0.29 0.27

83.5

229

33.2

179

26

0.40

0.31

Annealed 1040

Tensile strength Steel

Fatigue notch sensitivity of various cast steels Fatigue endurance limit

Steel

Table 10 Fatigue notch sensitivity factors for cast and wrought steels at various strength levels

Annealed 1040 cast 1040 wrought

Table 8 Properties of various classes of cast carbon and low-alloy steels Heat treatment(b)

amplitudes, with a constant strain rate triangle wave form of 2.5  104 s1 at 0.5 to 3.3 Hz). Constant-load amplitude fatigue crack growth properties for load ratios of R = 0 (Fig. 20a) indicate comparable properties for cast and wrought steel and slightly better properties for normalized-and-tempered cast carbon steel (SAE 1030) under load ratios of R = 1 (Fig. 20b). These tests were conducted in air at 10 to 30 Hz, depending on load ratio, initial stress intensity, and crack length. Variable load-amplitude fatigue tests block loading tests indicate equal total life for cast and wrought carbon steels (cast 1030 and wrought 1020, respectively) (Fig. 21). The slower crack growth rate in the cast material

576

(a) Notched tests run with theoretical stress-concentration factor of 2.2

Tensile strength, MPa(ksi)

Cast Wrought

Fig. 15

648 (94) 620 (90)

Yield strength, Elongation, % MPa(ksi)

386 (56) 386 (56)

25 27

Hardness, HB

187 170

Fatigue characteristics of normalized and tempered cast and wrought 1040 steels. Notched and unnotched specimens were tested in R.R. Moore rotating-beam tests

Steel Castings Properties / 959

Fig. 19

Low-cycle strain-controlled fatigue behavior of cast and wrought carbon steels in the normalized-and-tempered condition

Fig. 16

Fatigue endurance limit versus tensile strength for notched and unnotched cast and wrought steels with various heat treatments. Data obtained in R.R. Moore rotating-beam fatigue tests, theoretical stressconcentration factor = 2.2

Fig. 18

Monotonic tensile and cyclic stress-strain behavior of comparable cast and wrought normalized-and-tempered carbon steels

Table 11 Monotonic tensile and cyclic stress-strain properties of cast and wrought steels Property

Tensile strength, MPa(ksi)

Cast 8630

Yield strength, MPa(ksi)

952 (138) 869 (126)

Wrought 8640 952 (138) 855(124)

Elongation, Hardness, % HB

15

286

22

286

Fig. 17

Fatigue characteristics (S-N curves) for cast and wrought 8600-series steels, quenched and tempered to the same hardness, both notched and unnotched. Moore rotating-beam tests; stressconcentration factor K = 2.2

compensated for the longer crack initiation life of the wrought carbon steel. Section Size and Mass Effects. Mass effects are common to steels, whether rolled, forged, or cast, because the cooling rate during heat treating varies with section size and because the microstructure constituents, grain size, and nonmetallic inclusions increase in size from surface to center. Mass effects are metallurgical in nature and are distinct from the effect of discontinuities. Figure 22 shows how the mass of a component lowers the strength properties of wrought AISI 8630 steel plate. The properties are plotted for the ¼T location, halfway between the surface and the center of the plate. Tables 12 and 13 show some of the effects of section size on selected cast steel alloys.

Monotonic tension 0.2% yield strength, MPa (ksi) Ultimate strength, MPa (ksi) True fracture strength, MPa (ksi) Reduction in area, % True fracture ductility Modulus of elasticity E, GPa (ksi) Strain-hardening exponent n Strength coefficient K, MPa (ksi) Cyclic stress-strain 0.2% yield strength, MPa (ksi) Strength coefficient K0 , MPa (ksi) Strain-hardening coefficient n0

Cast 1030

Wrought 1020

303 (44)

262 (38)

496 (72)

414 (60.0)

750 (109)

1000 (145)

46 58 0.62 0.87 207 (3  104) 203 (2.9  104) 0.3

0.19

1090 (158)

738 (107)

317 (46)

241 (35)

710 (103)

772 (112)

0.13

0.18

The section size or mass effect is of particular importance in steel castings, because mechanical properties are typically assessed from test bars machined from standardized coupons having fixed dimensions and are cast separately from or attached to the castings. The removal of test bars from the casting is impractical, because removal of material for testing would destroy the usefulness of the component. Test specimens removed from a casting will not routinely exhibit the same properties as test specimens machined from the standard test coupon designs for which minimum properties are established in specifications. The mass effect discussed previously, that is, the difference in cooling rate between the test coupons and the

part being produced, is the fundamental reason for this situation. Several specifications provide for the mass effect by permitting the testing of coupons that are larger than normal and that have cooling rates more representative of those experienced by the part being produced. Among these specifications are ASTM E 208, A 356, and A 757. Discontinuity Effects. If the plane-strain fracture toughness, KIc, of a material is known at the temperature of interest, designers can determine the critical combination of flow size and stress required to cause failure in one load application. In addition, designers can calculate the remaining life of a component having a discontinuity, or they can compute the largest acceptable flow from knowledge of the crack growth rate, da/dN, of a material and other fracture mechanics parameters. In the absence of suitable plane-strain fracture mechanics data, approximations can be made on the basis of results obtained from various tensile, impact, and fatigue tests. Elevated-Temperature Applications. Steels operating at temperatures above ambient are subject to failure by a number of mechanisms other than mechanical stress or impact. These include oxidation, hydrogen damage, sulfide scaling, and carbide instability, which manifests itself as graphitization. The environmental factors involved in elevated-temperature service (370 to 650  C, or 700 to 1200  F) require that steels used in this temperature range be carefully characterized. As a consequence, four ASTM International specifications have been developed for cast carbon and low-alloy steels for elevated-temperature service. One of these specifications, ASTM A 216, describes carbon steels; the other three, A 217, A 356, and A 389, cover low-alloy steels. The two alloying elements common to nearly all the steel compositions used at elevated temperatures are molybdenum and chromium. Molybdenum contributes strongly to creep resistance. Depending on microstructure, it has been shown that 0.5% Mo reduces the creep rate of steels by a factor of at least 103 at 600  C (1110  F). Chromium also reduces the creep rate modestly at levels to approximately 2.25%. At higher

960 / Steel Castings

Fig. 20

Constant-amplitude fatigue behavior of normalized-and-tempered cast and wrought carbon steels. (a) Load ratio R = 0. (b) R = 1

chromium levels, creep resistance is somewhat reduced. Vanadium improves creep strength and is indicated in some specifications. Other elements that improve creep resistance include tungsten, titanium, and niobium. The effect of tungsten is similar to that of molybdenum, but on a weight percent basis, more tungsten is required in order to be equally beneficial. Titanium and niobium have been shown to improve the creep properties of carbon-free alloys, but because they remove carbon from solid solution, their effect tends to be variable. Temper embrittlement is normally characterized by an increase in the Charpy V-notch impact transition temperature. It is generally associated with quenched-and-tempered steels, but it may also occur in normalized-and-tempered steels and may develop during either

tempering or elevated-temperature service. The tempering or service temperature range in which embrittlement occurs is considered to be 425 to 595  C (800 to 1100  F). During heat treatment, embrittlement can be avoided by rapid cooling through this range. The elements generally considered to cause embrittlement are phosphorus, tin, antimony, and arsenic; manganese and silicon enhance the effect of phosphorus. An addition of 0.05% Mo to alloy steels is very effective in reducing temper embrittlement. All of the alloy steels normally used for elevated-temperature service contain at least this quantity of molybdenum; therefore, temper embrittlement during heat treatment is seldom a problem. However, in those cases in which castings must be cooled from the tempering temperature at a rate below

20  C (35  F) per hour, it may be necessary to control the amount of temper-embrittling elements present in the steel. When embrittlement occurs during service, no reduction in impact properties at elevated temperature is apparent. Instead, diminished impact toughness may be observed during periods of shutdown when the equipment is at ambient temperatures.

Mechanical Properties of Wear-Resisting Steels Mechanical properties for the compositions shown in Table 1 are illustrated in Fig. 2, 3, 23, and 24. Property impairment as a result of increasing metal section and the reduced response to heat treatment is indicated. All

Steel Castings Properties / 961

Fig. 21

Average blocks to specific crack lengths and fracture for comparable cast and wrought carbon steels in the normalized-and-tempered condition

and in the dulling and rapid wear of the tool. Machining is performed with cemented carbide tools using heavy, rigid equipment; slow, steady feed; and deep cuts. A well-equipped machine shop, through the use of grinding techniques, can perform all basic machine tool functions. Boring, planing, keyway cutting, and similar operations are efficiently done by grinding.

Mechanical Properties of Corrosion-Resisting Steels

Fig. 22

Section size effects on water-quenched and tempered wrought AISI 8630 steel in sizes over 25 mm (1 in.). The properties reported are those at the ¼T location (midway between surface and center).

grades, with the exception of the lean austenitic and the machinable grades, approach or exceed 827 MPa (120 ksi) in ultimate tensile strength in 25 mm (1 in.) sections. The chromium-alloyed grades and the 2% Mo grades show yield strengths of 414 MPa (60 ksi) or greater, while the highyield-strength alloy develops 655 MPa (95 ksi) in 25 mm (1 in.) metal sections. Elongation and impact strength data favor the 1% Mo grades in heavy metal sections, while the 1% Mo lean alloys exhibit the poorest properties in this regard. Low magnetic permeabilities can be attained quite readily and economically. The combination of gouging wear resistance and low magnetic permeability is used to good advantage in manganese steel magnet cover wear plates. Machinability. Austenitic manganese steels are difficult to machine largely because they harden under and in front of the tool. Even the smallest amount of tool chatter on the casting results in high hardness at the point of contact

Some of the high-alloy cast steels exhibit many of the same properties of cast carbon and low-alloy steels. Some of the mechanical properties of these grades (for example, hardness and tensile strength) can be altered by suitable heat treatment. The cast high-alloy grades that contain more than 20 to 30% Cr and Ni, however, do not show the phase changes observed in carbon and low-alloy steels during heating or cooling between room temperature and the melting point. These materials are therefore nonhardenable, and their properties depend on composition rather than heat treatment. Therefore, special consideration must be given to each grade of high-alloy cast steel with regard to casting design, foundry practice, and subsequent thermal processing (if any). Strength and Hardness. Representative room-temperature tensile properties, hardness, and Charpy impact values for corrosion-resistant alloys are given in Table 14 and Fig. 25. These properties are representative of the alloys rather than the specification requirements. Minimum specified mechanical properties for these alloys are given in ASTM standards A 351, A 743, A 744, and A 747. A wide range of mechanical properties are attainable in high-alloy steel grades, depending on the selection of alloy composition and heat treatment. Tensile strengths ranging from 476 to 1310 MPa (69 to 190 ksi) and hardnesses from 130 to 400 HB are available among the cast corrosion-resistant alloys.

Similarly, wide ranges exist in yield strength, elongation, and impact toughness. The straight chromium steels (CA15, CA40, CB30, and CC50) possess either martensitic or ferritic microstructures in the end-use condition (Table 2). The CA15 and CA40 alloys, which contain nominally 12% Cr, are hardenable through heat treatment by means of the martensite transformation and are often selected as much or more for their high strength as for their comparatively modest corrosion resistance. Castings of these alloys are heated to a temperature at which the structure is fully austenitic and then cooled at a rate (usually in air) adapted to the casting composition so that the austenite transforms to martensite. Strengths in this condition are quite high (for example, 1034 to 1379 MPa, or 150 to 200 ksi), but tensile ductility and impact toughness are limited. Consequently, martensitic castings are usually tempered at 315 to 650  C (600 to 1200  F) to restore ductility and toughness at some sacrifice in strength. It follows, then, that significant ranges of tensile properties, hardness, and impact toughness are possible with the martensitic CA15 and CA40 grades, depending on the choice of tempering temperature. The higher-chromium CB30 and CC50 alloys, on the other hand, are fully ferritic alloys that are not hardenable by heat treatment. These alloys are generally used in the annealed condition and exhibit moderate tensile properties and hardness (Table 14). Like most ferritic alloys, CB30 and CC50 possess limited impact toughness, especially at low temperatures. Three chromium-nickel alloys—CA6NM, CB7Cu, and CD4MCu—are exceptional in their response to heat treatment and the resultant mechanical properties. Alloy CA6NM is balanced compositionally for martensitic hardening response, but it cannot be hardened to the level of CA15. It has improved impact toughness and weldability. Alloy CB7Cu contains copper and can be strengthened by age hardening. These alloys are initially solution heat treated and then cooled rapidly (usually by quenching in oil or water); thus, the phases that would normally precipitate at slow cooling rates cannot form. The casting is then heated to an intermediate aging temperature at which the precipitation reaction can occur under controlled conditions until the desired combination of strength and other properties is achieved. The CB7Cu alloy possesses a martensitic matrix, while the CD4MCu alloy possesses a duplex microstructure, consisting of over 40% austenite in a ferritic matrix. Alloy CB7Cu is applied in the aged condition to obtain the benefit of its excellent combination of strength and corrosion resistance, but alloy CD4MCu is seldom applied in the aged condition because of its relatively low resistance to SCC in this condition compared to its superior corrosion resistance in the solution-annealed condition. The CE, CF, CG, CH, CN, and CK alloys are essentially not hardenable by heat treatment. To ensure maximum corrosion resistance, however, it is necessary that castings of these

in.



C

29  C (20  F)

51  C (60  F)

0.95 Mn, 0.35 Si, 0.010 P, 0.010 S, 25 ... . . . 79 (58) 65 51 100 73 75 (55) 58 76 90 68 62 (46) 47 127 90 68 75 (55) 56

Mn, 0.38 Si, 0.007 P, 0.014 64 (47) 45 40 40 49 (36) 18 40 40 50 (37) 16 40 40 28 (21) 8 0.045 (48) (43) (35) (42)

Al, 61 53 42 50

0.013 (45) (39) (31) (37)

51  C (60  F)

29  C (20  F)

46  C (50  F)

(continued)

0.14 Cr, 0.012 Mo, 0.04 Cu, (31) 0.79 0.69 (27) (25) 0.64 0.51 (20) (32) 0.81 0.71 (28) (26) 0.66 0.56 (22)

0.008 N2 62 56 50 49

45 38 34 32

0.03 Ni, 0.03 Cr, 0.015 Mo, 0.04 Cu, 0.016 N2 0.97 (27) 0.69 0.48 (19) 29 12 0.64 (7) 0.18 0.10 (4) 21 8 0.51 (5) 0.13 0.025 (1) 25 5 0.36 (1) 0.025 0 (0) 15 3

Ti, 1.82 Ni, (39) 0.99 (40) 1.02 (42) 1.07 (34) 0.86

S, 0.051 A1, 0.002 Ti, (33) 33 (24) (38) (13) 11 (8) (25) (12) 11 (8) (20) (6) 8 (6) (14)

46  C (50  F)

40 32 27 27

10 2 3 1

N2, 3.5 ... 1 1 ... ... 1 1 ... ... 1 1 ... ... 10 ... ... ... 10 ... ...

51  C (60  F)

(Cv) Shear area, %

0.03 Ni, 0.06 Cr, 0.01 Mo, 0.05 Cu, 0.011 O2, 0.0098 1.17 . . . ... ... ... 18 ... 1.02 . . . ... 0.56 (22) 15 ... 1.04 . . . ... 0.64 (25) 13 ... 0.86 (26) 0.66 ... ... 8 1 1.12 . . . ... ... ... 28 ... 1.19 . . . ... 0.64 (25) 25 ... 1.14 . . . ... 0.64 (25) 18 ... 1.09 (38) 0.97 ... ... 19 1 1.07 . . . ... ... ... 20 ... 1.22 . . . ... 0.43 (17) 22 ... 1.17 . . . ... 0.33 (13) 15 ... 1.12 (10) 0.25 ... ... 21 1 1.19 . . . ... ... ... 20 ... 1.07 . . . ... 0.36 (14) 38 ... 1.02 (22) 0.56 ... ... 35 10 1.12 (14) 0.36 ... ... 30 10 1.07 . . . ... ... ... 20 ... 0.86 . . . ... 0.41 (16) 28 ... 0.81 (24) 0.61 ... ... 28 10 0.99 (19) 0.48 ... ... 25 10

29  C (20  F)

0.007 S, 0.056 Al, 0.010 Ti, . . . . . . . . . . . . (46) . . . . . . 39 (29) (40) . . . . . . 42 (31) (41) 42 (31) . . . . . . (34) . . . . . . . . . . . . (44) . . . . . . 58 (43) (47) . . . . . . 43 (32) (45) 64 (47) . . . . . . (43) . . . . . . . . . . . . (42) . . . . . . 35 (26) (48) . . . . . . 24 (18) (46) 37 (27) . . . . . . (44) . . . . . . . . . . . . (47) . . . . . . 40 (29) (42) 41 (30) . . . . . . (40) 28 (21) . . . . . . (44) . . . . . . . . . . . . (42) . . . . . . 30 (22) (34) 39 (29) . . . . . . (32) 34 (25) . . . . . . (39)

46  C (50  F)

(Cv) Lateral expansion, mm (mils) ksi

MPa

600 552 565 565

483 455 448 455

87 80 82 82

70 66 65 66

476 441 434 448

317 317 296 296

69 64 63 65

46 46 43 43

53 53 50 50 54 53 51 52 55 53 52 48 54 52 53 48 57 50 48 46

ksi

Yield strength

ppm H2 (riser) 510 74 365 503 73 365 496 72 345 483 70 345 510 74 372 503 73 365 496 72 352 496 72 359 517 75 379 503 73 365 496 72 359 490 71 331 517 75 372 503 73 359 483 70 365 476 69 331 531 77 393 476 69 345 476 69 331 462 67 317

MPa

Ultimate tensile strength

30 35 35 31

35 36 39 35

33 33 34 32 32 33 32 30 34 34 32 33 35 36 35 36 34 32 35 34

Elongation, %

67 62 66 54

73 69 72 60

69 69 69 62 72 69 67 56 74 67 63 61 69 70 70 68 65 69 71 71

Reduction in area, %

163 166 159 163

156 143 143 143

149 146 146 146 149 146 146 143 146 146 146 143 ... ... ... ... ... ... ... ...

Hardness, HB

C

40 34 29 29

7 +7 16 16



(40) (30) (20) (20)

(+20) (+45) (+60) (+60)

... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...

( F)

Fractureappearance transition temperature50

(a) Abbreviations: N = normalize (air cool), T = temper, WQ = water quench, AC = air cool, SR = stress relief. Temperatures are listed in  F; the number following the stress relief or aging temperature indicates time (hours) at temperature. Property values for T = 51, 76, and 127 mm (2, 3, and 5 in.) sections were obtained at ¼ T distance below the plate surface. (b) NDTT, nil ductility transition temperature

A352-LCC(Ni Mod.): 0.205 C, N1750, WQ1650, 1 T1226WQ, 2 AGE750/48 3 5

A352-LCB: UNS J03003 0.18 C, 0.75 WQ1650, T1200WQ, 1 25 AGE800/40, 2 51 SR1100/40 3 76 5 127

F

(Cv) energy, J (ft  lbf)

0.35 Si, 0.010 P, . . . 75 (55) 60 72 (53) 60 75 (55) 50 61 (45) 83 (61) 60 80 (59) 60 79 (58) 50 76 (56) 69 (51) 60 85 (63) 60 68 (50) 50 77 (57) 99 (73) 60 80 (59) 50 79 (58) 50 79 (58) 75 (55) 60 66 (49) 50 66 (49) 50 58 (43)



NDTT(b)

A216-WCC, UNS J02503: 0.20 C, 0.91 Mn, N1650, T1150AC, 25 1 ... WQ1650, T1150AC 51 2 51 76 3 51 127 5 46 N1650, N1600, 25 1 T1150AC, WQ1650, 51 2 51 T1150AC 76 3 51 127 5 46 N1650, N1600, 25 1 T1150AC, WQ1650, 51 2 51 T1150AC, SR1125/15 76 3 51 127 5 46 N1650, N1600, 25 1 T1150AC, WQ1650, 51 2 51 T1150AC, SR1125/40 76 3 46 127 5 46 25 1 N1650, N1600, 51 2 T1150AC, WQ1650, 51 T1150AC, SR1125/15, 76 3 46 SR 1125/40 127 5 46

mm

Section size

Trends of property changes with section size of several cast steel grades

Grade and heat treatment(a)

Table 12

962 / Steel Castings

(continued) 0.010P, 0.008 S, 0.041 Al, 81 (60) 64 (47) 58 76 (56) 58 (43) 53 83 (61) 64 (47) 58 73 (54) 54 (40) 47

0.019 (43) (39) (43) (35)

Ti, 1.76 Ni, (44) 1.12 (41) 1.04 (41) 1.04 (39) 0.99

0.04 Cr, 0.22 (35) 0.89 (29) 0.74 (34) 0.86 (27) 0.69

Mo, 0.04 Cu, 0.76 (30) 0.69 (27) 0.84 (33) 0.48 (19)

0.009 N2, 62 53 49 39

0.044 (16) (16) (12) (11)

Al, 19 20 16 14

Cr, 0.34 Mo, 1.02 0.91 0.74 0.69 0.69 0.58 0.48 0.46

0.04 Cu (36) (27) (23) (18)

WQ1750, 100 90 82 43

627 648 648 648

91 91 91 91

T1150WQ, AGE700/48 94 84 772 112 70 64 814 118 60 50 758 110 27 21 724 105

0.012 Ti, 1.70 Ni, 0.075 Cr, 0.31 Mo, 0.035 Cu, 7.11 ppm H2 (Riser) (14) (20) 0.51 (15) 0.38 0.25 (10) 11 8 7 (15) (8) 0.20 . . . ... ... ... 5 4 3 (12) (4) 0.10 . . . ... ... ... 5 2 1 (10) (5) 0.13 . . . ... ... ... 5 ... ...

S, 0.055 Al, 0.012 Ti, 1.63 Ni, 0.13 (59) 76 (56) (43) 1.09 (40) (46) 57 (42) (33) 0.84 (29) (43) 53 (39) (35) 0.89 (27) (32) 37 (27) (26) 0.66 (19)

1.08 Mn, 0.43 Si, 0.010P, 0.010 S, 25 ... . . . 31 (23) 22 51 20 29 26 (19) 22 76 20 29 18 (13) 16 127 0 18 23 (17) 15

0.007P, 0.011 85 (63) 80 75 (55) 62 73 (54) 58 56 (42) 43

483 490 483 476

676 689 614 572

70 71 70 69

98 100 89 83

24 22 23 24

18 28 20 16

15 14 11 12

20 21 22 20

27 28 32 25

... ... ... ...

... ... ... ...

... ... ... ...

... ... ... ...

... ... ... ... ... ... ... ...

... ... ... ...

... ... ... ... ... ... ... ...

... ... ... ...

... ... ... ...

... ... ... ...

... ... ... ...

... ... ... ...

... ... ... ...

(a) Abbreviations: N = normalize (air cool), T = temper, WQ = water quench, AC = air cool, SR = stress relief. Temperatures are listed in  F; the number following the stress relief or aging temperature indicates time (hours) at temperature. Property values for T = 51, 76, and 127 mm (2, 3, and 5 in.) sections were obtained at ¼ T distance below the plate surface. (b) NDTT, nil ductility transition temperature

A487-Class 11N, 90-60: 0.25 C, N1750, T1200-1225Ac, 1 AGE700/48 2 3 5

A487-Class 11Q, 105–85: 0.26 C, 1.03 Mn, 0.49 Si, WQ1750, T1150WQ, 1 25 ... ... AGE700/48 2 51 150 101 3 76 120 89 5 127 80 62

157 1020 148 151 910 132 140 752 109 132 717 104

116 96 88 85

72 68 69 61

1082 1041 965 910

800 662 607 586

496 496 476 421

130 119 111 107

95 88 92 78

896 820 765 738

6.88 ppm H2 (riser) 45 40 655 38 32 607 33 28 634 21 14 538

A487-Class 12BO 120-95: 0.275 C, 1.02 Mn, 0.40 Si, 0.010P, 0.010 S, 0.042 Al, 0.009 Ti, 1.60 Ni, 0.06 Cr, 0.28 Mo, 0.035 Cu, 7.79 ppm H2 (riser) WQ1750, T1075WQ, 1 25 ... . . . 75 (55) 72 (53) 69 (51) (33) 0.84 (31) 0.79 0.74 (29) 100 100 99 AGE700/48 2 51 140 96 65 (48) 53 (39) 50 (37) (26) 0.66 (22) 0.56 0.51 (20) 70 53 47 3 76 100 73 68 (50) 53 (39) 47 (35) (29) 0.74 (20) 0.51 0.48 (19) 64 43 36 5 127 50 46 47 (35) 35 (26) 31 (23) (25) 0.64 (16) 0.41 0.33 (13) 40 23 18 A148-Gr 150-135, D51350: 0.295 C, 0.90 Mn, 0.27 Si, 0.009 P, 0.010 S, 0.048 Al, 0.008 Ti, 1.58 Ni, 0.42 Cr, 0.33 Mo, 0.03 Cu, 6.38 ppm H2 (riser) WQ1750, T980WQ, 1 25 ... . . . 47 (35) 46 (34) 45 (33) (19) 0.48 (16) 0.41 0.38 (15) 100 100 157 AGE700/48 2 51 80 62 40 (29) 34 (25) 31 (23) (10) 0.25 (7) 0.18 0.16 (6) 68 26 24 3 76 60 51 30 (22) 24 (18) 24 (18) (5) 0.13 (3) 0.076 0.076 (3) 35 18 16 5 127 10 23 24 (18) 23 (17) 23 (17) (10) 0.25 (10) 0.25 0.25 (10) 6 ... ...

A352-LCC-(Ni-Mo-Mod):0.18C, 0.92 Mn, 0.38 Si, N1750, WQ1650, 1 25 ... ... T1225WQ, 2 51 100 73 AGE750/48 3 76 90 68 5 127 80 62

Table 12

Steel Castings Properties / 963

in.

C



F

type: 0.09 1 2 101 3 96 5 68

C, 0.64 Mn, 111 150 121 140 92 90 79

101  C (150  F)

0.019 1.07 1.40 1.24 1.17

73  C (100  F)

51  C 73  C (60  F) (100  F)

101  C (150  F)

Cr, 0.01 Mo, 0.03 Cu, 0.006 N2, 4.87 ppm 0.41 (16) 50 32 10 0.13 (5) 43 23 6 0.33 (13) 42 24 7 ... ... 30 10 ...

101  C (150  F)

Shear area (Cv), %

2.80 Ni, 1.29 Cr, 0.46 Mo, 0.04 Cu, 0.010 N2 (36) 0.43 (17) 79 52 18 (38) 0.53 (21) 100 78 22 (24) 0.30 (12) 86 43 14 (20) 0.23 (9) 50 21 13

Ti, 2.44 Ni, 0.10 (42) 1.02 (40) (55) 1.09 (43) (49) 0.86 (34) (46) 0.64 (25)

51  C (60  F)

Lateral expansion, (Cv), mm (mils)

0.013 S, 0.038 Al, 0.005 Ti, 46 (34) 1.35 (53) 0.91 47 (35) 1.32 (52) 0.97 34 (25) 1.19 (47) 0.61 30 (22) 0.89 (35) 0.51

P, 0.013 S, 0.049 Al, 69 (51) 34 (25) 61 (45) 27 (20) 64 (47) 24 (18) 43 (32) . . . . . .

0.39 Si, 0.009 P, (82) 83 (61) (89) 88 (65) (68) 62 (46) (58) 43 (32)

0.010 (78) (67) (68) (57)

73  C (100  F)

Energy (Cv), J (ft  lbf) 51  C (60  F)

C, 0.68 Mn, 0.36 Si, 106 96 140 91 96 140 92 79 110 77



NDTT(b)

ksi

421 407 393 372

MPa

61 59 57 54

ksi

Yield strength

... ... ... ... 634 92 517 75 621 90 490 71 627 91 483 70

H2 (riser) 503 73 496 72 490 71 483 70

MPa

Ultimate tensile strength

... 29 31 26

34 35 33 34

Elongation, %

... 68 69 67

75 69 70 54

Reduction in area, %

... 196 196 196

... 143 143 137

Hardness, HB

C

F

65 45 40 20



76 105 87 125 68 90 51 60

53 43 40 29



Fractureappearance transition temperature50

(a) Abbreviations: N = normalize (air cool), T = temper, WQ = water quench, AC = air cool, SR = stress relief. Temperatures are listed in  F; the number following the stress relief for aging temperature indicates time (hours) at temperature. Property values for T = 51, 76, and 127 mm (2, 3, and 5 in.) sections were obtained at ¼ T distance below the plate surface. (b) NDTT, nil ductility transition temperature

A352-LC2-1, J31550, Ni-Cr-Mo 25 N1750, 51 WQ1650, T1200WQ, 76 AGE 800/64 127

A352-LC2 J22500:(a) J22500: 0.06 N1650, 25 1 WQ1650, T1150WQ 51 2 76 3 127 5

mm

Section size

Trends of property changes with section size of several cast steel grades at lower temperatures

Grade and heat treatment(a)

Table 13

964 / Steel Castings

Steel Castings Properties / 965 Table 14

Room-temperature mechanical properties of cast corrosion-resistant alloys Tensile strength

Alloy

2 for key.

ksi

MPa

ksi

Reduction in area, %

Hardness, HB

J

ft  lbf

827 120 793 115

689 100 689 100

24 22

60 55

269 225

94.9 27.1

70 20

980  C (1800  F), AC, T

1034 150

862 125

10

30

310

2.7

2

CB7Cu CB30

1040  C (1900  F), OQ, A 1310 190 1172 170 790  C (1450  F), AC 655 95 414 60

14 15

54 ...

400 195

33.9 2.7

25 2

CC50 CD4MCu

CE30

1040  C (1900  F), AC 1120  C (2050  F), FC to 1040  C (1900  F), WQ 1120  C (2050  F), FC to 1040  C (1900  F), A 1095  C (2000  F), WQ

CF3 CF3A CF8 CF8A

CA40

Typical values of ultimate tensile strength for austenitic manganese steels. See Fig.

MPa

Charpy impact energy

Elongation in 50 mm (2 in.), %

>955  C (1750  F), AC, T 980  C (1800  F), AC, T

CA6NM CA15

Fig. 23

Heat treatment(a)

Yield strength (0.2% offset)

CF20

669 97 745 108

448 558

65 81

18 25

... ...

210 253

... 74.6

... 55

V-notch Keyhole notch Keyhole notch V-notch Keyhole notch ... V-notch

896 130

634

92

20

...

305

35.3

26

V-notch

669

97

434

63

18

...

190

9.5

7

>1040  C (1900  F), WQ >1040  C (1900  F), WQ >1040  C (1900  F), WQ

531 600 531

77 87 77

248 290 255

36 42 37

60 50 55

... ... ...

140 160 140

>1040  C (1900  F), WQ

Keyhole notch V-notch V-notch Keyhole notch Keyhole notch Keyhole notch V-notch V-notch Keyhole notch Keyhole notch Keyhole notch V-notch Keyhole notch Izod Vnotch Keyhole notch

149.2 110 135.6 100 100.3 74

586

85

310

45

50

...

156

94.9

70





531

77

248

36

50

...

163

81.4

60





>1095 C (2000 F), WQ

CF3M CF3MA CF8M

>1040 C (1900 F), WQ >1040  C (1900  F), WQ >1065  C (1950  F), WQ

552 621 552

80 90 80

262 310 290

38 45 42

55 45 50

... ... ...

150 170 170

CF8C

>1065  C (1950  F), WQ

CF16F

162.7 120 135.6 100 94.9 70

531

77

262

38

39

...

149

40.7

30





531

77

276

40

52

...

150

101.7

75





>1095 C (2000 F), WQ

CG8M CH20

>1040 C (1900 F), WQ >1095  C (2000  F), WQ

565 607

82 88

303 345

44 50

45 38

... ...

176 190

108.5 40.7

80 30

CK20

1150  C (2100  F), WQ

524

76

262

38

37

...

144

67.8

50

476

69

214

31

48

...

130

94.9

70

CN7M

Specimen





1120 C (2050 F), WQ

(a) AC, air cool; FC, furnace cool; OQ, oil quench; WQ, water quench; T, temper; A, age

Fig. 24

Typical yield strength values for austenitic manganese steels. See Fig. 2 for key.

grades receive a high-temperature solution anneal. This treatment consists of holding the casting at a temperature that is high enough to dissolve all chromium carbides, which are damaging to intergranular corrosion resistance, and then cooling them rapidly enough to avoid reprecipitation of the carbides by quenching in water, oil, or air. Although this can be accomplished throughout in the lower-carbon grades (<0.08% C), heavy sections of alloys with higher carbon contents may have carbides

present at some distance below the surface where the cooling rate is slow. Subsurface carbides could be exposed by subsequent machining of the casting, resulting in lowered corrosion resistance in service. By virtue of their microstructures, which are fully austenitic or duplex without significant carbide precipitation, the alloys exhibit generally excellent impact toughness at low temperatures. The tensile strength range represented by these alloys typically extends from 476 to 669 MPa (69 to 97 ksi). As indicated earlier in this article, the alloys with duplex structures can be strengthened by balancing the composition for higher ferrite levels. The tensile and yield strengths of CF alloys with a ferrite number of 35 are typically 150 MPa (22 ksi) higher than those of full austenitic alloys. Fatigue properties can become a design factor in applications in which the casting experiences cyclic loading conditions in service. The resistance of cast stainless steels to fatigue depends on a sizable number of material, design, and environmental factors. For example, design factors of importance include the stress distribution within the casting (residual

and applied stresses), the location and severity of stress concentrators and the environment and service temperature. Material factors of importance include strength and microstructure. It is generally found that fatigue strength increases with the tensile strength of a material. Both fatigue strength and tensile strength generally increase with decreasing temperature. Under equivalent conditions of stress, stress concentration, and strength, evidence suggests that austenitic materials are less notch sensitive than martensitic or ferritic materials. Toughness. The fully austenitic corrosionresistant alloys exhibit excellent toughness. Figure 25 indicates the magnitude of toughness attainable in various grades in Charpy keyhole impact testing. Effects of Aging. Cast corrosion-resistant steels are extensively used at moderately elevated temperatures (up to 650  C, or 1200  F). Elevated-temperature properties are important selection criteria for these applications. In addition, room-temperature properties after service at elevated temperatures are increasingly considered because of the aging effect that these exposures may have. For example, cast alloys

966 / Steel Castings temperatures, may differ from those in the asheat-treated condition because of the microstructural changes that may take place at the service temperature. Microstructural changes in Fe-NiCr-(Mo) alloys may involve the formation of carbides and such phases as s, w, and Z (Laves). The extent to which these phases form depends on the composition as well as the time at elevated temperature. The martensitic alloys CA15 and CA6NM are subject to minor changes in mechanical properties and SCC resistance in NaCl and polythionic acid environments upon exposure for 3000 h at up to 565  C (1050  F). In CFtype Cr-Ni-(Mo) steels, only negligible changes in ferrite content occur during 10,000 h exposure at 400  C (750  F) and during 3000 h exposure at 425  C (800  F). Carbide precipitation, however, does occur at these temperatures, and noticeable Charpy V-notch energy losses have been reported. These effects are illustrated in Fig. 26 and 27 for CF8 cast corrosion-resistant steel. Above 425  C (800  F), microstructural changes in Cr-Ni-(Mo) alloys take place at an increased rate. Carbides and s-phase form rapidly at 650  C (1200  F) at the expense of ferrite (Fig. 28). Tensile ductility and Charpy V-notch impact energy are prone to significant losses under these conditions. Density changes, resulting in contraction, have been reported as a result of these high-temperature exposures.

Mechanical Properties of HeatResisting Steels

Alloy

CA15 (a) (b) (c) (d) CA40 (a) (b) (c) (d) CB30 CC50 (a) (b) (c)

Heat treatment

AC from 980 ⬚C (1800 ⬚F), T at 790C (1450 ⬚F) AC from 980 ⬚C (1800 ⬚F), T at 650C (1200 ⬚F) AC from 980 ⬚C (1800 ⬚F), T at 595C (1100 ⬚F) AC from 980 ⬚C (1800 ⬚F), T at 315C (600 ⬚F) AC from 980 ⬚C (1800 ⬚F), T at 760C (1400 ⬚F) AC from 980 ⬚C (1800 ⬚F), T at 650C (1200 ⬚F) AC from 980 ⬚C (1800 ⬚F), T at 595C (1100 ⬚F) AC from 980 ⬚C (1800 ⬚F), T at 315C (600 ⬚F) A at 790 ⬚C (1450 ⬚F), FC at 540 ⬚C (1000 ⬚F),AC As-cast (<1%Ni) As-cast (>2%Ni;>0.15%N) AC from 1040 ⬚C (1900F)(>2%Ni;>0.15%N)

Alloy

CE30 (a) (b) CF8 CF2 CF8M, CF12M CF8 CF16F CH20 CK20 CN7M

Heat treatment

As-cast WQ from 1065–1120 ⬚C(1950–2050 ⬚F) WQ from 1065–1120 ⬚C(1950–2050 ⬚F) WQ from above 1095 ⬚C(2000 ⬚F) WQ from 1065–1150 ⬚C(1950–2100 ⬚F) WQ from 1065–1120 ⬚C(1950–2050 ⬚F) WQ from above 1095 ⬚C(2000 ⬚F) WQ from above 1095 ⬚C(2000 ⬚F) WQ from above 1150 ⬚C(2100 ⬚F) WQ from above 1065–1120 ⬚C(1950–2050 ⬚F)

Fig. 25

Mechanical properties of cast corrosion-resistant steels at room temperature. (a) Tensile strength. (b) 0.2% offset yield strength. (c) Charpy keyhole impact energy. (d) Brinell hardness. (e) Elongation. Also given are the heat treatments used for test materials: AC, air cool; FC, furnace cool; WQ, water quench; A, anneal; T, temper.

CF8C, CF8M, CE30A, and CA15 are currently used in high-pressure service at temperatures to 540  C (1000  F) in sulfurous acid environments in the petrochemical industry. Other uses

are in the power-generating industry at temperatures to 565  C (1050  F). Room-temperature properties in the aged condition, that is, after exposure to elevated-service

Elevated-Temperature Tensile Properties. The short-term elevated-temperature test, in which a standard tension test bar is heated to a designated uniform temperature and then strained to fracture at a standardized rate, identifies the stress due to a short-term overload that will cause fracture in uniaxial loading. The manner in which the values of tensile strength and ductility change with increasing temperature is shown in Fig. 29 for alloy HP50WZ. Representative tensile properties at temperatures between 650 and 1095  C (1200 and 2000  F) are shown in Table 15 for several heat-resistant alloy steel grades. Creep and Stress-Rupture Properties. Creep is defined as the time-dependent strain that occurs under load at elevated temperature. Creep is operative in most applications of heat-resistant high-alloy castings at the normal service temperatures. In time, creep may lead to excessive deformation and even fracture at stresses considerably below those determined in room-temperature and elevated-temperature short-term tension tests. The designer must usually determine whether the serviceability of the component in question is limited by the rate or the degree of deformation. When the rate or degree of deformation is the limiting factor, the design stress is based on the minimum creep

Steel Castings Properties / 967

Transformation of d-ferrite to austenite and sphase upon exposure of a solution-treated CF8 casting to elevated temperature

Fig. 28

Fig. 26

Effects of time at elevated temperature on the tensile properties of static and centrifugal CF8 alloy castings. Parts had a ferrite number of 9 to 11 and contained 0.081% N.

Fig. 29

Fig. 27

Effect of time at elevated temperature on the room-temperature impact strength, ferrite number and stresscorrosion cracking (SCC) resistance of CF8 castings with a ferrite number of 9 to 11 and nitrogen content of 0.081%. Arrows indicate no failure in SCC testing after 336 h.

Tensile properties versus temperature for heatresistant alloy HP50WZ

rate and design life after allowing for initial transient creep. The stress that produces a specified minimum creep rate of an alloy or a specified amount of creep deformation in a given time (for example, 1% total creep in 100,000 h) is referred to as the limiting creep strength or limiting stress. The manner in which the minimum creep rate depends on the applied stress is illustrated in Fig. 30 by data for alloy HP50WZ. When fracture is the limiting factor, stressto-rupture values can be used in design (Fig. 31). The stress-to-rupture values can be combined with those for minimum creep rate, as shown in Fig. 32. It should be recognized that long-term creep and stress-rupture values (for example, 100,000 h) are often extrapolated from shorter-term tests. Whether these property

968 / Steel Castings Table 15 Representative short-term tensile properties of cast heat-resistant alloys at elevated temperatures Property at indicated temperature 650  C (1200  F) Ultimate tensile strength

Yield strength

870  C (1600  F)

MPa

ksi

MPa

... 10 ... 14 ... ... ... ... ... 5 ... ... 8

159 145 127 148 179 161 210 140 179 130 135 131 141

23 21 18.5 21.5 26 23 30.5 20 26 19 19.5 19 20.5

... 107 93 110 ... 101 ... 100 121 103 ... 103 121

MPa

HD HF HH (type I) HH (type II) HI HK HL HN HP HT HU HW HX

... ... ... ... 414 60 217 31.5 ... ... ... ... 417 60.5 222 32 ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... 292 42.5 193 28 ... ... ... ... ... ... ... ... 303 45 138 20

Fig. 30

Minimum creep rate versus temperature for alloy HP50WZ

MPa

ksi

Yield strength

Elongation, %

Alloy

ksi

Ultimate tensile strength

1095  C (2000  F) Ultimate tensile strength

Yield strength

ksi

Elongation, %

MPa

ksi

MPa

ksi

Elongation, %

... 15.5 13.5 16 ... 15 ... 14.5 17.5 15 ... 15 17.5

18 16 30 18 12 16 ... 37 27 24 20 ... 48

... ... ... 38 ... 39 ... 43 52 41 ... ... ...

... ... ... 5.5 ... 5.5 ... 6 7.5 6 ... ... ...

... ... ... ... ... 34 ... 34 43 ... ... ... ...

... ... ... ... ... 5 ... 5 6 ... ... ... ...

... ... ... ... ... 55 ... 55 69 ... ... ... ...

stressed

Fig. 32

Creep-rupture properties of alloy HK40. Scatter bands are þ20% of the central  tendency line. Although such a range usually encompasses data for similar alloy compositions, scatter of values may be much higher, especially at longer times and high temperatures.

Fig. 31

Rupture time versus stress and temperature for alloy HP50WZ

values are extrapolated or determined directly often has little bearing on the operating life of high-temperature parts. The actual material behavior is often difficult to predict accurately because of the complexity of the service stresses relative to the idealized, uniaxial loading conditions in the standardized tests and because of the attenuating factors such as cyclic loading, temperature fluctuations, or metal loss from corrosion. The designer should anticipate the synergistic effects of these variables. Thermal Fatigue Resistance. The design of components that are subject to considerable temperature cycling must also include consideration of thermal fatigue. This is particularly true if the temperature changes are frequent or rapid and nonuniform within or between casting sections. Fatigue is a condition in which failure results from alternating load applications in shorter times or at lower stresses than would

be expected from constant-load properties. Thermal fatigue denotes the conditions in which the stresses are primarily due to hindered thermal expansion or contraction. Good design helps minimize external restraints to expansion and contraction. Rapid heating and cooling may, however, impose temperature gradients within the part, causing the relatively cool elements of the component to restrain the hotter elements. Finite element analysis has shown that, for some industrial applications, these thermally induced stresses may exceed those resulting from mechanical loads. Nevertheless, such test results have been useful in considering alloy selection questions, identifying the superior thermal fatigue resistance of nickel-containing grades, and documenting the good performance of some HH-type compositions. Thermal Shock Resistance. Thermal shock failure may occur as a result of a single, rapid temperature change or as a result of rapid cyclic temperature changes, which induce stresses that are high enough to cause failure. Brittle ceramic materials are subject to such failure in a single temperature cycle, but only very

infrequently are conditions encountered that would cause failure in a single thermal cycle of the cast heat-resistant grades. The nickel-predominating grades are generally the most resistant to thermal shock. Resistance to Hot Gas Corrosion. The corrosion of heat-resistant alloys, that is, their attack by the environment at elevated temperatures, varies significantly with alloy type, temperature, velocity, and the nature of the specific environment to which the part is exposed. Table 16 presents a general ranking of the standard cast heat-resistant grades in various environments.

Ferrite in Cast Stainless Steels The CF alloys constitute the most technologically important and highest-tonnage segment of corrosion-resistant casting production. These 19Cr-9Ni alloys are the cast counterparts of the AISI 300-series wrought stainless steels (Table 2). In general, the cast and wrought alloys possess equivalent resistance to corrosive media, and they are frequently used in conjunction with each other. Important differences do exist, however, between the cast CF alloys and their wrought AISI counterparts. Most significant among these is the difference in alloy microstructure in the end-use condition. The CF-grade cast alloys have duplex structures (Table 2) and usually contain 5 to 40% ferrite, depending on the particular alloy. Their wrought counterparts are fully austenitic. The ferrite in cast stainless with duplex structures is magnetic, a point that is often confusing when cast stainless steels are compared to their wrought counterparts by checking their attraction to a magnet. This difference in microstructures is attributable to the fact that the chemical compositions of the cast and wrought alloys are not identical by intent. Significance of Ferrite. Ferrite is intentionally present in cast CF-grade stainless steels for three principal reasons: to provide strength, to improve weldability, and to maximize resistance to corrosion in specific environments. Strengthening in the cast CF-grade alloys is limited essentially to that which can be gained by incorporating ferrite into the austenitematrix phase. These alloys cannot be strengthened by thermal treatment, as with the cast martensitic alloys, nor by hot or cold working, as with the wrought austenitic alloys. Strengthening by carbide precipitation is also out of the question because of the detrimental effect of carbides on corrosion resistance in most aqueous environments. Thus, the alloys are effectively strengthened by balancing alloy composition to produce a duplex microstructure consisting of ferrite (up to 40% by volume) distributed in an austenite matrix. It has been shown that the incorporation of ferrite into 19Cr-9Ni cast steels improves yield and tensile strengths substantially without loss of ductility or impact toughness at temperatures below

Steel Castings Properties / 969 Table 16 Corrosion resistance of heat-resistant cast steels at 980  C (1800  F) in 100 h tests in various atmospheres Corrosion rating(a) in indicated atmosphere Alloy

Air

Oxidizing flue gas(b)

HA HC HD HE HF HH HI HK HL HN HP HT HU HW HX

U G G G S G G G G G G G G G G

U G G G G G G G G G G G G G G

Reducing flue gas(b)

Reducing flue gas(c)

Reducing flue gas (constant temperature)(d)

Reducing flue gas cooled to 150  C (300  F) every 12 h(d)

U G G G S G G G G G G G G G G

U S S ... U S S U S U G U U U S

U G G G S G G G G S G S S U G

U G G ... S G G G G S ... U U U U

(a) G, good (corrosion rate r < 1.27 mm/yr, or 50 mils/yr); S, satisfactory (r < 2.54 mm/yr, or 100 mils/yr); U, unsatisfactory (r > 2.54 mm/yr, or 100 mils/yr). (b) Contained 2 g of suifur/m3 (5 grains S/100 ft3). (c) Contained 120 g S/m3 (300 grains S/100 ft3). (d) Contained 40 g S/m3 (100 grains S/100 ft3)

resistance is impractical or impossible. In the case of SCC, the presence of ferrite pools in the austenite matrix is thought to block or make more difficult the propagation of cracks. In the case of intergranular corrosion, ferrite is helpful in sensitized castings that may become sensitized. The carbides precipitate at the discontinuous ferrite/austenite boundaries rather than at the continuous austenite/austenite grain boundaries, where they would increase susceptibility to intergranular attack. The presence of ferrite also places additional phase boundaries in the austenite matrix, and there is evidence that intergranular attack is arrested at austenite-ferrite boundaries. The most comprehensive study of the effect of ferrite on the corrosion resistance of cast stainless steels indicates that ferrite:

Fig. 33

Yield strength and tensile strength versus percentage of ferrite for CF8 and CF8M alloys. Curves are mean values for 277 heats of CF8 and 62 heats of CF8M.

425  C (800  F). The magnitude of this strengthening effect for CF8 and CF8M alloys at room temperature is illustrated in Fig. 33. Fully austenitic stainless steels are susceptible to a weldability problem known as hot cracking or microfissuring. The intergranular cracking occurs in the weld deposit and/or in the weld heat-affected zone and can be avoided if the composition of the filler metal is controlled to produce approximately 4% ferrite in the austenitic weld deposit. Duplex CF-grade alloy castings are immune to this problem. The presence of ferrite in duplex CF alloys improves resistance to SCC and generally to intergranular attack. Although failures of highalloy castings due to these two types of corrosion are not common, SCC and intergranular attack are concerns because they can occur unexpectedly, particularly in castings that have been sensitized by welding in the field, where postweld heat treatment to restore corrosion


Improves the resistance of CF alloys to chlo-

ride SCC
Improves resistance of these alloys to intergranular attack
Affords greater operating safety for the CF alloys with respect to both types of attack at ferrite contents exceeding 10% It is important to note, however, that not all studies have shown ferrite to be unconditionally beneficial to the general corrosion resistance of cast stainless steels. Whether or not corrosion resistance is improved by ferrite and to what degree depends on the specific alloy composition and heat treatment and the service conditions (environment and stress state). Ferrite Control. From the preceding discussion, it is apparent that controlled ferrite contents in predominantly austenitic cast chromium-nickel steels, notably, the CF alloys, offer certain property advantages and that the amount of ferrite present will depend primarily on the compositional balance of the alloy. The underlying causes for the dependence of ferrite content on composition are found in the phase equilibria for the Fe-Cr-Ni system. These phase

equilibria have been exhaustively documented and related to commercial stainless steels. The major elemental components of cast stainless steels are in competition to promote austenite or ferrite phases in the alloy microstructure. Chromium, silicon, molybdenum, and niobium promote the presence of ferrite in the alloy microstructure; nickel, carbon, nitrogen, and manganese promote the presence of austenite. By balancing the contents of ferrite-forming and austenite-forming elements within the specified ranges for the elements in a given alloy, it is possible to control the amount of ferrite present in the austenite matrix. The alloy can usually be made fully austenitic or with ferrite contents up to 30% or more in the austenite matrix. The relationship between composition and microstructure in cast stainless steels permits the foundry metallurgist to predict and control the ferrite content of an alloy, as well as its resultant properties, by adjusting the composition of the alloy. This is accomplished with the Schoefer constitution diagram for cast chromium-nickel alloys (Fig. 34). This diagram was derived from an earlier diagram developed by Schaeffler for stainless steel weld metal. Use of Fig. 34 requires that all ferrite-stabilizing elements in the composition be converted into chromium equivalents and that all austenite-stabilizing elements be converted into nickel equivalents by means of empirically derived coefficients representing the ferritizing or austenitizing power of each element. A composition ratio is then obtained from the total chromium equivalent, Cre, and nickel equivalent, Nie, calculated for the alloy composition according to the following: Cre ¼ %Cr þ 1:5ð%SiÞ þ 1:4ð%MoÞ þ %Nb  4:99 Nie ¼ %Ni þ 30ð%CÞ þ 0:5ð%MnÞ þ 26ð%N  0:02Þ þ 2:77

(Eq 1)

(Eq 2)

where the elemental concentrations are given in weight percent. Although similar expressions have been derived that take into account additional alloying elements and different compositional ranges in the Fe-Cr-Ni alloy system, use of the Schoefer diagram has become standard for estimating and controlling ferrite content in stainless steel castings. See ASTM A 800/A 800M for the standard practice of estimating ferrite content in certain ausyenitic Fe-Cr-Ni alloys. The Schoefer diagram possesses obvious utility for casting users and the foundry metallurgists. It is useful for estimating or predicting ferrite content if the alloy composition is known, and it is useful for setting nominal values for individual elements in calculating the furnace charge for an alloy in which a specified ferrite range is desired. Limits of Ferrite Control. Although ferrite content can be estimated and controlled on the basis of alloy composition only, there are limits to the accuracy with which this can be done. The reasons for this are many. First, there is an

970 / Steel Castings

Fig. 34

Schoefer diagram for estimating the ferrite content of steel castings in the composition range 16 to 26% Cr, 6 to 14% Ni, 4% Mo (max), 1% Nb (max), 0.2% C (max), 0.19% N (max), 2% Mn (max), and 2% Si (max). Dashed lines denote scatter bands caused by the uncertainty of the chemical analysis of individual elements. See text for equations used to calculate Cre and Nie.

Fig. 35

Schematic showing heating/cooling cycles used for various heat treatments

unavoidable degree of uncertainty in the chemical analysis of an alloy (note the scatter band in Fig. 34). Second, the ferrite content depends on thermal history in addition to composition, although to a lesser extent. Third, ferrite contents at different locations in individual castings can vary considerably, depending on section size, ferrite orientation, presence of alloying element segregation, and other factors. Measurements of ferrite content in stainless steel castings are also subject to significant limitations. Magnetic measurements of ferrite content are limited to small material volumes

and require simple casting geometries. In addition, careful calibration with primary and secondary standards is required for accuracy of measurement. Quantitative metallographic determinations of ferrite content on polished surfaces are essentially impossible to conduct in a nondestructive fashion with respect to the casting. The metallographic approach is also quite time-consuming, is limited by alloy etching characteristics and microscope resolution, and is complicated by the fact that it is a two-dimensional technique while ferrite pools and colonies in the alloy structure are three-dimensional. Both the foundry metallurgist and user of stainless steel castings should recognize that the aforementioned factors place significant limits on the degree to which ferrite content (either as ferrite number or ferrite percentage) can be specified and controlled in stainless steel castings. In general, the accuracy of ferrite measurement and the precision of ferrite control diminish as the ferrite number increases. As a working rule, it is suggested that  þ6 about the mean or desired ferrite percent be viewed as a limit of ferrite control under ordinary circumstances, with  þ3 possible under ideal circumstances.

Carbon and Low-Alloy Heat Treatment Heat treatment is an important step in the production of steel castings because it develops the mechanical properties of a hardenable steel. Several types of heat treatment are available. The essential elements of any heat treatment are the heating cycle and the cooling cycle. Figure 35 shows schematically a heating cycle and three different cooling cycles. The length of time that a casting is held at temperature and the cooling rate are important factors. The holding time should be long enough to complete the desired microstructural transformation. Annealing is practiced on low-carbon steels to provide a soft, readily machinable structure. Annealed castings have relatively low strength but good ductility. In a full anneal, the castings are heated above the upper critical temperature, held there long enough to complete the transformation to austenite, and then furnace cooled at a controlled rate to obtain a stress-relieved casting with a pearlite-ferrite structure that is ductile and readily machinable. There are variations of the annealing heat treatment for specialized purposes, for example, to achieve a spheroidized pearlite structure. The annealing temperatures of such specialized treatments may differ substantially from those of a full anneal. Full annealing and spheroidizing heat treatments are costly and should not be specified unless maximum ductility is actually required. Cooling is accomplished by reducing or by simply turning off the heat input to the furnace. When the castings have cooled to below the lower critical temperature (approximately 425



C, or 800  F), the transformation of austenite is usually complete. The castings can then be removed from the furnace and air cooled or even quenched. Furnace cooling to lower temperatures merely wastes furnace time and requires additional heat to bring the furnace up to temperature for the next load. Normalizing consists of heating the steel to a suitable temperature above the upper critical transformation temperature, holding long enough to complete the transformation to austenite, removing the work from the furnace, and cooling it in still air. Some castings are tempered after normalizing. The castings must be placed so that the air can circulate freely around every casting. If air flow is restricted, the operation will be more like annealing. On the other hand, accelerated cooling by fans or forced-air flow may produce a result more like quenching. The microstructure that results from normalizing is a mixture of ferrite and pearlite, associated with only low residual stresses and almost no distortion. Tensile strengths up to 655 MPa (95 ksi) can be obtained in this way. The normalizing and tempering operation is used to meet a number of standard specifications in this strength range. Because of the uniform structure obtained upon normalizing, machinability is good. The cost of normalizing makes this heat treatment attractive. It requires less furnace time than annealing, and its cooling cycle is less expensive than quenching. Quenching hardens the steel by heating it above the transition temperature (austenitized), as in annealing or normalizing (Fig. 35). The work is cooled more rapidly than in the other heat treatments—fast enough so that pearlite and ferrite do not have time to form. Water and oil are the media most commonly used for quenching steel castings. Water is used whenever possible. Higher-carbon steels require oil quenching. Some complicated shapes also demand oil quenching to minimize quench cracking. Oil quenches a little more slowly than water at all temperatures. At lower temperatures, the cooling curves for oil taper off; this means even slower rates than for water. Therefore, martensite forms more slowly in oil than in water. Certain organic chemicals can be added to water to produce a quenching solution that resembles oil in its heat-removal characteristics. The main advantage of these solutions is that they have the behavior of oil without the fire hazard. Their greatest disadvantage is that they coat the work and thus change the composition of the bath. The quench severity of these baths varies widely with small changes in composition. Tight control is necessary and sometimes difficult. The quenching of carbon steels is always followed by tempering in order to adjust the mechanical properties of the quenched steel. The higher tensile strength levels of carbon steels can be obtained only by quenching and tempering. The quenching-and-tempering

Steel Castings Properties / 971 operation produces the optimum combination of strength and toughness. Tempering is a heat treatment that follows quenching and sometimes normalizing. One purpose of tempering is to reduce the residual stresses that develop during cooling and transformation. Another objective of tempering is to modify the metallurgical structure of martensite and thus adjust strength and other mechanical properties to specified levels. Tempering consists of heating the work to a temperature below the transformation range, holding for a specified time, and finally cooling. Carbon steels are tempered in the range of 175 to 705  C (350 to 1300  F). The holding time at temperature may vary from 30 min to several hours. A longer time at a given tempering temperature or a higher tempering temperature for a given time produces greater tempering. Martensite softens more than pearlite at a given tempering temperature. Composition also affects the tempering response; carbide-forming elements cause the steel to exhibit greater resistance to tempering. Tempering below 595  C (1100  F) may cause temper embrittlement in certain steels. Tempering is usually not done in this temper embrittlement range, and when higher tempering temperatures are used, the work can be quenched from the tempering temperature to minimize time in the embrittling zone during cooling. Stress Relief. Because tempering is a stressrelief treatment, there is no need for a special stress relief after tempering. Stress relief is obtained by heating to temperatures above 260  C (500  F). A stress-relief treatment is sometimes required when operations are performed after heat treatment that leave residual stresses in the casting. Welding, induction hardening, and grinding are examples of such operations. The maximum temperature for stress relief is generally limited to 30  C (50  F) below the tempering temperature that had been used in heat treating the casting to prevent it from softening. Hydrogen Removal. Hydrogen has been found to cause low elongation and reduction in area in steel. If the steel contains 4 or 5 ppm of hydrogen, the ductility will be approximately 20% of that of a hydrogen-free steel. Hydrogen in steel is a mobile element. Above room temperature, the hydrogen will diffuse from the steel, and ductility will be restored. Hydrogen removal can be accelerated by heating to 205 to 315  C (400 to 600  F). This heat treatment is commonly referred to as aging. The aging time is proportional to section thickness. Generally, 25 mm (1 in.) equals 20 h. For heavy sections (250 mm, or 10 in.), hydrogen removal by aging becomes impractical because of time requirements.

High-Alloy Heat Treatment The heat treatment of stainless steel castings is very similar in purpose and procedure to the thermal processing of comparable wrought

materials. However, the differences in detail warrant separate consideration here. Martensitic alloys CA15 and CA40 do not require subcritical annealing to remove the effects of cold working. However, in workhardenable ferritic alloys, machining and grinding stresses are relieved at temperatures from approximately 260 to 540  C (500 to 1000  F). Casting stresses in the martensitic alloys noted previously should be relieved by subcritical annealing prior to further heat treatment. When these hardened martensitic castings are stress relieved, the stress-relieving temperature must be kept below the final tempering or aging temperature. Alloy CA6NM (UNS J91540) possesses better casting behavior and improved weldability; it equals or exceeds all of the mechanical, corrosion, and cavitation resistance properties of CA15; and it has largely replaced the older alloy. Both CA6NM and CA15 castings are normally supplied in the normalized condition at 955  C (1750  F) minimum and tempered at 595  C (1100  F) minimum. However, when it is necessary or desirable to anneal CA6NM castings, a temperature of 790 to 815  C (1450 to 1500  F) should be used. The alloy should be furnace cooled or otherwise slow cooled to 595  C (1100  F), after which it can be air cooled. When stress relieving is required, CA6NM can be heated to 620  C (1150  F) maximum, followed by slow cooling to prevent martensite formation. Homogenization. Alloy segregation and dendritic structures may occur in castings and may be particularly pronounced in heavy sections. Because castings are not subjected to the high-temperature mechanical reduction and soaking treatments involved in the mill processing of wrought alloys, it is frequently necessary to homogenize some alloys at temperatures above 1095  C (2000  F) to promote uniformity of chemical composition and microstructure. The full annealing of martensitic castings results in recrystallization and maximum softness, but it is less effective than homogenization in eliminating segregation. Homogenization is a common procedure in the heat treatment of precipitationhardening castings. Ferritic and Austenitic Alloys. The ferritic, austenitic, and mixed ferritic-austenitic alloys are not hardenable by heat treatment. They can be heat treated to improve their corrosion resistance and machining characteristics. The ferritic alloys CB30 and CC50 are annealed to relieve stresses and to reduce hardness by heating above 790  C (1450  F). The austenitic alloys achieve maximum resistance to intergranular corrosion by solution annealing. As-cast structures, or castings exposed to temperatures from 425 to 870  C (800 to 1600  F), may contain complex chromium carbides precipitated preferentially along grain boundaries in wholly austenitic alloys. This microstructure is susceptible to intergranular corrosion, especially in oxidizing solutions. (In partially ferritic alloys, carbides tend to

precipitate in the discontinuous ferrite side; thus, these alloys are less susceptible to intergranular attack than wholly austenitic alloys.) The purpose of solution annealing is to ensure the complete solution of carbides in the matrix and to retain these carbides in solid solution. Solution-annealing procedures for all austenitic alloys are similar and consist of heating to a temperature of approximately 1095  C (2000  F), holding for a time sufficient to accomplish complete solution of carbides, and quenching at a rate fast enough to prevent reprecipitation of the carbides—particularly while cooling through the range from 870 to 540  C (1600 to 1000  F). The temperatures to which castings should be heated prior to quenching vary somewhat, depending on the alloy. A two-step heat treating procedure can be applied to the niobium-containing CF8C alloy. The first treatment consists of solution annealing. This is followed by a stabilizing treatment at 870 to 925  C (1600 to 1700  F), which precipitates niobium carbides, prevents formation of the damaging chromium carbides, and provides maximum resistance to intergranular attack. Because of their low carbon contents, CF3 and CF3M as-cast do not contain enough chromium carbides to cause selective intergranular attack; therefore, these alloys can be used in some environments in this condition. However, for maximum corrosion resistance, these grades require solution annealing. Martensitic Alloys. Alloy CA6NM should be hardened by air cooling or oil quenching from a temperature of 1010 to 1065  C (1850 to 1950  F). Even though the carbon content of this alloy is lower than that of CA15, this fact in itself and the addition of molybdenum and nickel enable the alloy to harden completely without significant austenite retention when cooled as suggested. The choice of cooling medium is primarily determined by the maximum section size. Section sizes exceeding 125 mm (5 in.) will harden completely when cooled in air. Generally, alloy CA6NM is not susceptible to cracking during cooling from elevated temperatures. For this reason, no problem should arise in the air cooling or oil quenching of configurations that include thick as well as thin sections. A wide selection of mechanical properties is available through the choice of tempering temperatures. Alloy CA6NM is normally supplied, normalized, and tempered at 595 to 620  C (1100 to 1150  F). Reaustenitizing occurs upon tempering above 620  C (1150  F); the amount of reaustenitization increases with temperature. Depending on the amount of this transformation, cooling from such tempering temperatures may adversely affect both ductility and toughness through the transformation to untempered martensite. Even though the alloy is characterized by a decrease in impact strength when tempered in the range of 370 to 595  C (700 to 1100  F), the minimum reached is significantly higher than that of CA15. This improvement in impact

972 / Steel Castings toughness results from the presence of molybdenum and nickel in the composition and from the lower carbon content. The best combination of strength with toughness is obtained when the alloy is tempered above 510  C (950  F). The minor loss of toughness and ductility that does occur is associated with the lesser degree of tempering that takes place at the lower temperature and not with embrittlement, as may be the situation with other 12% Cr steels that contain no molybdenum. The addition of molybdenum to 12% Cr steels makes them unusually stable thermally and normally not susceptible to embrittlement in the annealed or annealed-and-cold-worked conditions, even when exposed for long periods of time at 370 to 480  C (700 to 900  F). The hardening procedures for CA15 castings are similar to those used for the comparable wrought alloy (type 410). Austenitizing consists of heating to 955 to 1010  C (1750 to 1850  F) and soaking for at least 30 min; the high side of this temperature range is normally employed. Parts are then cooled in air or quenched in oil. To reduce the probability of cracking in the brittle, untempered martensitic condition, tempering should take place immediately after quenching. Tempering is performed in two temperature ranges: up to 370  C (700  F) for maximum strength and corrosion resistance, and from 595 to 760  C (1100 to 1400  F) for improved ductility at lower strength levels. Tempering in the range of 370 to 595  C (700 to 1100  F) is normally avoided because of the resultant low impact strength. In the hardened-and-tempered condition, CA40 provides higher tensile strength and lower ductility than CA15 tempered at the same temperature. Both alloys can be annealed by cooling slowly from the range of 845 to 900  C (1550 to 1650  F). Precipitation-Hardening Alloys. It is desirable to subject precipitation-hardenable castings to a high-temperature homogenization treatment to reduce alloy segregation and to obtain more uniform response to subsequent heat treatment. Even investment castings that are slowly cooled from the pouring temperature exhibit more nearly uniform properties when they have been homogenized.

Corrosion-Resistant Applications Martensitic grades include CA15, CA40, CA15M, and CA6NM. Alloy CA15 contains the minimum amount of chromium necessary to make it essentially rustproof. It has good resistance to atmospheric corrosion as well as to many organic media in relatively mild service. Alloy CA40 is a higher-carbon modification of CA15 that can be heat treated to higher strength and hardness levels. A molybdenum-containing modification of CA15, alloy CA15M, provides improved elevated-temperature strength properties and improved low-temperature properties. Alloy CA6NM is an Fe-Cr-

Ni-Mo alloy of low carbon content. The presence of nickel offsets the ferritizing effect of the low carbon content so that strength and hardness properties are comparable to those of CA15 and impact strength is substantially improved. The molybdenum addition improves the resistance of the alloy in seawater. A wide range of mechanical properties can be obtained in the martensitic alloy group. Tensile strengths from 620 to 1520 MPa (90 to 220 ksi) and hardnesses as high as 500 HB can be obtained through heat treatment. The alloys have fair to good weldability and machinability if proper techniques are employed; CA40 is considered the poorest and CA6NM is the best in this regard. The martensitic alloys are used in pumps, compressors, valves, hydraulic turbines, propellers, and machinery components. Austenitic grades include alloys CH20, CK20, and CN7M. The CH20 and CK20 alloys are high-chromium, high-carbon, wholly austenitic compositions in which the chromium exceeds the nickel content. They have better resistance to dilute sulfuric acid than CF8 and have improved strength at elevated temperatures. These alloys are used for specialized applications in the chemical-processing and pulp and paper industries for handling pulp solutions and nitric acid. The high-nickel CN7M grade containing molybdenum and copper is widely used for handling hot sulfuric acid. This alloy also offers resistance to dilute hydrochloric acid and hot chloride solutions. It is used in steel mills as containers for nitric-hydrofluoric pickling solutions and in many industries for severe-service applications for which the high-chromium CFtype alloys are inadequate. The 6-Mo grades are used to resist marine corrosion and are used in water desalination facilities. Ferritic grades are designated CB30 and CC50. Alloy CB30 is practically nonhardenable by heat treatment. As this alloy is normally made, the balance among the elements in the composition results in a wholly ferritic structure similar to that of wrought type 442 stainless steel. By balancing the composition toward the low end of the chromium and the high ends of the nickel and carbon ranges, however, some martensite can be formed through heat treatment, and the properties of the alloy approach those of the hardenable wrought type 431. Alloy CB30 castings have greater resistance to most corrosives than the CA grades and are used for valve bodies and trim in general chemical production and food processing. Because of its low impact strength, however, CB30 has been supplanted in many applications by the higher-nickel-containing austenitic grades of the CF type. The high-chromium CC50 alloy has good resistance to oxidizing corrosives, mixed nitric and sulfuric acids, and alkaline liquors. It is used for castings in contact with acid mine waters and in nitrocellulose production. For best impact strength, the alloy is made with more than 2% Ni and more than 0.15% N. Austenitic-ferritic grades include CE30, CF3, CF3A, CF8, CF8A, CF20, CF3M,

CF3MA, CF8M, CF8C, CF16F, and CG8M. These alloys usually contain 5 to 40% ferrite, depending on the particular grade and the balance among the ferrite-promoting and austenite-promoting elements in the chemical composition. This ferrite content improves the weldability of the alloys and increases their mechanical strength and resistance to SCC. The amount of ferrite in a corrosion-resistant casting can be estimated from its composition by using the Schoefer diagram (Fig. 34) or from its response to magnetic measuring instruments. Alloy CE30 is a high-carbon, high-chromium alloy that has good resistance to sulfurous acid and can be used in the as-cast condition. It has been extensively used in the pulp and paper industry for castings and welded assemblies that cannot be effectively heat treated. A controlled ferrite grade, designated CE30A, is used in the petroleum industry for its high strength and resistance to SCC in polythionic acid. The CF alloys as a group constitute the major segment of corrosion-resistant casting production. When properly heat treated, the alloys are resistant to a great variety of corrosives and are usually considered the best general-purpose types. They have good castability, machinability, and weldability and are tough and strong at temperatures down to 255  C (425  F). Alloy CF8, the cast equivalent of type 304 stainless steel, can be viewed as the base grade, and all the others as variants of this basic type. The CF8 alloy has excellent resistance to nitric acid and all strongly oxidizing conditions. The higher-carbon CF20 grade is satisfactorily used for less corrosive service than that requiring CF8, and the low-carbon type CF3 is specifically designed for use where castings are to be welded without subsequent heat treatment. The molybdenum-containing grades CF8M and CF3M have improved resistance to reducing chemicals and are used to handle dilute sulfuric and acetic acids, paper mill liquors, and a wide variety of industrial corrosives. Alloy CF8M has become the most frequently used grade for corrosion-resistant pumps and valves because of its versatility in meeting many corrosive service demands. Because CF3M has a low carbon content, it can be used without heat treatment after welding. The niobium-stabilized CF8C alloy is the cast equivalent of type 347. Castings of this alloy, therefore, are used to resist the same corrosives as CF8 but where field welding or service temperatures of 650  C (1200  F) are involved. Higher mechanical properties are specified for grades CF3A, CF8A, and CF3MA than for the CF3, CF8, and CF3M alloys, because the compositions are balanced to provide a controlled amount of ferrite that will ensure the required strength. These alloys are being used in nuclear power plant equipment. The CF16F grade has an addition of selenium to improve the machinability of castings that require extensive drilling, threading, and the like. It is used in service similar to that for which CF20 is used. Type CG8M has a higher

Steel Castings Properties / 973 molybdenum content than CF8M and is preferred to the latter in service where improved resistance to sulfuric and sulfurous acid solutions and to the pitting action of halogen compounds is needed. Unlike CF8M, however, it is not suitable for use in nitric acid or other strongly oxidizing environments. Duplex alloys have higher yield strength than the austenitics. This difference gives the duplex alloys an economic edge in, for example, the chemical process industry; higher process flow rates and operating pressures are possible without a major equipment modification. Cost savings can also be realized when the higher strength allows the downgaging of the wall thickness of piping, heat exchanger tubing, tanks, columns, and pressure vessels. For rotating equipment, such as centrifuges, the mass of equipment can be reduced by using a duplex stainless steel. Further savings in motors and gearing results because of smaller loads. For some time, cast duplex valves and pumps have used the strength of these materials either to allow higher pressures or to lower costs by using thinner walls. Precipitation-hardening alloys include CB7Cu and CD4MCu. Alloy CB7Cu is a lowcarbon martensitic alloy that may contain minor amounts of retained austenite or ferrite. The corrosion resistance of CB7Cu lies between that of the CA types and the nonhardenable CF alloys, so it is used when both high strength and improved corrosion resistance are required. Castings of CB7Cu are machined in the solution-treated condition and then through hardened by a low-temperature aging treatment (480 to 595  C, or 900 to 1000  F). Because of this capability, the CB7Cu grade has found wide application in highly stressed, machined castings in the aircraft and food-processing industries. Type CD4MCu is a two-phase alloy with an austenite-ferrite structure that, because of its high chromium and low carbon contents, does not develop martensite when heat treated. Like the CB7Cu grade, CD4MCu can be hardened by a low-temperature aging treatment, but it is normally used in the solution-annealed condition. In this condition, its strength is double that of the CF grades, and its corrosion resistance is optimized. This alloy has corrosion resistance equal to, or better than, the CF types and has excellent resistance to SCC in chloride-containing media. It is highly resistant to sulfuric and nitric acids and is used for pumps, valves, and stressed components in the marine, chemical, textile, and pulp and paper industries, for which a combination of superior corrosion resistance and high strength is essential.

Heat-Resistant Applications Iron-chromium alloys include grades HA, HC, and HD. Alloy HA has limited application because of its low strength and limited resistance to gaseous corrosion at high temperature.

It has been used in valves, flanges, and fittings where light stresses are encountered. Alloys HC and HD can be used for loadbearing applications up to 650  C (1200  F) and where only light loads are involved up to 1040  C (1900  F). These grades, however, become embrittled by s-phase in the 650 to 870  C (1200 to 1600  F) temperature range. They are similar to each other in corrosion resistance, with HD exhibiting higher strength. Both are used in ore-roasting furnaces for such parts as rabble arms and blades, for salt pots and grate bars, and in high-sulfur applications in which high strength is not required. Iron-chromium-nickel alloys include HE, HF, HH, HI, HK, and HL. They are predominantly or completely austenitic and exhibit greater strength and ductility than the ironchromium alloys. Alloy HE has excellent corrosion resistance at high temperatures coupled with moderate strength. This combination of corrosion resistance and strength makes HE suitable for service to 1095  C (2000  F). Grade HE can be used in high-sulfur applications and is often found in ore-roasting and steel mill furnaces. The alloy is prone to s-phase formation, however, at temperatures of 650 to 870  C (1200 to 1600  F). Alloy HF is essentially immune to s-phase formation and can be used at temperatures to 870  C (1600  F). It is used for tube supports and beams in oil refinery heaters and in cement kilns, ore-roasting ovens, and heat treating furnaces. Grade HH exhibits high strength and excellent resistance to oxidation at temperatures to 1095  C (2000  F). Its composition can be balanced to yield a partially ferritic or a completely austenitic structure and a wide range of properties. Because of this, the composition of the HH alloy should be tailored to the application. The partially ferritic alloy, type I, has a somewhat lower creep strength and a higher ductility at elevated temperature than the wholly austenitic alloy type II. Type I is also more prone to s-phase formation between 650 and 870  C (1200 and 1600  F). Type II is preferred for application in this temperature range. Both types are widely used for furnace parts of many kinds but are not recommended for severe-temperature cycling service, such as that experienced by quenching fixtures. Grade HI is similar to the fully austenitic alloy HH, but its higher chromium content confers sufficient scaling resistance for use up to 1180  C (2150  F). Its major application has been in cast retorts for calcium and magnesium production. Alloy HK has high creep and rupture strengths and can be used in structural applications to 1150  C (2100  F). Its resistance to hot gas corrosion is excellent. It is often used for furnace rolls and parts as well as for steam reformer and ethylene pyrolysis tubing. Grade HL has properties similar to those of HK and exhibits the best resistance to corrosion in high-sulfur environments to 980  C (1800  F) of the alloys in this group. It is typically used in gas dissociation equipment.

Iron-nickel-chromium alloys include grades HN, HP, HT, HU, HW, and HX. These materials employ nickel as a predominant alloying (or base) element and remain austenitic throughout their temperature range of application. They are generally suitable for use to 1150  C (2100  F) and resist thermal fatigue and shock induced by severe temperature cycling. However, the nickel-base grades are not considered suitable for high-sulfur environments. Alloy HN has properties similar to those of HK. It is employed in brazing fixtures, furnace rolls, and parts. Grade HP is extremely resistant to oxidizing and carburizing atmospheres. It has good strength in the temperature range of 900 to 1095  C (1650 to 2000  F) and is often specified for heat treat fixtures, radiant tubes, and coils for ethylene pyrolysis heaters. Alloy HT can withstand oxidizing conditions to 1150  C (2100  F) and reducing conditions to 1095  C (2000  F). It is widely used to heat treat furnace parts subject to cyclic heating, such as rails, rolls, disks, chains, boxes, pots, and fixtures. It has also found application for glass rolls, enameling racks, and radiant tubes. Alloy HU has excellent resistance to hot gas corrosion and thermal fatigue, and it has good high-temperature strength. It is often used for severe applications, such as burner tubes, lead and cyanide pots, retorts, and furnace rolls. Grades HW and HX are extremely resistant to oxidation, thermal shock, and fatigue. Their high electrical resistivities make them suitable for the production of cast electrical heating elements. Both are highly resistant to carburization when in contact with tempering and cyaniding salts. The higher alloy content of HX confers better gas corrosion resistance, particularly in reducing gases containing sulfur, where HW is not recommended. Both grades are typically used for hearths, mufflers, retorts, trays, burner parts, enameling fixtures, quenching fixtures, and containers for molten lead.

ACKNOWLEDGMENT Parts of this article were adapted from “Steel Castings,” revised by Malcolm Blair, Properties and Selection: Irons, Steels, and High-Performance Alloys, Vol 1, ASM Handbook; John M. Svoboda, “Plain Carbon Steels,” “LowAlloy Steels,” and “High-Alloy Steels,” all from Castings, Vol 15, ASM Handbook; and Steel Castings Handbook, 6th ed., edited by Malcolm Blair and Thomas Stevens. REFERENCES 1. M. Blair, Corrosion of Cast Stainless Steels, Corrosion; Materials, Vol 13B, ASM Handbook, ASM International, 2005, p 78–87 2. J.D. Redmond, Selecting Second-Generation Duplex Stainless Steels, Chem. Eng., Oct 27 and Nov 24, 1986

974 / Steel Castings 3. Steel Castings Handbook, 6th ed., Steel Founders’ Society of America and ASM International, 1995 4. Steel Castings Handbook Supplement 2— Summary of Standard Specifications for Steel Castings, Steel Founders’ Society of America, 1999

5. Steel Castings Handbook Supplement 8—High Alloy Data Sheets—Corrosion Series, Steel Founders’ Society of America, 2004 6. Steel Castings Handbook Supplement 9— High Alloy Data Sheets—Heat Series, Steel Founders’ Society of America, 2004

7. R.I. Stephens, “Fatigue and Fracture Toughness of Five Carbon or Low Alloy Cast Steels at Room or Low Climatic Temperatures,” Research Report 94 (A and B), Steel Founders’ Society of America, 1982

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 975-986 DOI: 10.1361/asmhba0005330

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Selection and Evaluation of Steel Castings David Poweleit, Steel Founders Society of America

STEEL CASTINGS are used in safety-critical applications and in harsh, demanding environments to carry significant loads. They are commonly welded into a fabricated structure. Castings give users unlimited potential for steel geometry. Steel castings are used broadly in industrial equipment for end-use applications such as mining and construction. The railroad industry uses steel castings extensively for couplers, trucks, wheels, and corners. Fifth wheels for semitrucks are steel castings. The complex internal passages of pumps and valves are created through steel casting. Steel castings are used in energy production and are relied on for high-performance military applications. Steel castings are used in high-pressure service at nuclear power plants. Finally, steel castings serve many other ends where steel and geometry are important. New markets, such as building construction, are supported every day. The reason steel castings are attractive is their freedom of design. Any shape that can be imagined can be cast. Frequently, a casting cannot be made effectively, not because the design was too aggressive but because it was too timid. Often, a fabrication design that is inadequate is sent to the foundry to see if it can be made as a casting. This normally causes manufacturing problems, and changing the manufacturing process will not overcome inherent performance characteristics of a design. Cast parts are best manufactured when designed for the casting process. One reason that castings are seen as problems is because the foundry is often asked to make poor designs successful, and so the lead time is long and the cost is high. Good casting design allows weight to be reduced, cost to be lowered, and performance to be improved. Steel casting producers routinely test each heat of steel to make sure it meets the mechanical properties required in the material specification. The heat is also analyzed chemically to certify that it meets the standard. Other specialized tests can be required, such as low-temperature impact testing, when service performance requirements dictate. Steel castings are expensive sources of scrap steel if false economy is applied at the design stage. Good applications of steel castings are components with complex details that would

have many parts with high fabrication costs, poor material utilization, and reduced performance if assembled from multiple small wrought parts. Casting size and shape have limits based on section size or geometric limits. The flexibility of casting allows material to be placed where needed and material to be removed where it is not needed. Castings permit big, sweeping curves, nonuniform sections, and complex geometry that serve both functional and aesthetic needs. The properties of cast steel depend on the composition and heat treatment. Because designers often use yield strength as a basic property in design, material is often ordered to higher strength without considering the advantage in castings of using a lowerstrength material with optimum ductility and weldability. Since the load-carrying cross section can be increased to accommodate lower-strength material, the casting can be supplied in the highest ductility with strength levels that are compatible with the rolled structural shapes. This use of cast steel in its optimal condition ensures that the casting will perform safely and reliably and that excessive loads will cause failure to occur first in the rolled section familiar to the designer.

Design Cast steel is used in monstrous mining trucks as part of the chassis, because of durability and resistance to impact loads. Steel castings in truck frames have been leveraged to minimize stress concentrations, handle multidirectional loading, and relocate welds to lower-stressed regions. Steel is commonly thought of as a high-strength but high-weight material. While the density of steel is high, steel can offer equal strength-to-weight ratios compared to either titanium or aluminum, and at a fraction of the cost of titanium or with a superior modulus of elasticity compared to aluminum. Cast steel also offers wear resistance, corrosion resistance, heat resistance, and weldability. Materials such as maraging steels and modified martensitic precipitation-hardening steels can achieve these properties, and additional steel casting alloy development is always looking for means to improve mechanical performance. By taking advantage of creative geometry via casting,

the weight of a cast steel part can be further reduced by removing material where it is not needed on the part and using transitioning section thickness to have them sized for the load carried at each point. Details are found in the articles “Riser Design,” “Gating Design,” and “Shape Casting of Steel” in this Volume. Design Impact. Casting offers freedom of geometry, so casting part design can play a key role in the mechanical performance and function of the component. It is wise to consider the function and loads on the part independent of pre-existing part designs to take advantage of this design freedom. Sections of a cast part subject to higher stress can be enlarged, while lowstress regions can be reduced. This flexibility facilitates a part with optimum performance and reduces weight; both minimize cost. One way to ensure the best casting design is to work with a foundry on design. A foundry will have the technical expertise to partner on the casting, design, and materials selection. General Design Rules. There are several rules of thumb to the development of a good steel casting. First, reduce the number of isolated heavy sections and have smooth-flowing transitions. Smooth transitions between features promote laminar flow of the liquid metal into the casting and also improve the mechanical performance of the part by reducing stress raisers. An additional advantage of these smooth transitions is that designs for castability and for finite-element stress analysis often converge on the same optimal design. It is feasible to cast any geometry, but this may increase cost. Junctions within a casting should be designed to not add mass. Changing section thickness in a casting should be through smooth, easy transitions; adding taper and large radii helps. Reducing undercuts and internal geometry helps to minimize the cost of the pattern or die. The foundry and customer should agree on tolerances. Specifying as-cast tolerances is also important in minimizing cost. Postprocessing, such as machining, surface treatment, and how the part will be held in a fixture, also influences the final cost of the part. Datum points should be stated, and machine stock should be added to required locations. It is important to identify all datum targets in the initial design so that gates, risers, and other finishing operations do not affect

976 / Steel Castings the integrity of these. Draft is the amount of taper or the angle that must be allowed on all vertical faces of a casting tool to permit its removal from the mold without tearing the mold walls. Draft should be added to the design dimensions, but metal thickness must be maintained. The amount of draft recommended under normal conditions is approximately 1.5
. All of these rules can be reviewed with a foundry engineer. Good applications use the performance of steel with the flexible geometry of a casting. Steel castings offer high mechanical properties over a wide range of operating temperatures. Cast steel offers the mechanical properties of wrought steel and can be welded. Material Effects. The chemical composition and microstructure of a steel casting determine its mechanical properties. Heat treatment can change microstructure and provide a wide range of mechanical properties. The response to heat treatment for a given section is hardenability. A steel with a high hardenability will have uniform hardness in thicker sections than ones with low hardenability. In general, adding alloying elements increases cost and improves some properties but may reduce others. Most elements will increase the hardenability of steel. The effects of common alloying elements on steel properties are given in Table 1. When selecting a steel alloy, it is important to first know the required properties of the cast component. Carbon should be kept as low as possible to maximize weldability. Minimizing alloying elements to safely meet the performance requirements of the item will reduce cost. The foundry can provide assistance with materials selection to ensure the appropriate properties are considered when specifying the material, and they will know the availability and cost factors of the purchase. Design Considerations. Design requirements are typically determined in terms of strength or maximum stress. The design is commonly constrained by elastic modulus, fatigue, toughness, or ductility. Increasing the strength of steel normally reduces the ductility, toughness, and weldability. It is often more desirable in steel casting design to use a lower-strength grade and increase the section size or modify the Table 1

shape. However, if weight or size reduction is a factor to reduce operating cost over the life of a part, the initial higher-cost material may be attractive. The design freedom makes castings an attractive way to obtain the best fabrication and material performance and the needed component stiffness and strength. When designing a part, it is important to understand the design limits so that proper materials selection can be made. Stress, strain, fatigue, impact, wear, creep, and corrosion are common service conditions that can impose design limits. Stress results from a mechanical load carried by a component. The strength of a steel is the measure of its load-carrying ability. Loads can be applied and removed without deformation if they are small enough (elasticity). When a large enough load is applied, the material will stretch or compress and deform permanently (plasticity). Plastic deformation starts to occur when the yield strength is exceeded. The maximum load that can be applied before the material fractures is the ultimate tensile strength. Designers need to ensure that the part will not break or permanently deform. Thus, it is important to design a casting for stress levels below the yield strength or in the elastic region of the part. Typically ½ the yield strength is used for safety but 2/3 can be used with a thorough evaluation. Room-temperature properties are usually shown in tables, so the temperature dependence of the strength of steel must be considered for the intended design environment. The strength of steel depends on the composition and heat treatment. Steel is an iron-base alloy, mainly alloyed with carbon. Other alloying elements (Table 1) add strength but are also important in determining how effectively the steel grade will respond to heat treatment. Heat treatment rearranges the crystal structure of the iron and the distribution of carbon. Slow cooling rates produce coarse microstructures, which have lower strength. Cooling slowly in the furnace is called annealing and is not commonly used, except as an intermediate treatment to allow some grades to be machined. Cooling in still air is called normalizing and is the most common treatment to provide good strength and ductility. Rapidly cooling in water or oil is known as quenching. Steel must be reheated

Effects of alloying elements

Element

Carbon (C) Manganese (Mn) Silicon (Si) Nickel (Ni) Chromium (Cr) Molybdenum (Mo) Vanadium (V) Tungsten (W) Aluminum (Al) Titanium (Ti) Zirconium (Zi) Oxygen (O) Nitrogen (N) Hydrogen (H) Phosphorus (P) Sulfur (S)

Effect on Steel Properties

Increases strength but decreases toughness and weldability (most common and important) Similar, although lesser, effect as carbon Similar to carbon but with a lesser effect than manganese (important for castability) Improves toughness Improves oxidation and other corrosion resistance Improves hardenability and high-temperature strength Improves high-temperature strength Improves high-temperature strength Reduces the oxygen or nitrogen in the molten steel Reduces the oxygen or nitrogen in the molten steel Reduces the oxygen or nitrogen in the molten steel Negative effect by forming gas porosity Negative effect by forming gas porosity. Increases resistance to localized corrosion in stainless steels In high quantities, results in poor ductility Can increase strength but drastically reduces toughness and ductility Reduces toughness and ductility

or tempered after quenching to improve ductility. Quenching and tempering give the highest strength available from any grade. Varying the tempering temperature and time allows the production of a wide range of final strength levels with a quenched-and-tempered grade. Strain is the amount of elongation or compression in a loaded component. The increase in length per unit length is the elongation, and the change in area at the point of fracture of the bar is the reduction of area. The ability to absorb strain without fracture is critical to safety and reliability. Steel can be bent, twisted, compressed, or stretched without breaking. The amount of strain in a component will also determine its functionality. The elastic limit is the amount of stress that the material will withstand without permanent deformation. The yield strength is the amount of stress required to produce a small specified plastic deformation (often 0.2% in the United States). The ability to stretch or deform without cracking is called ductility. Brittleness is a characteristic of materials having limited plastic deformation; the yield strength and ultimate tensile strength are close. Ductility and brittleness are temperature-dependent. Low-alloy steels undergo a transition in character from ductile to brittle at low temperature. The nil ductility transition temperature is a measure of where brittle fracture occurs with no observable reduction of area. These properties and characteristics such as resilience and toughness are means of expressing material behavior to predict and judge performance. See the article “Steel Castings Properties” in this Volume. In general, increasing the strength of a steel grade reduces its ductility. While most designers think in terms of the material strength, most of the production of steel is in the lower-strength grades, which have good ductility. Heat treatment and carbon content influence strength and ductility, as shown in Fig. 1 and 2. Carbon contents are typically kept well below 0.30% to avoid problems with cracks in heat treating or welding. The ratio of stress to strain is the elastic modulus. Data are derived from tensile testing. The modulus of elasticity is based solely on the material composition; heat treatment does not affect modulus. The modulus of elasticity depends on the bonding energy between the atoms. Steels have a modulus of approximately 200 GPa (30  106 psi). Steels have the highest modulus of elasticity of commonly used materials. The larger the modulus, the smaller the deflection of a part; steel provides good stiffness. Fatigue is the failure of a component when it is repeatedly loaded, even at levels well below the yield strength of the steel. It is measured by repeated loading and unloading of several test bars at different stress levels and determining the number of cycles to failure. A typical result of stress versus cycles is shown in Fig. 3. Low-cycle fatigue is below 100,000 loadings, where ductility is needed. High-cycle fatigue is normally above 1 million loadings, and high strength is required. Smooth and notched samples are tested. The notched

Selection and Evaluation of Steel Castings / 977

Fig. 1

Room-temperature properties of cast low-alloy steels. QT, quenched and tempered; NT, normalized and tempered. Source: Ref 1

samples simulate surface irregularities in applications, both designed and accidental. Notched samples will fatigue at lower stress levels than unnotched, but cast samples had slightly higher notched fatigue strength than wrought alloys. Impact toughness measures the ability of steel to resist fracture or cracking during service. The ability of steel to resist cracking at low temperatures or during impact loading is known as fracture toughness. Toughness is measured by the amount of energy required to break the material at a certain temperature. A common test that measures toughness is the Charpy impact test. Like ductility, toughness tends to fall as the strength of the material increases. The reduction in toughness at higher carbon contents for two heat treatments is shown in Fig. 4. Wear occurs when materials rub against each other under load, and material is lost from the contacted surfaces. Gouging, abrasion, impact, cavitation, erosion, and corrosion must be considered in different types of applications. Typically, harder materials resist wear better. Since hardness increases with strength, higher-carbon, higher-strength steels are commonly used. It is important to retain adequate toughness at the high hardness levels to avoid cracking and premature failures. Creep occurs at elevated temperatures when the material permanently deforms at loads

Fig. 2

Stress-strain curves for selected steels showing influence of carbon content. Source: Ref 2

below the yield strength. Resisting creep requires complex alloys with relatively high carbon content. Generally, as the temperature or load increases, higher alloy contents are required for adequate performance. Creep rate is the rate of strain at a particular load, temperature, and time. The rate is not constant. Stress rupture is the time to failure under a given load at a particular temperature. Selection of a material for creep service must also take into account the oxidation or other high-temperature corrosion mechanisms that may limit the service life of the component. Corrosion is a chemical or electrochemical reaction between a material and its environment that produces a deterioration in the material or its function. It can be a general loss of material or a localized condition such as pitting, crevice corrosion, or selective attack. There are many standard laboratory tests for corrosion, but these rarely replicate service conditions or the nature of the environment. Materials evaluation should include a combination of laboratory testing, simulated service, and in-service monitoring. Corrosive conditions such as temperature, pH, oxygen concentration, chlorine and other chemicals, biological agents, and other variables

must be considered when selecting a material for service. Alloy Selection. The structure of wrought sections of steel is elongated in the direction of rolling, drawing, or forging. The strength and ductility are improved in the direction of elongation but reduced perpendicular to that direction (Fig. 5). The lack of directionality in steel castings gives them uniform properties in all directions. The cold forming of steel can also strengthen the steel but reduces ductility and toughness. Cast steel grades achieve the same trade-off of strength, ductility, and toughness by alloying and heat treatment. Therefore, casting grades with mechanical properties similar to wrought grades are designated differently (e.g., ASTM A 216 grade WCB is the cast counterpart to wrought 1020). Both ASTM A 915 (Table 2) and A 958 use grade names similar to their wrought counterparts; the grades may be heat treated to a variety of strength, ductility, and toughness levels. Selecting a steel alloy begins with understanding the design limits. Knowing the major mode of failure is key. Materials selection is part of the process of meeting the design

978 / Steel Castings

Fig. 5

Relation between mechanical properties of rolled steel slab and the inclination of the test speciment. Source: Ref 3

Fig. 3 Ref 3

Fatigue characteristic (S-N) curves for cast and wrought 8600-series steels quenched and tempered to the same hardness. Notched (Kt = 2.2) and unnotched specimens tested with R.R. Moore rotating-beam tests. Source:

Fig. 4

Room-temperature Charpy V-notch impact test values versus carbon content of cast steels in normalized-and-tempered condition. Tempering temperature: 650
C (1200
F)

criteria. Alloying elements can be added to improve performance; thus, starting with a carbon steel and building from there is the best practice. A material selection guide for five major design applications is shown in Fig. 6. This chart is meant to provide some initial guidance with reference to ASTM International standards designations. It is important to consult with a foundry to select the right material for each application. This is especially true of

higher-alloyed materials (outer rings in Fig. 6). Other alloys may be better, and the alloy and heat treatment can be tailored for specific conditions. The segments of the chart are discussed as follows. Structures (segment of the Fig. 6 guide) covers general applications governed by strength, deflection, and fatigue. Strength and ductility are still the main properties used to designate available steel grades. Increasing the strength of steel is easily achieved through more severe heat treatments or increases in the alloy content. The addition of alloying elements not only increases the strength that an alloy can achieve but also the section size of the part that can be effectively heat treated. The grades in Table 2 are common alloys of steel available for casting ASTM A 915 or A 958. The first column is the alloy designation, the second is the minimum requirements for the lowest-strength grade commonly available from that alloy, the third column is the minimum requirements for the highest-strength grade, and the last column is the calculated largest section that can be effectively heat treated through the section. It is apparent that higher-strength alloys have lower ductility but can be heat treated more effectively in larger sections. Higher-strength alloys also require more extensive weld procedures and may crack

in heat treatment, especially at high carbon contents. Fatigue resistance depends on the strength and ductility of the material, service conditions, corrosion, wear, residual stresses in the component, design (particularly in high-stress areas), and surface condition. The duty cycle, required service life, and analysis of the effects of failure of the component will determine the importance of fatigue resistance. Fatigue analysis is difficult, and component testing is not unusual to verify the design and part durability. Analytical tools such as computer modeling of service loads, along with the development of useful materials properties databases, reduce the number of design iterations. Properties such as strain-controlled cyclical properties, crack growth rates, integration of inspection standards, and life prediction improve designs by reducing the traditional requirements for factor of safety and allow more aggressive use of material. Castings allow the geometry to be tailored to the service requirements. Steels for structural applications can be found in ASTM A 27, A 148, A 747, A 915, and A 958. Pressure-containing applications have similar characteristics to structural applications. Specifications for pressure-containing applications have been developed to meet ASME Boiler and Pressure Vessel code. These steels are specified in ASTM standards and the related ASME code (SA): A 217/A 217M (SA-217/SA217M) for high-temperature service; A 487/A 487M (SA-487) and A 352/A 352M (SA-352) for low-temperature service; A 389 specially heat treated for high-temperature service; and

Selection and Evaluation of Steel Castings / 979

Fig. 6

Cast steel selection guide referencing ASTM International standards. Circle is segmented by primary application feature. Rings of greater diameter are more highly alloyed. Consult the standard for specific grades and classes of materials.

Table 2

Common cast steel properties

Grades similar to wrought grades Minimum low strength(a) ASTM A 915 and A 958 grade

SC1020 SC1030 SC1040 SC8620 SC8630 SC4130 SC4140 SC4330 SC4340

Tensile—yield

Minimum high strength

Median composition ideal critical diameter(b)

Tensile—yield

MPa

ksi

Elongation, %

MPa

ksi

Elongation, %

mm

in.

450–240 450–240 485–250 550–345 550–345 550–345 620–415 620–415 620–415

65–35 65–35 70–36 80–50 80–50 80–50 90–60 90–60 90–60

24 24 22 22 22 22 20 20 20

485–250 550–345 620–415 795–665 1035–930 1035–930 1105–1035 1450–1240 1450–1240

70–36 80–50 90–60 115–95 150–135 150–135 165–150 210–180 210–180

22 22 20 14 7 7 5 4 4

10 18 23 51 89 76 140 152 203

0.4 0.7 0.9 2.0 3.5 3.0 5.5 6.0 8.0

(a) The minimum tensile requirement of all grades by ASTM A 958 is 450–240 MPa (65–35 ksi). Values given are likely minimums. (b) The section size is not specified in either standard.

A 757/A 757M ferritic and martensitic for lowtemperature service. Similarly, there are European standards, such as BS EN 10213:2007 steel casting, for pressure purposes. Impact resistance, or toughness, is required when a part performs a safety-critical function, is subject to low temperatures, or is impact loaded in service. Impact toughness can be improved

through careful control of composition and heat treatment. Adding nickel is the common way to improve toughness. The toughness of all grades can be improved by lowering carbon, sulfur, and phosphorus and by using a quench-and-temper heat treatment. When toughness is needed, it should be verified by test. The steels specified for low-temperature service (ASTM A 352 and

A 757) require Charpy V-notch impact testing. Other grades can be tested as agreed upon by supplier and purchaser as a supplemental requirement as part of the purchase order or quality-assurance program. Always check specifications. In the case of corrosion-resistant cast steels for general applications, ASTM A 743/A 743M does not specify an impact strength, while ISO 11972 has ISO V-notch impact strength among the mechanical requirements, although verification is only necessary if required by the purchaser. High-temperature resistance and creep strength are required to carry loads at elevated temperatures. As the temperature increases, the required alloy content also increases. Commonly, chromium and molybdenum are added to the steel to improve elevated-temperature properties. Higher carbon content also helps. The preferred heat treatment of carbon and low-alloy steels is to normalize and temper. Weldable carbon steels for high-temperature service are found in ASTM A 216, and martensitic stainless and alloy steels for pressure and high-temperature service are in A 217. When the service temperatures exceed 650
C (1200
F), the alloyed steels are no longer adequate, and the cast heat-resistant grades containing high levels of chromium and nickel are used. These alloys are in ASTM A 297 and A 351. Wear (segment of Fig. 6) applies to mechanical wear and chemical corrosion. Mechanical wear and corrosion often exacerbate each other. The mechanism is called mechanically assisted corrosion or, in the case of fluids, erosioncorrosion. Severe wear or corrosion environments require high-alloy steels. Wear resistance is usually improved through the use of highhardness materials. Strength and hardness are related, so high-strength materials are commonly used when wear is a problem. Increasing carbon content also increases wear resistance. Special materials such as austenitic manganese alloys or high-chromium irons are used to give better wear resistance. Toughness must be adequate to avoid premature catastrophic failure. High-chromium irons offer good wear resistance in abrasion or even in corrosive environments. When impact loading is a part of the wear environment, austenitic manganese alloys work harden, allowing them to resist wear while maintaining high toughness. Wear materials are found in ASTM A 128, A 351, A 532, A 743, A 744, A 890, and A 494. Carbon or alloy steels may be used in less severe environments. Severe environments, such as saltwater and chemical processing, require high-alloy stainless steels or nickel-base alloys. Generally, higher chromium and molybdenum are needed as the environment becomes more severe. The proper selection of corrosion-resistant materials depends on the thorough knowledge of the end-use environment, and factors are often industry-specific; thus, a metallurgist or corrosion expert should be consulted. For more about corrosion and wear, see the Handbooks listed in the Selected References in this article.

980 / Steel Castings With cast steel alloys, one can increase various performance characteristics, such as corrosion resistance and wear resistance, through alloying and heat treatment. Mechanical properties such as strength and elongation can be adjusted. There are three keys to selecting the right steel casting alloy for optimized performance and cost:  Design the geometry of the steel casting to

carry the load uniformly.

 Start with carbon steel for most applications;

modify the heat treatment, and then add alloying elements to attain the required properties.  Know the design limits, normal and extreme conditions, design life, tolerances, and quality requirements for an application, and work with a foundry to design the part, the process, and select a material to exceed the requirements on a statistically acceptable basis that will assure productivity. Drawing Notes. Material should be identified as conforming to a specific specification, with the grade and/or class stipulated. Nondestructive examination, verification testing, inspection method, and criteria for acceptance should be called out in accordance with a national or industry standard or, in some cases, an internal specification. Porosity can be quantified as a maximum size per discontinuity, minimum separation between discontinuities, maximum number of discontinuities per area, and minimum size to count as a discontinuity; these are stipulated in the acceptance. Acceptance can also be based on some form of comparison to a standard. In-process welding is common to remove discontinuities. Generally, the weld would meet the same requirements as the cast material. Type and extent of reheat treatment after welding should be stated. Allowable draft, marking requirements, surface-quality referencing a comparator plate, allowance for projections where the gating system was cut off, incorporation of fillets and radii, and a comment to remove all burrs are common notes for steel casting drawings. Zoning a casting to have more stringent levels of examination in critical areas is optimal for minimizing cost while ensuring performance. Zoning can also be used in limiting weld repair or draft on the part. Most purchase specifications state that requirements for material and testing shall be done based on agreement of the purchaser and supplier. Alloy Designation. Cast steels do not follow the conventional AISI grade designation for wrought steels. Steel castings may have different alloy composition and heat treatment compared to a wrought counterpart. This is to provide similar mechanical properties and manufacturing characteristics while balancing the manufacturability of the steel as a casting in a mold. For instance, cast corrosion-resistant grades have 19% Cr and 9% Ni versus their wrought 300-series counterparts that have 18% Cr and 8% Ni. The wrought microstructure for 300-series stainless is fully austenitic, which makes the material nonmagnetic. The cast

microstructure has ferrite in the austenite; thus, it is actually magnetic. The ferrite provides strength, improves weldability, and maximizes corrosion resistance. The cast and wrought materials have similar resistance to a corrosive medium and can be used with one another. Several ASTM specifications, such as A 958, specify cast grades similar to wrought grades as SCxxxx, where xxxx is the wrought grade, so that designers can quickly reference a cast counterpart to the wrought steel. High-alloy steel grades are based on the Alloy Casting Institute system and the composition of the material. Grades start with either a “C” or an “H” for corrosion-resistant and heat-resistant grades, respectively. The second letter represents the nominal chromium-nickel content (Fig. 7), with later letters representing a higher percentage of nickel in the alloy. After the two letters, a number is provided to represent the carbon content. In corrosion-resistant alloys, the number is the maximum 1/100% carbon. In heat-resistant alloys, the number is the 0.1% midpoint of the range for carbon. Additional letters are then used to represent other major alloying constituents. An example of a corrosion-resistant grade would be CF8M. Breaking this down, the “C” would mean corrosion resistant, the “F” a nominal 19% Cr and 9% Ni, the “8” a maximum of 0.08% C, and the “M” that molybdenum is alloyed into this material. An example of a heat-resistant grade would be HK40. Breaking this down, the “H” would mean heat resistant, the “K” a nominal 26% Cr and 10% Ni, and the “40” a carbon ranging from 0.35 to 0.45%. Note that HY80 and HY100 from MIL-S-23008 are actually high-yield steel castings versus heatresistant grades.

Manufacturing Steel castings are made through the same casting processes used for other materials. However, due to the higher liquidus temperature of steel compared to some nonferrous melts and the mode of solidification, there are several attributes that are specific to the manufacturing of steel castings. Minimum Section Thickness. The rigidity of a section often governs the minimum thickness to which a section can be designed. There are cases, however, when a very thin section will suffice mechanically, and castability becomes the governing factor. In these cases, it is necessary that a limit of minimum section thickness be adopted in order for the liquid steel to completely fill the mold cavity. Liquid steel cools rapidly as it enters a mold. A minimum thickness of 6 mm (0.25 in.) is suggested for design use when conventional steel casting techniques are employed. Wall thickness of 1.5 mm (0.060 in.) and sections tapering down to 0.76 mm (0.030 in.) are common for investment castings. Draft is the amount of taper or the angle that must be allowed on all vertical faces of a pattern

Fig. 7

Chromium and nickel Institute standard corrosion-resistant cast steels. second letter in the designation.

content in Alloy Casting grades of heat-and These letters are the See text for details.

to permit its removal from the mold without tearing the mold walls or causing the tool to wear. Draft should be added to the design dimensions, but metal thickness must be maintained. Regardless of the type of pattern equipment used, draft must be considered in all casting designs. Draft can be eliminated by the use of cores; however, this adds significant costs. In cases where the amount of draft may affect the subsequent use of the casting, the drawing should specify whether this draft is to be added to or subtracted from the casting dimensions as given. The necessary amount of draft depends on the size of the casting, the method of production, and whether molding is by hand or machine. Machine molding will require a minimum amount of draft. Interior surfaces in green sand molding usually require more draft than exterior surfaces. The amount of draft recommended under normal conditions can range from 1 to 2
for sand casting but can be near 0
for investment casting. The draft allowance would normally be added to design dimensions. Mold features such as the parting line and cores are covered in full detail in other articles in this volume. Pouring. Cast steel is typically poured into the mold at a temperature of slightly less than 1650
C (3000
F). This higher pouring temperature limits the mold media that liquid steel can be poured into. It also means that steel needs to be melted with either an electric arc furnace or an induction furnace. An argon-oxygen decarburization vessel can be used when melting steel to lower sulfur content or increase the recovery of chromium; thus, it is sometimes used in conjunction with heavy-walled castings from carbon and low-alloy steels or with highalloy castings. The injection of reactive metals (calcium, magnesium, or rare earths) into the liquid steel is another means of lowering sulfur content. Carbon and low-alloy steels are deoxidized by adding aluminum, titanium, or zirconium. Of these, aluminum is used more frequently because of its effectiveness and low cost. Given that liquid steel has a greater affinity to react with the oxygen in the air, it is important to prevent turbulent flow or entrainment in the stream during pouring. The high temperature of liquid steel also makes venting of the

Selection and Evaluation of Steel Castings / 981 mold valuable, so that gases can escape as the liquid metal fills the cavity. Inclusion Shape Control. Nonmetallic inclusions that form during solidification depend on the oxygen and sulfur contents of the casting. Deoxidation decreases the amount of nonmetallic inclusions by eliminating the oxides, and it affects the shape of the sulfide inclusions. High levels of oxygen (>0.012% in the metal) form FeO, which decreases the solubility of manganese sulfides and causes the manganese sulfides to freeze early in solidification as globules. The use of silicon deoxidation alone normally causes the formation of globular (type I) sulfides. Decreasing the oxygen to between 0.008 and 0.012% in the metal, through the use of aluminum, titanium, or zirconium, increases the solubility of the manganese sulfides so that they solidify last as grain-boundary films. These grain-boundary sulfide films (type II inclusions) are detrimental to the ductility and toughness of the steel and cause increased susceptibility to hot tearing. If the oxygen content is very low (<0.008% in the metal), the deoxidizer forms complex sulfides that are crystalline and form early in solidification. These type III inclusions are less harmful than type II but more harmful than type I. The actual sulfide shape depends on the type and amount of deoxidizer. These are complex interactions, so care must be taken to avoid porosity and type II sulfides. If a high level of rare earth (RE) metals (RE/S > 1.5) is added, then galaxies of sulfides (type IV inclusions) form; these are only for rare earth additions. The most common inclusion modifier is calcium. Directional Solidification. Steel castings begin to solidify at the mold wall, forming a continuously thickening envelope as heat is dissipated through the mold-metal interface. The volumetric contraction that occurs within a cross section of a solidifying cast member must be compensated by liquid feed metal from an adjoining heavier section or from a riser that serves as a feed metal reservoir and is placed adjacent to, or on top of, the heavier section. The lack of sufficient feed metal to compensate for volumetric contraction at the time of solidification is the cause of shrinkage cavities. They are found in sections that, due to design, must be fed through thinner sections. The thinner sections solidify too quickly to permit liquid feed metal to pass from the riser to the thicker sections. Note that even though voids may lack material, they may not prevent the casting from achieving the final performance characteristics. Developing the means to get the liquid metal into a casting and provide enough feed metal is generally the responsibility of the foundry engineer. The size and location of these features can be analyzed through solidification and fill simulation software. To encourage directional solidification, castings should have well-positioned gates, gradually changing cross sections, and smooth-flowing geometry.

Steel Casting Processes. Steel castings are largely manufactured as sand or investment castings but can also be done with centrifugal casting and other specialized processes. Sand castings can be made via green sand or chemically bonded sand, which would include the shell casting process. Sand-mold-making processes can be automated to make a higher-production volume of parts. Sand typically produces a rougher surface (200 to 300 rms) and requires more open tolerances (þ þ1.5 mm, or 0.030 and 0.75 and  0.060 in., when over 500 mm, or 20 in.). Through the pit molding process, sand molding is capable of producing parts that would easily fit in one’s hand to parts larger than a car. Investment casting typically produces a smoother surface (10 to 80 rms), has fairly tight tolerances (þ þ0.075 0.025 mm for the first 25 mm and  mm for each additional 25 mm, or 0.0010 in. for the first 1 in. and 0.003 in. for each additional inch), and can produce parts over 50 L (2 ft3) in volume. Tooling for sand casting patterns was traditionally made from wood. This has largely been replaced by a composite material. Tooling for investment casting is a metal die for making the investment waxes. Both sand and investment casting processes for steel can take full advantage of the rapid tooling technologies that are being developed and used. Rapid prototyping techniques, such as stereolithography, selective laser sintering, laminated object manufacturing, and three-dimensional printing, can be leveraged to make short-run or prototype steel castings. Heat Treatment. Some components can be used as-cast, such as those made out of lowcarbon stainless steels, but the majority of steel castings are heat treated. Heat treating is employed to at least stress relieve and normalize the casting, since the metal is often slowly cooled within the mold. Thin sections of investment castings will experience rapid cooling. Differential cooling rates within a component will impart varying mechanical properties that may need to be addressed by stress relieving and normalizing. Heat treatment can modify the mechanical properties of the part, but this usually means a trade-off. For instance, strength and hardness can be improved, but ductility, toughness, and weldability will be degraded. Cast steel responds very well to heat treatment. Weldability. Most steels are readily weldable, and it is routine practice to weld castings in production. Therefore, it is common in many specifications to set weld practices between the customer and the supplier. Carbon and Low-Alloy Steels. The weldability of carbon and low-alloy steels is primarily a function of composition and heat treatment. Carbon steels having low manganese and silicon contents and a carbon content below 0.30% can be welded without any special precautions. When the carbon content exceeds 0.30%, preheating of the casting prior to welding may be advisable. The low-temperature preheat (120 to 205
C, or 250 to 400
F) reduces the rate at which heat is

extracted from the heat-affected zone (HAZ) adjacent to the weld. Preheating also helps to relieve mechanical stresses and to prevent underbead cracking, because hydrogen is still relatively mobile and can diffuse away from the last areas to undergo a metallurgical transformation. Preheating minimizes the chances of a hardening transformation occurring in the HAZ and thus reduces the hardness adjacent to the weld. The metallurgical notch that could result from the hardened area can be eliminated in this manner. Generally speaking, if the hardness of the HAZ after welding does not exceed 35 HRC or 327 HB, preheating of the casting for welding is not required. As additional alloying elements are added to the steel, the need for preheating increases. Most of the low-alloy steels, such as the SC8630, SC8730, or SC4130 steels, require some preheating. When properly preheated, such steels are readily welded. A comparison can be made between the weldability of two steels by comparing their carbon equivalents (CE). One such calculation is: CE ¼ %C þ %Mn=6 þ ð%Cr þ %Mo þ %VÞ=5 þ ð%Cu þ %NiÞ=15

Another includes silicon: CE ¼ %C þ ð%Mn þ %SiÞ=6 þ ð%Ni þ %CuÞ=15 þ ð%Cr þ %Mo þ %VÞ=5

Steels having the same CE will have approximately the same weldability and will require similar preheating and other precautions. Other factors (section thickness) will also affect weldability. The alloy steels of higher hardenability, for example, the cast 4300 series, require more care and must be preheated in the range of 205 to 315
C (400 to 600
F). In the case of some low-alloy steels, especially those subject to toughness requirements, the interpass temperature (maximum or minimum temperature of the weld bead before additional beads are laid) must be maintained at a prescribed low level to prevent embrittlement of the parent metal. High-Alloy Steels. Most of the high-alloy steels are readily weldable, especially if their microstructures contain small percentages of d-ferrite. Grades of stainless can become sensitized and lose their corrosion resistance if subjected to temperatures above 425
C (800
F). Great care must therefore be used in welding to be certain that the casting or fabricated component is not heated excessively; thus, a waterjet sprayer may be used between welds to maintain the interpass temperature, and stainless steels are almost never preheated. Control of heat input and interpass temperature would also apply to heat-resistant steels. Complete reheat treatment may be required after welding in stress-corrosion cracking service performance conditions. Heating the casting above 1065
C (1950
F) and then cooling it

982 / Steel Castings rapidly redissolves the carbides precipitated during the welding operation and restores corrosion resistance. If solution annealing cannot be used, alloying elements such as titanium and niobium can be added to form stable carbides. Another option is to lower the carbon content. As the alloy content of the high-alloy grades is increased, welding without cracking becomes more difficult, due to the tendency of forming microfissures adjacent to the weld. This tendency toward microfissuring increases as nickel and silicon contents increase and carbon content decreases. The microfissuring is reduced by extremely low sulfur contents. In welding these grades, low interpass temperatures, low heat inputs, and peening of the weld to relieve mechanical stresses are effective. Postprocessing of some sort is typically required on all steel castings. Machining, plating or coating, inspection, assembly, and packaging are common practices. Additional activities such as straightening can be included when driven by the design application. Cast steel is moderately machinable and would be similar to that of wrought steels. Cast steel is readily weldable. Coatings can be applied to cast steels to improve wear or corrosion protection, as one would do with wrought steels. The most common coating on steel castings is paint, and surface preparation is an integral part of the process.

Nondestructive Examination Nondestructive examination (NDE) is a valuable tool for ensuring manufacturing quality and part performance in steel castings. The existing quality-assurance methods are workmanship standards. The current design standards often require large factors of safety and significant testing of a part to failure, which is expensive. The inspection level is set conservatively to ensure no failures in the field. Understanding the advantages and limitations of NDE methods is important. Research has shown that simulation is a good way to predict part quality and know how the part will perform. Simulation also has the advantage of minimizing testing costs and reducing development time. Similar to welding procedures that rely on the same principles of metal solidification as castings, casting procedures call out testing requirements to ensure mechanical performance. First-article tests typically call out x-ray and magnetic particle inspection to ensure the quality level of the casting. Therefore, ensuring the quality and performance of a casting is very similar to that of a weldment. Since the same NDE practices are used for all metal manufacturing techniques, the challenges with repeatability and reliability are applicable to every process. Traditionally, nondestructive testing has been used to certify casting quality. Soundness is verified through the use of radiographic inspection. Surface quality is evaluated using magnetic particle inspection. More recently, the use of

computer simulation of the casting solidification, integrated with finite-element analysis of its performance, has been used to design optimal casting configurations. The development of these tools allows the designer to ensure that critical areas of the part meet requirements while ensuring the most economical means of manufacturing the whole part. Quality is a primary concern for casting designers and users. The casting process is more poorly understood and variable than many alternate manufacturing methods. Solidification introduces features in manufacturing that may limit component performance. Even with these limits on understanding and confidence, castings are used in the most demanding and critical applications. From the production of prosthetic hip and knee replacements, to single-crystal jet engine turbine blades, to crane hooks for lifting off-shore platforms, castings are used in the most difficult and demanding applications. From the producers’ point of view, many designers and customers are designing and specifying castings incorrectly due to fundamental misconceptions about castings and their performance. For steel castings, many of these misunderstandings can be resolved by relying on industry standards and practices that achieve high performance with high reliability. As designers and customers better grasp casting process technology and specification, better cast components can be developed. According to ASTM E 1316–2005, “Standard Terminology for Nondestructive Testing,” components have defects only when they fail to meet the specification requirements. No matter how large a crack in a part may be, from a specification and technical point of view, it is not considered a defect unless an inspection for cracks is specified and the crack exceeds the required acceptance criterion. No real part has perfectly flat surfaces, perfectly square corners, or perfectly sized and located features. The amount of deviation from the ideal that is tolerated is expressed in the tolerances for geometric design. All real parts have mechanical properties that are measurably different than the typical or specified minimum values for the grade. A conservative design approach ensures adequate performance. All real parts have imperfections and discontinuities. Material production processes deliver materials that contain nonuniform properties and other variations. Forming the part introduces other imperfections and discontinuities. Testing is used to ensure that the part will meet the performance requirements demanded. For each critical performance attribute to be evaluated, a test method should be identified and an acceptance criterion established. A part contains defects only when the part fails to meet the specified requirements. Much of the common discussion of defects in castings treats ordinary phenomena of solidification, such as shrinkage, as a defect. Shrinkage is only a defect if soundness is specified and the

shrinkage exceeds the imposed acceptance requirement, or if a functional requirement such as a pressure test is specified and not met. This inexact terminology confuses the user of castings by implying that castings are full of defects when, in fact, proper design and specification can easily produce a defect-free casting. Defect-free does not mean perfect, whatever perfect would mean. This situation is made worse by poor failure analysis, where the cause of failure is reported as the part feature that initiated the final fracture. Most failures in new designs are due to inadequate design, not poor-quality part production. Most field failures are due to product misuse and not poor-quality component manufacture. When a part is subjected to unsurvivable loads, it will fail through the most heavily loaded section, and the failure will initiate at the largest performance-limiting feature in the heavily loaded section. In both cases, it is possible to find the fracture-initiating feature, but it is fundamentally incorrect to report that it caused the failure and was therefore a defect. A discontinuity or NDE indication of voids does not necessarily cause a part to fail. If the discontinuity is less than the critical size, has a certain orientation, or is in a location of low stress, the feature will not affect the suitability for service of the part. It is helpful to recognize that all real parts have performance limitations of a finite design life. The performance-limiting features of the part may be a material property, physical property, or even a geometric feature. All manufacturing processes can limit part performance. The challenge is to use the characteristics of the manufacturing process to enhance the performance of cleverly designed parts. Quality is not freedom from defects but desired and predictable performance. This is not limited to the fitness for service but may also include aesthetics or manufacturability. Testing of parts can be either destructive or nondestructive. For example, a part can be loaded to failure, giving a measurement of its actual load-carrying capacity, but after testing, the part is no longer usable. Nondestructive testing evaluates parts without destroying their ability to be used. Since the term test often implies the determination of a property value of a specimen of material, many prefer not to use it for nondestructive methods. While the title of ASTM Annual Volume 3.03 uses the term nondestructive testing, the standard indicates that nondestructive examination, nondestructive evaluation, and nondestructive inspection are equivalent. In the standard, under the definition of test and inspection, it is indicated that examination is the preferred term. This article follows that practice, calling the techniques nondestructive examination. Unfortunately for designers and users as well as the producer, NDE standards are workmanship standards. Workmanship standards specify a level of acceptance based on the general capability of the process and not on the performance of the part. They are set not on an engineering

Selection and Evaluation of Steel Castings / 983 evaluation of the impact on part performance but on the subjective judgment of standards writers. Acceptance levels are not chosen based on application requirements but are arbitrary. The inherent subjectivity of NDE standards limits their usefulness. Visual Inspection. Casting specifications may contain ambiguous wording in regard to visual inspection, such as, “Castings are to be clean and free from injurious defects.” There is no definition for “defect” and no basis for judgment as to what was “injurious.” If the purchaser said, “A surface condition was injurious; it had to be removed and welded,” this would also be problematic. Or, if the “injurious defect” had to be “completely removed to sound metal,” there is no definition for “sound metal” and no basis for judging “completely.” Requirements of this type can be easily misunderstood and misapplied; they can cause no end of grief for the designer, user, and producer. ASTM A 703, A 781, A 957, and A 985 specifications have replaced such ambiguous wording with the requirement, “The surface of the casting shall be examined visually and shall be free of adhering sand, scale, cracks, and hot tears.” This preferred wording goes on to clarify the extent of removal required: “Unacceptable visual surface discontinuities shall be removed and their removal verified by visual examination of the resultant cavities.” Unfortunately, visual standards are subjective, and producers commonly overfinish the casting by grinding and welding without improving the surface or adding value for the customer. The designer and user can specify any surface they like, even for aesthetic reasons. Frequently, slight changes in the casting process alter the texture of the surface, and this change triggers concern for the user. Recent studies show that visual inspection in the producer’s operations is costly and variable. In a repeatability and reproducibility study, two small castings were twice given to two inspectors. After the castings were marked for additional grinding and welding, they were cleaned and returned for a second inspection. After each inspection, the marks were covered by white dots, and the dots counted. The same casting could have 45 dots of work or 0, depending on the visual inspection. In one case, an engineer, to avoid any unnecessary finishing, walked a single large casting through the process, and that casting had one-third the normal grinding and welding. In a similar study, the time spent grinding the castings was evaluated. Typically, 80% of the grinding was not required to process the part but to meet the surface-quality requirements. Unnecessary finishing increases cost and stretches out production times. Traditional measures of surface roughness are not normally useful in casting surface inspection, since the texture of the casting is more severe and long range than machined surfaces. New surface inspection techniques using laser topography or other optical means may be useful for visual inspection. It may be possible

to use these data-intensive surface measurements to develop a less subjective standard. Fundamentally, it is not clear what role surface roughness may play in performance. Current industry best practices are to clearly communicate the need and expectation for surface finish. Avoid any finishing or inspection of surfaces to be machined. Do not do any finishing on surfaces used as datum targets. Accept the normal shot-blasted cast surface unless otherwise required. In highly loaded areas of the casting, where bending and fatigue limit performance capability, more stringent standards may be required. In this case, it is common to apply magnetic particle or liquid penetrant inspection techniques. Magnetic particle inspection is used to detect surface features that disrupt the magnetic field in a magnetized part. Nonmagnetic materials cannot be inspected with this technique. For steel castings, magnetic particle inspection is a valuable way of identifying cracks that are so tight or fine that they escape visual inspection. It also picks up surface and slightly subsurface inclusions and porosity that typically are shallow and have little effect on performance. Magnetic particle methods for dry powder and wet inspection are set forth in ASTM E 109 and E 138. Acceptance criteria can be selected from a set of reference photographs in ASTM E 125 that depict the appearance of casting surface conditions revealed by the dry powder magnetic particle technique. Different types of discontinuities do not have equal effects on performance, and an effort should be made to assign different acceptance levels to areas of the casting, based on the stresses to which each area is subjected in service. Purchasers can specify ASTM A 903. Instead of pictures and subjective comparisons, this standard sets out dimensions and frequencies of indications for acceptance criteria when using magnetic particle and liquid penetrant inspections. A misuse of the standard is to tighten the acceptance criteria on a casting when inclusions or porosity are found on a machined surface. Since the inspection is unable to see below the surface, it does not detect porosity or inclusions at the machined surface. Worse, the tighter standard requires the producer to grind and weld a surface that will be removed by machining. This adds cost and delays production. When a producer is responsible for machining and finds a problem at the machined surface, no magnetic particle inspection is imposed. No NDE technique can routinely detect this undesirable condition. Ultrasonic examination may be able to detect this condition, but the casting surface would need to be machined in order to prepare it for examination. The producer makes changes in the casting process or adds a larger machine stock allowance to solve the problem. In fact, many users avoid this problem by purchasing a rough-machined casting from the producer. Magnetic particle inspection is valuable for first-article inspection to ensure that the casting

design and rigging do not allow cracking in production. If cracks are found, magnetic particle inspection is used to identify the castings to be welded. Magnetic particle inspection is also used on heavily loaded casting surfaces to ensure surface integrity. Magnetic particle inspection is used to audit the process by inspecting samples of normal production. Liquid penetrant inspection is another surface-feature detection method. It is not generally used on as-cast or shot-blasted surfaces because of the likelihood of obtaining false indications. Penetrant may be retained in surface roughness and give indications unrelated to actual surface conditions. The penetrant method is best suited for use on machined, ground, or very smooth as-cast surfaces. Liquid penetrant inspection is of particular importance for austenitic alloys because they are nonmagnetic; therefore, their surfaces cannot be examined by magnetic particle inspection. ASTM E 165 describes the standard method for conducting this test. A set of reference photographs for acceptance or rejection is contained in ASTM E 433. Acceptance criteria are found in ASTM E 125 for the dry powder magnetic particle technique. Each of the documents must specify actual dimensions, including maximum length of indications and number of indications per unit area. Liquid penetrant inspection is subject to many of the limits and misuse cited in magnetic particle testing, in addition to false indications. It is included in ASTM A 903, and this provides workmanship levels. All the specifications for surfaces must recognize the scale of the part. Frequently, the highest level with the smallest allowable features is specified for large castings. This is unrealistic, adds cost, and becomes the subjective standard of the customer’s inspector. Liquid penetrant inspection, like magnetic particle, should be used for first-article inspection, process auditing, crack detection, and to monitor heavily loaded areas on critical castings. Radiographic Inspection. Customers commonly see radiography or x-ray inspection as essential to the production of high-quality steel castings. This creates a permanent record for liability defense, especially on critical castings. Unfortunately, radiography as a technique overpromises and underdelivers. The subjective and nonreproducible application of radiography, with limits on resolution and indication location, severely impedes its usefulness in performance evaluation. There are three basic groups of reference radiographs issued by ASTM International for evaluation of steel castings: E 446 applies to castings up to 51 mm (2 in.) in thickness; E 186 is for heavy-walled 51 to 114 mm (2 to 4.5 in.) thick sections; and E 280 is for heavy-walled thickness of 114 to 305 mm (4.5 to 12 in.). Currently, reference radiographs become standards for acceptance and rejection after the purchaser and the producer have agreed, in the purchase order or contract, to the acceptable severity level for each individual type of discontinuity. The choice of discontinuity

984 / Steel Castings severity level should ideally be based on realistic evaluation of design and stress analysis criteria under anticipated service conditions. Generally, low severity levels are specified for pressurecontaining castings with high-pressure rating and wall sections of 25 mm (1 in.) or less. Likewise, low severity levels are specified for machinery or dynamically loaded castings subject to high fatigue and impact stresses and with wall sections of less than 13 mm (0.5 in.). As wall sections increase and the fatigue and impact stresses are reduced, severity levels become somewhat relaxed. For structural castings, which are not dynamically loaded, moderate severity levels are usually specified, and again, for heavier sections (approximately 76 mm, or 3 in.), higher severity levels are usually called for. Unfortunately, the workmanship basis of radiography and the subjective application of the acceptance criteria do not support any particular application of radiography for performance. None of the reference radiographs are based on any kind of test data, and the severity levels are not graded to any basis of acceptability as to service performance. Since the radiograph does not indicate where in the cross section the indication may be, it is inherently unable to predict performance. The radiographic standards are used as a reference point in communicating the purchaser’s requirements. Because the standards are subjective, they are not reproducible. The radiographic images are not to be evaluated based on gray scale, and most importantly, the reference radiographs are to be prorated to the actual casting image size. No standard method or approach is suggested for this prorating, which allows the standard to have the appearance of objectivity while being completely subjective. A recent study applying gage repeatability and reproducibility to only the reading of existing plate-casting radiographs demonstrates the subjective nature of this technique. One hundred twenty-eight x-rays were rated seven times for shrinkage type and level. Unanimous agreement was 37% of the x-rays on shrinkage type, 17% on shrinkage level, and 12.5% on both type and level. Agreement was higher if the castings were completely sound or very unsound. Two of the radiographers rated the x-rays twice. Comparing each radiographer’s second rating to their first, both radiographers gave different type ratings for at least 19% of the x-rays and different level ratings for at least 34% of the x-rays. Both radiographers also reversed accept/reject decisions for 10 to 15% of the x-rays. Thus, this shows the subjectivity of radiographic examination. For this reason, most market sectors select some radiography for the first-article evaluation. This allows the adequacy of the rigging and solidification soundness to be evaluated. In critical parts, radiography may be needed on each part. The casting process is frequently audited with occasional radiography for noncritical parts. For many applications, the imposition of level 3 communicates the desire for a

reasonable soundness that is ordinarily achievable in structural castings. For critical areas, a radiographic level of 1 may be required. Requiring quality levels in excess of those justified by actual service conditions adds needlessly to the cost of the casting. It should also be kept in mind that the entire casting need not necessarily require radiographic inspection, and that the same severity levels need not apply to all areas of the casting. Careful analysis or, at least, good judgment can effect sizable cost-savings. In any case, the areas to be radiographed with the required severity level should be indicated on the casting drawing. Ultrasonic Inspection. Although the ultrasonic method of inspection has not been in common use for as long as radiographic methods, it is nevertheless a valuable tool for examining heavy-wall castings for internal discontinuities. The first ASTM specification for ultrasonic inspection of steel castings was issued in 1970 and is for longitudinal-beam ultrasonic inspection of heat treated carbon and low-alloy steel castings. In general, this inspection method is not useful for austenitic steel castings due to the large grain size of these castings. It is well recognized that ultrasonic inspection and radiography are not directly comparable. However, the technique is invaluable in detecting discontinuities in heavy sections, where radiographic methods would be considerably slower. Since no image of the discontinuity is obtained, considerable judgment must be exercised in the interpretation of results. The addition of computer analysis allows a wider application of ultrasonic inspection. It is more quantitative (digital) than most of the other NDE techniques. This should allow the enhanced use of ultrasonic, especially to develop standards that ensure performance. At the least, ultrasonic inspection can assess the amount of sound wall that exists to carry the design load. It can also pick up areas of microporosity too fine to be detected by radiography. One approach in the examination of large, heavy-wall castings, when ultrasonic evaluation may not be acceptable to the purchaser, is to first inspect with ultrasonic to identify areas with indications and then check these areas with radiography. Another possibility, since radiography does not reveal the depth of a discontinuity, is to follow radiography with ultrasonic in order to determine and evaluate the depth of the discontinuity. Leak Testing. Fluid-handling components are often leak tested to ensure performance. The component manufacturer frequently does this test. If the casting fails, it becomes the responsibility of the producer to replace or rework the casting. Often, leaks are due to porosity too fine to be detected by radiography. It may be possible to use solidification analysis software and a minimum Niyama value (a porosity prediction model), which would evaluate the microporosity, to assess castings for leaks. Hardness testing is an efficient means to ensure that the processing of an individual

casting was done correctly. The hardness of cast steel has a good correlation to strength but it is not a good indicator of quality in austentic and duplex stainless steels. Hardness testing location(s) should be stipulated. Tests and Castings ASTM International requirements and commercial practices normally result in the producer performing mechanical tests on each heat of steel to verify conformance with specification requirements. Many purchasers misunderstand the significance of these tests and believe that the test bar results for each heat will be the same as the properties of the casting. This is not true in most products, certainly not in steel products and clearly not in castings. The routine mechanical tests that are performed for each heat are to verify the capability of the material, not to determine the properties of the product. The properties of bars cut from the product, which are cut from the casting, depend on the location, section size, heat treatment, and shape of the casting. If verification of mechanical properties of the product is required, a check of hardness in a critical area is done. If a large critical casting is produced, a large connected test coupon may be added to the casting for properties verification. In this case, the properties of the test coupon are not given in the material standard but are subject to agreement between purchaser and producer. If a test bar is removed from a casting and it fails to meet the specification requirements for the material grade, the casting is not defective. Only if the test bar from the heat fails to meet the properties is the material unacceptable. Test coupons are typically made from an ASTM International double-legged keel block in accordance with ASTM A 985/A 985M for investment castings and A 703/A 703M for pressure-containing parts (Fig. 8). There is good correlation between tensile test data for attached specimens and keel block specimens (Table 3). Analysis has shown that two tests provide a 95% probability that the test sample and actual strength are within  þ6.9 MPa (þ 1 ksi) for ultimate tensile strength (UTS),  þ11 MPa (þ  1.6 ksi) for yield strength, and that elongation is within þ3%. Table 3 demon strates that data taken from an integrally cast production casting are similar to that of a keel block casting. Larger castings have a great propensity to form discontinuities and be affected by section size and geometric and metallurgical effects on material properties. Foundry practices can be used to provide optimal properties. Analysis of CA6NM and 105/85 cast steel materials has shown that alloying reduces any impact of section thickness and geometry. This same analysis showed that shrinkage in a casting had little effect on yield strength. In theory, since shrinkage is along the centerline—a lowstress region—and fatigue failures typically occur from surface defects, the fatigue limit should likewise be minimally affected. Tests have shown that tensile properties are not greatly affected by different types of discontinuities. There is a mostly direct relationship between the severity level of a defect and the

Selection and Evaluation of Steel Castings / 985

Fig. 8

Keel block coupon according to ASTM A 703/A 703M and ASTM A 985/A 985M. Dimensions given in inches. See the ASTM standards for metric values.

Table 3 Number of tests required for various degrees of accuracy using double-legged keel block test specimens Acceptable variation

Tests required for indicated probability

Property tested

MPa

ksi

99%

95%

90%

80%

70%

Tensile strength

0.69 1.38 2.07 2.76 3.45 4.14 4.83 5.52 6.89 3.45 4.14 4.83 5.52 6.21 6.89 8.27 9.65 11.03 1% 2% 3% 1% 2% 3% 4% 5% 6%

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 1 0.5 0.6 0.7 0.8 0.9 1 1.2 1.4 1.6

266 72 32 18 12 8 6 5 3 34 24 18 14 11 9 6 5 4 17 5 2 60 15 7 4 3 2

166 42 19 11 7 5 4 3 2 20 14 10 8 6 5 4 3 2 10 3 2 35 9 4 3 2 ...

177 30 13 8 5 4 2 ... ... 14 10 8 6 5 4 3 2 ... 7 3 2 24 6 2 ... ... ...

72 18 8 5 3 2 ... ... ... 9 6 5 4 3 3 2 ... ... 5 2 ... 15 4 2 ... ... ...

47 12 6 3 2 ... ... ... ... 6 4 3 3 2 ... ... ... ... 3 2 ... 10 3 2 ... ... ...

Yield strength

Elongation in 50 mm (2 in.)

Reduction in area

degradation in tensile properties. For instance, gas porosity will reduce the UTS by a little over 20 MPa (3 ksi) per severity level. Of the various types of porosity, linear shrinkage has the biggest impact on strength. However, even this type of defect only reduces the UTS by a little

over 55 MPa (8 ksi) per class. Keeping in mind location, use, and factor of safety, it is easy to see why steel castings with discontinuities can survive well past testing requirement levels. Testing on a hanger bracket casting has shown that defects do not necessarily lead to

part failure. Brackets with severity level 2 and five discontinuities had very little influence on the static and fatigue properties. Only in a few cases, where the defect was at or near the surface, did the location of the failure change. However, even in these cases, the properties of the casting were not impacted. Obtaining test specimens from an actual part is costly and renders the part useless. Thus, a cast coupon or separately cast keel block is used to determine that the material meets the requirements of the specification. However, in some instances, such as material thickness over 76 mm (3 in.), the specification may require that a test specimen be taken from an actual casting. Mechanical properties obtained from test bars represent the quality of the steel but do not necessarily represent the properties of the castings themselves. Solidification and heat treatment affect the properties of the casting, and these are dependent on section thickness, size, and shape. General Rules. Testing a section from a part is expensive but provides a data point on properties for the part. Testing a part to failure can likewise provide valuable information but comes at a cost. If a part with known level and type of discontinuity from NDE is tested, then setting the level one step higher for production parts should ensure desired performance characteristics. Working with a quality foundry will help ensure that discontinuities are minimized in the casting through optimized manufacturing techniques. Selecting a source based on best value versus lowest cost always pays for itself over the long run. As stated earlier, using analysis software is not only cost-effective but a very good means to identify discontinuities. Using zoning to inspect a casting will ensure cost-effective inspection. Areas of the part that see higher performance demands are inspected more frequently and to a higher level than other sections. Thus, sections with greater discontinuities that do not affect the part performance will not prevent the part from passing inspection. The impact of a discontinuity on performance depends on the casting size and use. A good starting point for NDE of steel castings is ASTM A 903 with acceptance level III indication length. Radiography is good to use in critical sections with level III as a starting point, and a rating per type of discontinuity should be applied. Ultrasonic testing can be used in castings with section sizes over 150 mm (6 in.). A tensile test should be carried out for each heat to ensure the material mechanical properties. Hardness testing should be done on critical sections. A hardness test is relatively low in cost, and hardness correlates with other mechanical properties. Since casting and welding are similar processes, the same NDE can be used. In the future, modeling casting production and solidification, in particular, will be key. The ability to link properties based on solidification patterns, design with these real-world properties, and craft NDE standards that assure performance promises a new generation of highperformance cast steel components.

986 / Steel Castings

Purchasing To obtain the highest-quality product, the part should be designed from inception to take advantage of the flexibility of the casting process. The foundry must have the part drawing, solid model or pattern equipment, and know the number of parts to be made. To take advantage of the casting process, the foundry should also know which surfaces are to be machined and where datum points are located. Reasonable dimensional tolerances must be indicated and should be appropriate for the casting process used. Tolerances are normally decided by agreement between the foundry and purchaser. Close cooperation between the purchaser’s design engineers and the foundry’s casting engineers is essential to optimize the casting design, in terms of cost and performance. Many specifications, such as ASTM International and ISO, have other requirements based on agreement between the purchaser and supplier; thus, communication between the designer, purchaser, and foundry is important in maximizing the benefits of steel castings. Purchasers should consider sourcing the foundry as the prime contractor. This invokes quality in the casting up front and can help assure that the delivery schedule is met. When making inquiries or ordering parts, all pertinent information must be stated on both the inquiry and order. This information should include all of the following components:  Casting shape—either by drawing, electronic

  



file, solid model, or pattern. Drawings or solid models should include dimensional tolerances, indications of surfaces to be machined, and datum points for locating. If only a pattern is provided, then the dimensions of the casting are as predicted by the pattern. Material specification and grade (e.g., ASTM A 27/A 27M—95 grade 60-30 class 1) Number of parts Supplementary requirements (e.g., ASTM A 781/A 781M—95 S2 radiographic examination) a. Test methods (e.g., ASTM E 94) b. Acceptance criteria (e.g., ASTM E 186 severity level 2, or MSS SP-54-1995) Any other information that may contribute to the production and use of the part (finishing and packaging requirements, documentation requirements)

To produce a part by any manufacturing process, it is necessary to know the design of the part, the material to be used, and the testing required. Machining. The foundry engineer is responsible for giving the designer a cast product that is capable of being transformed by machining to meet the specific requirements intended for

the function of the part. To accomplish this goal, a close relationship must be maintained between the customer’s engineering and purchasing staff and the casting producer. Jointly and with a cooperative approach, the following points must be considered:  The molding process—its advantages and

limitations

 Machining stock allowance to assure cleanup

on all machined surfaces  Design in relation to clamping and fixturing devices to be used during machining  Selection of material specification and heat treatment  Quality of parts to be produced Layout. It is imperative that every casting design, when first produced, be checked to determine whether all machining requirements called for on the drawings may be attained. This may be best accomplished by having a complete layout of the sample casting to make sure that adequate stock allowance for machining exists on all surfaces requiring machining. For many designs of simple configuration that can be measured with a simple rule, a complete layout of the casting may not be necessary. In other cases, where the machining dimensions are more complicated, it may be advisable that the casting be checked more completely, calling for target points and the scribing of lines to indicate all machined surfaces.

2. H. Davis, G. Troxell, and G. Hauck, The Testing of Engineering Materials, 4th ed., McGraw-Hill, 1982, p 314 3. M. Blair and T.L. Stevens, Ed., Steel Castings Handbook, 6th ed., SFSA and ASM International, 1995 SELECTED REFERENCES  C. Beckermann et al., “Analysis of ASTM



    



Research and Development Current and future research will always strive to improve the performance and manufacturing of steel castings. On-going research today (2008) is looking at modeling discontinuities and integrating the analysis with finite-element analysis software to facilitate the prediction of part performance. Modeling capabilities are also expanding to fill and flow analysis. Materials development centers around better performance characteristics, such as corrosion resistance, or better mechanical properties, such as strength to weight. Improvements to manufacturing are working on faster tooling technology, pressurizing risers, new pattern materials, and automated grinding capability. REFERENCES 1. C.W. Briggs and R.A. Gezelius, The Effect of Mass upon the Mechanical Properties of Steel, ASM, Vol 26, 1938, p 367



  

 

X-Ray Shrinkage Rating for Steel Castings,” Paper presented at 54th SFSA T&O Conference (Chicago, IL), Nov 2000 C. Briggs, “The Evaluation of Discontinuities in Commercial Steel Castings by Dynamic Loading to Failure in Fatigue,” Steel Foundry Research Foundation, Rocky River, OH, 1967 J. Carpenter and B. Hanquist, Specifying Steel Castings—Keeping Alloy Composition in Mind, Mod. Cast. Corrosion: Environments and Industries, Vol 13C, ASM Handbook, ASM International, 2006 Corrosion: Fundamentals, Testing, and Protection, Vol 13A, ASM Handbook, ASM International, 2003 Corrosion: Materials, Vol 13B, ASM Handbook, ASM International, 2005 F. Peters et al., “Assessing Process and Product Variability,” Paper presented at 57th SFSA T&O Conference (Chicago, IL), Nov 2003 F. Peters et al., “Variability: Causes, Concerns, and Corrections,” Paper presented at 58th SFSA T&O Conference (Chicago, IL), Nov 2004 A. See and B. Hollandsworth, “Mechanical Properties of Test Bars Compared to Castings,” Paper presented at 55th SFSA T&O Conference (Chicago, IL), Nov 2001 Steel Castings Handbook, 6th ed., SFSA and ASM International, 1995 Steel Castings Handbook Supplement 3— Dimensional Tolerances, SFSA, 2003 Steel Castings Handbook Supplement 6— Repair Welding and Fabrication Welding of Carbon and Low Alloy Steel Castings, SFSA, 1985 Steel Castings Handbook Supplement 7— Welding of High Alloy Castings, SFSA, 2004 P. Wieser, Ed., Steel Castings Handbook, 5th ed., Edward Brother, 1980

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 989-991 DOI: 10.1361/asmhba0005306

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Nonferrous Casting—An Introduction CASTING OF NONFERROUS ALLOYS on a tonnage basis is dominated by aluminum, which is cast by ingot and continuous processes for primary mills and by all foundry (shape casting) processes. In terms of shape casting, the percentage of nonferrous shape casting tonnage is roughly as follows (Ref 1):
Aluminum high-pressure die casting, 54%
Aluminum sand and permanent mold





casting, 12% Zinc high-pressure die casting, 11% Aluminum lost foam casting, 10% Sand mold casting of copper alloys, 10% Magnesium high-pressure die casting, 3%

The articles in this Section, “Casting of Nonferrous Alloys,” describe the shape casting of aluminum, copper, and zinc alloys along with articles on the continuous casting of aluminum and copper. Casting of magnesium alloys is detailed in the article “Magnesium and Magnesium Alloy Castings” in this Volume. This article briefly reviews the melt processing and casting of other nonferrous alloys other than aluminum, copper, magnesium, and zinc. More details are also given in the articles in the next Section, “Nonferrous Castings.”

Nickel Alloys Foundry practice for nickel-base alloys is, for the most part, similar to that used for cast stainless steels. Requirements for alloy cleanliness involve increased use of the ladle as a refining vessel, similar to steel (see the article “Steel Melt Processing” in this Volume). Melting involves vacuum induction melting (VIM), which is followed by consumable electrode remelting methods of electroslag remelting (ESR) and/or vacuum arc remelting (VAR). Electroslag remelting is not widely used, but several producers of critical rotating components in the gas turbine industry have adopted the use of the triple-melt sequence (Ref 2). First, VIM produces an initial electrode with low oxygen and precise chemistry, followed by ESR. The ESR electrode will be clean and sound but may contain freckles. The final segregation-free structure is obtained through the application of a third melting process (VAR) to the ESR ingot. Investment Casting of Superalloys. A number of casting processes can provide near-net

shape superalloy cast parts, but essentially all components are produced by investment casting. The characteristic physical and mechanical properties and complex, hollow shape-making capabilities of investment casting have made it ideal in applying the high-temperature properties of superalloys. Dimensional tolerances for superalloy investment castings are generally 0.075 mm (0.003 in.), with section thicknesses as low as 1.25 mm (0.05 in.) or less, and excellent surface finishes can be obtained. Nickel-base superalloy castings are produced by investment casting under vacuum, while most cobalt-base superalloys are produced by investment casting in air. Improvements in properties have been made not only through control of composition but also through more precise control of microstructure. The absence of grain boundaries in single-crystal alloys permits elements such as carbon, zirconium, and boron to be deleted from the composition. The resulting increase in melting point in turn provides improved flexibility in alloy composition and heat treatment (Ref 2).

Titanium Alloys (Ref 3) Titanium castings have been cast in machined graphite molds, rammed-graphite molds, and proprietary investments used for precision investment casting. The principal technology that allowed the proliferation of titanium castings in the aerospace market was the investment casting method coupled with the adaptation of hot isostatic pressing (HIP) to the most critical castings (see the article “Titanium and Titanium Alloy Castings” in this Volume). Significant design complexity, tolerance, and surface-finish control have been achieved, and large parts approaching 1.5 m (5 ft) in diameter can be cast. Castings generally are produced by vacuum arc remelting titanium in a copper, water-cooled crucible and pouring into molds. If the initial investment can be justified, hearth-melting processes may find application in the future for the remelting of alloy prior to casting. Of necessity, the casting mold systems must be relatively inert to molten titanium. Proprietary lost wax ceramic shell systems were developed by the various foundries engaged in

titanium casting. Usually, the face coats of the ceramic shells are made with the proprietary coatings, and conventional refractory systems are used to add shell strength. Regardless of mold type (e.g., investment or rammed graphite), foundry practice focuses on methods to control both the extent of the reaction of the molten alloy and mold and the subsequent diffusion of reaction products inward from the cast surface. Depth of surface contamination can vary from nil on very thin sections to more than 1.5 mm (0.06 in.) on heavy sections. On critical aerospace structures, the brittle alpha case is removed by chemical milling. The depth of surface contamination must be taken into consideration in the initial wax pattern tool design for investment casting. The wax pattern and resultant casting are made slightly oversized, and final dimensions are achieved by careful chemical milling. Metal pouring temperature, mold temperature, casting force (if centrifugally cast), and thermal conductivity of the metal and the mold, as well as the cooling environment, are all factors in the production of a good investment-cast titanium part. Casting Processes. The molten metal resulting after remelting is poured into either an investment mold or a rammed-graphite mold as just described. Pattern making and investment-cast mold techniques are similar to those used with superalloy technology. Although rammed-graphite molds are different from investment molds, they are similar to conventional sand molds. Cores are used to cast hollow parts. Porosity continues to be a potential problem but one that is addressed for premium aerospace castings by the use of HIP, as described subsequently. Once the mold-metal reaction problem is addressed, the most significant problem—with respect to titanium and titanium alloy castings—is achieving sufficient levels of superheat in the molten metal to maximize flow and mold-fill characteristics. In many cases, either a centrifugal table or mold preheating (or both) is used to ensure proper mold filling. A significant difficulty related to large titanium castings is the problem of porosity. However, by use of HIP, internal soundness of titanium castings can be improved to the point that no porosity or small voids can be detected. Weld repair is used to close gross defects after HIP. Because strength increases and ductility (toughness) decreases as oxygen level increases,

990 / Casting of Nonferrous Alloys oxygen content is a matter of concern regarding titanium alloy castings. Control of oxygen levels in cast components is achieved mostly by selection of melt stock, but hearth melting can make an additional improvement possible. An ingot with a low oxygen content generally results in a casting with the lowest oxygen content. The exact role of revert in oxygen-content control and in alloy element segregation is not clear; both revert and virgin ingots are used. Oxygen also can be introduced to the casting from the surface mold-metal interaction. The rammed-graphite method is the oldest mold technique used to produce titanium castings. The method uses a mixture of graphite powder and associated binders and water additions that are rammed against a pattern to form a portion of the mold. Individual mold segments are then fired and assembled for casting. Most high-performance titanium alloy casting applications rely on the technique of investment casting.

Lead Melting of Lead Alloys. Lead and its alloys constitute a small but commercially important segment of the nonferrous metal casting industry. Principal lead alloys and their areas of importance include Pb-60Sn (solders), antimony-lead and calcium-lead (battery grids), and Pb-7Sn (terne plating). The melting and casting of pure lead and foundry alloys pose fewer problems than with other nonferrous systems. The major lead alloys have relatively low melting points (315 to 425  C, or 600 to 800  F) and are not susceptible to hydrogen absorption and the attendant gas porosity to any great extent. Dross formation can be a problem, especially with alloy constituents, so simple cover fluxes such as graphite or vermiculite are sometimes used. However, oxidation is relatively minor unless temperatures are excessive. In the terne plating industry, where lead-tin coatings are applied by hot dipping steel articles, principally for use in fuel systems, a cover flux is often used. This is similar to the flux used in galvanizing—a ZnCl2-NH4Cl mixture, often with other chloride salts such as KCl. The flux used in terne plating serves to preclean the steel, to absorb pretreatment pickling residue, to prevent spitting as the articles enter the plating bath, to alleviate bath dross formation, and to prewet the article being coated. In the die casting of lead alloy acid battery grids, no treatment is usually necessary at the die cast machine melting pot itself, although a melt cover can be used for heat insulation purposes and to prevent dross buildup. In remelt pots, where trim scrap and other scrap is melted, chloride-base drossing fluxes can be used. Pitch, sawdust, or wood chips are used to agglomerate the residues that accumulate on the surface.

The lead is usually first softened (refined) in the production of the newer Ca-Pb-Sn alloys used for maintenance-free battery grids from either primary metal or recycled scrap. Calcium is then added as an alloy, often under a protective cover such as sawdust, wood chips, or resin to prevent oxidation at the usual higher temperature at which alloying is facilitated. Refining of Lead. A number of fluxing procedures and refining processes are used in the smelting and refining of primary and secondary lead production. Several of these involve chemical flux-refining techniques somewhat analogous to those used with other nonferrous metals. Figure 1 is a flow chart showing a number of important sequential process steps in the refining of lead. These steps are briefly discussed here and are covered in greater detail in Ref 4. Copper Drossing. In copper drossing, the first step, the elements copper, iron, nickel, cobalt, and zinc can be removed, if present, by cooling the lead almost to its freezing point. The liquid solubility of these elements at that temperature is very low; thus, they can be removed as an intermetallic dross as they separate from the heavier lead melt. Copper will precipitate with any sulfur, arsenic, antimony, or tin, in that order. Lead sulfide may also form. These drosses are usually intimately intermingled with the lead metal phase. A molten phase (copper-lead sulfide) may form, and when the newer calcium lead battery scrap is remelted, calcium may be present in this phase, lowering the lead content. If oxides are present, there is also a slag phase, and silica can be added to increase its fluidity, resulting in the creation of iron, zinc, and lead silicates. It is also possible to remove copper from lead down to a level of 0.05% by stirring sulfur into the lead melt at temperatures near the freezing point. When silver or tin is present, copper can be reduced to even lower levels, but a reversion (copper reentering the metallic phase) can occur if stirring is excessive (over 10 min) or if temperature is increased (Fig. 2). Softening. In softening, the second process step, elements more readily oxidized than lead are removed by oxidation. These include tin, arsenic, antimony, and indium, whose free energies of oxide formation are greater than that of lead. The term softening relates to the effect these elements have on lead. When present, they contribute to solid-solution strengthening; thus, their removal results in softening. The Harris process of softening consists of oxidizing the impurities, converting them to sodium salts, and collecting them in a caustic soda melt slag phase that rejects molten lead, providing for relatively easy physical separation. Temperatures of best reactivity are fairly high (700 to 800  C, or 1290 to 1470  F), and thorough mixing of the metal with the salt flux is required. Tellurium, arsenic, and tin can be reacted directly with caustic soda to form lower-sodium salts. A further oxidation reaction, employing

Fig. 1

Flow chart showing steps in the lead-refining process. Impurities remaining are shown in brackets after each step. Source: Ref 4

sodium nitrate (NaNO3) as the oxidizing agent, forms the higher-sodium salts (arsenates, stannates, and antimonates). The intermixed metallic oxide products produced by these reactions are then removed and separated by various wet processing and precipitation methods. A simple softening process involves the refining of lead by oxygen at a somewhat reduced temperature (650  C, or 1200  F) directly in a refining kettle. A 22.5 Mg (25 ton) lead melt containing 0.6% Sb was refined to a level of just 0.02% by blowing 10 m3 (350 ft3) of oxygen for 1.5 h (Ref 5). Moreover, approximately 100 kg/m2 (20 lb/ft2) of antimony was oxidized with oxygen injection, versus a rate of only 10 kg/m2 (2 lb/ft2) using air. Because this is in excess of the expected stoichiometric ratio, the oxygen must also act catalytically. Desilvering is the term applied in lead refining to the removal of elements more noble than lead—silver, gold, and bismuth. These elements can be coprecipitated with lead and zinc by the Parkes process, which is the addition and mixing of zinc with the lead at a temperature just above the freezing point of lead (320  C, or 610  F). Because there is substantial metal value contained by silver- or gold-bearing precipitate residues, technologies (pressing,

Nonferrous Casting—An Introduction / 991

Fig. 3

Final lead refining using the modified Harris process. Source: Ref 4

fluxing techniques. Fractional oxidation or crystallization can yield liquid slag and solid crystals that are purer than the starting metal. Bismuth can also be precipitated from lead by adding alkali or alkaline earth metal reagents, such as calcium and magnesium (the Kroll-Betterton process). Magnesium metal and Pb-5Ca master alloy are stirred into the bismuth-containing molten lead at 420  C (790  F). As the lead is cooled back to its freezing point, CaMg2Bi2 crystals precipitate, leaving the lead refined of bismuth. The precipitate is intermixed with lead, and the latter can be removed by liquation by adding the mixture to the next batch and raising the temperature once again. Further reduction of bismuth is possible by stirring in antimony (or antimonial lead) in the presence of residual calcium and magnesium in the melt after the initial CaMg2Bi2 precipitate has been removed. The antimony addition forms a similar precipitate, which facilitates easier formation and substitution of bismuth for antimony in the precipitate at more dilute concentrations of bismuth. Final Refining. Before casting, a final refining step is carried out on the molten lead, using a modified Harris process at 450 to 500  C (840 to 930  F). Caustic soda or NaNO3 at the rate of 0.1% is stirred into the lead to decrease the zinc and antimony that may remain after previous refining steps to less than 1 ppm each, as shown in Fig. 3 (Ref 4). Arsenic, tin, calcium, magnesium, and iron will also be removed to very low levels with or before the antimony. The softened and highly purified lead thus produced is required in the production of battery alloys, whose electrochemical performance is highly dependent on the removal of impurities.

Fig. 2

REFERENCES

liquation, and zinc distillation, for example) have evolved for separating the noble metals from the precipitate. Zinc Removal. Zinc can be removed preferentially from the base lead metal by preferential oxidation or chlorination or by the Harris process without the use of an additional oxidizing agent:

1. Trends and Survey, AFS, 2002 2. M.J. Donachie and S.J. Donachie, Superalloys: A Technical Guide, ASM International, 2002 3. M.J. Donachie, Titanium: A Technical Guide, ASM International, 2000 4. T.R.A. Davey, The Physical Chemistry of Lead Refining, Proceedings of the Symposium on Lead-Zinc-Tin, AIME, 1980, p 477 5. J. Blanderer, The Refining of Lead by Oxygen, J. Met., Dec 1984, p 53

Decoppering of lead. (a) Effect of sulfur. Curve 1 represents the copper content of silver-free lead as a function of time while stirring in 0.1% S at 330  C (625  F), and curve 4 represents similar conditions for a lead containing 0.04% Ag. Curve 2 resulted when only 0.05% S was added, and curve 3 when a total of 0.1% S was added as ten equal additions of 0.01% S each, at 1 min intervals, approximating plant practice in which the sulfur is added gradually over 10 min. (b) Effect of silver. (c) Effect of tin. Source: Ref 4

Zn þ 2NaOH ! Na2 ZnO2 þ H2

Vacuum dezincing, however, provides much higher zinc recoveries than any of these techniques and is therefore preferred. Bismuth Removal. Bismuth is more noble than lead and therefore cannot be removed by chemical

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 992-1000 DOI: 10.1361/asmhba0005285

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Dross, Melt Loss, and Fluxing of Light Alloy Melts Daniel E. Groteke, Q.C. Designs, Inc. David V. Neff, Metaullics Systems Division, Pyrotek Inc.

DROSS, which is the oxide-rich surface that forms on melts due to exposure to air, is a term that is usually applied to nonferrous melts, specifically the lighter alloys such as aluminum or magnesium. Dross comprises a physical mixture of oxides and entrapped molten or semimolten metal. The dross that forms on aluminum melts is normally considered to be an undesirable material, which is supported by the definition found in most dictionaries: “Scum on molten metal.” While it is certainly the source of much grief in operations that melt aluminum or aluminum alloys, the solid form of dross is far from being undesirable and, in truth, is a valuable co-product that can have a major effect on the profitability of the producer’s operation. This comes from the simple fact that, in addition to the oxides present in the dross, it often contains large quantities of entrapped useful aluminum alloy. The alloy values can potentially be lost from the process stream during the normal removal and cleaning process unless great care is exercised at the generating source or the alloys are recovered in subsequent recycling efforts. The composition of dross is dependent on the base analysis of the alloys being melted, but the major constituent in almost every case is aluminum oxide, with other contaminants added from the other alloy constituents and process contributors. Any new aluminum surface, whether solid or molten, will quickly form a protective oxide when the surrounding medium contains even a very small trace amount of oxygen from any source (free oxygen, moisture, or reducible oxides). Molten surfaces are subject to the same oxidants but are also exposed to other potential contributors in the layer of oxide-rich materials present on the melt. Industries conducting melting operations on aluminum and aluminum alloys operate a wide variety of types of primary melters and holders to provide molten material to their process lines. All are active dross producers, with their own characteristics influencing both the rate and quantity of oxide-bearing material generated. Further, the operation of transferring the molten material between the melters, holders,

and the final solidification point is an additional source of dross generation that can, in many cases, exceed the volumes generated in the actual initial melting.

Dross Formation Many different crystallographic forms of aluminum oxide can exist that are associated with molten aluminum processing. Up to 16 different forms have been identified, although in principal only the major ones need to be considered. The first oxide film to form on molten pure aluminum is generally agreed to be gamma-alumina (g), which is a thin film that inhibits further oxidation. After a brief incubation period, the g film transforms to an alpha-alumina (a) form, which oxidizes and grows at a much greater rate. When alloying elements are added to the pure aluminum, both the rate of formation and the nature of the resultant oxide are changed dramatically. While the common alloying elements (copper, iron, manganese, chromium, zinc, silicon, etc.) in foundry cast alloys have little effect on the rate of oxidation, other elements can play a major role in changing both the rate and volume of dross formation. The most important of these is magnesium, which is present in almost all commercial alloys as either a trace or major alloying element. A second contributing element is lithium, which is never used as an alloying element in conventional casting alloys but may be present in trace amounts in the scrap streams that are used to generate the secondary foundry alloys. Since these scrap streams constitute the source for the bulk of materials poured in the foundry casting industry, lithium, when present, can be one of the more significant elements encouraging dross formation in commercial operations. Even small quantities of magnesium in the cast alloy can convert the dross to a mixed oxide of MgO Al2O3 and Al2O3, with the percentage of the mixed oxide (spinel) increasing rapidly until a level of 2% Mg is reached in the alloy. At that approximate level, the




composition of the oxide reverts to pure magnesia, and the rate of formation increases. The oxide and mixed oxide films are continuous over the surface of the melt but can grow at rates that are multiples of the growth on solid metals. This growth is supported by the entrained alloy trapped in the filmy oxides and additionally by capillary action transporting fresh aluminum through the relatively porous layer of material. In addition to the growth of oxide films and dross on the melt surfaces that are induced by exposure to general atmospheric conditions, there are a number of factors encountered in industrial melters that can supplement and even alter the growth factors. These include turbulence induced by the mode of melting energy, the generation of atmospheres that are chemically more reactive, and even the addition of oxidants or contaminants with the melt stocks. These are illustrated in Fig. 1 and 2. Turbulence caused by direct impingement of burner flames on the melt can be an especially damaging condition in accelerating the rate of dross formation. The movement on the metal surface ruptures the oxide film, exposing a new surface to oxidizing gases at elevated temperature. These gases contain potentially high levels of moisture from entrained humidity or products of combustion, further adding to the oxidation potential. Similarly, the melt stocks themselves can be a source of dross and contaminants as they carry in their own layer of oxide and surface moisture that reacts to generate additional oxide, while also adding hydrogen to the melt. Further, many items of die cast home scrap and returns will carry in moisture and organic compounds from die lubes, machine oil and grease, paint, and so on. These add carbides and nitrides to the suspended inclusion matter in the metal and ultimately wind up in the dross. Finally, excessive melting temperatures, long holding periods, improper burner adjustment or electrical heating controls, worn elements, and deteriorating furnace doors that permit air

Dross, Melt Loss, and Fluxing of Light Alloy Melts / 993 aspiration into the furnace will contribute to increased oxidation at the melt surface and growth of the dross layer.

Influence of Melter Type on Dross Generation

Fig. 1

Complex interactions that occur between the atmosphere above the melt, the entrapment of useful aluminum alloy by the crumpled oxide films, and the resultant buoyancy of oxide inclusions in the melt due to the adsorption of hydrogen onto the complex surfaces of the inclusion

Fig. 2

Addition of supplementary volumes of contained surface oxides that are on the metal stocks entering the furnace molten metal. The presence of organic compounds will similarly react with the molten bath to add carbides and nitrides to the inclusion level. These suspended inclusions and oxides by themselves have a higher density than the molten aluminum but are carried to the dross layer by hydrogen gas adsorbed on the inclusion surface or gas bubbles generated by charge contaminants.

The various requirements for alloy composition, metal volumes, and pouring temperatures have created a very wide range in the size and types of melters and holders used in the aluminum shape-casting (foundry) industry. The very large aluminum casters make use of direct molten-metal deliveries from the secondary or primary refiners to minimize the need for re-melting cold stocks. It is not uncommon for these deliveries to cover distances of more than 160 km (100 miles) from the smelter plant, and they are delivered on a very tight schedule. The shipments are made in large refractory-lined ladles with sealed tops to retain the heat and, in some cases, are held at temperature in the receiving plant with auxiliary burners to supply molten alloy on an as-needed basis for the melting department. Over-the-road shipments of quantities as small as 4500 kg (10,000 lb) have been made, but a more common delivery quantity is approximately 7000 or 14,000 kg (15,000 or 30,000 lb). Typically, the size of the shipment is limited more by carrier and highway load limits than anything else. Surprisingly enough, tight specifications can be held on delivery temperature, time, and volume. Specification arrival temperatures of a total range of 40  C (75  F) are not uncommon, with arrivals scheduled within 30 min of a target delivery window. The use of molten deliveries will normally add a requirement for very large holding furnaces (with a minimum capacity of twice the delivery volume) at the receiving point to balance usage against deliveries and to minimize temperature surges and cold deliveries that can wreak havoc with process control. The elimination of the remelting operation does produce major benefits in terms of minimizing both dross generation and melting energy requirements. Further, since the holding furnaces are designed with smaller burners (merely holding metal at temperature requires less energy than melting), they also minimize the generation of dross on the hearth during the holding period. These savings are partially offset by higher transportation costs because of the unique carriers involved and their high maintenance costs, but the resultant savings in energy, labor, and melting loss more than compensate. With energy expenses constituting a larger portion of total melt-room costs, the savings associated with molten deliveries are becoming ever more important. The other major savings with molten deliveries are in the area of reduced melting losses. Reduced holding costs have already been noted but equally important is the minimization of melting loss associated with the lower volume

994 / Casting of Nonferrous Alloys of cold material being melted. With moltenmetal deliveries, the primary source of dross is that created by turbulence during the transfer. Most operations using hot metal deliveries experience less than 1% melting loss for the volume of material received. Those operations too remote from suppliers to use molten deliveries rely on primary melters in their own operations to convert their feedstocks to molten form. A variety of melter types are used in these operations, with the choice dictated by available capital, energy costs, melt demand, labor rates, and the types of charge materials that will be used or generated internally. The melter types in use include induction melters (both channel and coreless), resistance-heated furnaces (usually all wet-bath charge types), reverberatory furnaces (of wetbath, dry hearth, or combination charging modes), and stack melters (where the charge is preheated and dried before entering the melting zone) (see the article “Reverberatory and Stack Furnaces” in this Volume). All furnace types have their proponents, depending on selection factors mentioned before, and in general will sustain melting losses (dross generation) ranging from a low of 0.5% up to 4.0% or greater for similar charge stocks. There are also significant differences in melt quality generated by the various furnace types, which are addressed later in this section. For molten deliveries, the aluminum alloy caster does not sustain the melting loss on the original feedstocks, because that is absorbed at the refining plant (the caster’s only loss is usually that associated with the metal transfers and the minimal normal hearth losses). Among the other furnace types, the choice is usually made with primary consideration given to the size and surface area of the materials being melted. Dry hearth furnace types are usually chosen when the materials being charged have minimal surface area (large sows or full stacks of ingot). Wet-bath furnaces have advantages when the charge materials are mixed in section thickness, with the thinner materials being able to be quickly submerged to minimize additional surface oxidation. However, this benefit can be lost if there is any moisture or lubricant associated with the charge materials, because these contaminants are a major source of dross and dissolved hydrogen when added to molten aluminum. Stack melters have the advantage of energy efficiencies gained from preheat and drying of the charge materials before entering the melting zone but sustain somewhat higher melting losses when melting thin-gage stocks and flashings because of increased oxidation before reaching the melting zone. The other notable disadvantage of stack melters is that their best performance is developed with a mixed charge of heavy and light melt stocks. Gates and runners from large foundry or die castings cover a significant amount of surface area for their weight and will require size reduction at

the trim press or cutoff station and subsequent “bundling” to increase the charge density. Without this activity, the number of low-density charges will lead to increase labor expense, potentially higher melt loss, and a slowdown in the melting operation. An additional area of advantage for stack melters is that the holding hearth is usually sized at approximately 2 times the melting rate of the unit. Other volume-melting furnace types will frequently be sized at a multiple of 5 and even 10 times the melting rate. The larger volume in the hearth is associated with the need to minimize temperature fluctuations associated with the introduction of cold melt stocks to the molten volume. Side-by-side comparisons of oxidation/ melt losses have shown an advantage for stack melters when compared to dry hearth furnaces, but the types of charge materials used could have introduced some bias. Adherence to proper operating conditions is absolutely necessary to achieve low melt loss in stack melters. Resistance furnaces generally have lower melting losses because of the elimination of combustion atmosphere-related melt losses and the absence of any flame-induced turbulence on the surface of the melt. Again, there are trade-offs for these benefits, and they are normally associated with higher energy costs for power. Induction furnaces have historically been the preferred choice for scrap forms with very high surface areas (dried machine turnings or flashings, etc.) because of their rapid submergence of the materials. Normally, very little surface dross can be seen on some aluminum melts in induction furnaces. While it may not be visible, the intense stirring is normally viewed as a factor in breaking these oxides into smaller and smaller inclusions, which may then be suspended in the melts. The need to remove these with downstream cleaning will somewhat negate the low initial dross losses. While surface dross/metal loss may be low, the resultant residual oxides associated with fine-scale charges may reach 6 to 10% metal loss in the bulk metal when ultimate metal weight cast is compared with metal weight melted. Because of the high surface area associated with chips, borings, turnings, flash, and so on and the subsequent entrainment of the oxide surfaces with the melt, other forms of direct melting of these materials are coming into greater favor, especially those with mechanical or molten-metal pump-assisted devices designed for rapid submergence.

Influence of Charge Materials on Melt Loss As indicated earlier, the primary factors influencing melting loss, other than equipment or processing-related variables, are the surface-area-to-mass ratios of the charge materials and the condition of the surface itself. Large heavy-section melt stocks (i.e., sows or even

ingot shapes) have the largest ratio of mass to surface area and sustain proportionally minimal melt loss. In-house remelting of clean scrap, shot buttons, gates and runners, trim scrap, and so on will increase the melt loss because of the increased surface area. If the same materials carry moisture; sand or molding materials; metal working, finishing, or die-casting die; and release lubricants as they enter the melt, the loss could be doubled. A good rule of thumb is that for every 1% contaminant contained— organics and moisture—2% melt loss will result. Charging turnings, borings, and machining chips will usually result in as much as 10 to 15% dross formation or melt loss because of the high surface-area-to-volume ratio of such items. Other scrap forms, such as painted scrap or oily and greasy clean-out scrap from machines, are not currently being introduced into the melting furnaces of most foundries because of the downstream need for bag houses and other air-pollution-control equipment. These materials are processed in secondary refining plants that are equipped with such equipment and are often predried to remove the bulk of contaminants before introduction into molten metal. While adding cost to the recycling operation, the early removal of moisture, oil, and organic materials is recognized as essential to minimize associated melt losses and inefficiencies.

Influence of Operating Practices on Melt Loss The practices used in melting aluminum can have a major impact on the rate of dross generation and the resultant melting losses. With most melting being done in fuel-fired reverberatory furnaces, relatively small variations in a number of process parameters can have a large impact on the melting losses sustained. One of the first would be the adjustment of the fuel/ air ratios used in the burners, with strongly oxidizing conditions producing the obvious result of an increased rate of dross formation if there is any movement of the melt surface to expose fresh aluminum surfaces. A second and equally important factor is the holding temperature of the melt after melting has been achieved. The impact of temperature increases of only 25 to 55  C (50 or 100  F) can be quite significant, as may be seen in Fig. 3. It should be kept in mind that thermocouples in a molten aluminum bath actually measure the temperature of the bath itself. It is impossible to measure absolutely the temperature of just the top surface, which in reality may be considerably hotter than the bulk molten volume. Consequently, the temperatures shown in the graph of Fig. 3 can indeed be reached at the melt surface, with attendant rampant oxidation and resultant increase in dross thickness. Similarly, the thickness of the layer of dross over the melt can have an impact because of its insulating effect and the required increase in firing rate of the burner system to maintain

Dross, Melt Loss, and Fluxing of Light Alloy Melts / 995 process temperatures. The temperature in the dross layer rises, increasing the rate of oxidation of the metal being aspirated from the hearth volume, and can produce a logarithmic increase in melting losses. The impact of increases in the dross thickness on melt rate is shown in Fig. 4 and is recognized by most melting operators. For this reason, the thickness of dross on melter surfaces is most commonly maintained at less than 40 mm (1.5 in.). Typically, the temperature of the dross layer is approximately 50 to 85  C (100 to 150  F) higher than the melt temperature. Thus, high melt temperatures incur the risk of ignition of the drosses, with spontaneous burning of the aluminum metal (called thermiting), causing potentially catastrophic results. If allowed to continue, temperatures attained during these reactions will exceed 1650  C (3000  F) and can result in melting of furnace refractories and the total loss of the furnace. The industry preference for wet-bath charging of light-gage scraps can be the source of an additional incremental increase in melting losses if the drosses formed on the surface of the melt are subsequently exposed to direct burner flames. The surface drosses are a “soupy” mixture of oxides and entrained metal and can contain up to 80 or even 90% of useful alloy. Without prompt removal from the surface, or treatment to separate the metallics from

Fig. 3

Logarithmic increase in aluminum oxidation as the holding temperature is increased

the oxides and dirt, they will oxidize, consuming valuable metallics. Keys to reducing dross formation at the melt surface include not only careful control of charging practices and temperature control but also increasing opportunity for better heattransfer capability. Reducing the melt surface temperature allows a greater temperature difference between the heat source and the melt, resulting in greater heat transfer and hence faster melting practice. This is accomplished by convectional burners, by melt stirring, and, in larger furnaces, by application of molten-metal pumping. This allows heat to be more effectively transmitted to and throughout the bath from the radiation heat transfer provided by the heat source into the roof and upper sidewalls, which “radiate” heat into the melt. Dry hearth furnaces are immune to some of these problems because the surface oxides from the charge materials remain on the sloping hearth since the melted metal drains into the furnace. If wet or oily gates, runners, and shot buttons are placed on the dry hearth, they will not develop the same problems as if they were charged into a molten bath. They will, however, contribute a higher oxide load to the hearth, which can prevent complete drainback of the molten alloy and expose it to oxidation. It is for these reasons that the more efficient aluminum melting operations will emphasize skimming of the furnace hearths on a regular basis. Skimming should be conducted on an as-needed basis, usually once per shift, unless wet drosses are created. Recommended practices with the wetter drosses are to skim on a minimum of a 4 h schedule.

argon or nitrogen, and the latter is more common because of economics. Gas mixtures may also be used, usually adding an active halogen component—chlorine or fluorine compounds— in smaller amounts, normally only 5 to 10%, to the base inert gas. The purpose of this addition is to provide a better separation of solid particulate (i.e., aluminum oxide inclusions) from the liquid metal. Gas fluxing does not reduce dross; in fact, it increases dross! The principal function of gas fluxing is hydrogen removal, not dross treatment or recovery (see the article “Aluminum Fluxes and Fluxing Practice” in this Volume). In-Furnace Treatment with Solid Fluxes. Once formed on the surface of the melt, it is imperative that the dross be removed to maintain normal melt rates and to avoid further loss of the contained metallics. Methods used in the industry usually combine treatment of the dross on the hearth or charge well with routine maintenance on the sidewalls and bottom of the furnace. This is normally done by employing several different types and application of solid fluxes, which are chemical compounds and mixtures of chemical compounds designed to perform specific functions. Mixtures may include reactive ingredients as well as filler materials, which provide extension of volume and/or reduce cost. Table 1 lists several important chemical constituents that are included in many fluxes. The specific kinds of chemical fluxes include:

Furnace Fluxing

The three are commonly employed in the industry and vary in application, depending on the type and size of the melting unit. Cover fluxes are used, as the name implies, to cover the melt. These fluxes are comprised of salt compounds, which form eutectic mixtures that are liquid at normal aluminum melting temperatures. A typical base composition is 47.5% sodium chloride, 47.5% potassium chloride, and 5% sodium aluminum fluoride,

Fluxing is a word applied in a broad sense to a number of methods of treating a melt. It is important to understand what it means and where it is used. There are two types of fluxing that the metal caster commonly refers to in typical melting and handling operations. Gas fluxing is the application of a gas treatment to the melt. The gas may be inert, such as

Table 1 Typical fluxing compounds employed Compound

Fig. 4

Impact of a layer of hearth dross on the melt rate of a reverberatory melter

 Cover fluxes  Wall-cleaning fluxes  Drossing fluxes

Fluidizer (F) thickener (T)

Oxide surfactant

Chemically active

Exothermic

Gas released

Element added

AlF3 CaCl2 MgCl2 MnCl2 KF NaF NaCl KCl

F F F F F F F F

... ... ... ... ... ... ... ...

X ... ... X ... X ... ...

... ... ... ... ... ... ... ...

... ... ... ... ... ... ... ...

... ... ... ... ... Na ... ...

CaF2 Na3AlF6 Na2SiF6

T T T

X X X

... ... ...

... ... ...

... ... ...

... ... ...

KNO3

...

...

X

X

...

...

C2Cl6 K2CO3 Na2CO3

... ... ...

... ... ...

X X X

... ... ...

Cl2, AlCl3 CO2 CO2

... ... ...

K2TiF6 KBF4

... ...

... ...

X X

... ...

... ...

Ti B

996 / Casting of Nonferrous Alloys or cryolite. Covering fluxes become liquid on contact with the melt, help to agglomerate dirt and oxides, and provide some degree of protection of the melt against further oxidation and absorption of hydrogen. Wall-cleaning fluxes are used to help in softening the buildup on furnace walls to permit easier removal by mechanical cleaning. They are usually applied with a lance or flux gun or may simply be broadcast manually onto the walls. Slightly exothermic, these fluxes, under closed-door and high-fire conditions, increase the localized temperature of the oxide buildup and thereby soften or loosen the attached oxides, permitting easier mechanical removal and subsequent skimming. If the oxide buildup has been converted over time to the insidious form of aluminum oxide called corundum, it cannot be removed chemically. However, in its early stages sufficient loosening of the mildly adherent product can be accomplished to allow removal from furnace sidewalls with the combination of fluxing salts and hard work. Drossing Fluxes. Recognizing that a high percentage of useful metallics can be lost when wet drosses are removed from the furnace, many operations will attempt to reduce the metallic content by in-furnace treatment of the drosses with exothermic fluxes. These contain oxidants and fluorides to generate temperature and help promote a separation of the oxides

from the metallics. The exothermic reaction does consume aluminum values while generating heat and can be the source of a loss of up to 20% of the contained metallics in the dross, if the flux is applied too liberally. Drossing fluxes are designed to assist in separating the entrapped aluminum from the adherent oxide skin envelope, which surrounds the aluminum in the dross layer. The structure of dross can be likened to M&M candies (Mars, Inc.), as in Fig. 5, or a heavy mix of “corn flakes” of oxide films floating in a bath of milk. As discussed earlier, the nature of the dross is dependent on the materials being charged and the style of melting furnace. The fluxes used generally are reactive, which rely on good contact with the physical dross to achieve the separation, and the flux may contain an exothermic compound designed to add heat to accelerate the separation process. The application of heat and physical mixing increases the localized surface temperature and the fluidity and helps to loosen the loosely adherent oxide skin or shell, opening up this skin to allow the metallic aluminum to coalesce back into the bath. Usage practices for both types of flux vary widely in the industry, but all require agitation and intimate mixing of the flux and dross to achieve a good separation. The initial step is to spread the flux over the layer of dross. Application rates vary, but some flux manufacturers

recommend starting at approximately 1 kg/m2 (2 lbyd2) of hearth area. Following that uniform application, most operators rabble the dross and flux mixture to begin the separation. The furnace doors are then closed and the burners turned on high fire for a short period of time. The increase in temperature promotes flux reactions with the dross, improving the separation. Normally, the door is then reopened, and the operator then carefully rabbles, rakes, and chops through the dross layer with an appropriate tool to further mix the flux with the dross. The combined material is then removed from the furnace as quickly as possible to minimize thermal losses that can be quite excessive during the fluxing and cleaning operation. Care must be taken to allow the now-freed molten aluminum to drain from the dross/flux mixture to achieve best recovery of the entrapped metal. A typical operation is shown in Fig. 6. One should fully understand that the drossing flux process does not chemically reduce aluminum oxide; this can only be done electrolytically. Rather, correct application of the proper flux composition and application practice results in the physical separation or breaking apart of a weakly-bound oxide layer entrapping fresh aluminum, as depicted in Fig. 5, allowing the metallic liquid aluminum alloy droplets to coalesce and return to the main bath below the dross surface. The objective of any fluxing practice is to transform white dross, or metal-rich dross, into a black powdery dross that is low in metal content. Figures 7 and 8 exhibit this physical appearance comparison portraying the visual difference between untreated and treated dross, respectively. Metal-rich dross appears bright and shiny, certainly laden with metallic aluminum, whereas treated dross will appear duller and powdery. Typically, the aluminum content of untreated dross averages 85 to 90%. Initial fluxing treatment can recover approximately half of this amount, if done aggressively, rendering the remainder still economically and technically viable for secondary metal recovery in-house, with auxiliary equipment, or with an outside

Fig. 6 Fig. 5

Schematic view of dross shell and entrapped aluminum

dross

Furnace operator cleaning the sidewalls of the furnace and rabbling the flux into the hearth

Dross, Melt Loss, and Fluxing of Light Alloy Melts / 997

Fig. 9

Fig. 7

Typical range of dross fluxing aluminum recovery

Unfluxed white dross, which is aluminum-rich

Fig. 10

Pile of white dross collected after the turbulent fill of a transfer ladle of A380 alloy in a die casting plant. The aluminum content of the material is +95%.

of the same composition. Additionally, this returned metal that has already been paid for replaces the need for purchasing new metal units and minimizes the inventory cost of metal units that are in the recycling loop.

Drosses Generated in Metal Transfer

Fig. 8

Fluxed black powdery dross low in aluminum

dross processor. With either auxiliary in-house or outside dross processing, an additional 30 to 60% absolute metal content recovery is still possible. Figure 9 depicts the general recoveries and reduced metallic content in treated dross that may be achieved with aggressive furnace fluxing as a first step, followed by secondary fluxing treatment off-line in another dross recovery process. It should be noted that these furnace fluxing results are optimum percentages, achieved only with very careful practice. Recovery values may be decreased by 50% or more with “drag-

out” losses if good alloy is entrained in the granular dross being removed. Realization of the metal recovery in a drossfluxing or secondary process has a significant economic impact. Secondary processing may be done either with in-plant or external operations at recycling plants. External operations are far more costly than in-plant treatment because larger volumes of material require transportation, energy requirements are higher, and additional melting losses are incurred. Metal recovered is of the same alloy content and can be immediately returned to a furnace

An additional source of dross that can contribute significantly to the total metal losses in any operation that melts and transfers aluminum alloys are those drosses generated by turbulent metal transfer. When molten aluminum alloys are poured or pumped with a free fall of the metal stream, large quantities of air can be ingested into the falling stream as it enters the liquid bath. A comparable analogy is the generation of the “head” of foam that occurs when a glass of beer is poured. The white drosses generated by any turbulent transfer of metal can be quite significant and moreover are among the richest in contained aluminum alloy of all drosses generated in aluminum processing (Fig. 10). The entering stream of metal brings one layer of oxide to the pool but further ruptures the surface oxide on the waiting bath and generates new oxides at the turbulent point of entry. The problem is compounded by the fact that not all of the oxides generated in the turbulent transfer are collected in the dross layer on the surface of the melt. Often, the smaller oxides have not had enough time to coalesce into the dross layer and may be contained in the molten

998 / Casting of Nonferrous Alloys metal and carried downstream to the finishing operations that make cast or wrought product. There, they can be the source of lower mechanical properties, porosity in the final product, or even grow into hard spots with detrimental and costly losses in machining operations.

Economic Implications of Dross Treatment of the drosses found on any aluminum melt surface varies widely from operation to operation. Some foundries perform a very careful fluxing and cleansing process on the material, and others merely skim the material into a convenient container provided by the local scrap dealer. These drosses constitute a large portion of the melting loss sustained by all operating aluminum melters and will typically be in the range of 1 to 2% of the weight of clean and dry material, but may be as high as 6 to 10% for some of the more contaminated charge materials. Thus, the drosses generated will vary widely from operation to operation. The most efficient aluminum melters ship out drosses for reprocessing with metal contents of 40 to 50% Al, while those shops doing a poorer job of fluxing and treatment of the drosses in-house (by far the majority) may be shipping out materials with metal content approaching 75 or even 85%. Most aluminum melters generate many thousands of pounds of drosses during the course of a year’s operation, and these metalrich materials leave their plants through a variety of channels. Regardless of the departure channel, the drosses and skimmings will typically wind up at a remote processor who will incur container and transportation charges, the costs to remelt and process the material, the need to landfill the reclamation salts used to improve the recovery processes, the unavoidable metal losses associated with the processing and contained in the salt wastes, and the transportation costs back to the generating source, which will reuse the recovered material in its operation. It is for these reasons and costs that the melt shop generating the original dross will receive compensation for the materials shipped out at rates that vary from 10 to 40% of the true value of the metallics leaving the plant.

In-Plant Enhancement or Recovery of Dross Metallics Historically, most in-house aluminum dross processing was rejected by the foundry as not being part of their core business. Indeed, foundry ingot suppliers maintained buy-back contracts to acquire drosses generated. Several methodologies have been employed by secondary smelters and dross processors whose primary business is to process dross and recover metal values. References 1 to 3 provide further information about out-of-foundry dross processing. Within the foundry, there have been

simple technologies employed to recover dross, especially the little used “dross buggy,” described subsequently. Today (2008), however, there are newer systems available so that the foundry can efficiently and economically process its own drosses and still provide desirable residues for the dedicated dross processor or the secondary smelter. Since aluminum dross can contain upward of 95% Al metal, interspersed in a “lacy” network of aluminum oxide skim, there is a strong incentive to be able to return a good portion of this contained aluminum metal, which is indeed recoverable. With the exception of some wet skimmings, most drosses will contain a percentage of white dross that has the potential to thermite (a reaction where the contained aluminum values burn in the presence of air), consuming any aluminum metal values that are left. Once the thermite reaction has been allowed to start, it can only be stopped by removing all sources of oxygen, cooling the reacting mass to a point where the thermite reaction is quenched, or separating the remaining aluminum, which is the fuel for the reaction. This is the basic approach adopted by several methods of in-plant enhancement/recovery. Dross coolers are not commonly used in the foundry or die casting industry but have seen wide application in other aluminum melting operations. The dross from large aluminum melters can be treated in place with the proper fluxing action and then raked off periodically from the bath surface for further processing. However, dross must not be allowed to cool merely by removing it from the furnace and placing it on the floor or in a dross pan. If this is done, the residual heat contained in the dross pile will actually increase in temperature and thermite. When this occurs, any metallic aluminum that has been entrapped in the dross layer will burn to an oxide and therefore be lost from possible secondary metal recovery. For this reason one of the ways such a reaction can be quenched is to use a chamber dross cooler. The chamber dross coolers are usually large, tightly shielded steel clamshell devices that can be plumbed to provide an argon atmosphere. The shielding is required to maintain a tight seal against a vacuum induced by the thermiting reaction, which will attempt to pull in additional oxygen to supply the reaction. The argon atmosphere keeps oxygen from the cooling dross and stops further burning. Dross is pulled from the generating furnace into the transfer pan directly from the furnace and is then placed in the dross cooler, and the chamber is sealed. A rapid transfer is required to minimize losses that will occur if the thermiting reaction is allowed to proceed unchecked. Temperature controls and argon flow can be set and monitored to determine the appropriate cycle time. When the dross has cooled, some metallic aluminum can be recovered in the bottom of the pan or may be drained through an open hole into a lower ingot mold.

Depending on the extent of the normal thermiting reaction, the recoveries at the secondary plant that recycles the uncooled dross may produce recovery gains of as much as 10 to 20% from the cooled mass. Dross Presses. Another method of external dross processing outside the furnace is to use a dross press. The dross is raked off the furnace while still hot and then transported (quickly) in a specially designed dross press mold. The mold is a heavy-walled steel bucket that is placed in the press, as shown in Fig. 11. A press head contoured to the shape of the bucket is then lowered and pressed with force into the dross mass. The action is much like a juice press but without the twist/rotation. The pressing action squeezes the recoverable metallic molten aluminum from the dross, and the metal drains through a small hole in the bottom of the dross transporter into an ingot or sow mold placed below the press. This sow metal can then be sent directly back to the generating furnace, because it has the same composition as when it was taken from the melting furnace. Typical total recoveries in this process are approximately 50%. That is, the metallic content of the dross as removed from the furnace is approximately 80 to 85% as-raked, and after processing in the dross press will produce the squeezed product and metal-dross skulls that have been quenched in the heavy steel molds. Recoveries with the equipment are definitely a function of temperature losses in the transfer, and the process is subject to freezeups with cold transfers of material to the press, although the pressing operation takes only a few minutes.

Fig. 11

Working area of a dross press, with segmented pressing head, transfer vessel, and recovery pan placed below the unit

Dross, Melt Loss, and Fluxing of Light Alloy Melts / 999 One of the secondary benefits of the process is that the steel molds provide a real quench function to extinguish any thermiting reactions that may be occurring in the dross mass. Some operators view this as the primary benefit from the pressing operation and can raise recoveries from the processed dross by as much as 5 to 10%. Dross Buggy. It was mentioned previously that removing the aluminum fuel from a mass of white dross was an effective way to stop the thermite reaction that frequently follows a cleaning operation on an aluminum melter. Several manufacturers have made equipment to accomplish this on a volume basis.

Fig. 12

Schematic of the operating positions of the unit pictured in Fig. 13. In this application, all rotary movement of the reaction vessel is done manually, with only the mixing functions automated and timed. Courtesy of Q.C. Designs, Inc.

One type of equipment required to promote the separation includes an insulated ladle that must be preheated and used at the melting furnace to collect the dross as it is raked from the furnace. The ladle is then transported to the mixing station, where an additional flux addition is made, and the hot mass is stirred by a large paddle. As the metal content separates, it either drains into a collector pan below the ladle or is allowed to pool in the bottom of the ladle and then may be tapped into a sow mold. The recovered metal can be recycled back into the melt furnace, since it will normally be of the same composition as the original melt. This technique is quite simple; however, many metal casters are moving toward more advanced techniques. Dross Boss (Q.C. Designs, Inc.). In recent years, some modifications and adaptations of the concepts from the dross buggy have resulted in a new group of in-plant dross processors designed specifically to serve aluminum melting furnaces generating small to midsized volumes of dross. Process equipment is available to recover a high percentage of the contained alloy from quantities as small as 4.5 kg (10 lb) of hot dross and as large as 225 kg (500 lb). All sizes are available with varying levels of automation, with some smaller units capable of being manually stirred. As with the dross buggy, the new equipment family can only be used with hot dross transferred directly from the melting furnace. The specially designed reaction vessels do not require preheating and will accept a direct transfer of dross from the melter or processing station. The valuable metallics are recovered

after a small addition of exothermic flux is made to the dross mass, which is then stirred to develop a gravity separation of the materials. When the reaction and separation are complete, the recovered metal units are drained into either a collection ingot pan or transferred directly back to the melter in molten form. Either transfer method provides a major benefit in energy savings to minimize the cost of remelting the recovered material and associated melting losses, with 40 to 45% savings if transferred back at 425  C (800  F) and a total savings with molten transfers. While recoveries of up to 80% of the contained aluminum values have been achieved, a more normal recovery will be in the 60 to 70% range of the weight of charged dross and flux. The residual demetallized material will typically have between 25 and 35% contained aluminum and still have enough value to make it a desirable material for recycling and use in the chemical or construction industry. The midsized unit shown schematically in Fig. 12 and pictured in Fig. 13 will handle 70 kg (160 lb) of dross from a reverberatory well and is cycled hourly with approximately 45 kg (100 lb) of dross from a chip-melter installation. Smaller forms of systems are also in service to process the crucible skimmings generated by flux injection/degassing of transfer ladles in sand, permanent mold, and die casting operations. The automatic unit pictured schematically in Fig. 14 will accommodate a maximum charge of 25 kg (60 lb) of crucible skimmings, while recovering +50% of the weight of material transferred as good alloy. A partially automated unit of smaller size recovers more than 3.2 kg (7 lb) of metal per cycle from an average of more than 100 ladles processed per day. The largest number of applications for this equipment are found in service of wet-well

Fig. 14

Fig. 13

Dross Boss unit with 70 kg (160 lb) of hot dross capacity serving a well on a reverberatory melter. The average recovery of metallics is +50% on the weight of dross and flux charged, and the recovered metal is returned in molten form to the well.

Schematic view of an automatic skimming station process unit. The operator is required only to seal the drain hole and transfer the hot dross to the reaction vessel. All other operations of mixing, draining the recovered metal, dumping the demetallized dross, and cleaning the reaction vessel are controlled by a programmed logic controller unit. Courtesy of Q.C. Designs, Inc.

1000 / Casting of Nonferrous Alloys furnaces, where both the hearth and well drosses can be combined to yield quantities of dross that can be conveniently transferred to the process units.

Ways to Reduce Dross Formation The best way to initially deal with dross formation is to minimize it. This can be accomplished by a number of common-sense and conscientious maintenance procedures. The following list enumerates many, but probably not all, of the useful means to minimize dross formation that aluminum casting foundries can employ:  Use clean, dry charge materials.  Use charge materials with high mass-

to-surface-area ratios to minimize the amount of oxide introduced into the furnace.  Purchase molten alloy when feasible.

 Use fluxes to cover the melt and reduce        

oxidation. Minimize melt cycles by stirring and pumping to increase melting rates. Keep furnaces covered as much as possible, especially when holding metal, to reduce oxidation. Keep burner flames from direct impingement on the molten metal. Ensure proper fuel/air ratios in burners to minimize oxidizing conditions. Hold molten alloy at as low a temperature as possible. Transfer molten alloy as little as possible, and minimize turbulence or cascading during tap-out or pumped metal transfers. Employ the use of a dross reclamation system. Coat tools with an appropriate nonwetting wash.

 Keep furnace doors in good condition.  Maintain positive pressure in reverberatory

melting furnaces.

 Ensure that thermocouples are in good work-

ing order and properly calibrated to correctly monitor/control desired temperatures.

REFERENCES 1. V. Kevorkijan, Evaluating the Aluminum Content of Pressed Dross, JOM Feb 2002, p 34 2. D. Groteke and D. Neff, “A Guide to Reducing and Treating Aluminum Dross,” NADCA Publication 526, 2006 3. R. Peterson and L. Newton, Review of Aluminum Dross Processing, Light Met., 2002, p 1029

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1001-1008 DOI: 10.1361/asmhba0005286

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Aluminum Alloy Ingot Casting and Continuous Processes Elwin Rooy, Elwin Rooy & Associates

INGOT CASTING is the vital conduit between molten metal provided by primary production and recycling and the manufacture of aluminum and aluminum alloy products. Extrusions, forgings, sheet, plate, and foil begin as billet and fabricating ingot. Sand, permanent mold, investment, and pressure die castings typically originate in alloyed remelt ingot. A number of ingot casting processes have been developed to assure the soundness, integrity, and homogeneity required by these downstream manufacturing processes. The importance of ingot casting technology to subsequent wrought and shape casting processes is measured by the wide range of material properties and characteristics that can be consistently and reliably developed in these products.

process but must be scalped for indirect extrusion. Ingots for Rolling. Terminologies differ but ingots cast to provide starting stock for sheet, plate, and foil are called sheet ingots or rolling ingots and are of essentially rectangular cross section (Fig. 3). Sizes vary widely and are controlled by coil size considerations and hot mill widths and break-down mill openings. Maximum ingot dimensions are approximately 66 cm (26 in.) in thickness, 200 cm (80 in.) in width, and up to 915 cm (360 in.) in length.

Ingot Forms Remelt Ingot. The metal castings industry requires ingot (Fig. 1) that can be remelted to provide specification-compliant, on-composition molten metal and alloys. Sizes vary according to the volume of operations and the furnace sizes and types that are used for remelting. Typical ingot sizes are 13.5, 22.5, 315, 675, and 990 kg (30, 50, 700, 1500, and 2200 lb). Shape is usually determined by the ingot supplier, but there is general industry conformity. Dimensions, especially thickness relative to width and length, and casting practices are designed to minimize or avoid surface cracks and open shrinkage voids. These and other defects that permit water contamination and retention represent significant safety risks in furnace charging and melting. Ingots up to 22 kg (50 lb) are often notched so that pieces can be broken to facilitate charging and more accurately top off crucible furnace charges. Billets (Fig. 2) are round, cross-sectional lengths used extensively for extrusion. Sizes conform to extrusion press container dimensions, which typically range from 15 to 28 cm (6 to 11 in.) in diameter. Billets with as-cast surfaces are acceptable in the direct extrusion

Fabricating Ingot. Most often consumed in forging, fabricating ingot may be of rectangular or circular cross section. In either case, initial fabricating steps usually involve preworking of the ingot by a series of upsets and draws to assure thoroughly wrought metallurgical structures before performing final hand (open die) or die forging. The use of extruded stock from billet and plate from rolling ingot is also common. Round cross-sectional ingots for forging range in diameters up to 97 cm (38 in.). Fabricating ingot may also be the starting form for rolled aluminum bar and rod and for rolled and drawn wire. Particle Ingot and Powder. Granulated, pebbled, and flake forms are used in chemical processes as reactants and catalysts. Aluminum shot is used for deoxidation in steel production. Atomized aluminum powder is used in applications that include paint, roofing materials, explosives, and powder metallurgy parts.

Molten-Metal Processing Molten metal must be composition controlled and processed before ingots are cast. Alloying, degassing to reduce dissolved hydrogen, treatment to separate and remove nonmetallic contamination, grain refinement, and, in some cases, the addition of elements influencing

Fig. 1

Ingot for foundry remelting. (a) Automatic ingot stacker at casthouse. (b) Stack of aluminum ingot at die casting foundry. Courtesy of Light Metal Age

Fig. 2

Vertical direct chill billet casting machine. Courtesy of Light Metal Age, Dec 1998

1002 / Casting of Nonferrous Alloys

Fig. 3

Rolling ingot from direct chill casters with mold frame in background. Courtesy of Loma Machine

metallurgical structure are all within the province of ingot cast shop operations. The principles of degassing, flux treatment, filtration, and grain refinement have been covered in the Section “Molten-Metal Processing and Handling” in this Volume. Because of the scale and generally common design of most ingot casting facilities, a degree of commonality in process types for addressing the treatment of molten metal has evolved. Large reverberatory furnaces up to 135 Mg (300,000 lb) capacity are used for melting and/or holding. Alloying is accomplished through the addition and solution of metallurgical metals and master alloys. Compositions are confirmed by chemical analysis. In-furnace treatments for adjusting magnesium concentration, separating entrained oxides and other nonmetallics, and reducing dissolved hydrogen levels include the use of solid and gaseous fluxes. In-line processes for the efficient removal of dissolved hydrogen and included matter are almost universally in use. Rotary degassers capable of reducing sparging gas bubble sizes for the most efficient hydrogen removal are preferred over diffusers or other systems. Even micrometer-sized nonmetallics may be removed by the most efficient molten-metal filters, which include deep-bed, rigid media, and porous ceramic designs. Additions of grain refiners are also typically made in-line using rod feeders that introduce highly refined 9.5 mm (3=8 in.) diameter master alloys of aluminum, titanium, titanium-boron, or carbides.

Ingot Casting Processes Nearly all ingots in the form of billet, sheet, and fabricating ingot are produced by the direct chill process or one of its many variations. These process variations are also used to produce a percentage of remelt ingot, including T-cross-sectional lengths as an alternative to open-mold sow, but open-mold casting of these products is more common. Firms specializing in sophisticated complex mold tooling have contributed to standardization across the industry, but specialization based on proprietary designs, developments, and practices continues to complicate discussions of ingot casting process technology. Process principles are less controversial and serve the purpose of this discussion.

Open-Mold Casting Alloyed and unalloyed remelt ingots are predominantly cast in open steel or cast iron molds. Larger forms (>315 kg, or 700 lb), termed sows, are the international standard for unalloyed primary aluminum. Sow may be ladle poured directly from the reduction plant. The pouring practice can be automated by in-line and rotary conveyor systems. Smaller remelt ingot forms (14 to 23 kg, or 30 to 50 lb), termed pig, are more typically cast on conveyor systems. In either case, automation can extend from metering pouring devices to identification, stacking, and bundling after solidification.

Fig. 4

Vertical direct chill casting. (a) General arrangement of process. (b) Mold assembly. Source: Ref 1, 2

Direct Chill Process As originally developed, the direct chill (DC) process is the vertical discontinuous casting of ingot in water-cooled aluminum molds (Fig. 4). The lateral dimensions and shape of the ingot are formed by the mold, while the bottom or butt of the ingot solidifies against a closely fitting contoured bottom block positioned within the mold before molten metal is introduced. Molten metal flows into the cavity formed by the mold and bottom block through a pin or float-controlled spout. As solidification begins, the bottom block is withdrawn from the mold and continues to lower at controlled rates. In this way, cooling water flowing over the surface of the mold or through passages in the mold directly impinges on the ingot surface. Water-cooling patterns may be complex. In some cases, the mold is internally cooled through elaborate passageways and flow-control devices. In others, the mold is externally

Aluminum Alloy Ingot Casting and Continuous Processes / 1003 cooled through water impingement. The pattern of water contact with the emerging ingot can also be manipulated to increase or decrease heat flux as a function of the ingot geometry. Sheet ingot molds are contoured to generate parallel rolling faces in the desired ingot thickness under dynamic steady-state conditions for a given casting speed. However, the solidification that occurs at the start of casting is analogous to that of the permanent mold process for engineered castings, so that the bottom of the ingot is thicker (butt swell) than desired. At the same time, stresses generated as casting progresses deform the ingot butt ends upward, a condition referred to as butt curl. Both conditions can be minimized by moderating the severity of water cooling at the start of the cast and by providing insulation over the center of the bottom block, which results in the retention of higher postsolidification metal temperature. Control of the water-cooling rate has been achieved through controlled variations in dissolved carbon dioxide. Additional developments affecting degrees of butt swell and butt curl are contoured bottom blocks and molds whose convexity can be dynamically altered during the cast. Mold lubrication is required to prevent attack or adherence by the metal being cast. Manually applied greases have been largely replaced by injecting selected oils through a network of small-diameter holes in the upper area of the mold or through grooved rings. A DC casting complex comprises a mold table on which the mold is mounted and through which water is directed to the mold, a platen-mounted bottom block, and a lined casting pit. Maximum length is determined by the depth of the casting pit and the mechanism by which the bottom block movement is controlled. In most cases, the bottom block, attached to a platen, is raised and lowered hydraulically, although cable systems continue to be in limited use. When the desired ingot length is reached, the mold table is tilted or slid away so that the solidified ingot can be removed from the casting pit (Fig. 2, 5b). Removal is usually by overhead cranes with sling or tong attachments. When the ingot has been removed from the casting pit, the mold table is repositioned, and the bottom block is raised into the mold for the next casting cycle. The use of multiple molds and bottom blocks is normal for all production DC units. Large sheet ingot may be cast seven or more at a time, and the largest billet stations are equipped with tooling for multiples exceeding 100.

DC Process Variations Level-Pour DC (FDC) Process with Header. Mold designs, following the successful development of the DC process, include mounting an insulating header above the water-cooled aluminum mold (Fig. 6). Its purpose is to increase aluminostatic pressure and

to promote thermal differentiation during solidification. A corollary benefit is the ability to introduce molten metal to the mold under level-pour conditions through the insulating body rather than through downspouts. The insulating header and mold are separated by metal rings that provide passage for the mold lubricant. The oil ring arrangement also minimizes heat flux between the insulating header and water-cooled mold. The vertical FDC process is most extensively used in the production of billet and large-diameter fabricating ingot. Hollow Round FDC. Hollow round ingot, commonly used for the production of extruded tube, is cast using a sequence of cores lowered with the ingot or by a water-cooled mold piece that forms the inside diameter. Air-Injected FDC. An increase in the overhang of the insulating header and the circumferential injection of air at the header-mold interface effectively reduce mold contact. The result is improved surface quality and a reduction in surface and subsurface disturbances in metallurgical structure. Originally developed in Japan, the process and its variations have had profound influences on billet production. Horizontal Direct Chill (HDC) Casting. The discontinuous nature of DC and FDC casting and the cost of deep casting pits inevitably led to development efforts for the horizontal casting of ingot using DC process principles (Ref 1). In concept, a headered mold turned 90 from vertical is fed with processed alloyed molten metal through a refractory or ceramic insulating panel. A conventional bottom block is used to start casting, but thereafter, ingot is drawn from the mold at controlled rates by a conveyor system. Clamps and a flying saw cut ingot to desired lengths on the conveyor (Fig. 7). The first successful use of the HDC process was the production of multiple-extrusion billet strands (Fig. 8a). Some time later, the process was adapted to produce foundry remelt ingot of extraordinary density and quality relative to open-mold ingot (Fig. 8b). T-bar in weights corresponding to remelt sow is also cast by the HDC process (Fig. 9), although T-bar ingot is more typically produced by the DC process. The ultimate achievement in HDC was its adaptation for full cross-sectional sheet ingot. A significant concern in this development was the gravity-induced variation in metallurgical structures across the thickness dimension. While these variations would result in unacceptable product characteristics in certain alloys and products, extensive industrial experience and testing have proven the process highly advantageous for many common alloy sheet products, with no appreciable effects on product performance. Electromagnetic casting provides ingot surfaces that permit the as-cast application of sheet ingot in hot rolling or reduce scalping requirements. An induction coil shaped to provide the desired ingot cross section provides an

Fig. 5

Direct chill caster with (a) mold frame in position and (b) mold frame partially tilted up. Courtesy of Light Metal Age, Feb 1995

Fig. 6

Insulating header above the water-cooled mold. Source: Ref 2

electromagnetic field through which molten metal passes. Eddy currents induced in the molten column are strongly repulsed by the electromagnetic field. As a result, molten metal is suspended and lowered into the cooling water curtain without mold contact. While a nonmagnetic mold or shroud may still be employed for safety reasons, there is normally no mold contact, and solidification conditions that affect conventional DC process surface and subsurface structures are avoided. Ingot surfaces are typically smooth, and alternating segregation bands (Fig. 10) and oxide folds are avoided.

Particle Processes Aluminum and its alloys can be valuable as reactants and catalysts in chemical processes. Depending on the application, it may be desirable to maximize or minimize surface-tovolume ratio. For this reason, processes have been developed to produce a variety of particle

1004 / Casting of Nonferrous Alloys

Fig. 8

Horizontal direct chill caster. (a) Four-strand billet caster. (b) Twenty-strand foundry alloy ingot caster. Courtesy of Light Metal Age, Feb 1995. Source: Ref 3

Fig. 7

Mold package for horizontal casting. Source: Ref 2

Fig. 10

Fig. 9

T-bar ingot-handling system for a continuous horizontal universal caster. Courtesy of Hertwich Engineering-Austria

ingot sizes and shapes. Most particle products are unalloyed, but purity can be crucially important. Many particle products are cast from refined metal of >99.99% purity. Most particle products are sorted by size and dried following production. Highly automated systems separate and package powder fractions. Pebbled ingot is formed by introducing droplets of molten aluminum into flowing water. The droplets can be formed in a vibrating mold wash-coated pan in which holes have been drilled. Droplet size is controlled by the

opening dimension in the distribution pan and its position relative to the water stream. Ovality is controlled by the cooling water stream velocity. Granulated Ingot. Granules are produced by the deposition of molten aluminum droplets on a moving water-saturated porous fiber belt. The granules are characterized by flat and rounded surfaces. Flake Ingot. Flakes of aluminum particles are produced by droplets of molten aluminum falling onto cooling water flowing over an

Characteristic alternating bands of direct chill ingot

inclined surface. An alternate process drops a stream of molten aluminum onto a rapidly revolving water-covered cylinder. Deoxidation Products. Particles used in steelmaking for deoxidation are called deox shot and are formed by pouring molten aluminum onto a rotating disk above a cooling water tank. The particles leaving the disk under centrifugal pressure may be directed into the water tank or pass through a screen that reduces particle size before quenching. Aluminum powder is produced by air-inspirated atomization or by the mechanical or pneumatic disruption of a molten aluminum stream.

Continuous Processes While the DC process and its variations dominate in the production of ingot for wrought

Aluminum Alloy Ingot Casting and Continuous Processes / 1005 product manufacture, the development of continuous processes (Ref 4) has resulted in significant production volume with equally significant advantages. Continuous processes provide commercial alternatives to conventional ingot casting. Among the more important incentives for continuous operation are:     

Reduce capital investment Improve recovery Eliminate process steps Conserve energy Promote product uniformity

Alternatives to conventional ingot for rolled products, sheet, and foil include strip and slab casting. Each offers production rates at significantly reduced capital cost when compared with the equipment, facilities, and mills required for conventional hot rolling. A number of intermediate processing steps are eliminated. Neither process requires scalping, which reduces throughput recovery and contributes to increased melt losses. No stress relief, homogenization, preheating for hot rolling, and reheating during hot rolling are required. These processing steps, not always mandatory for all products, involve repetitive energy-intense heating and cooling. There are no end-crop losses that are experienced in conventional hot rolling. Typically, edge trim is eliminated or greatly reduced. Because strip and slab casting are continuous, ingot structural differences associated with starting and stopping in vertical DC casting that represent nonsteady-state solidification conditions are avoided, and product uniformity is improved by sustained solidification under steady-state conditions. Twin-roll strip casting (Fig. 11) is a continuous process in which molten metal is introduced through refractory or ceramic spout tips into the gap between counterrotating watercooled rolls, producing strip at conventional reroll gages (3.18 to 11.4 mm, or 0.125 to 0.450 in.) that typically serve as stock for subsequent cold rolling or, in some cases, may be rolled in-line to finished sheet gages. Depending on size, design, and operational practices, the average annual capacity of a strip caster is approximately 18,000 Mg (40 million lb). Casting speeds range widely but average approximately 2.0 to 2.5 m/min (7 to 8 ft/ min). Strip cast stock is the principal source of aluminum foil. Rolls must be effectively coated to prevent adherence of the solidified strip. This is accomplished by automated spray applications. Solidified sheet is coiled directly and sheared to coil size without interruption. Widths are established by edge dams or other methods. Thickness is established by the roll gap. Molten metal contacts the rolls behind the roll nip so that solidification precedes the exit point and an element of extrusion is included in the solidification process. The amount of setback of the feed tip through which

Fig. 11

Typical arrangement of a twin-roll horizontal caster used in the aluminum industry. Source: Ref 5

Fig. 12

Block caster

molten metal flows into the roll gap is a significant and controllable process variable. Because of extremely high heat-extraction rates, strip casting is alloy-sensitive, and highly alloyed compositions or those that require specific microstructures may not be suitable for production by these processes. This is especially true for dispersion-hardened compositions in which severe centerline segregation is encountered. Slab Casting. Molten metal solidified between moving belts or blocks produces a slab of 10 to 45 mm (0.4 to 1.8 in.) thickness that is reduced in-line by high-torque, slow-speed rolling mills to produce either hot-band or finishgage sheet. Molten metal is introduced through refractory or ceramic feed tips. Thickness results from the positioning of upper and lower belts/blocks. Width is determined by the position of edge dams. As in the case of twin-roll casters, the process is continuous. Slab casters (Fig. 12) demonstrate production rates an order of magnitude greater

(180,000 Mg/yr, or 400 million lb/yr) than strip casters and, as opposed to strip casting, are relatively alloy-tolerant. Casting speeds approximate 4 m/min (13 ft/min). The metallurgical structure of the slab and the reroll gage product is similar to that of metal produced from the hot rolling of conventionally cast ingot. In the twin-belt system (Fig. 13), tension and elaborate guide and support rolls are used to maintain flatness and minimize distortion. Cooling is accomplished by arrays of cooling water applicators and removal devices located on the reverse sides of each belt. The twin-block caster employs interlocking precision-machined blocks that form the rolling surfaces. The blocks are water cooled and dried during the cycle time that blocks are not in contact with the slab. Despite the precision of the blocks, molten aluminum nevertheless penetrates the seams formed by adjacent blocks, and while penetration is limited, sheet cosmetics after rolling are affected.

1006 / Casting of Nonferrous Alloys cross section and belt from either the 12:00 or 3:00 positions, using level transfer in the former and spout control in the latter. The belt is held over the cavity through tension and guides for up to 235 of wheel travel. The solidified bar is stripped from the wheel cavity and guided into an in-line multistand mill to produce coilable rod at typical diameters of 9.5 to 12.7 mm (0.375 to 0.500 in.). For electrical conductance alloys, conductivity is established by composition and through control of mill entry temperature and the reduction/cooling sequence in the mill. Production rates can exceed 5700 kg/h (12,500 lb/h). Continuous Extrusion Processes. Concepts for the direct production of extrusions from either molten metal or compacted scrap have been explored and developed.

Fig. 13

Fig. 14

Twin-belt slab caster. Courtesy of Light Metal Age. Source: Ref 6

Wheel-belt caster

Comparisons with Conventional Product. In all cases, metallurgical structure results from alloy composition, solidification conditions, and the rate of postsolidification cooling. Process differences and resulting differences in metallurgical structure may be considered advantageous or disadvantageous, depending on the product and its application. Solidification conditions characteristic of strip casting result in structures that are more susceptible to centerline segregation. The degree of segregation can be moderated by adjustments in casting rate, cast thickness, tip setback, and cooling as well as through downstream thermal treatments, but it remains a factor in any comparison with DC cast product. By comparison, solidification rates in slab casting approximate those of conventional DC casting, resulting in similar properties. Crystallographic texture, which dominates anisotropy and formability, results from the relationship of relative amounts of hot and cold reductions in the forming of wrought products.

For this reason, there are significant differences between continuous strip and slab processes, in which the extent of hot reduction is necessarily limited, and sheet and foil originating in thick ingot that undergoes extensive hot working to the reroll gage. Wheel-Belt Processes. Rod for electrical applications is almost universally produced in continuous wheel-belt processes (Fig. 14), which generate a trapezoidal bar that is rolled in-line to rod dimensions in triaxial multistand mills. Alloys for mechanical applications may also be produced by specialized wheel-belt casting machines. There are a large number of caster designs and a wide range of production capabilities. Casting rates are controlled by wheel material and diameter, belt material, and cooling. Belt material and its length are associated with longer run-times. Not all casters are capable of producing high-strength, heat treatable alloys. Molten aluminum is typically introduced into the cavity formed by the water-cooled wheel

Solidification in the DC Process The conditions under which solidification occurs in the conventional DC process are complex and involve mold length, cooling water flow rates, method of mold cooling, water distribution, mold length, casting rate, ingot size, metal temperature, and alloy. Molten metal initially flowing into the space created by the mold and bottom block solidifies on contact. The thin shell of solidified aluminum on the mold surface contracts, creating an air gap between the shell and mold wall. The heat mass of molten metal is sufficient to remelt the solidified shell, causing molten metal to again flow against the mold surface. This liquation sequence is repetitive until the shell thickness increases sufficiently that the heat mass at the solidification front no longer prevents its continuous growth. The first metal to solidify is relatively pure compared to alloy composition, according to Hoopes law. Metal flowing into the air gap after remelting of the ingot shell is alloy-enriched, resulting in a region of reverse segregation. The result is, in addition to normally occurring macrosegregation across the ingot cross section, an ingot surface and subsurface affected by alternating alloy-depleted and alloy-enriched zones (Fig. 10) also containing oxide films. A stable, solidified shell is formed before exiting the mold, at which point thermal extraction is greatly accelerated by direct contact with the flow of cooling water. The DC process is capable of producing dense metallurgical structures. The depth of the liquid-solid zone at the solidification interface is limited by steep thermal gradients, and there is an unlimited source of thermally differentiated liquid metal for shrinkage compensation. The ratio of thermal gradient and rate of solid metal formation also favors fine dendrite arm spacing and enhances grain-refiner effectiveness. Inclusions and hydrogen porosity are avoided by in-line metal treatment and the control of turbulence during mold filling and subsequent molten-metal flow.

Aluminum Alloy Ingot Casting and Continuous Processes / 1007

Postsolidification Processes Rolling and fabricating ingot or billet may be thermally treated to reduce residual stresses, homogenize the microstructure, and remove the surface disturbed zone and dimensional irregularities from scalping. Stress Relief. Residual stress levels can be sufficient to cause failure by cracking or splitting during casting. Failure results when residual stresses created by the differences in postsolidification cooling rates across the ingot cross section exceed the strength of the alloy at the temperature where failure occurs. The tendency for such failure is increased by the elevated-temperature strength of the alloy, ingot thickness, and severity of cooling. One of the techniques developed to reduce the risk of failure during casting is the removal of cooling water from the ingot surface at a point below the depth of the molten crater, which allows reheating and promotes the homogeneity of temperatures across the ingot thickness. For many thick or large-diameter highstrength alloys, postsolidification stress relief through furnace treatments approximating annealing is necessary. Homogenization. Depending on composition, ingot may also be homogenized using long, higher-temperature heating cycles and controlled cooling from homogenization temperature. The intent is the uniform distribution of soluble phases across grains and the control of the morphology of certain insoluble phases and other effects based on the limited diffusion of insoluble phases. Principles are those applicable to the elimination or minimization of microsegregation by diffusion. Homogenization may be incorporated in preheating procedures for fabricating operations. Scalping. As-cast ingot surfaces are typically characterized by surface and subsurface variations in structure and chemistry, by dimensional variations, and, in some cases, by surface defects such as oxides, oxide/salt patches, and shallow cracks. Scalping is performed to remove the disturbed zone—defined as the surface layer comprised of alternate segregation and oxide folds—the oxide surface formed during solidification, surface defects, and to provide dimensional regularity. The only significant exception is the use of billet with as-cast surfaces in direct extrusion. In direct extrusion, the surface of the billet is retained in the press at the end of the extrusion cycle. Electromagnetically cast ingot is still typically scalped, but scalping depth is sharply reduced. Removal of the oxide that forms at the surface before and during solidification improves sheet/plate finish and reduces hot mill roll coating.

charging, this discussion of safety is limited to ingot casting. It is not intended that the discussion be comprehensive nor form the exclusive basis for safe practices. Every aluminum ingot casting facility involves molten aluminum and large quantities of water in its operations. Conditions may result in the risk of explosions, ranging from minor to violent reactions, affecting equipment, facilities, and the safety of personnel. The most serious safety concerns are related to the DC process and its variations. There are a number of ways molten aluminum may become exposed to the cooling water. Bleedouts occur when the ingot shell melts below the bottom of the mold. This could result from improper mold cooling, excessive metal temperature, or excessive casting rate. Ingot may adhere to the mold wall, resulting in a hangup. Lubrication failure and mold condition are probable factors. When the solidified body of metal falls or when it is forced from the mold, molten aluminum may flow directly into the casting pit. Equipment malfunctions can contribute to these and other unsafe conditions. Rules and practices have been developed to minimize these risks. Most are derived from extensive testing in which quantities of molten aluminum were intentionally introduced into water-filled containers. Testing permitted the assessment of the influence of metal quantities, addition rates, composition and temperature, water temperature and cleanliness, drop height, stream dimensions, and container materials and coatings. The analysis of large numbers of these tests resulted in conclusions regarding variables that either promoted or diminished the probability of explosion occurrence. A generalized summary of conclusions is instructive:  Explosions were not experienced if metal

stream diameter was less than 70 mm (23/4 in.).

 Violent explosions are most likely to occur 

     

Safety



While there are substantial concerns over the safety of molten-metal transport and furnace



when molten aluminum flows into water at depths between 50 and 760 mm (2 and 30 in.). Explosions were more likely to occur when drop height ranged from 38 to 1220 mm (1½ to 48 in.). No explosions occurred with drop heights exceeding 3 m (10 ft). Explosion tendencies are inversely related to water temperature. Explosion tendencies increase with metal temperature. Explosion tendencies are reduced when water is oil-contaminated. Oxidized iron or steel surfaces promote explosions. The presence of hydroxides on container surfaces promotes the occurrence of explosions. The presence of lithium, lead, and bismuth in the alloy increases the risk of explosions. Container walls with inorganic coatings reduce the risk of explosions. Any form of impact concurrent with the introduction of molten aluminum into the

water-filled container increases the risk of explosions.  Explosions do not occur (with the exception of lithium-containing alloys) unless liquid metal contacts the container surface. Ingot casting operations sensitive to these conclusions have adopted practices that reduce the probability of explosions. Casting pit walls, platens, and bottom block assemblies are coated with selected inorganic coatings and/or compounds. Many of these that are based on petroleum or coal tars are no longer used for environmental reasons. Current recommendations can be obtained from The Aluminum Association, which has sponsored programs for improving ingot casting safety and also summarizes explosion-related incidents for the aluminum industry. Coatings are routinely inspected and repaired. Pumps removing water from the casting pit are operated to maintain a minimum water depth of 914 mm (36 in.) (over debris that may collect there). Debris is routinely removed. Casting crews are trained to recognize the development of unsafe conditions during the cast and to use appropriate abortive procedures. Fail-safe engineering solutions involving cooling water supply and cylinder operation have been designed and implemented. REFERENCES 1. C.M. Adam, Overview of D.C. Casting, Proceedings of the 1980 Conference on Aluminum-Lithium Alloys, The Metallurgical Society, 1981, p 39–48 2. “Alcoa Casting Systems,” Report 22, Alcoa Technology Marketing Division, Oct 1974 3. H. Zeillinger and A. Beevis, Universal Continuous Horizontal Caster for Ingot, Billet, or Bar, Light Met. Age, June 1997, p 20 4. R.E. Spear and K.J. Brondyke, Continuous Casting of Aluminum, J. Met., April 1971 5. B.Q. Li, Producing Thin Strip by Twin-Roll Casting, Part I: Process Aspects and Quality Issues, J. Met., May 1995, p 29 6. P. Regan, Recent Advances in Aluminum Strip Casting and Continuous Rolling Technology—Implications for Aluminum Can Body Sheet Production, Light Met. Age, Feb 1992, p 58 SELECTED REFERENCES  M.G. Chu and J.E. Jacoby, Macrosegregation

Characteristics of Commercial Size Aluminum Alloy Ingot Cast by the Direct Chill Method, Light Metals 1990, TMS, 1990  D.P. Cook, B.Q. Li, and J.W. Evans, Electromagnetic Casting: Mathematical and Physical Models, Light Metals 1989, TMS, 1989  J.M. Drezet and M. Plata, Thermomechanical Effects During Direct Chill and Electromagnetic Casting of Aluminum Alloys, Part I, Light Metals 1995, TMS, 1995

1008 / Casting of Nonferrous Alloys  S.G. Epstein, Causes and Prevention of Molten

 B. Frischnecht and K.P. Maiwald, Roll

 R. Sautebin and W. Haller, Industrial Appli-

Aluminum-Water Explosions, Proceedings of the Fourth International Aluminum Extrusion Technology Seminar, 1988  S.G. Epstein, Causes and Prevention of Molten Aluminum-Water Explosions, Light Metals 1991, TMS, 1991  J.P. Faunce, F.E. Wagstaff, and H. Shaw, New Casting Method for Improving Billet Quality (Air-Slip), Light Metals 1984, TMS, 1984  H.G. Fjaer and E.K. Jensen, Mathematical Modeling of Butt Curl Deformation of Sheet Ingots, Light Metals 1995, TMS, 1995

Caster Applications and Developments, Light Metals 1988, TMS, 1988 Guidelines for Handling Molten Aluminum, The Aluminum Association Y. Ishii, HDC Process for Small Diameter Ingot, Light Metals 1989, TMS, 1989 I. Jin, L.R. Morris, and H.D. Hunt, Centre Line Segregation in Twin Roll Cast Aluminum Alloy Slab, Light Metals 1982, TMS, 1982 A. Mo et al., Modelling the Surface Segregation Development in DC Casting, Light Metals 1994, TMS, 1994

cation of Electromagnetic Casting, Light Metals 1985, TMS, 1985  R.E. Spear and K.J. Brondyke, Continuous Casting of Aluminum, J. Met., April 1971  W. Szczpiorski and R. Szczpiorski, The Mechanical and Metallurgical Characteristics of Twin-Belt Cast Aluminum Strip, Light Metals 1991, TMS, 1991  R.P. Taleyarkhan, S.H. Kim, and C.L. Knaff, Fundamental Studies on Aluminum-Water Explosion Prevention in Direct Chill Casting Pits, Light Metals 2001, TMS, 2001

  



ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1009-1018 DOI: 10.1361/asmhba0005287

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Aluminum Shape Casting ALUMINUM CASTING ALLOYS are the most versatile of all common foundry alloys and generally have the highest castability ratings. As casting materials, aluminum alloys have the following favorable characteristics:
Good fluidity for filling thin sections
Low melting point relative to those required

for many other metals


Rapid heat transfer from the molten alumi







num to the mold, providing shorter casting cycles Hydrogen is the only gas with appreciable solubility in aluminum and its alloys, and hydrogen solubility in aluminum can be readily controlled by processing methods. Many aluminum alloys are relatively free from hot-short cracking and tearing tendencies. Chemical stability Good as-cast surface finish with lustrous surfaces and little or no blemishes

Aluminum is one of the few metals that can be shape cast by essentially all existing processes, including pressure die casting, permanent mold, clay/water-bonded sand, chemically bonded sand, plaster mold, and investment casting. Important variations include molding and pattern distinctions such as lost foam (evaporative pattern), shell and V-mold, and process derivatives such as squeeze casting, lowpressure permanent mold, vacuum riserless casting, and semisolid forming based on rheocasting/ thixocasting principles. Of these casting methods, high-pressure die casting of aluminum is the dominant nonferrous casting process (Fig. 1). This article provides an overview on the common methods of aluminum shape casting. Details on melting and melt treatment are covered in the article “Dross, Melt Loss, and Fluxing of Light Alloy Melts” in this Volume. Structure control and grain refinement are detailed in the following articles in this Volume:

Melting and Melt Handling Aluminum and aluminum alloys can be melted in a variety of ways. Coreless and channel induction furnaces, crucible and openhearth reverberatory furnaces fired by natural gas or fuel oil, and electric resistance and electric radiation furnaces are all in routine use. The nature of the furnace charge is as varied and important as the choice of furnace type for metal casting operations. The furnace charge may range from prealloyed ingot of high quality to charges made up exclusively from low-grade scrap. Even under optimum melting and melt-holding conditions, molten aluminum is susceptible to three types of degradation:
With time at temperature, adsorption of

hydrogen results in increased dissolved hydrogen content up to an equilibrium value for the specific composition and temperature.
With time at temperature, oxidation of the melt occurs; in alloys containing magnesium, oxidation losses and the formation of complex oxides may not be self-limiting.
Transient elements characterized by low vapor pressure and high reactivity are reduced as a function of time at temperature; magnesium, sodium, calcium, and strontium, upon which mechanical properties directly or indirectly rely, are examples of elements that display transient characteristics.

Turbulence or agitation of the melt and increased holding temperature significantly increase the rate of hydrogen solution, oxidation, and transient element loss. The mechanical properties of aluminum alloys depend on casting soundness, which is strongly influenced by hydrogen porosity and entrained nonmetallic inclusions. Reductions in dissolved hydrogen content and in suspended included matter are normally accomplished through treatment of the melt before pouring, as discussed in the following articles in this Volume:
“Degassing”
“Molten-Metal Filtration”
“Aluminum Fluxes and Fluxing Practice”

Gravity Casting. Regardless of the types of melting and holding furnaces and the particular gravity casting process used, there is great concern for reducing or eliminating dissolved hydrogen and entrained oxides. These procedures are less frequently employed for pressure die casting, in which concerns are focused on the dominant process-related causes of casting unsoundness, namely, entrapped gas and pouring injection-associated inclusions. Sensitivity to melt quality varies with the casting process and part design and necessitates special consideration of relevant criteria for each application. In general, the melt is processed to achieve hydrogen reductions and


“Aluminum and Aluminum Alloy Castings”
“Modification of Aluminum-Silicon Alloys”
“Grain Refinement of Aluminum Casting

Alloys”


“Refinement of the Primary Silicon Phase in

Hypereutectic Aluminum-Silicon Alloys”

Fig. 1

Nonferrous casting processes by tons poured. Source: From data of AFS 2002 survey.

1010 / Casting of Nonferrous Alloys the removal of oxides to meet specific casting requirements. Modification and grain-refiner additions are made as appropriate to the given alloy and end product. Die Casting. Different melt preparation practices are employed in die casting operations because process-related conditions are more dominant in the control of product quality than those controlled by melt treatment. For this reason, degassing for the removal of hydrogen, grain refinement, and modification or silicon refinement in the case of hypereutectic silicon alloys are often intentionally neglected. The movement toward higher-integrity die castings has brought into focus the importance of the same melt-quality parameters established and used in the gravity casting of aluminum alloys. In high-production die casting operations, the consumption of internal and external scrap is of primary importance in reducing base metal costs for the predominantly secondary alloy compositions that are consumed. Scrap crushing, shredding, and pretreatment of various types precede melting, often in efficient induction systems. Oxides entrained in the melt as a result of this sequence of operations are dealt with through the use of salts and/or reactive gas fluxing. Melt treatment is typically confined to this and to rudimentary fluxing in holding furnaces to remove gross oxide and to facilitate the maintenance of minimum furnace cleanliness. A concern in die casting is the formation of complex intermetallics that are insoluble at melt-holding temperatures and/or precipitate under holding conditions or during transfer to and injection from the hot or cold chamber. These intermetallics (sludge) affect furnaces, transfer systems, and, by inclusion, the quality of the castings produced. Die casters are familiar with composition limits that prevent sludge formation. A common rule is that iron content plus two times manganese content plus three times chromium content should not exceed the sum of 1.7%. This limit is arbitrary and inexact, it is often assigned values from 1.5 through 1.9%, and it is subject to the specific composition and actual minimum process temperature. Pouring. It is critically important that the metal be drawn and poured according to the best manual or automatic procedures. These procedures avoid excessive turbulence, minimize oxide generation and entrainment, and limit regassing of hydrogen. Frequent skimming of the melt surface from which metal is drawn may be necessary to minimize oxide contamination in the ladle. Siphon ladles that fill from below the melt surface are used for these purposes, but most often, coated and preheated ladles of simple design are employed. The process of repetitive drawing and skimming inevitably degrades melt quality, and this necessitates reprocessing if required melt-quality limits are exceeded. Pouring should take place at the lowest position possible relative to the pouring basin or sprue opening. Once pouring is initiated, the sprue must be continuously filled to

minimize aspiration and to maintain the integrity of flow in gates and runners. Countergravity mold-filling methods inherently overcome most of the disadvantages of manual pouring. Proprietary casting processes based on low pressure, displacement, or pumping mechanisms may be considered optimum for preserving the processed quality of the melt through mold filling, but some important and relevant considerations apply. Melt processing by fluxing is more difficult in some cases because the crucible or metal source may be confined. If, as in the case of low-pressure casting, the passage for introducing metal into the mold is used repetitively, its inner surface becomes oxide contaminated and a source of casting inclusions. In other countergravity casting, mold intrusion into the melt and devices employed to displace or pump metal to the cavity may be the source of turbulence, moisture reactions, and the possibility of hydrogen regassing. Automated pouring systems are common in the die casting industry (see the article “Automation in High-Pressure Die Casting” in this Volume). Robotized ladle transfer, as well as metered pumping, may nevertheless incorporate features and reflect provisions to protect molten-metal quality through sound drawing, transfer, and pouring techniques. The same techniques have application in die casting operations in which these operations are performed manually. Hot chamber operation offers apparent metal transfer advantages over cold chamber operation. Recent developments in the use of siphons or vacuum legs to the cold chamber in pressure die casting offer new and interesting opportunities for upgrading the quality of the metal deliverable to the die cavity. Gating and Risering Principles. The methods for introducing metal into the casting cavity, for minimizing degradation in metal quality, and for minimizing the occurrence of shrinkage porosity in the solidifying casting differ among the various casting processes, primarily as a function of process limitations. However, the objectives and principles of gating and risering are universally applicable: Somewhat different techniques in gating and risering are used for different alloys. In general, riser size and the need for stronger thermal gradients increase with more difficult-to-cast alloys. Crack sensitivity or hot shortness forces compromises in the steps normally taken to achieve directional solidification. Extensive localized chilling may aggravate crack formation. In these cases, more uniform casting section thickness, larger fillets, more gradual section thickness changes, larger risers and, in some cases, riser insulation, and more graduated chilling offer the best prospects for success. In alloys that are more difficult to feed but are relatively insensitive to cracking at elevated temperature, establishing thermal gradients by selective chilling (and heating as in permanent mold casting) usually provides good casting results. Examples are alloys high in eutectic

content, as well as purer compositions, in which the solidification range is limited. In these cases, localized areas of shrinkage are inevitable in the absence of adequate thermal gradients accompanied by effective risering. Gating of Die Castings. Conventional pressure die casting uses gating principles that are different from those for gravity casting. Although the fundamental principles employed in gravity casting remain desirable, injection under high pressure at significant metal velocity precludes application of many of the rules that govern gravity casting processes. In pressure die casting, gates and runners are the means by which molten metal is transferred from the shot chamber or injection system to the die impression. The gating system must be designed to permit the attainment of cavity pressurization without reducing cycle time by its own mass and solidification time. It must also be of minimum size to maximize the gross-to-net-weight ratio and to minimize trimming and finishing costs. The objective of gating in die casting is filling of the die cavity by establishing uninterrupted frontal flow. This requires the prevention of excessive turbulence and mixing within the die cavity, and it minimizes the entrainment of air and volatiles derived from the injection system and from lubricants contained in the die cavity. Runners are usually semiellipsoidal and decrease in cross-sectional area in the direction of metal flow. Their respective cross-sectional areas should exceed corresponding gate dimensions. At no time should gate thickness exceed that of the casting. Gates are normally severely tapered at the entry point(s) to facilitate removal during trimming with minimal risk of damage to the casting. The greatest progress has been made in research dedicated to die design and the design of metal entry conditions, including sprue, runner, gating, and parameters of injection. Based on carefully analyzed applied research as well as modeling of the die-filling sequence, mathematical and instrumented programs are now available to die casters for the development, modification, and control of gating design and operation. Much less sophisticated mathematical relationships and nomographs have been employed for many years to determine the relative dimensions of plunger diameters, runners, gates, and process parameters, including fill rate and velocity, plunger velocity, gate velocity, and system pressures.

Casting Process Selection Selection of a casting process may depend on several different factors, such as:
Casting

process considerations: requirements for fluidity, resistance to hot tearing, minimization of shrinkage tendencies
Casting design considerations: draft, wall thickness, internal passages. (For example,

Aluminum Shape Casting / 1011












parts with undercuts and complex internal passageways can usually be made by sand, plaster, or investment casting but may be impractical or impossible to produce in permanent mold or pressure die casting.) Mechanical property requirements: strength and ductility, hardness, fatigue strength, toughness, impact strength, specification limits Physical property requirements: electrical and thermal conductivity, specific gravity, expansion characteristics Process requirements: machinability, brazeability, weldability, impregnation, and chemical finishing Service requirements: pressure tightness, corrosion resistance, wear resistance, elevated-temperature strength, dimensional and thermal stability Economics: volume, productivity, process yield, material costs, tooling costs, cost of machining, welding, and heat treatment

In most cases, dimensions, design features, and material property requirements limit the range of candidate processes. Typical design details for aluminum castings are shown in Fig. 2.

Fig. 2

Dimensional tolerances are compared in Tables 1 to 3, for different shape casting processes. Table 4 compares general characteristics of sand casting, permanent mold casting, and die casting of aluminum. Although aluminum alloys can be produced by any of the available casting methods, highpressure die casting is the dominant method (Fig. 1). This process is very common to nonferrous metals with lower-temperature melting points. It is ideally suited for high production rate and volume production of dimensionally accurate parts with excellent surface finish. One of the important reasons for the success of die castings has been the development of high-speed precision equipment. Another is the extension of die casting technologies to larger castings with heavier wall thicknesses. When two or more casting methods are feasible for a given part, the method used very often is dictated by costs. Quality factors are also important in the selection of a casting process. When applied to castings, the term quality refers to both degree of soundness (freedom from porosity, cracking, and surface imperfections) and levels of mechanical properties (strength and ductility). The following

tabulation presents characteristic ranges of cooling rate for the various casting processes: Casting processes

Cooling rate 

C/s



F/s

Dendrite arm spacing mm

in

Plaster, 0.05–0.2 0.09–0.36 0.1–1 0.004–0.04 dry sand Green sand, 0.1–0.5 0.18–0.9 0.05–0.5 0.002–0.02 shell Permanent 0.3–1 0.54–1.8 0.03–0.07 0.001–0.003 mold Die 50–500 90–900 0.005–0.015 0.002–0.0006 Continuous 0.5–2 0.9–3.6 0.03–0.07 0.001–0.003

However, it should be kept in mind that in die casting, although cooling rates are very high, air tends to be trapped in the casting, which gives rise to appreciable amounts of porosity at the center. Extensive research has been conducted to find ways of reducing such porosity; however, it is difficult if not impossible to eliminate completely, and die castings often are lower in strength than low-pressure or gravity-fed permanent mold castings, which are more sound in spite of slower cooling. Expendable Mold Methods. Sand casting is the major type of expendable mold method. The advantages of typical sand casting are

Suggested design details for aluminum alloys. Recommended dimensions are averages, and the use of either larger or smaller numerical factors may result in more difficult casting or defects. Source Van Hoen, Vol. III.

1012 / Casting of Nonferrous Alloys Table 1

Dimensional limits and tolerances for aluminum castings produced by different processes

Sand castings

Diameter of cored holes: Tolerance  þ0.75 mm (0.030 in.). If cored hole is to be used for clearance and a tolerance of þ0.75 mm (0.030 in.) is necessary, hole should be ordered  undersized and machined to tolerance. Location of cored holes: þ1.25 mm (0.050 in.) approximate minimum for large-volume  acceptance (þ 0.38 mm, or 0.015 in.) for precision sand castings); can be obtained either by direct tolerance or by reducing hole size correspondingly. If more than one diameter is located from a common centerline, a concentricity tolerance for all diameters can be maintained within  þ0.75 mm (þ þ0.015 in., for precision sand 0.030 in.) (þ 0.38 mm, or  method), except as noted under “diameter of cored holes.” A tolerance of þ0.25 mm  (0.010 in.) (commonly specified) cannot be met consistently. Straightness: þ0.75 mm (0.030 in.) per 25 mm (1 in.) of length. Min of 0.13 mm (0.005 in.)  for precision sand castings. Allowance for machining: 1.8 mm (0.070 in.): on long, thin castings allowance should be increased to 3.18 mm (0.125 in.) Parallelism: Same as under “Straightness” Permanent mold castings

Maximum length of core supported at one end: 10  core diam Draft on outer surfaces: 1 min, 3 preferred Draft in recesses: 2 min, 5 preferred Draft on cores: ½ min, 2 preferred Diameter of core: 6.4 mm (¼ in.) min Allowance for machining: 0.8 mm (1/32 in.) min for castings up to 255 mm (10 in.) long 1.2 mm (3/64 in.) desirable; 1.2 mm (3/64 in.) min for castings over 255 mm (10 in.) long, 1.6 mm (1/16 in.) desirable; 1.6 mm (1/16 in.) for surfaces formed by sand cores Minimum radius of fillet: Average thickness of joining walls Straightness: 0.25 to 150 mm (0.010 to 6 in) plus 0.030 mm (0.0012 in.) per 25 mm (1 in.) of length up to 915 mm (36 in.) Die castings

Draft on outside surface: ½ min Draft in cored holes: 0.50 mm (0.020 in.) per 25 mm (1 in.) of depth for 2.5 to 3.2 mm (1/10 to 1/8 in.) diam; 0.41 mm (0.016 in.) per 25 mm (1 in.) of depth for 3.2 to 6.3 mm

versatility in a wide variety of alloys, shapes, and sizes. Alloys considered hot short because of cracking tendencies during solidification are more easily cast in green sand since molds offer reduced resistance to dimensional contraction during solidification. Lower mold strength is also advantageous for parts with widely varying section thicknesses and intricate designs. These advantages are diminished in chemically bonded molds, which display greater rigidity than green sand molds. There are only practical limitations in the size of the parts that can be cast. Minimum wall thickness is normally 4 mm (0.15 in.), but thicknesses as little as 2 mm (0.090 in.) can be achieved. Suggested tolerances for sand castings are in Table 5. Sand castings are relatively lower in dimensional accuracy with poor surface finish; basic linear tolerances of  þ30 mm/m (þ 0.03 in./in.), with a minimum tolerance of 20 mm/m (0.020 in./in.), and surface finishes of 6 to 13 mm (250 to 500 min.) root mean square (rms) are typical. Chemically bonded sands offer improved surface finish and dimensional accuracy and relatively unlimited shelf life, depending on the type of binder used. Achievable dimensional tolerances can be substantially improved using precision methods in the forming and assembly of dry sand mold components. Surface quality can be improved by using a finer grade of facing sand in the molding process. Strength is typically lower as a result of slow solidification rates.

(1/8 to ¼ in.) diam; 0.30 mm (0.012 in.) per 25 mm (1 in.) of depth for 6 to 25 mm (¼ to 1 in.) diam; 0.30 mm (0.012 in.) per 25 mm (1 in.) of depth. plus 0.05 mm (0.002 in.) per 25 mm (1 in.) of diam for diameters over 25 mm (1 in.) Plaster mold castings

Maximum length of core supported at one end: 5  diam of core Draft on cores: Zero draft often permissible, ½ min, otherwise, with 2 preferred Diameter of core: 6.3 mm (¼ in.) min Minimum radius of fillet: Sharp corners can be cast Flatness: 0.038 mm (0.0015 in.) per 25 mm (1 in.) of length Shell mold castings

Draft on outside surfaces: ½ min, 2 preferred Straightness: 0.05 mm (0.002 in.) per 25 mm (1 in.) of length Investment castings

Maximum length of core supported at one end: 5  diam of core (on some parts 10  diam) Location of cored hole: 0.10 mm (0.004 in.) per 25 mm (1 in.) of dimension Concentricity: 0.05 mm (0.002 in.) per 25 mm (1 in.) Straightness: 0.13 mm (0.005 in.) per 25 mm (1 in.) of length Centrifugal permanent mold castings

Diameter of core: 3.2 mm (1/8 in.) min Draft on core: ½ min for 1  diam of core, 1½ preferred; 1 min for 2  diam, 2 preferred; 1½ for > 2  diam, 3 preferred Contour surfaces: 0.13 mm (0.005 in.) min for surfaces up to 50 mm (2 in.) long, 0.25 mm (0.010 in.) preferred; 0.25 mm (0.010 in.) 25 mm (1 in.), min for surfaces over 50 mm (2 in.) long; 0.025 mm (0.001 in.) for each additional 25 mm (1 in.) 0.76 mm (0.030 in.) max Minimum radius of fillet: 0.13 mm (0.005 in.) fillet radius equal to web thickness preferred Angularity: 0 50 per 25 mm (1 in.) up to 25 mm (1 in.); 0 50 per 25 mm (1 in.) for each additional 25 mm (1 in.) to 300 max

Other expendable mold methods include:
Lost foam casting with unbonded sand mold

compacted around an expendable polystyrene pattern
Shell mold casting, which surpasses ordinary sand castings in surface finish and dimensional accuracy (table and cool at slightly higher rates; however, equipment and production costs are higher, and the size and complexity of castings that can be produced are limited.
Plaster casting with a permeable (aerated) or impermeable plaster mold. The high insulating value of the plaster allows castings with thin walls to be poured. Minimum wall thickness of aluminum plaster casting is typically 1.5 mm (0.060 in.), Plaster molds have high reproducibility, permitting castings to be made with fine details and close tolerances; basic linear tolerances of  þ5 mm/m (þ 0.005 in./in.) are typical. The surface finish of plaster castings is excellent; aluminum castings attain finishes of 1.3 to 3.2 mm (50 to 125 min.) rms. Suggested tolerances are in Table 6.
Investment casting of aluminum most commonly employs ceramic molds and expendable patterns of wax, plastic, or other low-temperature melting materials. Aluminum investment castings can have walls as thin as 0.40 to 0.75 mm (0.015 to 0.030 in.) basic linear tolerances of  þ5 mm/m

(þ 5 mils/in), and surface finishes of 1.5 to 2.3 mm (60 to 90 min.). Because of porosity and slow solidification, the mechanical properties of many aluminum investment castings are typically lower than those demonstrated by other casting processes. The interest of the aerospace and other industries in the combination of accurate dimensional control with controlled mechanical properties has resulted in the use of improved technologies to produce premium-quality castings by investment methods. Castings in the premium strength range can be achieved with molten-metal treatments, gating, and solidification conditions that are not typical for conventional investment castings. Investment casting applications include: instrument parts, impellers, compressor vanes, gears, ratchets, pawls, scrolls, speed brakes, wing tips, and aircraft pylons. Permanent Mold Casting. In principle, permanent mold casting is analogous to expendable mold casting processes. In this case, the molds are machined cast, wrought or nodular iron, cast steel, or wrought steel and can be reused repetitively until damage or wear necessitates repair or replacement. Intricate details and undercuts can often be cast using segmented steel cores. Permanent mold tooling is typically more expensive than that required for sand casting and other expendable mold processes and is justified by the volume of production.

Aluminum Shape Casting / 1013 Table 2

Dimensional tolerances of aluminum alloy castings Tolerance for dimensional limits, mm (in.)

Casting process

Under 25 mm (1 in.)

Over 25 mm (1 in.), add per 25 mm (1 in.)

Across parting line Sand þ0.38 (þ  0.015) Die . . .. Permanent mold  þ0.38 (þ 0.015) Plaster mold þ0.25 (þ  0.010) Shell mold þ0.25 (þ0.010)   Centrifugal(a) þ0.25 (þ  0.010)(b) Investment . . ..

. . .. . . .. +0.051 (+0.002) +0.038 (+0.0015) +0.030 (+0.0012) +0.025 (+0.001)(c) . . ..

Between points for one part of mold Sand þ0.38 (þ0.015)   Die . . .. Permanent mold  þ0.38 (þ 0.015) Plaster mold þ0.13 (þ0.005)   Shell mold þ0.13 (þ  0.005) Centrifugal þ0.25 (þ0.010)(b)   Investment . . ..

. . .. . . .. +0.025 (+0.001) +0.038 (+0.0015) +0.030 (+0.0012) +0.025 (+0.001)(c) . . ..

Sketch of dimensional requirement

Between points produced by core and mold Sand þ0.38 (þ0.015) . . ..   Die . . .. . . .. Permanent mold þ0.38 (þ0.015) +0.05 (+0.002)   Plaster mold þ0.25 (þ +0.038 (+0.0015)  0.010) Shell mold þ0.25 (þ0.010) +0.038 (+0.0015)   Centrifugal(a) þ0.38 (þ +0.025 (+0.001)(e)  0.015)(d) Investment . . .. . . ..

(a) Using permanent molds. (b) For dimensions under 50 mm (2 in.)  þ0.13 mm (þ 0.005 in.) min. (c) For dimensions over 50 mm (2 in.) +0.25 mm (+0.010 in.) per 25 mm (1 in.) min. (d) For dimensions under 50 mm (2 in.)  þ0.25 mm (þ 0.010 in.) min. (e) For dimensions over 50 mm (2 in.) þ0.38 mm (þ0.015 in.) per 25 mm (1 in.) min.  

While the principles and mechanics of gravity casting are similar, the metallurgical structure of permanent mold castings reflects the refinement of higher solidification rates. Typical and specified minimum mechanical properties, including ductility, are higher than those of expendable mold castings. The improved mechanical properties of permanent mold castings provide part of the justification for selecting this process over competing gravity casting options. There is a downsprue into which molten metal is introduced from a pouring basin, from which the metal flows into runners, risers, infeeds, and casting cavity. Directional solidification is promoted by selective chilling of mold sectors by air, mist, or water. An insulating coating is used to protect the mold from the molten aluminum and to facilitate removal of the casting from the mold after solidification is complete. Typical mold dressings or washes are suspensions of talc, various metal oxides such as zirconia, chromia, iron oxide and titania, colloidal graphite and calcium carbonate in water, and sodium silicate. The thickness and thermal characteristics of these coatings are used to locally increase or decrease heat absorption during solidification. As a result of wear, these coatings must be periodically repaired or replaced to ensure consistent process performance and casting results. Mold surfaces are periodically blasted with dry ice or

mild abrasives to remove coatings and scale, after which new mold coatings are applied. The permanent mold process is less alloy-tolerant than most expendable mold processes. The most popular permanent mold alloys display superior castability, such as those of the Al-Si, Al-Si-Mg, and Al-Si-Cu (Mg) families. Mold rigidity is a challenge in the casting of hot-short alloys in which liquidus-solidus range and elevated-temperature strength combine to increase the tendency for cracking during and after solidification. Determined efforts to cast even the most difficult foundry alloys, such as low-iron aluminum-copper alloys, have nevertheless been successful, and alloys with limited castability are routinely cast in permanent molds. Suggested tolerances are in Table 7. Permanent mold castings can be produced in sizes ranging from less than 0.5 kg (1 lb) to more than several hundred pounds. Surface finish typically varies 3.8 to 10 mm (150 to 400 min.). Basic linear tolerances are approximately  þ10 mm/m (þ 0.01 in./in.), and minimum wall thicknesses are approximately 2.5 mm (0.100 in.). Centrifugal Casting. Centrifugal force in aluminum casting involves rotating a mold or a number of molds filled with molten metal about an axis. Baked sand, plaster, or graphite molds have been used, but iron and steel dies are most common. Centrifugal castings are generally, but not always, denser than conventionally poured castings and offer the advantage of greater detail.

Wheels, wheel hubs, motor rotors, and papermaking and printing rolls are examples of aluminum parts produced by centrifugal casting. Aluminum alloys suitable for permanent mold, sand, or plaster casting can be cast centrifugally. Squeeze Casting. Although a number of process developments have been referred to as squeeze casting, the process by which molten metal solidifies under pressure within closed dies positioned between the plates of a hydraulic press is the only version of current commercial interest. Squeeze casting has been successfully used for a variety of ferrous and nonferrous alloys in traditionally cast and wrought compositions. Applications of squeeze-cast aluminum alloys include reciprocating engine pistons, brake rotors, automotive and truck wheels, and structural automotive frame components (Fig. 3). Squeeze casting is simple and economical, is efficient in its use of raw material, and has excellent potential for automated operation at high rates of production. Semisolid forming incorporates elements of casting, forging, and extrusion. It involves the near-net shape forming of metal parts from a semisolid raw material that incorporates a uniquely nondendritic microstructure. Semisolid forming is more costly than conventional casting but offers unique properties and consistently excellent quality. In addition, the viscous nature of semisolid alloys provides a natural environment for the incorporation of thirdphase particles in the preparation of reinforced metal-matrix composites.

Pressure Die Casting of Aluminum Pressure die casting came into existence in the early 1820s in response to the expanding need for large volumes of cast print type. Iron or steel dies had been used in casting print type in lead-base alloys in the 17th century. Iron molds were also used in colonial times to cast pewter. Intensive efforts to employ iron and steel molds in the casting of aluminum resulted in commercial “permanent mold” operations by the first decade of the 20th century. Progress in die casting aluminum was limited until the development of the cold chamber process in the 1920s. In pressure die casting, molten metal is injected under pressure into water-cooled dies. Pressure is maintained until the part has solidified. Molten metal usually enters the mold by the action of a hydraulic ram in a containment chamber (shot chamber), resulting in rapid filling of the mold cavity. While lubrication is required to facilitate the separation of the casting from the die surface, the dies are otherwise uninsulated. Dies are usually machined from high-quality tool steels. The die casting process has undergone significant changes through the evolution of machine design and instrumentation as well as process

1014 / Casting of Nonferrous Alloys Table 3 Typical tolerances of aluminum alloy castings Metal

Aluminum-silicon (400 series) Aluminum-copper (200 series) Aluminum-silicon (300 series) Aluminum-silicon (400 series) Aluminum-copper (200 series) Aluminum-silicon (300 series) Aluminum-magnesium (500 series) Aluminum-magnesium (500 series) Aluminum-copper (200 series) Aluminum-silicon (400 series) Aluminum-silicon (300 series) Aluminum-silicon (400 series) Aluminum-silicon (300 series) Aluminum-silicon (300 series) Aluminum-magnesium (500 series) Aluminum-silicon (400 series) Aluminum-silicon (400 series) Aluminum-copper (200 series) Aluminum-silicon (300 series) Aluminum-magnesium (500 series) Aluminum-silicon (300 series) Aluminum-silicon (400 series) Aluminum-copper (200 series) Aluminum-copper (200 series) Aluminum-magnesium (500 series) Aluminum-silicon (300 series) Aluminum-silicon (400 series) Aluminum-copper (200 series) Aluminum-magnesium (500 series) Aluminum-silicon (300 series) Aluminum-silicon (300 series) Aluminum-magnesium (500 series) Aluminum-copper (200 series) Aluminum-silicon (300 series) Aluminum-silicon (300 series) Aluminum-magnesium (500 series) Aluminum-silicon (400 series) Aluminum-copper (200 series) Aluminum-silicon (300 series) Aluminum-magnesium (500 series) Aluminum-silicon (400 series) Aluminum-copper (200 series) Aluminum

From dimension

To dimension

Lower tolerance

Upper tolerance

Premium tolerance

Parting line increment

Process(a)

mm

in.

mm

in.

mm

mm

mm

mm

Centrifugal Centrifugal Centrifugal Ceramic Ceramic Ceramic Ceramic EPC/full mold EPC/full mold EPC/full mold EPC/full mold Gravity die Gravity die Green sand Green sand Green sand Investment Investment Investment Permanent mold Permanent mold Permanent mold Permanent mold Plaster Plaster Plaster Plaster Plaster Plaster Plaster Pressure die casting Pressure die casting Pressure die casting Pressure die casting Shell Shell Shell Shell No-bake/air set/CO15 No-bake/air set/CO17 No-bake/air set/CO18 No-bake/air set/CO28 Sand

25 25 25 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 0 152

1 1 1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 6

305 305 305 610 610 610 610 254 254 254 254 152 152 152 152 152 254 254 254 152 152 152 152 2540 2540 2540 2540 2540 2540 2540 152 152 152 152 152 152 152 152 152 152 152 152 152 1016

12 12 12 24 24 24 24 10 10 10 10 6 6 6 6 6 10 10 10 6 6 6 6 100 100 100 100 100 100 100 6 6 6 6 6 6 6 6 6 6 6 6 6 40

2.5 2.5 2.5 0.25 0.25 0.25 0.25 0.38 0.38 0.38 0.38 0.64 0.64 0.38 0.38 0.38 0.25 0.25 0.25 0.23 0.23 0.23 0.23 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.13 0.13 0.13 0.13 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.81 0.81

in.

0.1 0.1 0.1 0.01 0.01 0.01 0.01 0.015 0.015 0.015 0.015 0.025 0.025 0.015 0.015 0.015 0.01 0.01 0.01 0.009 0.009 0.009 0.009 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.005 0.005 0.005 0.005 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.032 0.032

3.8 3.8 3.8 1.5 1.5 1.5 1.5 0.89 0.89 0.89 0.89 1.3 1.3 0.5 0.5 0.5 1.3 1.3 1.3 0.30 0.30 0.30 0.30 1.3 1.3 1.3 1.3 1.3 1.3 1.3 2.0 2.0 2.0 2.0 0.36 0.36 0.36 0.36 0.36 0.36 0.36 0.36 0.81 0.81

in.

0.15 0.15 0.15 0.06 0.06 0.06 0.06 0.035 0.035 0.035 0.035 0.05 0.05 0.02 0.02 0.02 0.05 0.05 0.05 0.012 0.012 0.012 0.012 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.08 0.08 0.08 0.08 0.014 0.014 0.014 0.014 0.014 0.014 0.014 0.014 0.032 0.032

1.3 1.3 1.3 0.13 0.13 0.13 0.13 0.05 0.05 0.05 0.05 0.25 0.25 0.51 0.51 0.51 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.38 0.038 0.038 0.038 0.038 0.038 0.038 0.025 0.025 0.025 0.025 0.23 0.23 0.23 0.23 0.23 0.23 0.23 0.23 ... ...

in.

0.05 0.05 0.05 0.005 0.005 0.005 0.005 0.002 0.002 0.002 0.002 0.01 0.01 0.02 0.02 0.02 0.005 0.005 0.005 0.005 0.005 0.005 0.005 0.0015 0.0015 0.0015 0.0015 0.0015 0.0015 0.0015 0.001 0.001 0.001 0.001 0.009 0.009 0.009 0.009 0.009 0.009 0.009 0.009 ... ...

0 0 0 0.25 0.25 0.25 0.25 0 0 0 0 0.38 0.38 0.792 0.76 0.76 0 0 0 0.38 0.38 0.38 0.38 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.51 0.51 0.51 0.51 ... ...

in.

0 0 0 0.01 0.01 0.01 0.01 0 0 0 0 0.015 0.015 0.0312 0.03 0.03 0 0 0 0.015 0.015 0.015 0.015 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.02 0.02 0.02 0.02 ... ...

First/base tolerance mm/m

10 10 10 6.0 6.0 6.0 6.0 3.0 3.0 3.0 3.0 15.6 15.6 25 25 25 1.0 1.0 1.0 15.6 15.6 15.6 15.6 2.0 2.0 2.0 2.0 2.0 2.0 2.0 6.0 6.0 6.0 6.0 5.0 5.0 5.0 5.0 20 20 20 20 ... ...

in./in.

0.01 0.01 0.01 0.006 0.006 0.006 0.006 0.003 0.003 0.003 0.003 0.015625 0.015625 0.025 0.025 0.025 0.001 0.001 0.001 0.015625 0.015625 0.015625 0.015625 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.006 0.006 0.006 0.006 0.005 0.005 0.005 0.005 0.02 0.02 0.02 0.02 ... ...

(a) EPC, evaporative pattern casting. Source: American Foundry Society, 2008

Table 4 Comparison of sand casting, die casting, and permanent mold casting processes for aluminum alloys Casting process Factor

Cost of equipment Casting rate Size of casting External and internal shape Minimum wall thickness Type of cores Tolerance obtainable Surface finish Gas porosity Cooling rate Grain size Strength Fatigue properties Wear resistance Overall quality Remarks

Sand casting

Permanent mold casting

Die casting

Lowest cost if only a few items required Lowest rate Largest of any casting method Best suited for complex shapes where coring required 3.0–5.0 mm (0.125–0.200 in.) required; 4.0 mm (0.150 in.) normal Complex baked sand cores can be used

Less than die casting 11 kg/h (25 lb/h) common; higher rates possible Limited by size of machine Simple sand cores can be used, but more difficult to insert than in sand castings 3.0–5.0 mm (0.125–0.200 in.) required; 3.5 mm (0.140 in.) normal Reusable cores can be made of steel, or nonreusable baked cores can be used Best linear tolerance is 10 mm/m (10 mils/in.)

Highest 4.5 kg/h (10 lb/h) common; 45 kg/h (100 lb/h) possible Limited by size of machine Cores must be able to be pulled because they are metal; undercuts can be formed only by collapsing cores or loose pieces 1.0–2.5 mm (0.040–0.100 in.); depends on casting size

Poorest; best linear tolerance is 300 mm/m (300 mils/in.) 6.5–12.5 mm (250–500) min.) Lowest porosity possible with good technique 0.1–0.5  C/s (0.2–0.9  F/s) Coarse Lowest Good Good Depends on foundry technique Very versatile as to size, shape, internal configurations

4.0–10 mm (150–400 min.) Best pressure tightness; low porosity possible with good technique 0.3–1.0  C/s (0.5–1.8  F/s) Fine Excellent Good Good Highest quality ...

Steel cores; must be simple and straight so they can be pulled Best linear tolerance is 4 mm/m (4 mils/in.) 1.5 mm (50 min.); best finish of the three casting processes Porosity may be present 50–500  C/s (90–900  F/s) Very fine on surface Highest, usually used in the as-cast condition Excellent Excellent Tolerance and repeatability very good Excellent for fast production rates

Aluminum Shape Casting / 1015 Table 5

Suggested dimensional tolerances for sand castings Type A dimension: between two points in same part of mold, not affected by parting plane or core Tolerance, þ, mm (in.) 

Specified dimension mm (in.)

Critical

Up through 152 (6) Over 152 to 305 (6 to 12)

0.76 (0.030) 30 + 3.0 mm/m (0.030 + 0.003 in./in.) over 152 mm (6 in.) 48 + 2.0 mm/m (0.048 + 0.002 in./in.) over 305 (12 in.)

Over 305 (12)

Type B dimension: across parting plane. A-type dimension plus following:

Noncritical

1.0 (0.040) 40 + 4.0 mm/m (0.040 + 0.004 in./in.) over 152 mm (6 in.) 64 + 2.0 mm/m (0.064 + 0.002 in./in.) over 305 (12 in.) Type C dimension: affected by core. A-type dimension plus following:

Projected area of casting, A1  A3, cm2 (in.2)

Additional tolerance for parting plane mm (in.)

Projected area of casting affected by core, A3  G, cm2 (in.2)

Up through 65 (10) Over 65 to 325 (10 to 50) Over 325 to 645 (50 to 100) Over 645 to 1615 (100 to 250) Over 1615 to 3225 (250 to 500)

0.5 (0.020) 0.89 (0.035) 1.1 (0.045) 1.5 (0.060) 2.3 (0.090)

Up through 65 (10) Over 65 to 325 (10 to 50) Over 325 to 645 (50 to 100) Over 645 to 3225 (100 to 500) Over 3225 to 6450 (500 to 1000)

D dimension: draft

Additional tolerance for core, mm (in.)

0.5 (0.020) 0.89 (0.035) 1.1 (0.045) 1.5 (0.060) 2.3 (0.090)

F dimension: allowance for finish Draft, deg

Location

Outside wall Recesses Cores

Critical

2 3 2

Noncritical

Maximum dimension, mm (in.)

3 5 3

Up through 152 (6) Over 152 to 305 (6 to 12) Over 305 to 455 (12 to 18) Over 455 to 610 (18 to 24)

E dimension: minimum wall thickness, 3.8 mm (0.150 in.)

Nominal allowance, mm (in.)

1.5 2.3 3.0 4.6

(0.060) (0.090) (0.120) (0.180)

Minimum diameter of cored holes, 6.4 mm (0.250 in.)

development and controls. The demand for larger, more complex castings with improved quality and lower cost has led to the development and promotion of specialized die casting machines capable of higher rates of production and improved performance. There are two basic concepts: hot and cold chamber operation. In the hot chamber process, the shot chamber and piston are immersed in molten metal. Metals such as magnesium and zinc that do not aggressively attack the materials of construction can be efficiently cast by this method, with production rate advantages. Despite intensive efforts to develop hot chamber process designs and materials that could be used in aluminum casting, none have been commercially successful. Except in rare cases, all aluminum die casting is performed in cold chamber equipment in which the shot chamber is filled with each cycle, and the chamber and

piston assembly are not continuously in contact with molten aluminum. Die casting machine designs are also differentiated by parting plane orientation. In practice, the dies are mounted on platens that can operate in either vertical or horizontal directions. Early die casting was typically performed with vertical die movement. Today (2008), with exceptions, vertical die casting is restricted to rotor production. Locking pressure defines the capacity of the machine to contain the pressure generated during the injection cycle. The larger the plan area of the casting and the greater the hydraulic pressure applied, the greater the required locking pressure. Die casting machines are designed with locking pressures from as little as 25 tonnes to more than 4500 tonnes corresponding to injection pressures of up to 280 MPa (40 ksi).

The principles of directional solidification and gravity-based gating and risering are essentially inapplicable to die casting. Gates and runners are used to convey metal from the shot chamber to the die cavity. Geometrical considerations are observed to minimize turbulence, air entrapment, and fragmentation of the metal stream, but no effort other than the use of sustained pressure is used to promote internal soundness. Techniques are used to intensify pressure during the solidification phase to decrease the volume fraction of internal porosity. With metal velocities exceeding 30 m/s (100 ft/s) and solidification rates exceeding 1000  C/s (1800  F/s), the greatest quality concern is entrapped gases, including combustion or volatilization components of the lubricant and turbulence-related inclusions. Rapid filling of the mold (20 to 100 ms) and rapid solidification under pressure combine to produce a consistently dense, fine-grained, and highly refined surface structure with excellent properties, including fatigue strength. Internal unsoundness affects bulk properties, the acceptability of machined surfaces, and pressure tightness. Impregnation is routine for die castings that must contain gases or liquids under substantial pressure. Internal unsoundness generally prevents full heat treatment and welding because of the risk of blister formation when die castings are exposed to elevated temperatures. Lower-temperature thermal treatments for stabilization or hardening are routinely used. In special cases and in restricted casting areas, limited welding can also be performed. The die casting process is the least alloy-tolerant of important commercial casting processes. Solidification conditions require alloys of superior castability and that display good resistance to cracking at elevated temperatures. The highly castable alloys of the aluminum-silicon family are the most common. Of these, alloy 380.0 and its variations comprise approximately 85% of total die casting production. These compositions provide attractive combinations of cost, strength, hardness, and corrosion resistance in the as-cast state, with excellent fluidity and resistance to hot cracking. Aluminum-silicon alloys lower in copper, such as 360.0, 364.0, 413.0, and 443.0, offer improved corrosion resistance and excellent castability. Hypereutectic aluminum-silicon alloys, including 390.0, have become more important in wear-resistant applications. Magnesium content is usually controlled at low levels to minimize oxidation and the generation of oxides in the casting process. Most commonly used die casting alloys specify restrictive magnesium limits. Nevertheless, aluminum-magnesium alloys can be die cast. Alloy 518.0, for example, is specified when the highest corrosion resistance and the brightest, most reflective finish are required. Iron contents of 0.7% or greater are preferred to maximize elevated-temperature strength, to facilitate ejection, and to minimize soldering

1016 / Casting of Nonferrous Alloys Table 6 Suggested dimensional tolerances for plaster castings Type A dimension: between two points in same part of mold, not affected by parting plane or core Specified dimension mm (in.)

Critical

Up through 25 (1) Over 25 (1)

Tolerance,  þ, mm (in.)

0.13 (0.005) 5.0 + 1.0 mm/m (0.005 + 0.001 in./in.) over 25 mm (1 in.)

Type B dimension: across parting plane. A-type dimension plus following: Additional tolerance for parting plane, mm (in.)

Projected area of casting, A1  A3, cm2 (in.2)

Up through 65 (10) Over 65 to 325 (10 to 50) Over 325 to 645 (50 to 100) Over 645 (100)

0.13 0.25 0.51 0.76

(0.005) (0.010) (0.020) (0.030)

D dimension: draft

Noncritical

0.25 (0.010) 10 + 2.0 mm/m (0.010 + 0.002 in./in.) over 25 mm (1 in.)

Type C dimension: affected by core. A-type dimension plus following: Projected area of casting affected by core, A3  G, cm2 (in.2)

Additional tolerance for core, mm (in.)

Up through 65 (10) Over 65 to 325 (10 to 50) Over 325 to 645 (50 to 100) Over 645 (100)

0.13 (0.005) 0.51 (0.020) 0.76 (0.030) 1.1 (0.045)

F dimension: allowance for finish Draft, deg

Location

Outside surface Recesses Cores

Critical

0 0 0

Noncritical

Maximum dimension, mm (in.)

Nominal allowance, mm (in.)

2 2 2

Up through 125 (5) Over 125 to 305 (5 to 12) Over 305 to 455 (12 to 18)

0.51 (0.020) 0.76 (0.030) 1.0 (0.040)

E dimension: minimum wall thickness, 1.5 mm (0.060 in.)

to the die face. Iron content is usually 1  þ 0.3%, but greater concentrations are also used. Improved ductility through reduced iron content has been an incentive resulting in widespread efforts to develop a tolerance for iron as low as approximately 0.25%. These efforts focus on process refinements, design modifications, and improved die lubrication. At higher iron concentrations, there is a risk of exceeding solubility limits of coarse Al-Fe-Cr-Mn segregate at molten-metal temperatures. Sludging or precipitation of segregate is prevented by chemistry controls related to metal temperature. A common rule is: Fe + 2 (Mn) + 3(Cr) + 1.7, where element values are expressed in weight percent. Aluminum alloys can be die cast to tolerances of  þ4 mils/in. (þ 0.004 in./in.) and commonly have finishes as fine as 1.3 mm (50 min.). Parts are cast with walls as thin as 1.0 mm (0.040 in.). Cores, which are made of metal, are restricted to simple shapes that permit drawing or removal after solidification is complete. Suggested tolerances are in Table 8.

Minimum diameter of cored holes, 6.4 mm (0.250 in.)

Premium Engineered Castings A premium engineered casting is one that provides higher levels of quality and reliability than found in conventionally produced castings. Premium engineering includes intimately detailed design and control of each step in the manufacturing sequence. The results are minimally variable premium strength, ductility, soundness, dimensional control, and finish. Castings of this classification are notable primarily for mechanical property performance that reflects extreme soundness, fine dendrite arm spacing, and well-refined grain structure. Premium engineered casting objectives require the use of chemical compositions competent to display superior properties (see the article “Aluminum and Aluminum Alloy Castings” in this Volume). Most of the compositions designated as premium engineered alloys had their origins in the 1950s and early 1960s. Alloy A356.0, registered in 1955, and B356.0, in 1956, were derivatives of 356.0 alloy, which

was originally developed in 1930. Similarly, alloy C355.0 dates from 1955, while the parent 355.0 was first used in 1930. Alloys 359.0 and 354.0 were developed in 1961. Alloy A357.0 (1962) had its origins in Tens-50 alloy that was also first registered in 1961. Premium engineered aluminum castings represent the culmination of decades of research and development involving molten-metal treatment, methods for the removal of included matter, measurement or assessment of dissolved hydrogen, gating development, microstructural modification and refinement, casting processes, mold materials, and the development of highstrength, ductile alloys. Each of these developments was necessary to advance casting capabilities to meet an increasingly challenging range of application requirements. Melt processing and solidification of aluminum are discussed in more detail in other articles in this Volume. Fluxing and degassing are key aspects in the production of premium aluminum castings. More details on the melting, melt treatment, and melt handling of aluminum are discussed in other articles in this Volume. To a large extent, melt quality has been assessed by variations of the Straube-Pfeiffer test in which the relationship of hydrogen solubility and pressure was qualitatively measured (see the article “Approaches to Measurement of Melt Quality” in this Volume). The absence of entrained oxides was assessed by their influence on hydrogen precipitation under reduced pressure. A semiquantitative approach using sample densities was developed and used extensively as a process control tool. For greater sensitivity, controlled vibration during sample solidification at absolute pressures of 2 to 5 mm Hg (0.04 to 0.10 psi) was employed. Real-time measurement of dissolved hydrogen by partial pressure diffusion in molten metal supplemented vacuum test results. Validation of hydrogen assessments was provided by solid-state extraction techniques. In terms of solidification, simulation models and programs help put into practice the principles of directional solidification for more complex cast parts. The relationship of properties and dendrite arm spacing also has significantly influenced the premium engineered casting effort in terms of both strength and ductility. Mold Materials. Investment casting produced small parts in which dimensional accuracy and surface finish were important criteria. Plaster molds were used in the production of dimensionally accurate cast parts such as impellers and tire molds. There were corollary pattern, mold material, and mold processing developments. Differences in the molding and heat extraction characteristics of differing sands were measured, and at the same time, supplier developments in dry sand binders were studied and evaluated. In permanent mold, variations in mold wash chemistry and application were used with air, mist, and water cooling to control solidification. On occasion, copper and ceramic inserts were

Aluminum Shape Casting / 1017 Table 7 Suggested dimensional tolerances for permanent and semipermanent mold castings Type A dimension: between two points in same part of mold, not affected by parting plane or moving parts Specified dimension mm (in.)

Critical

Up through 25(1) Over 25(1)

Tolerance, þ, in. 

0.25 (0.010) 10 + 1.5 mm/m (0.010 + 0.0015 in./in.) over 25 mm (1 in.)

Type B dimension: across parting plane. A-type dimension plus following:

Noncritical

0.38 (0.015) 15 + 2.0 mm/m (0.015 + 0.002 in./in.) over 25 mm (1 in.)

Type C dimension: affected by moving parts. A-type dimension plus following: Additional tolerance, mm (in.)

Additional tolerance for parting plane mm (in.)

Projected area of casting, A1  A3, cm2 (in.2)

Up through 65 (10) Over 65 to 325 (10 to 50) Over 325 to 645 (50 to 100) Over 645 to 1615 (100 to 250) Over 1615 to 3225 (250 to 500)

0.25 0.38 0.51 0.64 0.76

Projected area of casting affected by moving part, A3  G, cm2 (in.2)

(0.010) (0.015) (0.020) (0.025) (0.030)

Up through 65 (10) Over 65 to 325 (10 to 50) Over 325 to 645 (50 to 100) Over 645 to 3225 (100 to 500) Over 3225 to 6450 (500 to 1000)

D dimension: draft

Metal core or mold

0.25 0.38 0.38 0.56 0.81

(0.010) (0.015) (0.015) (0.022) (0.032)

Sand core

0.38 (0.015) 0.64 (0.025) 0.76 (0.030) 1.0 (0.040) 1.5 (0.060)

E dimension: minimum wall thickness

Draft, deg Location

Outside surface Recesses Cores

Critical

2 2 1

Noncritical

3 5 2

Maximum dimension, mm (in.)

Minimum wall thickness, mm (in.)

Up through 76 (3) Over 76 to 152 (3 to 6) Over 152 (6)

3.5 (0.140) 4.1 (0.160) 4.8 (0.188)

F dimension: allowance for finish Nominal allowance, mm (in.) Maximum dimension, mm (in.)

Metal core or mold

Up through 152 (6) Over 152 to 305 (6 to 12) Over 305 to 455 (12 to 18) Over 455 to 610 (18 to 24)

0.76 (0.030) 1.1 (0.045) 1.5 (0.060) 2.3 (0.090)

Sand core

1.5 2.3 3.0 4.6

(0.060) (0.090) (0.120) (0.180)

Minimum diameter of cored holes, 9.5 mm (0.375 in.)

designed into the mold to promote solidification directionality. The use of steel chills, contoured sections, and mold plates was normal in sand casting. The use of all available mold components from most insulating (foamed plaster) to most rapid heat extraction (water-cooled copper) could be incorporated in mold designs to reproducibly alter solidification in order to promote the highest possible degree of internal soundness. Computer simulations of mold filling dynamics and solidification that incorporate differences in heat extraction through finite-element analysis are gradually supplanting art and instinct in process designs.

The result is a composite mold design of significant complexity using dry sand and at least several other mold materials for each configuration. Mold Filling. The inevitable degradation of melt quality that occurs in drawing and pouring through conventional methods was recognized as a significant barrier to achieving premium engineered quality levels and properties. Very significant efforts had been made to evaluate different pouring and gating approaches. These included siphon ladles, sprue/gate/runner ratios, runner overruns, dross traps, pouring cup designs, strainers and screens, and nonvortexing cross-sectional geometry of downsprues and

runners. The use of optimal gating designs and rigorously controlled practices are integral components of premium casting technology. Several developments have altered mold-filling options; these developments are described in the following paragraphs. First, the low-pressure casting process achieved commercial importance in the United States in the 1950s. A number of challenges that were unaddressed by commercial low-pressure systems were successfully met. A camcontrolled back-pressure method based on gross casting weight was used to retain residual metal levels at the top of the feed tube. This feature prevented the inclusion-spawning characteristic of normal low-pressure cycles. In-gate filtration and screening methods were also devised. The range of part designs and alloys that were cast would be considered unusual today (2008), when the low-pressure process has become principally known for automotive wheel production. Instead, the low-pressure method was considered a means of nonturbulent mold filling with a number of additional advantages that included reduced gross/net weight and lower pouring temperature. Some examples were diesel engine and compressor pistons, air conditioner compressor bodies, bearings, furniture parts, and missile fins. Geometric symmetry, which is normally a criterion for low-pressure production, was not considered a prerequisite, and many of the castings that were produced used conventional risering rather than exclusively relying on the in-feed for shrinkage compensation. These developments were adapted to premium engineered plaster and dry sand parts. High-speed rotors and impellers were excellent examples, but many other premium engineered casting configurations were made by low-pressure mold filling. Second, countergravity mold-filling methods were developed involving the use of mechanical or induction pumps. Third, various techniques have been developed and are in use for filling molds quiescently by displacement of molten metal. Fourth, solidification time was not a significant factor in expendable mold production when extensive chilling was used, but it was always a factor in permanent mold. For this reason, and to overcome other low-pressure process limitations, vacuum riserless casting (VRC) was developed in the early 1960s. Rather than pressurize a contained molten reservoir, the application of a vacuum on the mold cavity drew metal from the bath through a short fill tube. The metal source was exposed for periodic treatment, the distance from subsurface metal entry to the casting cavity was minimal, dies were extensively chilled, and the process could be highly automated. While only relatively small and simple shapes were produced by the VRC method, productivity and mechanical properties were exceptional. More than 20 million air conditioner pistons and millions of rocker arms were produced by this process.

1018 / Casting of Nonferrous Alloys

Fig. 3

Automotive parts produced by squeeze casting

Table 8 Suggested dimensional tolerances for die castings Type A dimension: between two points in same part of die, not affected by parting plane or moving parts Specified dimension, mm (in.)

Up through 25 (1) Over 25 to 305 (1 to 12) Over 305 (12)

Critical

0.10 (0.004) 4.0 + 1.5 mm/m (0.004 + 0.0015 in./in.) over 25 mm (1 in.) 20.5 + 1.0 mm/m (0.0205 + 0.001 in./in.) over 305 mm (12 in.)

Type B dimension: across parting plane. A-type dimension plus following: Projected area of casting, A1  A3, cm2 (in.2)

0.13 0.20 0.30 0.38

(0.005) (0.008) (0.012) (0.015)

D dimension: draft

Up through 2.5 (0.1) Over 2.5 to 13 (0.1 to 0.5) Over 13 to 25 (0.5 to 1.0) Over 25 to 76 (1.0 to 3.0) Over 76 to 230 (3.0 to 9.0)

6 3 2 1

/32

1

Projected area of casting affected by moving part, A3  G, cm2 (in.2)

Up through 65 (10) Over 65 to 130 (10 to 20) Over 130 to 325 (20 to 50) Over 325 to 645 (50 to 100)

Additional tolerance for moving part, mm (in.)

0.13 0.20 0.30 0.38

(0.005) (0.008) (0.012) (0.015)

E dimension: minimum wall thickness, 2.3 mm (0.090 in.) Draft, deg

Critical

Noncritical

0.25 (0.010) 10 + 2.0 mm/m (0.010 + 0.002 in./in.) over 25 mm (1 in.) 32 + 1.0 mm/m (0.032 + 0.001 in./in.) over 305 mm (12 in.)

Type C dimension: affected by moving parts. A-type dimension plus following:

Additional tolerance for parting plane, mm (in.)

Up through 325 (50) Over 325 to 645 (50 to 100) Over 645 to 1290 (100 to 200) Over 1290 to 1935 (200 to 300)

Depth of draw, mm (in.)

Tolerance, þ, mm (in.) 

Noncritical

18 6 3 1½ 1

F dimension: allowance for finish Maximum dimension, mm (in.)

Nominal allowance, mm (in.)

Up through 125 (5) Over 125 to 305 (5 to 12) Over 305 to 455 (12 to 18)

0.51 (0.020) 0.76 (0.030) 1.0 (0.040)

Minimum diameter of cored holes, 3.56 mm (0.140 in.)

Fifth, the level pour process for premium engineered castings was developed. This process had its origins in the direct-chill process that was developed for fabricating ingot. It was natural that the shared concerns for metal distribution and solidification principles would result in the synthesis of process concepts. Aluminum-tin alloy bearings, which were typically hollow or solid cylinders, could be produced by the direct-chill process, but segregation and safety concerns led to a variation in which metal was introduced to the bottom of a permanent mold through a moving pouring cup that traversed the length of the mold. Quiescent flow and the continuous layering of molten metal provided improved internal quality. Excellent soundness was obtained without the use of the extensive risering normally required for these long-solidification-range alloys, pouring temperature was reduced, and the solidified structure was more chemically homogeneous than in conventionally cast parts. With determined engineering, the same approach was successfully used for more complex configurations with more challenging metallurgical requirements. The cost of doing so in permanent mold was not typically justified. However, aircraft and aerospace parts in the limited quantities normally associated with sand casting offered exciting opportunities for level pour technology, and a large number of prototype and production parts were produced by this method. In its final form, the assembled mold was lowered on a hydraulic platform through a trough arrangement that provided nonturbulent flow of metal through entry points that paralleled the vertical traverse of the mold. Metal flow was controlled by the dimensions of the entries and the lowering rate, which could be modulated for cross-sectional variations as a function of mold travel. Characteristic of the direct-chill process on which it was based, the level pour process features quiescent moltenmetal flow, minimized feeding distances, and reduced pouring temperatures. ACKNOWLEDGMENTS Portions adapted from:
J.G. Kaufman and E. Rooy, Aluminum Cast-

ing Alloys: Properties, Processes, and Applications, American Foundry Society and ASM International, 2004
A. Kearry and E. Rooy, Properties and Selection: Nou Ferrous Alloys and SpecialPurpose Materials, Vol 2, ASM Handbook, ASM International, 1990
E. Rooy, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 743
Van Hoen, Aluminum, Vol. III Fabrication and Finishing, American Society for Metals, 1967

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1019-1025 DOI: 10.1361/asmhba0005288

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Copper Continuous Casting* Derek E. Tyler, Olin Brass (Retired) Richard P. Vierod, Olin Brass

COPPER ALLOY PRODUCTS such as strip, billet, rod, or tube are continuous cast, defined as the continuous solidification and withdrawal of product from an open-ended shaping mold. Methods include both vertical and horizontal casting, depending on product size, shape, and volume. Casting vertically has certain inherent technical advantages. The symmetry of cooling promotes a uniform and predictable solidification growth pattern and uniform axial loading on the freshly solidified shell as it is withdrawn from the mold. In tube or hollow section casting, the vertical process has particular merit. The disadvantages of vertical casting are mostly logistic: difficulty in handling long lengths of section; cut-off can be more difficult to engineer and control; and it is generally a semicontinuous operation. Horizontal casting requires lower capital investment, is compatible with lower production rates, and is a continuous operation. This article briefly reviews the history and methods of copper alloy continuous casting; the information is drawn from the very detailed and extensive coverage of the subject in Ref 1 and the numerous publications of equipment supply companies such as Rautomead, SMS Meer, and so on. Although a relatively simple process, the metallurgical complexities of continuous casting involve the thermal and mechanical interactions between the mold and the moving solidifying shell of the casting. The variations of alloy chemistry, physical, thermal, and mechanical properties invoke detailed changes in the process protocols. The heat transfer within the mold is a key factor in the successful production of a quality product. The equipment suppliers have their own proprietary designs for the mold assembly, and some of the various design details of the dies and cooling assemblies are described in Ref 1. Typically, the molds (dies) are fabricated from copper alloy components that may or may not be sleeved with high-quality graphite inserts. When sleeves are used, precise machining and polishing of the surfaces is essential to provide intimate contact between graphite and

plate cooler to maximize heat transfer. Precise control of the withdrawal parameters then becomes of paramount importance.

History Many of the techniques developed for copper alloys have been adapted for aluminum and steel. In a general timeline of developments (Table 1), the breakthrough in continuous casting of nonferrous alloys can be credited to Eldred in 1930 (Ref 2). He developed a machine using graphite as the mold material

to cast copper rods and later a number of copper-base alloys. Prior to that, the continuous casting of metals had been practiced for almost a century earlier with patents by Sellers and Laign (Ref 3, 4). In 1938, Poland and Lindner were granted a U.S. patent (Ref 7) for a vertical casting machine very similar to Eldred’s. The graphite mold was cooled by a close-fitting metal water jacket (Fig. 1). The potential of graphite as a suitable mold material was quickly appreciated by various companies, such as American Smelting and Refining Company (Ref 8) and Flocast (Ref 17). The Asarco process (Ref 8), patented

Table 1 Historical timeline of continuous casting Year

Development related to continuous casting of copper

1840 1843

The first recorded patent in the nonferrous field was by Sellers for the manufacture of lead pipes (Ref 3). About the same time, Laign filed a patent in America for a method of continuous casting nonferrous metal tube (Ref 4). U.S. patent was granted to Swedish engineer Pehrson for the first horizontal closed-head system for continuous casting of cast iron bars (Ref 5). The Hazelett process was introduced as the first ingotless rolling plant for continuous casting of copper wire rod (Contirod process) and production of continuous cast and sheared copper anode plate for electrolytic refining (Contilanod) (Ref 6). Eldred developed a machine using graphite as the mold material to cast copper rods and later a number of copper-base alloys (Ref 2). Poland and Lindner were granted a U.S. patent for a vertical casting machine very similar to Eldred’s (Ref 7). The Asarco process was patented by the American Smelting and Refining Company for the continuous casting of phosphorus-deoxidized copper (Ref 8). The Properzi process was introduced in Italy to continuously cast and roll lead rod used for the manufacture of lead pellets for shotgun cartridges. The plant is used today (2008) for the large-scale production of aluminum rod and copper rod (Ref 9). An inexpensive machine for the continuous casting of bronzes was developed at the Tin Research Institute, England (Ref 10). United Wire, Edinburgh, patented their Unicast system for continuous casting brass and bronze rods (Ref 11). The Swiss company Alfred Wertli introduced the first industrial horizontal continuous caster for the production of cast iron rods and later expanded into continuous casting plants for a full range of copper-base alloys and shapes—rod, billet, and strip (Ref 12). Technica-Guss of Wurzburg, West Germany, introduced horizontal continuous casting systems tailored to individual customer requirements, producing strip, billets, round bars, tubes, and profiles in a range of copper-base alloys (Ref 13). The Southwire Company of Carrolton, GA, introduced continuous casting of copper wire by the Southwire continuous rod system, using a high-speed casting wheel mold (Ref 14). Outokumpu O.Y. of Finland introduced and patented the Outokumpu upward casting process for producing copper rod (Ref 15). Rautomead Dundee introduced horizontal and vertical continuous casting equipment based on the all-graphite system, with the Unicast principle of integrated melt, stabilization, and cast from a single crucible (Ref 16).

1914 Mid-1920s

1930 1938 1930s Late 1930s

1950s Early 1950s 1957

1960s

1964 1969 1978

*Portions adapted with permission from: Robert Wilson, A Practical Approach to Continuous Casting of Copper-Based Alloys and Precious Metals, Institute of Materials (IOM Communications Ltd.), 2000

1020 / Casting of Nonferrous Alloys (2008) is a subsidiary of SMS Meer. The company is a supplier of horizontal and vertical casting plants covering large installations for billet and strip casting to smaller plants for strip, wire, and tube casting of copper alloys and precious metals.

Vertical Continuous Casting (Ref 1, 18)

Fig. 1

Poland and Lindner vertical caster. Source: Ref 1

by American Smelting and Refining Company, was primarily designed for the continuous casting of phosphorus-deoxidized copper, but it is widely used today (2008) for a range of copper-base alloys. The Properzi process (Ref 9) was introduced in Italy in the late 1930s to continuous cast and roll lead rod used for the manufacture of lead pellets for shotgun cartridges. The plant is used today (2008) for the highvolume production of aluminum rod and copper rod. Among the first production vertical casting units to be introduced was the TRI equipment developed by the Tin Research Institute in England in 1950 (Ref 10). The equipment consisted of an induction melting unit that feeds a tundish, which was attached to a graphite mold in a tapered steel water jacket. Withdrawal of the product was by means of two grooved rolls situated below the mold. Following the advent of the TRI system, a number of vertical casting processes appeared, such as the Unicast process introduced by United Wire of Edinburgh. The TRI and Unicast equipment filled a need for equipment to produce tin bronze in the form of rod and tube and also a selection of brasses. Since the 1950s, United Wire plants have been installed worldwide, particularly in Britain, France, Italy, and the United States. Rautomead has become a major supplier of such equipment In 1957, the Swiss company Alfred Wertli (Ref 12) introduced the first industrial horizontal continuous caster for the production of cast iron rods and later expanded into continuous casting plants for a full range of copper-base alloys and shapes—rod, billet, and strip. In the 1960s, Technica-Guss (Ref 13) of Wurzburg, West Germany, introduced horizontal continuous casting systems tailored to individual customer requirements, producing strip, billets, round bars, tubes, and profiles in a range of copper-base alloys. In 1989, Technica became a member of the Mannesmann Demag AG group as Demag Technica GmbH and today

As noted, the Unicast system was developed in the 1950s for vertical continuous casting of bronze and brass rod approximately 16 to 19 mm (0.63 to 0.75 in.) in diameter as feedstock for the manufacture of fine wire mesh used for papermaking machines. The unit consisted of an integrated continuous casting plant in which the complete process of melting, alloying, holding, and casting takes place in one selfcontained furnace (Ref 1). The Unicast furnace differs from the modern version only in that refractory brick insulation is used throughout instead of low-thermal-mass insulation. The integrated melting, homogenizing, and casting in an all-graphite system produces a high-quality product with minimal residualelement impurities and low oxygen level, capable of being drawn to very fine wire. The initial casting machines manufactured and used internally by United Wire were vertical casters of approximately 1 tonne capacity with coiling equipment to give workable coils for subsequent rolling to 5 mm2 (0.008 in.2) and then drawn down to suitable wire sizes. United Wire operates these casting units today (2008) on a range of brasses, bronzes, and nickel-silver alloys, producing rod of high quality. A vertical continuous casting plant for copper alloy tubes and bars consists of a channeltype induction furnace, positioned on a tilting frame and feeding into a water-cooled graphite mold, with microprocessor-controlled withdrawal system and automatic tube cut-off. Mold or die change is made without emptying the holding furnace by incorporating a back-tilting mechanism, which is also a safety feature in vertical casting, because the melting unit can be tilted off the casting position in case of malfunction. The product range in tubes is 20 to 125 mm (0.8 to 5.0 in.) outside diameter and in bars 12 to 80 mm (0.5 to 3.1 in.) diameter. Strand lengths are generally in the range of 3 to 4 m (10 to 13 ft).

Vertical Semicontinuous Slab Casters Since the original invention of Unicast and TRI vertical continuous casting, the most commonly used plant for slab casting is still the proven design and low investment costs of vertical semicontinuous casters (Fig. 2). The layout consists of a casting pit with integrated casting cylinder, by which the casting table and products are moved up and down. The liquid metal can be supplied from the melting or holding

furnace via launder into the molds mounted on top of the oscillating casting machine. It is common for multiple molds to be mounted in the casting table; this allows for very high production rates. At the end of the casting cycle, the produced slabs are pulled out of the pit with an overhead crane and a hydraulic tool for clamping of the slab (Fig. 2b). This disadvantage can be avoided by use of a vertical semicontinuous caster with incorporated discharge device (Fig. 3). This layout shows a larger casting pit with integrated casting cylinder and casting table, to which the discharge device is attached by a swivel joint. The liquid metal solidifies in the mold on top of the oscillating casting machine. When the casting cycle is completed, the cast slabs are pushed from the vertical position slightly to the direction of the discharge device, and the discharging operation starts. The vertical semicontinuous slab casters can be used to cast a wide range of alloys and sizes. In addition to copper and brass, copper nickel and other copper alloys are produced. Modern casters incorporate components such as an automatic melt-level control system to provide automation and stable casting conditions for a consistently good product quality. Also, adjustable slab molds can be used, by which the casting width can be varied and thus the inventory of the molds reduced (Fig. 4). Most often, copper alloy molds without sleeves are used for slab casting.

Vertical Continuous Slab Casting When casting copper slabs, the highest production rates and consistent quality product can be achieved with a fully continuous vertical caster. The general layout is similar to the semicontinuous mode; a holding furnace supplies the liquid metal via an automatic melt-level control into the molds, which are mounted on top of the oscillating casting machine (Fig. 5). The solidified strands pass through the secondary water-cooling box and are transported by a withdrawal unit. After cutting, the products are tilted by a tilting basket into a horizontal position and further transported by a roller conveyor and lifting system. Appropriate sensors detect the melt level in the mold, and the flow of liquid metal is regulated automatically by an electronic control. The holding furnace is typically induction heated and has a specific shape for vertical casting, with an attached launder section or forehearth, thus supplying the metal by the shortest route into the mold without splashing. The solidified slabs are firmly clamped and transported by the withdrawal unit. The vertical movable saw clamps to the strands and travels together with the strands during the cutting operation. After cutting, the products are taken over by a tilting basket and tilted into a horizontal position for further transport on subsequent roller tables.

Copper Continuous Casting / 1021

Fig. 2

Vertical semicontinuous casting plant. (a) Layout without discharge device. (b) Slab discharging by overhead crane. Courtesy of Demag Technica GmbH of SMS

Fig. 4

Adjustable molds for vertical semicontinuous casting of slabs. Courtesy of Demag Technica GmbH of SMS

Upcasting Methods Fig. 3

Vertical semicontinuous casting plant with discharge device. Courtesy of Demag Technica GmbH of SMS

The Outokumpu Upcasting Method. The upward casting process was introduced and patented in 1969 by Outokumpu O.Y., Finland

(Ref 15), with the first production unit coming into operation in 1970 for casting oxygen-free, small-diameter copper rod. This system has all the technical advantages of casting in the vertical mode and, for small-diameter rod, none of the disadvantages. The method consists of a graphite die partially immersed in molten metal, with the upper part surrounded by a water-cooled jacket (Fig. 6) (Ref 1). The assembly is located just above the metal top surface, with the graphite die only just immersed into the liquid and maintained precisely in position by an electronic level-sensing control. The action of vertical pulsed withdrawal of the rod raises the metal beyond the lower extremity of the cooler, and solidification takes place. In the melting and transfer system, Outokumpu exposes the liquid metal to graphite or charcoal, resulting in deoxidation of the melt to a level of the order of 5 ppm oxygen. The machine operates on a multidie system, casting, for example, 12 mm (0.5 in.)

1022 / Casting of Nonferrous Alloys

Fig. 5

Vertical continuous casting. Courtesy of Demag Technica GmbH of SMS

diameter rods at speeds on the order of 3 m/min (10 ft/min). Rautomead Upwards Vertical Casting. Rautomead International, Dundee, introduced a modified upwards vertical casting process (Ref 16) based on graphite melt containment technology and using submersed dies with inert gas protection. The equipment is used primarily for the production of small-diameter, high-purity copper rod with oxygen levels on the order of 5 ppm at casting speeds on the order of 4.0 to 4.5 m/min (13 to 14.5 ft/min). The machine is also adaptable to alloy systems such as bronzes and brasses in rod form and also tube. By utilizing an all-graphite containment system and incorporating a specially designed graphite filter bed, deoxidation of copper to 5 ppm oxygen is ensured. Pressure Upcast System. A pressure Upcaster (Ref 19) was developed as a production unit at Dundee Institute of Technology (now University of Abertay, Dundee). During the casting operation, an inert gas applied to the sealed steel furnace casing exerts pressure on the molten metal in the graphite crucible, raising it into the graphite die where it solidifies and is withdrawn through a water-cooled jacket vertically upward in a conventional pulsed mode. On reverting to atmospheric pressure, metal drains to the crucible. The equipment is primarily intended for casting small-section high-purity copper rod in the range of 1.5 to 10 mm (0.06 to 0.40 in.) in diameter. A process using some of the same principles was explored in the 1970s by Chase Brass to produce copper and brass rod. A higher throughput was targeted, and the cast rod was reduced in-line to wire. The process remains in use to produce copper bus bar.

Horizontal Continuous Casting

Fig. 6

Principle of upward casting. Source: Ref 1

Casting in the horizontal mode facilitates product handling and generally occupies less space. There are inherent problems as opposed to vertical casting that mainly relate to gravity-induced directional cooling, which, in most cases, can be accommodated in the process protocols. The horizontal plants produce billets for subsequent extrusion in copper or brass in section sizes between 80 and 400 mm (3.1 and 16 in.) in diameter, operating as single- or multistrand machines. Smaller horizontal casters are used for the production of tube, bar, and sections in a full range of copper alloys. The Swiss company Alfred Wertli (Ref 12) introduced the world’s first industrial horizontal continuous caster for the production of cast iron rods and later expanded into continuous casting plants for a full range of copper-base alloys. State-of-the-art horizontal continuous casting lines for copper alloys are mostly designed to cast two narrow strips up to 450 mm (18 in.) wide or one strip up to 800 mm (32 in.) wide. Thin-strip withdrawal machines are designed to cast two strands simultaneously, or each strand can be independently withdrawn.

Copper Continuous Casting / 1023 The molds can also be configured for billet casting 100 to 400 mm (9 to 16 in.) in diameter, bar and tube casting in the size range of 25 to 350 mm (1.0 to 14 in.) in diameter, and for small-diameter rod and wire 12 to 25 mm (0.5 to 1.0 in.) in diameter. The original Wertli (Ref 12) concept consists of a channel-type induction furnace and holding furnace, together with graphite die and cooler assembly and runout track with withdrawal machine and cut-off device. Molten metal flows from the melting furnace to a holding or casting furnace that acts as a reservoir of molten metal, maintaining the required casting temperature. Water-cooled graphite dies are attached to the holding furnace or crucible. During the continuous casting operation, metal flows into the graphite casting die, where it solidifies. The solidified strands are intermittently withdrawn in a pull-pause sequence by means of withdrawal equipment. After leaving the graphite die, which is housed within the primary cooler, the cast strands pass through a secondary cooler in the form of a water sparge that removes the surplus heat contained in the solidified casting. The crucible can be manufactured in a refractory ceramic or from graphite. Integral ceramic crucibles are used extensively in induction melting and casting furnaces. These are the most energy-efficient furnaces and consist of melting units feeding a casting unit or a single induction-heated casting unit. The design varies depending on the application. The metal type and production rate will determine the crucible capacity and power rating. Frequency is chosen to suit these parameters and is selected from 150, 250, 500, 1000, 3000, and 10,000 Hz. The high frequencies apply to small crucible capacity, decreasing for the larger installations. Induction melting and casting furnaces use either integral or removable crucible assemblies, depending on the casting operation. Precast ceramic crucibles with graphite support carriers are used in either induction-heated or resistance-heated furnaces (Ref 1). The full range of Rautomead graphite resistance-heated furnaces use this basic design, ranging from small table-top units to installations with crucible capacity of 2500 kg (5500 lb) (copper). Graphite can only be used in a nonoxidizing atmosphere; therefore, the crucible and die assembly must be housed in a sealed furnace and protected with an inert gas, either nitrogen or argon. Most high-grade coppers, brasses, tin bronzes, phosphor bronzes, and aluminum bronzes can be successfully cast in an allgraphite crucible and die assembly (Ref 1). The volume of the crucible is dependent on the application and may vary in capacity from several tonnes to 1 kg (2 lb) or less. A crucible liner or protecting sleeve is frequently fitted, particularly with larger crucibles. This liner is manufactured in graphite and protects the main crucible against abrasion and oxidation. The seal between crucible and graphite die is often made by means of a grafoil gasket sheet or washer. Grafoil consists of graphite

in flexible lamellar form, which is compressible, forming a gastight seal and providing a liquidtight seal. Graphite baffles can be fitted within the crucible and held in position between the lower and upper graphite sleeves or liners (Ref 1). A baffle with suitable perforations provides an upper and lower chamber to facilitate melting and homogenization of the charge in the upper section prior to this metal entering the casting die. Another, most important function is to allow sufficient time for deoxidation of the melt and thus avoid attack on the graphite die. Unicast Horizontal Casting System. In the early 1970s, the first Unicast horizontal casting plant was installed by Timex Corporation, Dundee, for continuous casting brass rod for watch case manufacture, operating under conditions similar to the United Wire vertical casters. This operation was extremely successful, utilizing as feedstock 100% internally generated brass scrap. The recycled scrap from trim and machining operations on case manufacturing, together with high-quality press shop and screw machine residue, made the process economically viable. The chemistry of the product could be closely controlled, reducing trace element impurities to limits unobtainable on purchased stock. Since its inception in 1978, Rautomead, Dundee, has manufactured a wide range of continuous casting machines for the nonferrous metals industries, primarily for copper-base alloys and precious metals. The Rautomead resistanceheated all-graphite system is based on the United Wire Unicast technology. The design has been refined, particularly in the area of refractory insulation heating element configuration and unit modular construction. The majority of the machines operate in the horizontal mode, with a few special-purpose machines casting vertically downward. The construction of the Rautomead machines is an integrated all-graphite melt-and-cast system. The casters range from small table-top units with crucible capacities of 2 to 50 kg (4.5 to 110 lb) (copper) to large billet and strip casters with crucible capacities to 2500 kg (5500 lb) (copper).

Strip Casting The horizontal strip casting process was originally developed by Technica-Guss, which is now part of SMS Meer. It has been widely used in the industry for more than 30 years as a nearnet shape casting process for alloys that are considered difficult to hot roll, such as nickelsilver mint alloys and phosphor bronze for electronic applications. Horizontal continuous casters are manufactured for casting a range of strip widths. Typical strip widths are 450 mm (18 in.), 650 mm (26 in.), or larger, with a common thickness being between 14 and 20 mm (0.6 and 0.8 in.). The produced coil can be automatically discharged onto a subsequent discharge table.

A majority of horizontal strip casters are equipped with an in-line milling machine by which the top and bottom of the cast surface is cleaned of oxides and segregations so that a perfect coil is produced for direct passing into the cold rolling mill. Modern strip casters comprise high-performance coolers with temperature-sensoring systems to survey the solidification and apply nitrogen for reduction of surface oxides. Furthermore, a tight temperature control in the holding furnace is essential for stable casting conditions and product quality, which is realized with the stepless power control of the inductor by the inverter technique. A typical casting speed of a horizontal strip caster is up to 200 mm/min (8 in./min), resulting in a production of approximately 6,000 tons/year for strips 650 by 16 mm (26 by 0.6 in.). Due to the limited production rate, many companies who are using this casting process operate several machines; even eight machines are in operation in one company. Horizontal strip casters are used exclusively for the production of copper alloys and, in very few exceptions, worldwide for copper. To overcome the limitations of the thin-strip casting process regarding output and copper casting, SMS Meer developed a new concept for a vertical strip caster (Ref 18). Wertli Strip Casting. The Wertli strip casting plants include in-line operations such as milling equipment, traveling shear, coiler, and die and track friction. Hydraulically amplified electric drives are used to achieve forces in the range of 40 to 80 kN (9000 to 18,000 lbf) while maintaining motion accuracy. The Wertli drive concept is designed to handle such forces and accelerations by using backlash-free lowratio gears together with a high-precision servo motor with hydraulic amplification. To achieve a mechanically backlash-free drive, backlashfree gears are used between the driving motor and the rollers that drive the strands. Slippage between the cast strand and the drive roller is to be avoided if a precise strand motion must be maintained over long periods of casting. The tight gripping of the strands is achieved by hydraulic press-down cylinders. A cooling system with an enhanced water-cooled surface increases heat transfer (approximately 10%) compared to a conventional copper plate cooler (Ref 1). Hazelett Casting Process. The Hazelett steel belt casting invented in 1920 has been developed for various product forms and today (2008) is used in the production of copper wire, rod, strip, and anode. The metal is usually melted by induction and is delivered via a tundish to a straight-through mold formed by tensioned steel belts and edge dam blocks. Fast-film heat extraction from the mold is achieved by a proprietary design for the application and removal of high-flow-rate water cooling. The use of special coatings on the belt is also important. Strip up to 1.25 m (4.10 ft) wide and Contirod, a rectangular cast bar at

1024 / Casting of Nonferrous Alloys 6 to 60 tonnes/h depending on plant capacity, can be achieved. The Contilanode process is used for producing high-quality copper anode. The cast anode plate can be maintained geometrically to within close limits. Hanger lugs are cast in shaped recesses and thus become an integral part of the anode body.

Wheel Casting Properzi Wheel Casting Technology. Properzi first used wheel casting technology on copper in the 1950s and commercially introduced the continuous casting and rolling process for copper rod in 1963. Molten copper is poured into a revolving casting wheel from a gas-fired melting and refining furnace. The copper rim of the wheel is grooved to receive the molten metal, which is then retained in the groove by a steel belt. The solidified metal leaves the wheel and passes through a rolling mill without interruption. The casting wheel has a “U” profile, a shape that evolved to control solidification and heat transfer as the metal traversed the cooling segments of the wheel. A Cu-Cr-Zr alloy mold is used for the casting of all electrolytic tough pitch (ETP) copper. The position, alignment, and adjustment of the individual cooling spray nozzles located around the wheel are of the utmost importance in controlling the solidification and uniformity of the grain structure of the cast bar. A layer of acetylene soot, applied to both cavity and band, serves as a release agent and insulator, which provides uniformity of heat transfer. During each rotation of the wheel, the soot is stripped by high-pressure water sprays, then reapplied. The cast “D” section passes through two twohigh break-down roll stands, followed by six to eight three-high roll stands to yield product for rod, narrow strip, trolley wire, and other applications. Southwire Continuous Casting Rod Process. Following some research and development with Properzi in approximately 1960, the Southwire Company of Georgia, United States, introduced a continuous casting process for the high-speed production of ETP copper rod. The process has been described in detail in several published papers (Ref 20–23). The SCR process, as it is known, incorporates a continuous melting, holding, casting, rolling, pickling, and coiling system. The casting wheel provides a trapezoidalshaped casting groove in the periphery of a copper alloy ring. This ring is closed by an endless steel belt through an arc of approximately 180 to 210 , the belt being held in place by idler wheels and tensioners. The casting groove and the contact side of the steel band are coated with a controlled layer of soot that serves as a release agent and provides uniformity of heat extraction. The cast bar passes to the rolling mill through a trimming and descaling

Fig. 7

Schematic of Ohno continuous casting. Source: Ref 1

operation. The mill itself is composed of a number of roughing, intermediate, and finishing two-roll stands. The alternating vertical and horizontal shaft stands produce a repetitive series of oval-to-round reductions. In the continuous casting of ETP copper with oxygen content in the melt of approximately 400 ppm, the dissolved oxygen reacts with the impurities present during solidification, precipitating these out of the solid solution and resulting in improved annealability and electrical conductivity of the product. The claim for success of the SRC process is the ability to control the amount of superheat to a very close range immediately prior to casting, generally approximately 25  C (45  F) above the liquidus. The use of a high-purity cathode and the close control of temperature results in solidification in a columnar grain pattern with good bar quality. In subsequent rolling in the SCR process, the high temperature and severe initial reductions in the first pass cause dynamic recrystallization. Chia (Ref 24) described the mode of solidification on SCR tough pitch copper rod.

Ohno Continuous Casting Process The Ohno continuous casting concept is based on the application of the Ohno separation theory of solidification (Ref 25), with the continuous cast ingot consisting of unidirectional solidified structure with no equiaxed crystals. The process is described by Ohno and McLean (Ref 26). The patented process (Ref 27) differs from conventional techniques in that molten metal is poured into a heated mold rather than into a cooled mold or die. The mold is heated externally and its temperature maintained above the solidification point of the metal being cast. As a result, no metal nucleates on the mold surface. The Ohno process has been adopted by Furukawa, Japan (Ref 28), for the production of

oxygen-free high-purity copper rod. The rod has a structure characterized by longitudinal crystals or may even develop into a single crystal in some growth conditions. This special structural material is used in high-resolution audio signal transmission, having low impurities, no grain boundaries transverse to the direction of signal transmission, smooth surface finish, and excellent physical properties. The production casting equipment used is essentially as shown in Fig. 7, with some refinements, including a melting furnace and a casting furnace with precise metal-level control, ensuring constant metastatic pressure on the solidification front. The high-purity copper charge is deoxidized using carbonaceous material in the melting furnace before transfer to the casting furnace. REFERENCES 1. R. Wilson, A Practical Approach to Continuous Casting of Copper-Based Alloys and Precious Metals, Institute of Materials (IOM Communications Ltd.), 2000 2. B.E. Eldred, U.S. Patent 1,868,099, 1932 3. G.E. Sellers, U.S. Patent 1908, 1840 4. L. Laign, U.S. Patent 3023, 1843 5. A.H. Pehrson, U.S. Patent 1,088,171, 1914 6. Hazelett Process, Iron Steel Eng., Vol 43 (No. 6), 1966, p 105 7. Poland and Lindner, U.S. Patent 2,136,394, 1938 8. A. Kreil et al., Asarco Process, Met. Rev., Vol 5, 1960, p 413–446 9. Properzi Process, Met. Rev., Vol 6 (No. 22), 1961 10. E.C. Ellwood, J. Inst. Met., Vol 84, 1955– 1956, p 319–326 11. I.E. Ewen, United Wire Unicast Process, U.K. Patents 894,783, 894,784, and 934,484 12. T.P. Wertu, Alfred Wertli AG, Winterthur, Switzerland 13. Technica Guss, Wurzburg, West Germany

Copper Continuous Casting / 1025 14. Southwire Revolutionizes Non-Ferrous Rod Production with SCR System, Vol. 33, Southwire Mag., June 1975 15. M. Rantaneno, Upward Continuous Cast of Copper Wire, Wire Ind., July 1976 16. Rautomead International, Dundee, Scotland 17. A. Krell et al., Flocast Process, Met. Rev., Vol 5, 1960, p 413–446 18. “Copper and Copper Alloy Rolled Products—Trends in Equipment/Auxiliaries and Applications,” Seminor at Hotel Le Royal Meridiea (Mumbai), Jan 18–19, 2006 19. R. Wilson, Pressure Upcast, U.K. Patent GB 2,236,498B, 1992, and U.S. Patent 5,090,471, 1992

20. U. Sinha and R. Adams, Southwire Continuous Rod Process: Innovations for Quality Improvements, Wire J. Int., June 1993 21. U. Sinha and R. Adams, “Southwire Continuous Rod: A Method to Produce HighQuality Rods for Fine Wire Drawing and Special Applications,” Conference Indian Copper Development Centre and Winding Wires Manufacturers Association of India, Oct 1988 22. G.T. Hudson, “The Production of Copper Rod by SRC Process,” internal paper, Southwire Company, Carrolton, GA. 23. L.C. Richards et al., “Continuous Casting— Its History, Impact and Future,” Metals Week Copper Conference, Dec 10, 1989

24. H. Chia, International Conf., Inst. Wire and Mach. Assoc. (Torremolinos, Spain), April 1979 25. A. Ohno, Solidification, The Separation Theory and Its Practical Application, Springer Verlag, New York 26. A. Ohno and A. McLean, Ohno Continuous Casting, Adv. Mater. Process., Vol 4, 1995, p 43–45 27. A. Ohno, Japan Patent 1,049,148; U.S. Patent 4,515,204; and Germany Patent 3,246,470 28. K. Nakano, “Continuous Casting of Copper and Copper Alloys,” R&D Division of Furukawa Electric Company, Japan

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1026-1048 DOI: 10.1361/asmhba0005303

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Casting of Copper and Copper Alloys ALL COPPER ALLOYS can be successfully cast in sand. Sand casting allows the greatest flexibility in casting size and shape and is the most economical casting method if only a few castings are made (die casting is more economical above
50,000 units). Virtually all copper alloys can be cast successfully by the centrifugal casting process. Castings of almost every size from less than 100 g to more than 22,000 kg (<0.25 to >50,000 lb) have been made. Permanent mold casting is best suited for tin, silicon, aluminum and manganese bronzes, and yellow brasses. Dies casting is well suited for yellow brasses, but increasing amounts of permanent mold alloys are also being die cast. Size is a definite limitation for both methods, although large slabs weighing as much as 4500 kg (10,000 lb) have been cast in permanent molds. Brass die castings generally weigh less than 0.2 kg (0.5 lb) and seldom exceed 0.9 kg (2 lb). The limitation of size is due to the reduced die life with larger castings. Because of their low lead contents, aluminum bronzes, yellow brasses, manganese bronzes, low-nickel bronzes, and silicon brasses and bronzes are best adapted to plaster mold casting. For most of these alloys, lead should be held to a minimum because it reacts with the calcium sulfate in the plaster, resulting in discoloration of the surface of the casting and increased cleaning and machining costs. Size is a limitation on plaster mold casting, although aluminum bronze castings that weigh as little as 100 g (0.25 lb) have been made by the lost wax process, and castings that weigh more than 150 kg (330 lb) have been made by conventional plaster molding. This article describes the casting characteristics and practices of copper alloys. Successful production of copper and copper alloy castings depends on three important factors:  An understanding of casting and solidifica-

tion characteristics of copper and its various alloys  Adherence to proper foundry practices, including melting practices (e.g., selection of melting furnace and molten-metal treatments), pouring practices, and gating and risering techniques  Proper selection of the casting process, which, in turn, depends on the size, shape, and technical requirements of the product

This article addresses each of these factors. For more detailed information, see Ref 1 and the other references listed at the end of this article.

Castability The castability of alloys is generally influenced by their shrinkage characteristics and their freezing range (which is not necessarily related directly to shrinkage). Castability should not be confused with fluidity, which is only a measure of the distance to which a metal will flow before solidifying. Fluidity is thus one factor determining the ability of a molten alloy to completely fill a mold cavity in every detail. Castability, on the other hand, is a general term relating to the ability to reproduce fine detail on a surface. Colloquially, good castability refers to the ease with which an alloy responds to ordinary foundry practice without requiring special techniques for gating, risering, melting, sand conditioning, or any of the other factors involved in making good castings. High fluidity often ensures good castability, but it is not solely responsible for that quality in a casting alloy. Table 1 presents foundry characteristics of selected standard alloys, including a comparative ranking of both fluidity and overall

castability for sand casting; number 1 represents the highest castability or fluidity ranking. The effect of the freezing range on castability is discussed in the next section, “Control of Solidification.” Copper alloys are classified as high-shrinkage or low-shrinkage alloys. The former class includes the manganese bronzes, aluminum bronzes, silicon bronzes, silicon brasses, and some nickel-silvers. They are more fluid than the low-shrinkage red brasses, more easily poured, and give high-grade castings in the sand, permanent mold, plaster, die, and centrifugal casting processes. With high-shrinkage alloys, careful design is necessary to promote directional solidification, avoid abrupt changes in cross section, avoid notches (by using generous fillets), and properly place gates and risers; all of these design precautions help avoid internal shrinks and cracks. Turbulent pouring must be avoided to prevent the formation of dross becoming entrapped in the casting. Liberal use of risers or exothermic compounds ensures adequate molten metal to feed all sections of the casting. Pure copper is extremely difficult to cast, as well as being prone to surface cracking and porosity problems. Casting characteristics of copper can be improved by small additions of elements such as beryllium, silicon, nickel, tin, zinc, chromium, and silver. The copper-base

Table 1 Foundry properties of the principal copper alloys for sand casting UNS No.

Common name

Shrinkage allowance, %

C83600 C84400 C84800 C85400 C85800 C86300 C86500 C87200 C87500 C90300 C92200 C93700 C94300 C95300 C95800 C97600 C97800

Leaded red brass Leaded semired brass Leaded semired brass Leaded yellow brass Yellow brass Manganese bronze Manganese bronze Silicon bronze Silicon brass Tin bronze Leaded tin bronze High-lead tin bronze High-lead tin bronze Aluminum bronze Aluminum bronze Nickel-silver Nickel-silver

5.7 2.0 1.4 1.5–1.8 2.0 2.3 1.9 1.8–2.0 1.9 1.5–1.8 1.5 2.0 1.5 1.6 1.6 2.0 1.6

Approximate liquidus temperature 

C

1010 980 955 940 925 920 880 ... 915 980 990 930 925 1045 1060 1145 1180



F

Castability rating(a)

Fluidity rating(a)

1850 1795 1750 1725 1700 1690 1615 ... 1680 1795 1810 1705 1700 1910 1940 2090 2160

2 2 2 4 4 5 4 5 4 3 3 2 6 8 8 8 8

6 6 6 3 3 2 2 3 1 6 6 6 7 3 3 7 7

(a) Relative rating for casting in sand molds. The alloys are ranked from 1 to 8 in both overall castability and fluidity; 1 is the highest or best possible rating.

Casting of Copper and Copper Alloys / 1027 casting alloy family can be subdivided into three groups according to solidification (freezing) range. The three groups are as follows. Group I alloys are alloys that have a narrow freezing range, that is, a range of 50  C (90  F) between the liquidus and solidus. These are the yellow brasses, manganese and aluminum bronzes, nickel bronzes, (nickel-silvers), manganese (white) brass alloys, chromium-copper, and copper. Liquidus/solidus temperatures for these alloys are shown in Table 2. Group II alloys are those that have an intermediate freezing range, that is, a freezing range of 50 to 110  C (90 to 200  F) between the liquidus and solidus. These are the berylliumcoppers, silicon bronzes, silicon brass, and copper-nickel alloys. Liquidus/solidus temperatures for these alloys are shown in Table 3.

Table 2 Solidification ranges for group I copper alloys Liquidus temperature Alloy type



UNS No.

Copper Chrome copper Yellow brass

C81100 C81500 C85200 C85400 C85700 C85800 C87900 Manganese bronze C86200 C86300 C86400 C86500 C86700 C86800 Aluminum bronze C95200 C95300 C95400 C95410 C95500 C95600 C95700 C95800 Nickel bronze C97300 C97600 C97800 White brass C99700 C99750

C

1083 1085 941 941 941 899 926 941 923 880 880 880 900 1045 1045 1038 1038 1054 1004 990 1060 1040 1143 1180 902 843



F

1981 1985 1725 1725 1725 1650 1700 1725 1693 1616 1616 1616 1652 1913 1913 1900 1900 1930 1840 1814 1940 1904 2089 2156 1655 1550

Solidus temperature 

C

1064 1075 927 927 913 871 900 899 885 862 862 862 880 1042 1040 1027 1027 1038 982 950 1043 1010 1108 1140 879 818



F

1948 1967 1700 1700 1675 1600 1650 1650 1625 1583 1583 1583 1616 1907 1904 1880 1880 1900 1800 1742 1910 1850 2027 2084 1615 1505

Group III alloys have a wide freezing range, well over 110  C (200  F), even up to 170  C (300  F). These are the leaded red and semired brasses, tin and leaded tin bronzes, and highleaded tin bronzes. Liquidus/solidus temperatures for these alloys are shown in Table 4. Control of Solidification. Production of consistently sound castings requires an understanding of the solidification characteristics of the alloys as well as knowledge of relative magnitudes of shrinkage. The actual amount of contraction during solidification does not differ greatly from alloy to alloy. Its distribution, however, is a function of the freezing range and the temperature gradient in critical sections. Manganese and aluminum bronzes are similar to steel in that their freezing ranges are quite narrow— approximately 40 and 14  C (70 and 25  F), respectively. Large castings can be made by the same conventional methods used for steel, as long as proper attention is given to placement of gates and risers—both those for controlling directional solidification and those for feeding the primary central shrinkage cavity. Tin bronzes have wider freezing ranges (
165  C, or 300  F, for C83600). Alloys with such wide freezing ranges form a mushy zone during solidification, resulting in interdendritic shrinkage or microshrinkage. Because feeding cannot take place properly under these conditions, porosity results in the affected sections. In overcoming this effect, design and riser placement, plus the use of chills, are important. Another means of overcoming interdendritic shrinkage is to maintain close temperature control of the metal during pouring and to provide for rapid solidification. These requirements limit section thickness and pouring temperatures, and this practice requires a gating system that will ensure directional solidification. Sections up to 25 mm (1 in.) in thickness are routinely cast. Sections up to 50 mm (2 in.) thick Table 4 Solidification ranges for group III copper alloys Liquidus temperature Alloy type

Table 3 Solidification ranges for group II copper alloys

Alloy type

UNS No.

Beryllium-copper C81400 C82000 C82200 C82400 C82500 C82600 C82800 Silicon brass C87500 Silicon bronze C87300 C87600 C87610 C87800 Copper-nickel C96200 C96400

Liquidus temperature

Solidus temperature





C

1093 1088 1116 996 982 954 932 916 916 971 971 916 1149 1238



F

2000 1990 2040 1825 1800 1750 1710 1680 1680 1780 1780 1680 2100 2260

C

1066 971 1038 899 857 857 885 821 821 860 860 821 1099 1171



F

1950 1780 1900 1650 1575 1575 1625 1510 1510 1580 1580 1510 2010 2140

Leaded red brass

UNS No.

C83450 C83600 C83800 Leaded semired brass C84400 C84800 Tin bronze C90300 C90500 C90700 C91100 C91300 Leaded tin bronze C92200 C92300 C92600 C92700 High-leaded tin bronze C92900 C93200 C93400 C93500 C93700 C93800 C94300



C

1015 1010 1004 1004 954 1000 999 999 950 889 988 999 982 982 1031 977 ... 999 929 943 ...



F

1860 1850 1840 1840 1750 1832 1830 1830 1742 1632 1810 1830 1800 1800 1887 1790 ... 1830 1705 1730 ...

Solidus temperature 

C

860 854 843 843 832 854 854 831 818 818 826 854 843 843 857 854 ... 854 762 854 900



F

1580 1570 1550 1550 1530 1570 1570 1528 1505 1505 1518 1570 1550 1550 1575 1570 ... 1570 1403 1570 1650

can be cast, but only with difficulty and under carefully controlled conditions. A bronze with a narrow solidification (freezing) range and good directional solidification characteristics is recommended for castings having section thicknesses greater than approximately 25 mm (1 in.). It is difficult to achieve directional solidification in complex castings. The most effective and most easily used device is the chill. For irregular sections, chills must be shaped to fit the contour of the section of the mold in which they are placed. Insulating pads and riser sleeves sometimes are effective in slowing down the solidification rate in certain areas to maintain directional solidification.

Melting and Melt Control Furnaces for melting copper casting alloys are either fuel fired or electrically heated. They are broadly classified into three categories:  Crucible furnaces (tilting or stationary)  Open-flame (reverberatory) furnaces  Induction furnaces (core or coreless)

Selection of a furnace depends on the quantity of metal to be melted, the degree of purity required, and the variety of alloys to be melted. Environmental restrictions also influence furnace selection. In addition, the melting of copper alloys also involves various melt treatment to improve melt quality. These melt treatments include:     

Fluxing and metal refining Degassing Deoxidation Grain refining Filtration

Each of these is described in the sections that follow. It should be noted that some of these process methodologies pertain not only to foundry melting and casting but also to smelting, refining, and, in certain cases, mill product operations. Fuel-Fired Furnaces. Copper-base alloys are melted in oil- and gas-fired crucible and openflame furnaces. Crucible furnaces, either tilting or stationary, incorporate a removable cover or lid for removal of the crucible, which is transported to the pouring area where the molds are poured. The contents of the tilting furnace are poured into a ladle, which is then used to pour the molds (Fig. 1, 2). These furnaces melt the raw materials by burning oil or gas with sufficient air to achieve complete combustion. The heat from the burner heats the crucible by conduction and convection; the charge melts and then is superheated to a particular temperature at which either the crucible is removed or the furnace is tilted to pour into a ladle. While the molten metal is in the crucible or ladle, it is skimmed, fluxed, and transferred to the pouring area, where the molds are poured.

1028 / Casting of Nonferrous Alloys The other type of fuel-fired furnace is the open-flame furnace, which is usually a large rotary-type furnace with a refractory-lined steel shell containing a burner at one end and a flue at the other. The furnace is rotated slowly around the horizontal axis, and the rotary movement helps to heat and melt the furnace charge. Melting is accomplished both by the action of the flame directly on the metal and by heat transfer from the hot lining as this shell rotates. These furnaces usually tilt so that they can be charged and poured from the flue opening. At the present time (2008), these furnaces are not used often because of the requirement that a baghouse be installed to capture all the flue dust emitted during melting and superheating. While these furnaces are able to melt large amounts of metal quickly, there is a need for operator skill to control the melting atmosphere within the furnace. Also, the refractory walls become impregnated with the melting metal, causing a contamination problem when switching from one alloy family to another.

Fig. 1

Typical lift-out type of fuel-fired crucible furnace, especially well adapted to foundry melting of smaller quantities of copper alloys (usually less than 140 kg, or 300 lb)

Fig. 2

Typical lip-axis tilting crucible furnace used for fuel-fired furnace melting of copper alloys. Similar furnaces are available that tilt on a central axis

Electric Induction Furnaces. In the past 30 years, there has been a marked changeover from fuel-fired melting to electric induction melting in the copper-base foundry industry. While this type of melting equipment has been available for more than 60 years, very few were actually used due to the large investment required for the capital equipment. Because of higher prices and the question of availability of fossil fuels and because of new regulations on health and safety imposed by the Occupational Safety and Health Administration (OSHA), many foundries have made the changeover to electric induction furnaces. When melting alloys in group III, fumes of lead and zinc are given off during melting and superheating. The emission of these harmful oxides is much lower when the charge is melted in an induction furnace because the duration of the melting cycle is only approximately 25% as long when melting the same amount of metal in a fuel-fired furnace. By the use of electric induction melting, compliance with OSHA regulations can be met in many foundries without the need for expensive air pollution control equipment. The two types of electric induction furnaces are the core type, better known as the channel furnace, and the coreless type. Core Type. This furnace (Fig. 3) is a large furnace used in foundries for pouring large quantities of one alloy when a constant source of molten metal is required. This furnace has a primary coil, interfaced with a laminated iron core, surrounded by a secondary channel, which is embedded in a V- or U-shaped refractory lining located at the bottom of a cylindrical hearth. Here, the channel forms the secondary of a transformer circuit. This furnace stirs and circulates molten metal through the channel at all times, except when the furnace is emptied and shut down. When starting up, molten metal must be poured into the furnace to fill up the “heel” on the bottom of the bath. Because these furnaces are very efficient and simple to

Fig. 3

Cutaway drawing of a twin-channel induction melting furnace

operate, with lining life in the millions of pounds poured, they are best suited for continuous production runs in foundries making plumbing alloys of group III. They are not recommended for the dross-forming alloys of group I. The channel furnace is at its best when an inert, floating, cover flux is used and charges of ingot, clean remelt, and clean and dry turnings are added periodically. Coreless Type. This furnace has become the most popular melting unit in the copper alloy foundry industry. In earlier years, the coreless furnace was powered by a motor generator unit, usually at 980 Hz. The present coreless induction furnaces draw 440 V, 60 cycle power and, by means of solid-state electronic devices, convert the power to 440 V and 1000 or 3000 Hz. These furnaces are either tilting furnaces (Fig. 4) or crucible lift-out units (Fig. 5, 6). A coreless induction furnace comprises a water-cooled copper coil in a furnace box made of steel or Transite. The metal is contained in a crucible or in a refractory lining rammed up to the coil. Crucibles used in these furnaces are made of clay graphite; silicon carbide crucibles cannot be used because they become overheated when inserted in a magnetic field. Clay graphite crucibles do a good job of conducting the electromagnetic currents from the coil into the metal being melted. Induction furnaces are characterized by electromagnetic stirring of the metal bath. Because the amount of stirring is affected by both power input and power frequency, the power unit size and frequency should be coordinated with the furnace size in order to obtain the optimal-sized equipment for the specific operation. In general, the smaller the unit, the higher the frequency and the lower the power input. Large tilting units are used in foundries requiring large amounts of metal at one time. These furnaces, if over 4.5 Mg (10,000 lb) capacity, operate at line frequency (60 Hz). They are very efficient and will melt large quantities of metal in a very short time if powered with the propersized power unit. Stationary lift-out furnaces are often designed as shown in Fig. 5. Here, the crucible sits on a refractory pedestal, which can be

Fig. 4

Cross section of a tilt furnace for high-frequency induction melting of brass and bronze alloys. Crucible is of clay graphite composition.

Casting of Copper and Copper Alloys / 1029

Fig. 5

Cross section of a double push-out furnace. Bilge crucibles are placed on refractory pedestals and raised and lowered into position within the coils by hydraulic cylinders.

metal in which chemical compounds or mixtures of such compounds are employed. These compounds are usually inorganic. In some cases, metallic salts are used in powder, granulated, or solid tablet form and may often melt to form a liquid when used. They can be added manually or can be automatically injected, and they can perform single or, in combination, various functions, including degassing, cleaning, alloying, oxidation, deoxidation, or refining. The term fluxing also includes the treatment of nonferrous melts by inert or reactive gases to remove solid or gaseous impurities. Fluxing practice in copper alloy melting and casting encompasses a variety of different fluxing materials and functions. Fluxes are specifically used to remove gas or prevent its absorption into the melt, to reduce metal loss, to remove specific impurities and nonmetallic inclusions, to refine metallic constituents, or to lubricate and control surface structure in the semicontinuous casting of mill alloys. The last item is included because even these fluxes fall under the definition of inorganic chemical compounds used to treat molten metal.

Types of Fluxes

Fig. 6

Foundry installation of high-frequency induction lift swing furnaces

raised or lowered by a hydraulic cylinder. This unit, also called a push-out furnace, operates by lowering the crucible into the coil for melting and then raising the crucible out of the coil for pickup and pouring. While one crucible is melting, the other crucible can be charged and ready to melt when the knife switch is pulled as the completed heat is being pushed up for skimming and pouring. The other common type of coreless induction melting is the lift swing furnace (Fig. 6). Here, the coil (and box) is cantilevered from a center post to move up or down vertically and swing horizontally about the post in a 90 arc. Because there are two crucible positions, one

crucible can be poured, recharged, and placed into position to melt, while the other is melting. When the metal is ready to pour, the furnace box is lifted (by hydraulic or air cylinder), pivoted to the side, and lowered over the second crucible. The ready crucible is then standing free and can be picked up and poured, while melting is taking place in the second furnace.

Fluxing of Copper Alloys The term fluxing is used in this article to represent all additives to, and treatments of, molten

Fluxes for copper alloys fall into five basic categories: oxidizing fluxes, neutral cover fluxes, reducing fluxes (usually graphite or charcoal), refining fluxes, and semicontinuous casting mold fluxes. Oxidizing fluxes are used in the oxidationdeoxidation process; the principal function here is control of hydrogen gas content. This technique is still practiced in melting copper alloys in fuel-fired crucible furnaces, where the products of combustion are usually incompletely reacted and thus lead to hydrogen absorption and potential steam reaction (see the section “Degassing of Copper Alloys” in this article). The oxidizing fluxes usually include cupric oxide or manganese dioxide (MnO2), which decompose at copper alloy melting temperatures to generate the oxygen required. Figure 7 illustrates the effectiveness of oxidizing fluxes in reducing porosity due to hydrogen and in improving mechanical properties for a tin bronze alloy. Neutral cover fluxes are used to reduce metal loss by providing a fluid cover. Fluxes of this type are usually based on borax, boric acid, or glass, which melts at copper alloy melting temperatures to provide a fluid slag cover. Borax melts at approximately 740  C (1365  F). Such glassy fluxes are especially effective when used with zinc-containing alloys, preventing zinc flaring and reducing subsequent zinc loss by 3 to 10%. The glassy fluid cover fluxes also agglomerate and absorb nonmetallic impurities from the charge (oxides, molding sand, machining lubricants, and so on). As with aluminum alloys, fluxes containing reactive fluoride salts (CaF2 and NaF) can strip oxide films in copper-base alloys, thus permitting entrained

1030 / Casting of Nonferrous Alloys metallic droplets to return to the melt phase. Table 5 indicates the effectiveness of this type of flux in reducing melt loss in yellow and high-tensile brass. For red brasses, however, it may not be proper to use a glassy flux cover, because such a cover will prevent or limit beneficial oxidation of the melt (see the section “Degassing of Copper Alloys” in this article). Use of a glassy cover flux can sometimes result in reduced alloy properties (Ref 2). Oxide films in aluminum and silicon bronzes also reduce fluidity and mechanical properties. Fluxes containing fluorides, chlorides, silica, and borax provide both covering and cleaning, along with the ability to dissolve and collect these objectionable oxide skins. Chromium and beryllium-copper alloys oxidize readily when molten; therefore, glassy cover fluxes and fluoride salt components are useful here in controlling melt loss and achieving good separation of oxides from the melt. Reducing fluxes containing carbonaceous materials such as charcoal or graphite are used on higher-copper lower-zinc alloys. Their principal advantage lies in reducing oxygen absorption of the copper and reducing melt loss. Lowsulfur, dry, carbonaceous flux materials should always be used with copper alloys to avoid gaseous reactions with sulfur or with hydrogen from contained moisture. However, carbonaceous materials will not agglomerate

Fig. 7

nonmetallic residues or provide any cleaning action when melting fine or dirty scrap. For this reason, a glassy cover must also be used in the latter case. Table 5 indicates the beneficial effects of a glassy cover when melting brass turnings. Melt-Refining Fluxes. It is possible to remove many metallic impurity constituents from copper alloys through the judicious use of fire refining (oxidation). According to standard free energy of reaction, elements such as iron, tin, aluminum, silicon, zinc, and lead are preferentially oxidized before copper during fire refining (Ref 3), and there is an order of preference for their removal (Fig. 8). These metallic impurities are thus rendered removable if the oxide product formed can be adequately separated from the melt phase itself. A wet cover flux such as borax is useful with fire refining because it will agglomerate the impurity metal oxides formed and minimize the metal content of the dross. The need to refine specific metallic impurities is highly dependent on and variable with the specific alloy system being refined. An alloying element in one family of copper alloys may be an impurity in another, and vice versa. In red brass (Cu-5Zn-5Pb-5Sn; alloy C83600), the elements lead, tin, and zinc are used for alloying, while aluminum, iron, and silicon are impurities. In aluminum bronzes, on the other

hand, lead, tin, and zinc become contaminants, while aluminum and iron are alloying elements. Foundries typically do little melt refining, leaving this assignment to the secondary smelter supplier of their foundry ingot. However, there may be certain instances when additional refining capability is necessary within the foundry or mill. Table 6 gives the results of fire refining a melt of C83600 with aluminum, silicon, and iron contaminants under a variety of flux covers. Fire refining (oxidation) can be used to remove impurities from copper-base melts in approximately the following order: aluminum, manganese, silicon, phosphorus, iron, zinc, tin, and lead. Nickel, a deliberate alloying element in certain alloys but an impurity in others, is not readily removed by fire refining, but nickel oxide can be reduced at such operating temperatures. Mechanical mixing or agitation during fire refining improves the removal capability by increasing the reaction kinetics. Removal is limited, however, and in dilute amounts (<0.05 to 0.10%), many impurities cannot be removed economically. Oxygen-bearing fluxes can be effective in removing certain impurities, although they are less efficient than direct air or oxygen injection. Fig. 9 demonstrates the effect of increasing the copper oxide content of a flux in removing iron and zinc from phosphor bronze. Lead has been removed from copper alloy melts by the application of silicate fluxes or slags. The addition of phosphor copper or the use of a phosphate or borate slag flux cover and thorough stirring improves the rate of lead removal, as shown in Fig. 10 (Ref 5). Sulfur, arsenic, selenium, antimony, bismuth, and tellurium can occur as impurities in copper

Effect of amount of flux used on the porosity and mechanical properties of cast tin bronze alloy. Source: Ref 2

Table 5 Effect of various slags and covers on losses in the melting of high-tensile and yellow brass Metal temperature Alloy

Yellow brass

High-tensile brass

Source: Ref 2

Melting conditions

No lid or cover Charcoal Glassy cover flux No lid or cover Charcoal Cleaning cover flux



C

1085 1087 1100 1090 1095 1090



F

Melting time, min

Gross loss, %

Net loss, %

1985 1989 2012 1994 2003 1994

50 49 62 65 60 54

2.8 1.1 0.9 2.5 1.9 0.6

1.8 0.6 0.4 1.2 0.7 0.3

Fig. 8 Ref 4

Effect of fire refining (oxygen blowing) on the impurity content of molten copper. Source:

Casting of Copper and Copper Alloys / 1031 Table 6

Effect of fire refining under various fluxes on impurity levels in leaded red brass 



50 kg (110 lb) heats were melted under 1 kg (2.2 lb) of flux at 1150 C (2100 F). Melt No.

Flux

1

Borax

2

Borax-25% sand

3

Borax-20MnO

4

70CaF2-20Na2SO410Na2CO3

5

30NaF-20CaF2-20KF20Na2SO4-10Na2CO3

Impurities in refined metal, %

Refining time, min

Oxygen used, liters per kilogram of metal

Amount of zinc in refined metal, %

0 10 20 0 10 20 0 10 20 0 10 20

0 1.5 2.9 0 1.5 2.9 0 1.5 3.1 0 1.5 3.1

5.49 5.66 5.00 6.30 6.24 6.45 7.40 7.35 7.33 7.47 7.07 6.69

0.05 0.08 0.015 0.05 0.011 0.02 0.14 0.13 0.017 0.12 0.007 0.12 0.21 0.26 0.05 0.19 0.017 0.12 0.10 0.08 <0.005 0.02 <0.005 <0.005

0.38 0.38 0.29 0.37 0.42 0.41 0.47 0.52 0.51 0.62 0.55 0.45

0 10 20

0 1.8 3.5

7.35 6.74 7.20

0.23 0.005 <0.005

0.70 0.83 0.88

Al

Si

0.38 0.17 0.08

Fe

Source: Ref 4

Fig. 9

Influence of oxygen content of flux in reducing (a) iron and (b) zinc impurities in phosphor bronze. Source: Ref 2

Fig. 10

Removal of lead from oxidized Cu-0.1Pb melts at 1150  C (2100  F) by different slags (fluxes) 2% of charge weight. Melt stirred with nitrogen at 2 L/min. Source: Ref 5

alloy scrap, foundry ingot, and prime metal through incomplete refining of metal from the ore, electronic scrap, other scrap materials, or cutting lubricant. These impurities can largely be controlled by application and thorough contacting with fluxes containing sodium carbonate (Na2CO3) or other basic flux additives such as potassium carbonate (K2CO3). Figure 11 demonstrates the ability of Na2CO3 fluxes plus fire refining in eliminating arsenic, bismuth, and antimony from copper. Sulfur is a harmful impurity in copper-nickel or nickel-silver alloys. It can be removed from these materials by an addition of manganese metal or magnesium. Aluminum is often a contaminant in copper alloy systems, particularly the leaded tin bronzes and red brasses. Porosity and lack of pressure tightness result when the aluminum content is as little as 0.01%. Aluminum can be removed by a flux containing oxidizing agents to oxidize the aluminum, and fluoride salts to divert the Al2O3 from the melt and render it removable. Silicon can also be removed, but only after the aluminum has reacted. As much as 0.3% contaminant can be reduced to less than 0.01%, ensuring pressure tightness, by using a flux consisting of 30% NaF, 20% CaF2, 20% Na3AlF6, 20% Na2SO4, and 10% Na2CO3 at an addition rate of 1 to 1.5% for 10 min at approximately 1100  C (2010  F) (Ref 7). As usual, the flux must be intimately mixed with the melt to ensure good reactivity. The melt should then be allowed to settle, and the flux residue or slag layer should be thoroughly skimmed. Borax is useful as a flux constituent for refining to provide adequate fluidity and to agglomerate flux-reacted impurity oxides without excessive entrapment and loss of alloying elements. The borax mineral fluxes razorite and colemanite are commonly used in secondary smelting practice in converting copper alloy brass and bronze scrap to specified-composition foundry ingot. Chlorine fluxing also has potential for refining impurities from copper-base melts, particularly magnesium, aluminum, manganese, zinc, iron, lead, and silicon. However, very little chlorine refining is practiced commercially; the process may be cost effective only when removing aluminum. Mold Fluxes. Certain mold-lubricating fluxes have been used in the direct chill semicontinuous casting of brass and copper alloys into semifinished wrought shapes. These fluxes serve to protect the metal from oxidation during casting. They also act as lubricants so that the solidifying skin separates easily from the mold wall as the solidifying billet or slab moves downward from the mold during casting. Especially in brass alloys, zinc flaring and zinc oxide (ZnO) formation on the melt surface reduce lubricity, causing tearing and other undesirable skin defects during solidification that are detrimental to subsequent forming operations. Fluxing compounds are used on

1032 / Casting of Nonferrous Alloys the melt surface feeding the mold to alleviate these problems. Fluxes used to alleviate this problem usually contain borax, fluoride salts, soda ash, and eutectic salt mixtures to ensure that the flux is molten as the cast continues. In the direct chill casting of higher-copper alloys, graphite may also be present in such fluxes. Because solid flux particles can cause inclusions in the solidifying skin, the flux must be free of coarse particles and must melt quickly.

Degassing of Copper Alloys In the melting and casting of many copper alloys, hydrogen gas absorption can occur because of the generous solubility of hydrogen in the liquid state of these alloys. The solid solubility of hydrogen is much lower; therefore, the gas must be rejected appropriately before or during the casting and solidification process to avoid the formation of gas porosity and related defects (excessive shrinkage, pinholing, blowholes, and blistering). These defects are almost always detrimental to mechanical and physical properties, performance, and appearance. Different copper alloys and alloy systems have varying tendencies toward gas absorption and subsequent problems. Other gases, particularly oxygen, can cause similar problems (see the section “Deoxidation of Copper Alloys” in this article).

Fig. 11

Sources of Hydrogen There are many potential sources of hydrogen in copper, including the furnace atmosphere, charge materials, fluxes, external components, and reactions between the molten metal and the mold. Furnace Atmosphere. The fuel-fired furnaces sometimes used in melting can generate free hydrogen because of the incomplete combustion of fuel oil or natural gas. Charge Materials. Ingot, scrap, and foundry returns may contain oxides, corrosion products, sand or other molding debris, and metalworking lubricants. All these contaminants are potential sources of hydrogen through the reduction of organic compounds or through the decomposition of water vapor from contained moisture. Fluxes. Most salt fluxes used in copper melt treatment are hygroscopic. Damp fluxes can therefore result in hydrogen pickup in the melt from the decomposition of water. External Components. Furnace tools such as rakes, puddlers, skimmers, and shovels can deliver hydrogen to the melt if they are not kept clean. Oxides and flux residues on such tools are particularly insidious sources of contamination because they will absorb moisture directly from the atmosphere. Furnace refractories, troughs and launders, mortars and cements, sampling ladles, hand ladles, and pouring ladles

Effect of fire refining and use of Na2CO3 flux on removal of arsenic, bismuth, and antimony impurities from copper. Source: Ref 6

also are potential sources of hydrogen, especially if refractories are not fully cured. Metal/Mold Reactions. If metal flow is excessively turbulent during the pouring process, air can be aspirated into the mold. If the air cannot be expelled before the start of solidification, hydrogen pickup can result. Improper gating can also cause turbulence and suctioning. Excessive moisture in green sand molds can provide a source of hydrogen as water turns to steam.

Gas Solubility Hydrogen is the most obvious gas to be considered in copper alloys. Figure 12 shows the solubility of hydrogen in molten copper. As in aluminum and magnesium alloys, solubility is reduced in the solid state; therefore, the hydrogen must be removed prior to casting or rejected in a controlled manner during solidification. Alloying elements have varying effects on hydrogen solubility (Fig. 13).

Fig. 12

Solubility of hydrogen in copper. Source: Ref 8

Fig. 13

Effect of alloying elements on the solubility of hydrogen in copper. Source: Ref 9

Casting of Copper and Copper Alloys / 1033 Table 7 Summary of gases found in copper alloys Alloy family

Gases present

Remarks

Pure copper

Water vapor, hydrogen

Approximate hydrogen/water vapor ratio of 1. Higher purity increases the amount of water vapor and lowers hydrogen. Lead does not affect the gases present. Higher tin lowers total gas content. Increased zinc increases the amount of hydrogen, with a loss in water vapor. The presence of 5 wt% Ni in alloy C95800 causes CO to occur rather than water vapor. Lower aluminum leads to higher total gas contents. Approximate hydrogen/water vapor ratio of 0.5. Increased zinc decreases hydrogen and increases water vapor. All three gases are present up to 4 wt% Ni, after which only CO and hydrogen are present. Hydrogen increases with increasing nickel up to 10 wt% Ni but is decreased at 30 wt% Ni.

Cu-Sn-Pb-Zn Water vapor, alloys hydrogen

Fig. 14

Copper-oxygen phase diagram. Source: Ref 9

Oxygen also presents a potential problem in most copper alloys. In the absence of hydrogen, oxygen alone may not cause problems, because it has limited solubility in the melt. However, it forms a completely miscible liquid phase with the copper in the form of cuprous oxide (Fig. 14). During solidification, the combination of cuprous oxide and hydrogen can give rise to casting porosity resulting from the steam reaction (discussed later). Sulfur gases have significance in primary copper through the smelting of sulfide ores and in the remelting of mill product scrap containing sulfur-bearing lubricants. Sulfur dioxide is the most probable gascous product. Foundry alloys and foundry processing usually do not experience sulfur-related problems unless high-sulfur fossil fuels are used for melting. Carbon can be a problem, especially with the nickel-bearing alloys. The nickel alloys have extensive solubility for both carbon and hydrogen. Carbon can be deliberately added to the melt, along with an oxygen-bearing material such as nickel oxide. The two components will react to produce a carbon boil, that is, the formation of CO bubbles, which collect hydrogen as they rise through the melt (Ref 10). If not fully removed, however, the residual CO can create gas porosity during solidification. Nitrogen does not appear to be detrimental or to have much solubility in most copper alloys. However, there is some evidence to suggest that nitrogen porosity can be a problem in cast copper-nickel alloys (Ref 11). Water vapor can exist as a discrete gascous entity in copper alloys (Ref 8, 12, 13). Water vapor is evolved from solidifying copper alloys, which always have some residual dissolved oxygen. When the oxygen becomes depleted, hydrogen is produced as a separate species. The type and amount of gas absorbed by a copper alloy melt and retained in a casting depend on a number of conditions, such as melt temperature, raw materials, atmosphere, pouring conditions, and mold materials. Table 7 lists the gases that can be found in a number of copper alloy systems.

Aluminum bronzes

Water vapor, hydrogen, CO

Silicon brasses and bronzes

Water vapor, hydrogen

Coppernickels

Water vapor, hydrogen, CO

Source: Ref 11

Testing for Gases There are essentially three ways to determine the presence of gas in a copper alloy melt. The easiest and simplest is a chill test on a fracture specimen. In this method, a standard test bar is poured and allowed to solidify. The appearance of the fractured test bar is then related to metal quality standards (Ref 14). The second method is a reduced-pressure test similar to the Straube-Pfeiffer test used for aluminum alloys. In recent years, considerable development and definition have been given to this test as adapted for copper alloys, because the different classes of copper alloys have different solidification characteristics that affect the response to the test (Ref 8, 15–17). The objective of the test is to use a reduced pressure only slightly less than the dissolved gas pressure so that the surface of the sample mushrooms but does not fracture (Fig. 15). The test apparatus is shown schematically in Fig. 16. The key to establishing controllable results, that is, unfractured but mushroomed test sample surfaces, is to provide wave front freezing. This is accomplished by proper selection of vessel materials to achieve the correct solidification characteristics. Because the freezing ranges of copper alloys vary from short to wide, different materials must be used. The various material combinations that have proved successful are described in Ref 8.

Degassing Methods Oxidation-Deoxidation Practice. The steam reaction previously mentioned is a result of both hydrogen and cuprous oxide being

present in the melt. These constituents react to form steam, resulting in blowholes during solidification as the copper cools: 2H þ Cu2 O ! H2 OðsteamÞ þ 2Cu

The proper deoxidation of a copper alloy melt will generally prevent this steam reaction, although excessive hydrogen alone can still cause gas porosity if it is not expelled from the casting before the skin is completely solidified. Fortunately, there is a mutual relationship between hydrogen and oxygen solubility in molten copper (Fig. 17). Steam is formed above the line denoting equilibrium concentration but not below. Consequently, as the oxygen content is raised, the capacity for hydrogen absorption decreases. Therefore, it is useful to provide excess oxygen during melting to preclude hydrogen entry and then to remove the oxygen by a deoxidation process to prevent further steam reaction during solidification. This relationship gave rise to the Pell-Walpole oxidation-deoxidation practice for limiting hydrogen, especially when fossil-fuel-fired furnaces were used for melting. The melt was deliberately oxidized using oxygen-bearing granular fluxes or briquetted tablets to preclude hydrogen absorption during melting. The melt was subsequently deoxidized to eliminate any steam reaction with additional hydrogen absorbed during pouring and casting. Zinc Flaring. The term zinc flare applies to those alloys containing at least 20% Zn. At this level or above, the boiling or vaporization point of copper-zinc alloys is close to the usual pouring temperature, as shown in Fig. 18 (Ref 9). Zinc has a very high vapor pressure, which precludes hydrogen entry into the melt. Zinc may also act as a vapor purge, removing hydrogen already in solution. Furthermore, the oxide of zinc is less dense than that of copper, thus providing a more tenacious, cohesive, and protective oxide skin on the melt surface and increasing resistance to hydrogen diffusion. In zinc flaring, the melt temperature is deliberately raised to permit greater zinc vapor formation. Because some zinc loss may occur, zinc may have to be replenished to maintain the correct composition. Inert Gas Fluxing. With gas fluxing, an inert collector or sparger gas such as argon or nitrogen is injected into the melt with a graphite fluxing tube. The bubbling action collects the hydrogen gas that diffuses to the bubble surface, and the hydrogen is removed as the inert gas bubbles rise to the melt surface. The reaction efficiency depends on the gas volume, the depth to which the fluxing tube or lance is plunged, and the size of the collector gas bubble generated. Again, finer bubble sizes have higher surface-area-to-volume ratios and therefore provide better reaction efficiencies. Figure 19 depicts the amount of purge gas necessary to degas a 450 kg (1000 lb) copper melt. Figure 20 shows the response of an aluminum bronze alloy to nitrogen gas purging. The curve for oxygen purging is also shown.

1034 / Casting of Nonferrous Alloys

Fig. 15

Effect of pressure on the appearance of copper alloy reduced-pressure test samples containing the same amount of gas. (a) Pressure of 7 kPa (55 torr) results in surface shrinkage. (b) At 6.5 kPa (50 torr), a single bubble forms. (c) Boiling and porosity occur at 6 kPa (45 torr).

Fig. 16

Fig. 17

Schematic of the reduced-pressure test apparatus used to assess amounts of dissolved gas in copper alloys

Hydrogen/oxygen equilibrium in molten copper. Source: Ref 8

Fig. 18

Influence of zinc content on boiling point or vapor pressure in copper alloys. Source: Ref 9

Solid Degassing Fluxes. Other materials can be used to provide inert gas purging. A tabletted granular flux such as calcium carbonate (CaCO3), which liberates CO2 as the collector gas upon heating, has been successful in degassing a wide variety of copper alloys. This type of degassing is simpler than nitrogen but must be kept dry and plunged as deep as possible into the melt with a clean, dry plunging rod, preferably made of graphite. This type of degasser may also have the advantage of forming the collector gas by chemical reaction in situ. This results in an inherently smaller initial bubble size than injected gas purging for better reaction efficiency. Such an advantage can be inferred from Fig. 21, although after 5 min of degassing, the end results are the same. Other solid degassers in the form of refractory metals or intermetallic compounds, such as CaMn-Si, nickel-titanium, titanium, and lithium, are effective in eliminating porosity due to nitrogen or hydrogen by their ability to form stable nitrides and hydrides. Again, maintaining dry ingredients, deep plunging, and stirring or mixing will enhance their effectiveness. For best results and optimal reaction efficiency, such degassers should be wrapped or encapsulated in copper materials to provide controlled melting when plunged. Alternatively, copper master alloys containing these elements are available. For both inert gas fluxing and solid degassing additions, the sparging gas reaction should not be so violent as to splash metal and create an opportunity for gas reabsorption. Furthermore, a melt can be overdegassed; an optimal amount of residual gas remaining in the melt helps to counter localized shrinkage in long-freezingrange alloys such as leaded tin bronzes. Vacuum degassing is not generally applied to copper alloys, although it can be very effective. However, the cost of the equipment necessary is relatively high, and there may be a substantial loss of more volatile elements having a high vapor pressure, such as zinc and lead. Furthermore, significant superheat (up to 150  C, or 300  F) may be required to accommodate the temperature drop during the degassing treatment, further aggravating the vapor losses.

Fig. 19

Amount of purge gas required to degas a 450 kg (1000 lb) copper melt. Source: Ref 17

Casting of Copper and Copper Alloys / 1035 Phosphorus Deoxidation. Most copper alloys are deoxidized by a phosphorus reduction of the cuprous oxide. Although several other oxygen scavengers are possible according to the free energy of oxide formation, phosphorus is usually the easiest, most economical, and least problematic deoxidant. The phosphorus is usually added in the form of 15% phosphor copper master alloy, either in granular shot or briquetted waffle form. Care must be taken to ensure that the deoxidant is dry. The deoxidant is often added to the bottom of the ladle before pouring so that during pouring the deoxidant reacts with the cuprous oxide contained in the melt. The turbulence created during pouring is sufficient to ensure adequate mixing. The phosphorus-copper deoxidant should not be simply thrown onto the surface of the ladle after pouring, because little mixing will result. When the deoxidant is added directly to the furnace, however, it should be completely stirred into the melt using clean, dry tools, and pouring should begin as soon as possible so that the effect is not lost. Use of phosphorus deoxidation results in the formation of a liquid slag of cuprous phosphate:

Fig. 20

Effect of nitrogen flow rate during purging on the residual gas pressure remaining in an aluminum bronze melt. The curve for oxygen purging is also shown. Source: Ref 16, 18

phosphorus or lithium deoxidants can react with green sand containing moisture. Hydrogen is liberated and absorbed as the metal comes into contact with the sand. This can be minimized by correct deoxidant additions (discussed in the following section in this article). In addition, finer facing sands or mold coatings (inert or reactive such as sodium silicate or magnesia) can be used; these confine any reaction to the mold/metal interface area and retard hydrogen penetration into the solidifying skin of the cast metal.

Deoxidation of Copper Alloys

Fig. 21

Comparison of the effectiveness of solid degassing flux versus nitrogen purging. Source: Ref 18

Auxiliary Degassing Methods. There are other ways to degas a copper alloy melt than using a specific treatment. The technique of lowering the melt superheat temperature, if possible, and holding the melt (in a dry, minimal gas environment) provides outgassing simply by lowering the equilibrium liquid-state solubility of the gas. During casting, the use of chills to provide directional solidification, particularly for the long-freezing-range group III alloys such as the red brasses and tin bronzes, results in less tendency for gas porosity. In the mold-metal reaction, previously degassed and deoxidized metal containing excess

All copper alloys are subject to oxidation during most melting operations. Oxygen reacts with copper to form cuprous oxide, which is completely miscible with the molten metal. A eutectic is formed at 1065  C (1950  F) and 3.5% Cu2O, or 0.39% O, as shown in the copper-oxygen phase diagram in Fig. 14. Therefore, cuprous oxide exists within the melt as a liquid phase and is not generally separated by gravity alone. If not removed, this liquid phase will cause discontinuous solidification during casting, resulting in considerable porosity and low mechanical strength. Thus, some type of deoxidation process is required. In addition, proper deoxidation of all melts enhances fluidity and therefore castability. Of course, deliberate oxidation treatments (the oxidation-deoxidation process previously described) are still employed. These are designed to preclude hydrogen pickup in copper alloy melts.



2P þ 4Cu þ 70 ! 2Cu2 O P2 O5

This product easily separates from the rest of the melt; therefore, the phosphorus effectively scavenges the oxygen, delivering the product to the surface dross phase, where it can be easily skimmed. Phosphorus is usually added at a rate of 57 g (2 oz) of 15% master alloy per 45 kg (100 lb) of melt. In cases where oxidation of the melt is deliberately employed to reduce hydrogen absorption, double this amount can be required for full deoxidation. This is usually sufficient to deoxidize completely a melt saturated with 0.39 wt% O. However, the recovery of phosphorus is less than 100% and may be as little as 30 to 60% (Ref 19). It is desirable to maintain a residual level of at least 0.01 to 0.015% P in the melt, especially during pouring, so that reoxidation and potential steam reaction problems are alleviated. Each foundry must determine the proper addition rate for a given set of conditions. Furthermore, with foundry returns, a certain residual amount of phosphorus is usually already present. A routine addition of phosphor copper for deoxidation could then actually result in excessive phosphorus. If the phosphorus content is too high during melting or pouring, the lack of oxygen may invite hydrogen entry and result in the steam reaction during casting. Adding more phosphorus to control the steam reaction can therefore actually aggravate the condition. In addition, when the phosphorus content is 0.03% and beyond, excessive metal fluidity can result in penetration of the molding sand or burn-in during casting. Phosphorus-copper is an effective deoxidant for the red brasses, tin bronzes, and leaded

1036 / Casting of Nonferrous Alloys bronzes. However, phosphorus should not be used for deoxidizing high-conductivity copper alloys, because of its deleterious effect on electrical conductivity (as is discussed in the next section), and it should not be used for coppernickel alloys. In these materials, the presence of phosphorus results in a low-melting constituent that embrittles the grain boundaries. A silicon addition of 0.3% and a magnesium addition of 0.10% serve to deoxidize and desulfurize copper-nickel melts. For nickel-silver alloys (Cu-Ni-Zn), the use of 142 g (5 oz) copper-manganese shot per 45 kg (100 lb) of melt, and 57 g (2 oz) manganese coupled with 85 g (3 oz) of 15% P-Cu, is a recommended deoxidation technique. The yellow brasses, silicon bronzes, manganese bronzes, and aluminum bronzes usually do not require deoxidation per se, because of the oxygen-scavenging effects of their respective alloy constituents. High-Conductivity Copper (Ref 19–22). Where the high-copper alloys (pure copper, silver-copper, cadmium-copper, tellurium-copper, beryllium-copper, chromium-copper) are employed and electrical conductivity is a desirable property, phosphorus-copper cannot be used as a deoxidant. Moreover, the strong oxide formers beryllium and chromium serve as their own deoxidants. Figure 22 shows the effects of a variety of elements on the electrical conductivity of copper. Clearly, phosphorus even in small amounts significantly decreases conductivity; therefore,

Fig. 22

alternative deoxidants must be used. Fortunately, both boron and lithium are capable of deoxidizing high-conductivity copper without appreciably affecting electrical conductivity. Boron Deoxidation. Boron is available either as a copper-boron master alloy or as calcium boride (CaB6). The boron probably forms a copper-borate slag of the general form 2Cu2O B2O3 in much the same fashion as phosphorus produces a cuprous oxide phosphate slag (Ref 19). Theoretically, the boron combines with 60% more oxygen than the stoichiometric amount required to form B2O3 and therefore appears to be superefficient. However, practical experience has shown that this theoretical efficiency is not always achieved and that lithium is more effective, as shown in Fig. 23 (Ref 19). Therefore, lithium is often preferred, although there are greater precautions attendant upon its use. Lithium Deoxidation (Ref 19–23). Lithium has the advantage of serving as both a deoxidant and a degasser because it reacts readily with both oxygen and hydrogen. Lithium is soluble in molten copper but insoluble in the solid state. There is very little residual lithium contained in the casting or in scrap for remelt. Because lithium metal is very reactive in air, bulk lithium metal must be stored in oil. For foundry applications, lithium is supplied in sealed copper cartridges. These cartridges must be stored in a safe, dry environment and must be preheated (to above 105  C, or 225  F) before use to drive off any surface moisture.



Effect of alloying elements on the electrical conductivity of copper. Source: Ref 20

These preheated cartridges should then be carefully yet firmly and quickly plunged to the bottom of the reacting vessel (furnace or ladle) to achieve full intimate contact and reactivity with the bulk of the melt. Only clean, dry, preheated plunging tools should be used for this task. Graphite rods are usually preferred. Lithium-copper cartridges are generally available in various sizes ranging from 2.25 g (0.09 oz) for 23 kg (50 lb) melts to 108 g (4 oz). Thus, lithium additions can be made at maximum efficiency. The specific chemical reactions that can occur with lithium include: 2Li þ Cu2 O ! 2Cu þ Li2 O ðdeoxidationÞ Li þ H ! LiH ðdegassingÞ LiH þ Cu2 O ! 2Cu þ LiOH ðrecombinationÞ 2Li þ H2 O ! Li2 O þ H2 ðreaction during pouringÞ

The lithium oxide (Li2O) and lithium hydroxide (LiOH) products separate cleanly as a low-density fluid slag suitable for skimming (Ref 20). The lithium hydride that forms initially if hydrogen is present is unstable at normal copper melting temperatures and recombines with cuprous oxide (recombination). When lithium is added in excess of the amount of cuprous oxide present, it will react with moisture in the air during pouring and can therefore generate sufficient hydrogen to regas the melt. This can result in unanticipated additional gas porosity and unsoundness during solidification. Because lithium is such an effective deoxidant, it can also reduce residual impurity oxides (FeO, P2O5, and so on) in high-conductivity copper melts. This allows these elements to redissolve in the molten metal to the extent of their solubility limit and thus reduce electrical conductivity according to Fig. 22. Furthermore, lithium can form intermetallic compounds with silver, lead, tin, and zinc when the residual lithium exceeds that required for deoxidation. These intermetallic compounds, while reducing solid solubility, may improve mechanical properties and electrical conductivity.

Fig. 23

Deoxidant efficiency in copper alloy melts. Source: Ref 19

Casting of Copper and Copper Alloys / 1037 Occasionally, it may be desirable to practice a duplex deoxidation treatment using the less expensive phosphorus, followed by a lithium treatment (Ref 20, 22, 23). Care must be taken to do this quickly and not allow phosphorus reversion to occur by letting the copper phosphate deoxidation slag remain on the melt for an extended period of time. Magnesium Deoxidation. Magnesium behaves similarly to lithium but may be stored in air rather than oil. It is actually a stronger deoxidant in terms of its free energy of oxide formation (Ref 24), and it is used to deoxidize (and desulfurize) copper-nickel alloys. The deoxidation product (magnesia) is a stable refractory, unlike lithium compounds, but it forms a tenacious oxide skin and can result in inclusions in copper casting alloys. Testing for Proper Deoxidizer Addition. As stated previously, each foundry should assess its own casting practice for a given alloy and set of melting conditions and should determine the optimal addition of deoxidizer. However, there are two tests that can be used to determine whether a given amount of deoxidizer is adequate. In the first test, a test plug or shrink bar of metal approximately 75 mm (3 in.) in diameter by 75 mm (3 in.) deep is poured. If a shrinkage cavity results, the metal is deoxidized and ready for pouring, which should then be done immediately. Shrinkage will not occur until approximately 0.01% residual phosphorus is present (Ref 24). A puffed-out or mushroomed cap on the test plug indicates that deoxidation is incomplete and that more should be added. The second test involves a carbon or graphite rod immersed in the melt. When the rod surface reaches the molten-metal temperature, if there is oxygen present, the rod will vibrate because of the reaction (2C + O2 ! 2CO) occurring on the bar surface. The intensity of the vibration is a function of the oxygen content, and an experienced foundryman can readily determine the point at which the reaction becomes negligible, that is, when the melt is sufficiently deoxidized. The vibration decreases near the level of 0.01% residual phosphorus, as expected for a deoxidized melt (Ref 24). Relative Effectiveness of Copper Deoxidizers. Various elements capable of scavenging oxygen from copper alloy melts have been described. The theoretical relative capabilities of several deoxidizers are:

In practice, selection of the deoxidizer must be based on actual efficiency, economics, ease of use, and the specific metallurgical requirements of the alloy in question.

Grain Refining of Copper Alloys In general, the grain refinement of copper alloys is not practiced as a specific molten-metal processing step per se, because a certain degree of refinement can be achieved through normal casting processes. As with aluminum alloys, grain refinement in copper alloys can be achieved by rapid cooling, mechanical vibration, or the addition of nucleating or grain-growth-restricting agents. Further, many commercial copper alloys have sufficient solute (zinc, aluminum, iron, tin) to achieve constitutional supercooling during solidification. In this case, grain nucleation and growth are naturally retarded. Commercially pure copper can be grain refined by small additions (as little as 0.10%) of lithium, bismuth, lead, or iron, which provide constitutional supercooling effects (Ref 25, 26). Copper-zinc single-phase alloys can be grain refined by additions of iron or by zirconium and boron (Ref 27). In the latter case, the probable mechanism is the formation of zirconium boride particle nuclei for grain formation. In one case, the vibration of a Cu-32Zn-2Pb-1Sn alloy improved yield and tensile strengths by approximately 15%, with a 10% reduction in grain size from the unvibrated state. In general, the a copper-zinc alloys (<35% Zn) exhibit grain size reduction and greater improvement in properties, while the a-b alloys do not. Copper-aluminum alloys have been effectively grain refined with additions of 0.02 to 0.05% B; the effective nucleating agent is boron carbide (B4C). Figure 24 illustrates the improvement in mechanical properties achieved by grain refining a Cu-10Al alloy.

Tin bronze alloys have been successfully grain refined by the addition of zirconium (0.02%) and boron master alloys (Ref 28). However, pressure tightness is reduced because in these long-freezing-range alloys, finer grain size concentrates porosity because of gas entrapment.

Filtration of Copper Alloys The filtration process consists of passing the molten metal through a porous device (a filter) in which the inclusions contained in the flowing metal are trapped or captured. The filter material itself must have sufficient integrity (strength, refractoriness, thermal shock resistance, and corrosion resistance) so that it is not destroyed by the molten metal before its task is accomplished. Consequently, most filter media are ceramic materials in a variety of configurations. Although the bulk of filter technology has been directed toward aluminum alloys, the use of filtration is increasing for copper alloy melts. Inclusions in Copper Alloys. In addition to the usual inclusions arising from oxides, fluxing salts, and intermetallics, copper oxide inclusions and phosphorus pentoxide (from deoxidation) may be present in copper alloys if the melt is not allowed to settle or if it is inadequately skimmed before pouring and casting. Filter Applications. The filtration of copper casting alloys primarily involves the use of ceramic foam sections in the gating system. Oxide inclusions have been successfully removed from aluminum bronze alloys using such a filter deployment system (Ref 29). Investment castings can be successfully filtered using a filter section in the pouring cup or, in the case of larger castings, molded directly into the wax runner bars. Ceramic filters for copper alloys are usually alumina, mullite, or zirconia.

Amount of deoxidizer required to remove 0.01% oxygen Deoxidizer

Carbon Phosphorus Cu-15P Boron Cu-2B Lithium Magnesium

Reaction products

g/100 kg

oz/100 lb

CO CO2 P2O5 2Cu2P2O5 P2O5 2Cu2P2O5 B2O3 B2O3 Li2O MgO

7.5 3.8 7.5 5.6 49.3 36.8 4.4 224.5 8.7 15.0

0.12 0.06 0.12 0.09 0.79 0.59 0.07 3.60 0.14 0.24

Fig. 24

Effect of boron-refined grain size on the mechanical properties of Cu-10Al alloy. Test specimens were removed from the center or the top of the ingot as indicated. Source: Ref 28

1038 / Casting of Nonferrous Alloys

Melt Treatments for Group I to III Alloys

the pouring temperatures in Table 8 for group I to III alloys. More detailed information on fluxing and other melt practices can be found in earlier sections of this article.

For group I to III alloys, the melting procedure, flux treatment, and pouring temperature (Table 8) vary considerably from one alloy family to another. It should also be noted that for any copper alloy, the temperature at which the metal is poured into the mold is higher than the liquidus temperature. Compare the liquidus temperatures in Tables 2 to 4 with those of

Group I Alloys Pure Copper and Chromium-Copper. Commercially pure copper and high-copper alloys are very difficult to melt and are very susceptible to gassing. In the case of

Table 8 Pouring temperatures of copper alloys Light castings Alloy type

Group 1 alloys Copper Chromium-copper Yellow brass

Manganese bronze

Aluminum bronze

Nickel bronze

White brass Group II alloys Beryllium-copper

Silicon brass Silicon bronze

Copper-nickel Group III alloys Leaded red brass

Leaded semired brass Tin bronze

Leaded tin bronze

High-leaded tin bronze

UNS No.



C

Heavy castings 

F



C



F

C81100 C81500 C85200 C85400 C85800 C87900 C86200 C86300 C86400 C86500 C86700 C86800 C95200 C95300 C95400 C95410 C95500 C95600 C95700 C95800 C97300 C97600 C97800 C99700 C99750

1230–1290 1230–1260 1095–1150 1065–1150 1150–1175 1150–1175 1150–1175 1150–1175 1040–1120 1040–1120 1040–1095 1150–1175 1120–1205 1120–1205 1150–1230 1150–1230 1230–1290 1120–1205 1065–1150 1230–1290 1205–1225 1260–1425 1315–1425 1040–1095 1040–1095

2250–2350 2250–2300 2000–2100 1950–2100 1950–2150 1950–2150 1950–2150 1950–2150 1900–2050 1900–2050 1900–2000 1950–2150 2050–2200 2050–2200 2100–2250 2100–2250 2250–2350 2050–2200 1950–2100 2250–2350 2200–2240 2300–2600 2400–2600 1900–2000 1900–2000

1150–1230 1205–1230 1010–1095 1010–1065 1010–1095 1010–1095 980–1065 980–1065 950–1040 950–1040 950–1040 980–1065 1095–1150 1095–1150 1095–1175 1095–1175 1175–1230 1095–1205 1010–1205 1175–1230 1095–1205 1205–1315 1260–1315 980–1040 980–1040

2100–2250 2200–2250 1850–2000 1850–1950 1850–2000 1850–2000 1800–1950 1800–1950 1750–1900 1750–1900 1750–1900 1800–1950 2000–2100 2000–2100 2000–2150 2000–2150 2150–2250 2000–2200 1850–2200 2150–2250 2000–2200 2250–2400 2300–2400 1800–1900 1800–1900

C81400 C82000 C82400 C82500 C82600 C82800 C87500 C87800 C87300 C87600 C87610 C96200 C96400

1175–1220 1175–1230 1080–1120 1065–1120 1050–1095 995–1025 1040–1095 1040–1095 1095–1175 1095–1175 1095–1175 1315–1370 1370–1480

2150–2225 2150–2250 1975–2050 1950–2050 1925–2000 1825–1875 1900–2000 1900–2000 2000–2150 2000–2150 2000–2150 2400–2500 2500–2700

1220–1260 1120–1175 1040–1080 1010–1065 1010–1050 1025–1050 980–1040 980–1040 1010–1095 1010–1095 1010–1095 1230–1315 1290–1370

2225–2300 2050–2150 1900–1975 1850–1950 1850–1925 1875–1925 1800–1900 1800–1900 1850–2000 1850–2000 1850–2000 2250–2400 2350–2500

C83450 C83600 C83800 C84400 C84800 C90300 C90500 C90700 C91100 C91300 C92200 C92300 C92600 C92700 C92900 C93200 C93400 C93500 C93700 C93800 C94300

1175–1290 1150–1290 1150–1260 1150–1260 1150–1260 1150–1260 1150–1260 1040–1095 1040–1095 1040–1095 1150–1260 1150–1260 1150–1260 1175–1260 1095–1205 1095–1230 1095–1230 1095–1205 1095–1230 1095–1230 1095–1205

2150–2350 2100–2350 2100–2300 2100–2300 2100–2300 2100–2300 2100–2300 1900–2000 1900–2000 1900–2000 2100–2300 2100–2300 2100–2300 2150–2300 2000–2200 2000–2250 2000–2250 2000–2200 2000–2250 2000–2250 2000–2200

1095–1175 1065–1175 1065–1175 1065–1175 1065–1175 1040–1150 1040–1150 980–1040 980–1040 980–1040 1040–1175 1040–1150 1050–1150 1065–1175 1040–1095 1040–1121 1010–1150 1040–1150 1010–1150 1040–1150 1010–1095

2000–2150 1950–2150 1950–2150 1950–2150 1950–2150 1900–2100 1900–2100 1800–1900 1800–1900 1800–1900 1900–2150 1900–2100 1920–2100 1950–2150 1900–2000 1900–2050 1850–2100 1900–2100 1850–2100 1900–2100 1850–2000

chromium-copper, oxidation loss of chromium during melting is a problem. Copper and chromium-copper should be melted under a floating flux cover to prevent both oxidation and the pickup of hydrogen from moisture in the atmosphere. In the case of copper, crushed graphite should cover the melt. With chromium-copper, the cover should be a proprietary flux made for this alloy. When the molten metal reaches 1260  C (2300  F), either calcium boride or lithium should be plunged into the molten bath to deoxidize the melt. The metal should then be poured without removing the floating cover. Yellow Brasses. These alloys flare, or lose zinc, due to vaporization at temperatures relatively close to the melting point. For this reason, aluminum is added to increase fluidity and keep zinc vaporization to a minimum. The proper amount of aluminum to be retained in the brass is 0.15 to 0.35%. Above this amount, shrinkage takes place during freezing, and the use of risers becomes necessary. Other than the addition of aluminum, the melting of yellow brass is very simple, and no fluxing is necessary. Zinc should be added before pouring to compensate for the zinc lost in melting. Manganese Bronzes. These alloys are carefully compounded yellow brasses with measured quantities of iron, manganese, and aluminum. The metal should be melted and heated to the flare temperature or to the point at which zinc oxide vapor can be detected. At this point, the metal should be removed from the furnace and poured. No fluxing is required with these alloys. The only addition required with these alloys is zinc. The amount required is that which is needed to bring the zinc content back to the original analysis. This varies from very little, if any, when an all-ingot heat is being poured, to several percent if the heat contains a high percentage of remelt. Aluminum Bronzes. These alloys must be melted carefully under an oxidizing atmosphere and heated to the proper furnace temperature. If needed, degasifiers can be stirred into the melt as the furnace is being tapped. By pouring a blind sprue before tapping and examining the metal after freezing, it is possible to tell whether it shrank or exuded gas. If the sample purged or overflowed the blind sprue during solidification, degassing is necessary. As discussed earlier in this article, degasifiers remove hydrogen and oxygen. Also available are fluxes that convert the molten bath. These are in powder form and are usually fluorides. They aid in the elimination of oxides, which normally form on top of the melt during melting and superheating. Nickel-silvers, also known as nickel bronzes, are difficult alloys to melt. They gas readily if not melted properly because the presence of nickel increases the hydrogen solubility. Then, too, the higher pouring temperatures shown in Table 8 aggravate hydrogen pickup. These alloys must be melted under an oxidizing atmosphere and quickly superheated to the proper furnace temperature to allow for

Casting of Copper and Copper Alloys / 1039 temperature losses during fluxing and handling. Proprietary fluxes are available and should be stirred into the melt after tapping the furnace. These fluxes contain manganese, calcium, silicon, magnesium, and phosphorus and do an excellent job in removing hydrogen and oxygen. White Manganese Brass. There are two alloys in this family, both of which are copper-zinc alloys containing a large amount of manganese and, in one case, nickel. They are manganese bronze-type alloys, are simple to melt, and can be poured at low temperatures because they are very fluid (Table 8). They should not be overheated, since this serves no purpose. If the alloys are unduly superheated, zinc is vaporized and the chemistry of the alloy is changed. Normally, no fluxes are used with these alloys.

Group II Alloys Beryllium-Coppers. These alloys are very toxic and dangerous if beryllium fumes are not captured and exhausted by proper ventilating equipment. They should be melted quickly under a slightly oxidizing atmosphere to minimize beryllium losses. They can be melted and poured successfully at relatively low temperatures (Table 8). They are very fluid and pour well. Silicon Bronzes and Brasses. The alloys known as silicon bronzes, UNS alloys C87300, C87600, and C87610, are relatively easy to melt and should be poured at the proper pouring temperatures (Table 8). If overheated, they can pick up hydrogen. While degassing is seldom required, if necessary, one of the proprietary degasifiers used with aluminum bronze can be successfully used. Normally, no cover fluxes are used here. The silicon brasses (UNS alloys C87500 and C87800) have excellent fluidity and can be poured slightly above their freezing range. Nothing is gained by excessive heating, and, in some cases, heats can be gassed if this occurs. Here again, no cover fluxes are required. Copper-Nickel Alloys. These alloys— 90Cu-10Ni (C96200) and 70Cu-30Ni (C96400)—must be melted carefully because the presence of nickel in high percentages raises not only the melting point but also the susceptibility to hydrogen pickup. In virtually all foundries, these alloys are melted in coreless electric induction furnaces, because the melting rate is much faster than it is with a fuel-fired furnace. When ingot is melted in this manner, the metal should be quickly heated to a temperature slightly above the pouring temperature (Table 8) and deoxidized either by the use of one of the proprietary degasifiers used with nickel bronzes or, better yet, by plunging a 0.1% Mg stick to the bottom of the ladle. The purpose of this is to remove all the oxygen to prevent any possibility of steam-reaction porosity from occurring. Normally, there is little

need to use cover fluxes if the gates and risers are cleaned by shotblasting prior to melting.

Group III Alloys These alloys, namely leaded red and semired brasses, tin and leaded tin bronzes, and highleaded tin bronzes, are treated the same in regard to melting and fluxing and thus can be discussed together. Because of the long freezing ranges involved, it has been found that chilling, or the creation of a steep thermal gradient, is far better than using only feeders or risers. Chills and risers should be used in conjunction with each other for these alloys. For this reason, the best pouring temperature is the lowest one that will pour the molds without having misruns or cold shuts. In a well-operated foundry, each pattern should have a pouring temperature that is maintained by use of an immersion pyrometer. Fluxing. In regard to fluxing, these alloys should be melted from charges comprising ingot and clean, sand-free gates and risers. The melting should be done quickly in a slightly oxidizing atmosphere. When at the proper furnace temperature to allow for handling and cooling to the proper pouring temperature, the crucible is removed or the metal is tapped into a ladle. At this point, a deoxidizer (15% phosphor copper) is added. The phosphorus is a reducing agent (deoxidizer). This product must be carefully measured so that enough oxygen is removed, yet a small amount remains to improve fluidity. This residual level of phosphorus must be closely controlled by chemical analysis to a range between 0.010 and 0.020% P. If more is present, internal porosity may occur and cause leakage if castings are machined and pressure tested. In addition to phosphor copper, pure zinc should be added at the point at which skimming and temperature testing take place prior to pouring. This replaces the zinc lost by vaporization during melting and superheating. With these alloys, cover fluxes are seldom used. In some foundries in which combustion cannot be properly controlled, oxidizing fluxes are added during melting, followed by final deoxidation by phosphor copper.

Production of Copper Alloy Castings Copper alloy castings are produced by sand, shell, plaster, investment, permanent mold, die, centrifugal, and continuous casting. Each of these casting/molding methods is briefly reviewed in the following section, “Casting Process Selection.”

Pouring Temperature and Practice Pouring Temperature. Temperature ranges for pouring the principal copper casting alloys are given in Table 8. It should not be inferred from the breadth of most of these ranges that

pouring temperature is not critical; as noted in Table 8, the ranges are intended for pouring various section thicknesses. For castings with minimum section thickness , the metal should be poured at a temperature near the high side of the range. Conversely, for castings that have all heavy sections, pouring temperatures should be near the low side of the range. Under any conditions, identical castings should be poured at the same temperature, insofar as possible. It is generally advisable to allow a variation of no more than 55  C (100  F) during the pouring of a specific mold or when pouring several molds from the same ladle. The casting process used also influences the pouring temperature for a specific alloy. In die casting, for instance, a temperature near the low side of a given range is used, in the interest of longer die life. Pouring Practice. On the basis of the degree of care required when they are poured into molds, copper alloys can be classified into two groups:  Alloys that form tight, adherent, nonfluid

slags or oxides. Typical are group I aluminum bronze and manganese bronze alloys.  Alloys that form fluid slags or oxides. These include most of the alloys in general use— those containing various combinations of copper, tin, lead, and zinc. Copper alloys in the first group (those that form tight, adherent, nonfluid slags) require great care in pouring. Their general behavior can be compared to that of aluminum casting alloys, and similar pouring techniques are recommended. Good pouring practice for this first group of alloys includes attention to:  Molten alloys with tight oxide films should

never be stirred. After the ladle is filled or the crucible is removed from the furnace, and before pouring, the metal should be carefully skimmed but not stirred or mixed—thus minimizing oxide entrapment.  In addition to avoiding stirring of molten aluminum bronze and manganese bronze, other forms of agitation should also be avoided. If the metal is melted in a tilting furnace and must be transferred to a ladle for pouring, the distance the metal must drop should be minimized by holding the ladle close to the furnace lip.  Pouring should be smooth and even to avoid splashing and separated metal streams. With careful pouring of aluminum bronze and manganese bronze, it is possible to form an aluminum oxide “glove” around the metal stream, which will protect the molten metal from further oxidation. Alloys in this first group have a very narrow freezing range (Table 2), so that they solidify in much the same way as does a pure metal. The total shrinkage is concentrated in the region of the casting that solidifies last. These alloys are thus prone to piping and gross shrinkage

1040 / Casting of Nonferrous Alloys cavities. Risering is commonly used to prevent shrinkage from occurring in the casting. The metal is poured well above the liquidus so that the entire mold cavity is filled and so that solidification occurs from the bottom to the top, with feeding from a riser. Copper alloys in the second group (those that form fluid slags) are generally less affected by turbulence in pouring than are those in the first group. Although turbulence in pouring can cause casting defects in any alloy, the fact that the oxides that are formed with this group of alloys separate readily from the molten metal means less likelihood of oxide entrapment in the casting and greater likelihood of the escape of entrained air bubbles.

Casting Process Selection Among the more important factors that influence the selection of a casting method are (Ref 30):    

The number of castings to be processed The size and/or weight of the casting The shape and intricacy of the product The amount and quality of finish machining needed  The required surface finish  The prescribed level of internal soundness (pressure tightness) and/or the type and level of inspection to be performed Table 9

 The permissible variation in dimensional

accuracy for a single part and part-to-part consistency through the production run  The casting characteristics of the copper alloy specified Table 9 summarizes some of the technical factors that go into the choice of casting method for casting alloys. Additional information can be found in the descriptions of individual casting processes that follow. Sand casting accounts for approximately 75% of U.S. copper alloy foundry production (Ref 30). The process is relatively inexpensive, reasonably precise, and highly versatile. It can be used for castings ranging in size from a few ounces to many tons. Further, it can be applied to simple shapes as well as castings of considerable complexity, and it can be used with all of the copper casting alloys. There are a number of variations of the sand casting process. In green sand casting—the most widely used process—molds are formed in unbaked (green) sand, which is most often silica bonded with water and a small amount of clay to develop the required strength. The clay minerals absorb water and form a natural bonding system. Various sands and clays may be blended to suit particular casting situations. Other variations of sand casting include dry sand, cement-bonded sand, sodium silicatebonded sand (the CO2 process), and resinbonded sand molding.

Technical factors in the choice of casting method for copper alloys

Casting method

Copper alloys

Size range

General tolerances

þ1/32 in. up to 3 in.; þ3/64 in.; 3–6   in.; add  þ0.003 in./in. above 6 in.; add þ0.020 to þ0.060 in.   across parting line Same as sand casting

Sand

All

All sizes, depends on foundry capability

Nobake

All

All sizes but usually > 10 lb Typical maximum mold area = 550 in.2, typical maximum thickness = 6 in. Depends on foundry capability; best,
50 lb Best max thickness,
2 in.

Shell

All

Permanent mold

Coppers, high-copper alloys, yellow brasses, high-strength brasses, silicon bronze, highzine silicon brass, most in bronzes, aluminum bronzes, some nickel-silvers Limited to C85800, C86200, C86500, C87800, C87900, C99700, C99750, and some proprietary alloys Coppers, high-copper alloys, silicon bronze, manganese bronze, aluminum bronze, yellow brass

Die

Plaster

Best for small, thin parts; max area  3 ft2 Up to 800 in.2 but can be larger

Investment

Almost all

Fraction of an ounce to 150 lb, up to 48 in.

Centrifugal

Almost all

Ounce to 25,000 lb. Depends on foundry capacity

Source: Ref 30

Sand casting imposes few restrictions on product shape. The only significant exceptions are the draft angles that are always needed on flat surfaces oriented perpendicular to the parting line. Dimensional control and consistency in sand castings ranges from approximately þ0.8 to 3.2 mm (þ0.030 to 0.125 in.). Within   this range, the more generous tolerances apply across the parting line. Surface finish ranges between approximately 7.7 and 12.9 mm (300 and 500 min.) root mean square (rms). Shell Molding. Resin-bonded sand systems are also used in the shell molding process, in which prepared sand is contacted with a heated metal pattern to form a thin, rigid shell. The shell molding process is capable of producing quite precise castings and nearly rivals metal mold and investment casting in its ability to reproduce fine details and maintain dimensional consistency. Surface finish, at approximately 3.2 mm (125 min.) rms, is considerably better than that from green sand casting. Shell molding is best suited to small-tointermediate-sized castings. Relatively high pattern costs (pattern halves must be made from metal) favor long production runs. On the other hand, the fine surface finishes and good dimensional reproducibility can, in many instances, reduce the need for costly machining. While still practiced extensively, shell molding has declined somewhat in popularity, mostly because of its high energy costs compared with no-bake resin-bonded sand methods;

þ0.005–0.010 in. up to 3 in.; add  þ0.002 in./in. above 3 in.; add  þ0.005–0.010 in. across parting  line.

Minimum section thickness

Ordering quantities

150–500 min. rms

1/8–1/4

All

1–3

Same as sand casting 125–200 min. rms

All

1–3

3/32

100

2–3

100–1000, depending on size

2–3

Surface finish

in.

Same as sand casting in.

Relative cost, (1 low, 5 high)

Usually  þ0.010 in.; optimum þ0.005 in., þ0.002 in.   part-to-part

150–200 min. rms. best
70 min. rms

1/8–1/4

þ0.002 in./in.; no less than 0.002  in. on any one dimension; add þ0.010 in. on dimensions  affected by parting line One side of parting line, þ0.015 in.  up to 3 in.; add þ0.002 in./in.  above 3 in.; add 0.010 in. across parting line, and allow for parting line shift of 0.015 in. þ0.003 in. less than 1/4 in.; þ0.004   in. between 1/4 to ½ in.; þ0.005  in./in. between ½–3 in.; add þ0.003 in./in. above 3 in.  Castings are usually rough machined by foundry.

32–90 min. rms

0.05–0.125 in.

>1000

1

63–125 min. rms. best
32 min. rms

0.060 in.

All

4

63–125 min. rms

0.030 in.

>100

5

Not applicable

1/4

All

in.

in.

1–3

Casting of Copper and Copper Alloys / 1041 however, shell-molded cores are still very widely used. Plaster Molding. Copper alloys can also be cast in plaster molds to produce precision products of near-net shape. Plaster-molded castings are characterized by surface finishes as smooth as 0.8 mm (32 min.) rms and dimensional tolerances as close as þ0.13 mm (þ0.005 in.) and   typically require only minimal finish machining. Compared to other casting methods, plaster molding accounts for a very small fraction of the copper castings market. Investment casting, also known as precision casting or the lost wax process, is capable of maintaining very high dimensional accuracy in small castings, although tolerances increase somewhat with casting size. Dimensional consistency ranks about average among the casting methods; however, surface finishes can be as fine as 1.5 mm (60 min.) rms, and the process is unsurpassed in its ability to reproduce intricate detail. Investment casting is better suited to castings under 45 kg (100 lb) in weight. Because of its relatively high tooling costs and higher-thanaverage total costs, the process is normally reserved for relatively large production runs of precision products and is not often applied to copper alloys. Permanent Mold Casting. As the name implies, permanent mold casting makes use of reusable metal molds, or dies, in place of the sand-based molds used in conventional foundries. The molds are “permanent” in the sense that they can be used thousands of times. Permanent mold castings are characterized by good part-to-part dimensional consistency and very good surface finishes (approximately 1.8 mm, or 70 min.). Any traces of metal flow lines on the casting surface are cosmetic rather than functional defects. Permanent mold castings exhibit good soundness. There may be some microshrinkage, but mechanical properties are favorably influenced by the characteristically fine grain size of the casting. The ability to reproduce intricate detail is only moderate, however, and for products in which very high dimensional accuracy is required, plaster mold or investment processes should be considered instead. Permanent mold casting is more suitable for simple shapes in midsized castings than it is for very small or very large products. Die costs are relatively high, but the absence of molding costs makes the overall cost of the process quite favorable for medium to large production volumes. Die casting involves the injection of liquid metal into a multipart die under high pressure. Pneumatically actuated dies make the process almost completely automated. Die casting is best known for its ability to produce high-quality products at very low unit costs. Very high production rates offset the cost of the complex heat-resisting tooling required, and with low labor costs, overall casting costs are quite attractive.

The process can be used with several copper alloys, including yellow brass, C85800; manganese bronzes, C86200 and C86500; silicon brass, C87800; the special die-casting alloys C99700 and C99750; plus a few proprietary compositions. These alloys can be die cast because they exhibit narrow freezing ranges and high b-phase contents. Rapid freezing is needed to complement the fast cycle times of the process. Rapid freezing also avoids the hot shortness associated with prolonged mushy solidification. Beta phase contributes the hot ductility needed to avoid hot cracking as the casting shrinks in the unyielding metal mold. Highly intricate copper alloy products can be made by die casting (investment casting is even better in this regard). Dimensional accuracy and part-to-part consistency are unsurpassed in both small (<25 mm, or 1 in.) and large castings. The attainable surface finish, often as good as 0.76 mm (30 min.) rms, is better than with any other casting process. Die casting is ideally suited to the mass production of small parts. Extremely rapid cooling rates (dies are normally water cooled) result in very fine grain sizes and good mechanical properties. Leaded alloys C85800 and C99750 can yield castings that are pressure tight, although lead is incorporated in these alloys more for its favorable effect on machinability than for its ability to seal porosity. Centrifugal castings are produced by pouring molten metal into a mold that is being rotated or revolved. Both centrifugal and semicentrifugal castings can be described as castings that are spun on their own axes during the castings operation. The axis of rotation may be horizontal or inclined at any angle up to the vertical position. Molten metal is poured into the spinning mold cavity, and the metal is held against the wall of the mold by centrifugal force. The speed of rotation and metal pouring rate vary with the alloy and size and shape being cast. The castings with larger diameter than the axial length are cast vertically, while pieces with smaller lengths are cast horizontally. A wide range of castings, such as bearings, bushings, and gears of all types for applications in general machine production, road building equipment, farm machinery, and steel mill and marine applications, are produced. Centrifugal castings are made in sizes ranging from approximately 50 mm to 3.7 m (2 in. to 12 ft) in diameter and from a few inches to many yards in length. Size limitations, if any, are likely as not based on the melt shop capacity of the foundry. Simply shaped centrifugal castings are used for items such as pipe flanges and valve components, while complex shapes can be cast by using cores and shaped molds. Pressureretaining centrifugal castings have been found to be mechanically equivalent to more costly forgings and extrusions. Continuous casting is a process whereby molten copper alloy is fed through an openended graphite mold yielding a bar, tube, or

shape of the required cross section. This process, which is performed on a continuous basis, can be accomplished either vertically or horizontally, with the molten metal drawn from the molten-metal reservoir or lundish at a point below the surface of the bath. The solidified form is cooled and withdrawn at a controlled rate from the water-cooled mold by rollers, and the material is cut to length with a traveling saw. Continuous casting is used to produce bearing blanks and other long castings with uniform cross sections. It is the principal method used for the large-tonnage production of semifinished products such as cast rods, tube rounds, gear and bearing blanks, slabs, and custom shapes. The extremely high cooling and solidification rates attending continuous casting can, depending on the alloy, produce columnar grains. The continuous supply of molten metal at the solidification interface effectively eliminates microshrinkage and produces high-quality, sound products with very good mechanical properties. With its simple die construction, relatively low equipment cost, high production rate, and low labor requirements, continuous casting is a very economical production method.

Gating (Ref 31) The major function of a gating system is to deliver clean metal from the ladle into the mold cavity without adversely affecting the quality of the metal. Secondary considerations are the ease of molding, removal of gates, and high casting yield. However, these factors should not dictate a design that contributes to the production of castings of unacceptable quality. The Pouring Basin. The production of highquality castings requires not only proper melting and molding operations and properly designed pattern equipment but also an understanding of the principles of gating so that clean metal can be delivered to the mold cavity with a minimum amount of turbulence. A pouring basin allows a sprue to be filled quickly and maintains a constant head throughout the pour (Fig. 25).

Fig. 25

Section of a typical sand mold with pouring basin

1042 / Casting of Nonferrous Alloys When the weight of poured metal in a mold exceeds 14 kg (30 lb), use of a pouring basin offers many advantages. The pourer can better direct the flow of metal from the ladle into the basin, with less chance of spillages; also, the sprue need not be located near the edge of the mold. The pouring ladle can be brought within 25 to 50 mm (1 to 2 in.) of the basin, and a continued flow rate may be more easily maintained through a larger pouring head. If there are any brief interruptions in pouring the metal into the basin, the surplus metal will take up the slack until pouring has resumed. The major disadvantage of the pouring basin is that the yield is lowered, thereby requiring more metal to be recycled. Sprue. The correct sprue size is the single most important part of the gating system. If a wrong size is selected, or an improper taper is used, the damage done to the metal in the mold cavity is extensive and cannot be corrected regardless of the quality of the runner and gating systems. Because most molds under approximately 14 kg (30 lb) of poured weight are made on a high-production scale in flasks of 102 to 152 mm (4 to 6 in.) in cope height, a fairly standard sprue size may be used for all copper-base alloys. The top third of the sprue should be the pouring part, with approximately a 50 mm (2 in.) diameter opening. The remaining portion of the sprue should taper down to 13 to 22 mm (½ to 7/8 in.) in final diameter, depending on the pouring rate to be used. Figure 26 shows a sketch of a sprue that will do an excellent job of conveying brass or bronze into the gating system. There are many charts and formulas available to determine the entry diameter of a tapered sprue, but for the most part, this diameter should be just sufficient to provide approximately a 10 to 20 slope on the side of the sprue. When the sprue height is over 305 mm (12 in.), the top diameter of the sprue is much more important and should be approximately 50% larger than the diameter at

Fig. 26

Funnel sprue, sprue basin, and chokes (in drag runner) for reducing turbulence

the base of the sprue. When designing a pouring system for sprues of 102 to 152 mm (4 to 6 in.) in height, it is best to select the desired pouring rate first in order to determine the proper sprue base to be used. The Sprue Base. Because the velocity of the stream is at its maximum at the bottom of the sprue and is proportional to the square root of the height of the fall of the metal, it is mandatory that a sprue base or well be used as a cushion for the stream flowing down the sprue. The base also helps change the vertical flow of metal into a horizontal flow with the least amount of turbulence. Recommended sprue basin sizes are approximately twice as deep as the drag runner and two to three times as wide as the base of the sprue. In most cases, a well 25 to 38 mm (1 to 1½ in.) deep with a width of 38 to 50 mm (1½ to 2 in.) on each side is usually adequate for the majority of sprues being used for most normal pouring rates. Little damage is done if the sprue base is larger than necessary, except that the overall casting yield will be lowered slightly. Chokes should be used only when the proper pouring rate cannot be controlled by the correct sprue size. If clean metal is delivered into the sprue, a strainer core serves the sole function of retarding the metal flow rate. Conventional strainer cores, whether of tinned steel, mica, glass fiber, or ceramic, usually reduce the flow of metal by approximately 70%, depending on the size and number of holes that are open to the sprue area. The best strainer is one that has only one hole with a diameter of the correct sprue size. This avoids the turbulence caused by the metal being divided into many streams as it enters the runner. In no instance should a strainer be placed into the top of the sprue; if one must be used, the only suitable place is just above the sprue base at the parting line. Tinned steel strainers are the least acceptable because the remelted runners can introduce iron and hard spots to the copper-base alloys if they are not properly skimmed during melting. The mica and glass fiber strainers are popular because they can be laid on the parting line just above the sprue base before the mold is closed, thereby requiring no “prints” or recesses, as do the thicker ceramic or sand strainer cores, which are usually approximately 3.2 mm (1/8 in.) thick. A choke in the runner pattern is often the only consistent way to achieve a proper pouring rate. In no instance should the choke be put in the gate area. When necessary, it should be placed in the drag runner as close to the sprue as possible (Fig. 26). The chokes should have a smooth radiused contour and be located at the bottom of the drag runner. Choke depth may vary from 1/4 to 3/4 of the total runner depth, with a cross-sectional area not exceeding 3/4 of the area of the sprue base. The chokes should be located within an inch of the sprue base to ensure rapid filling of the sprue and maintenance of full capacity throughout the pour. This also permits dissipation of the turbulence before the stream reaches the gates.

Pouring rate depends on many factors, such as weight of the casting, section size , height of the sprue, and alloy system. Most alloys in group III for small work weighing 14 kg (30 lb) or less are poured with a hand ladle at approximately 1.8 kg/s (4 lb/s). Light memorial plaque castings are being successfully poured at 4.5 kg/s (10 lb/s), while many automatic pouring units operate at mold pouring rates of 3.6 to 4.5 kg/s (8 to 10 lb/s). Alloys in group I, if the poured weight is under 14 kg (30 lb), should be poured at 0.9 to 1.8 kg/s (2 to 4 lb/s) in order to obtain a clean, nonturbulent metal flow in the mold. Sprue exit diameters required for specific flow rates and various sprue heights are shown in Fig. 27. Table 10 shows flow rates from the bottom of the sprue for a number of commonly used sprue heights and diameters. As an example, for a gross casting weight of 14 kg (30 lb) or less and a sprue height of 102 to 152 mm (4 to 6 in.), a sprue diameter of 13 to 19 mm (½ to 3/4 in.) is adequate to obtain a flow rate of 0.9 to 1.8 kg/s (2 to 4 lb/s). It should not be necessary to use a 22 to 29 mm (7/8 to 11/8 in.) diameter sprue base unless pouring plaque work or using automatic pouring. A quite popular sprue for most production work is the 16 to 19 mm (5/8 to 3/4 in.) diameter size, which will deliver enough hot metal to fill most molds up to 14 kg (30 lb) weight in 8 to 10 s. The total pouring time in seconds may be calculated by dividing the total weight of the mold poured (castings plus gates and risers) by the flow rate at the base of the sprue, or: Total weight of castings including gates and risersðlbÞ Flow rate at base of sprueðlb=sÞ ¼ Calculated pouring time in seconds

Runners and Gates. For alloys in groups I and II, it is mandatory that all runners be placed in the drag and as much of the casting as possible be placed in the cope. In this way, all runners will be completely filled before any metal enters the gates, and the metal will drop the least amount or will rise to enter the mold cavity from the gates. Although this practice is also excellent for alloys in group III, experience has shown that quality castings may be obtained by using more traditional casting techniques because the alloys in group III are less sensitive to drossing and have a tendency to self-heal when dross is formed in the gating system. Runners should be as rectangular in shape as possible, and their total maximum cross-sectional area should be two to four times that of the tapered sprue or the choke, if chokes are used in the runner system. Care must be taken to ensure that the cross section of the runners is adequate in order to prevent premature chilling. Experience has shown that a rectangular runner with the wide side laying horizontal works best. The next best is a square runner, and the least desirable is a rectangular runner with the wide side being vertical, although sometimes space limitations necessitate use of this type of runner in order to obtain the

Casting of Copper and Copper Alloys / 1043

Fig. 27

Fig. 28

Typical single-cavity gating system with tapered runner

Fig. 29

Method of running a pump impeller with a well at the end of the runner

Fig. 30

Recommended multiple-cavity gating system with stepped runner

Flow rates of copper-base alloys through tapered sprues of varying diameter and height

Table 10 Flow rates of copper-base alloys through tapered sprues of varying diameter and height Sprue Area

Flow rate for sprue height, mm (in.) Diameter

102 (4)

152 (6)

305 (12)

610 (24)

1220 (48)

mm2

in.2

mm

in.

kg/s

lb/s

kg/s

lb/s

kg/s

lb/s

kg/s

lb/s

kg/s

lb/s

129 194 284 387 506 645

0.2 0.3 0.44 0.60 0.785 1.0

13 16 19 22 25 29

½

0.82 1.27 1.81 2.49 3.40 4.30

1.8 2.8 4.0 5.5 7.5 9.5

0.91 1.50 2.04 2.95 3.86 4.76

2.0 3.3 4.5 6.5 8.5 10.5

1.36 2.04 2.95 4.08 5.22 7.71

3.0 4.5 6.5 9.0 11.5 17.0

1.81 2.72 4.08 5.67 7.48 9.30

4.0 6.0 9.0 12.5 16.5 20.5

2.72 4.08 6.12 8.16 11.11 13.61

6.0 9.0 13.5 18.0 24.5 30.0

5/8

/4

3

7/8

1 11/8

proper ratios. The rectangular runner should be approximately twice as wide as it is deep. The cross-sectional area of the runner must be reduced by that of each gate as it is passed, so that metal enters the mold cavity simultaneously from each gate (Fig. 28). Because backfilling is seldom desired from the runner system, a well at the end of the runner can be used (Fig. 29), particularly if the runner does not have any taper. A good example of multiple-cavity gating may be seen in Fig. 29. X-ray movies of metal flow in sand molds show that relatively uniform gate discharge rates are achieved only when stepped or tapered runners are used. Multiple gates are shown in Fig. 28 to 30. The preferred location is in the cope just above the runner at the parting line. A rectangular flat gate is more desirable than a square gate, and a gate that has its wide dimension in the vertical plane is the least desirable, just as is the case for runners. In order to avoid a pressurized gating system, it is important that the total gate area be at least as large as the total runner area.

If a pattern has an excessive amount of small castings, it may be necessary to have the total gate area many times the runner area in order to obtain a sufficient gate to each casting. This deviation is acceptable because the gating system remains unpressurized. Figure 28 shows a good gating system that produces the minimum amount of turbulence. Gates should enter the casting cavity at the lowest possible level in order to avoid the erosion and turbulence associated with a falling stream of molten metal. To ensure nonturbulent filling of the casting nearest to the sprue, its gate should be at least 50 mm (2 in.) away from the base of the sprue. Regardless of the excellence of a gating system design, castings of acceptable quality will not be produced if the ladle is not positioned as close as is practical to the pouring basin or sprue, and if the sprue is not filled quickly and kept at full capacity throughout the pour. Knife and Kiss Gating Systems. Special applications of gating systems work in many cases for specific castings. Knife and kiss

gating are popular when group III alloys are used but not recommended for groups I and II because these alloys form too much dross with this system and cannot be fed adequately to eliminate surface shrinkage. Advantages are a high casting yield, easy removal of runner systems, and minimum grinding of gates. The major disadvantage is that many small castings become detached during shakeout, necessitating their manual retrieval from mechanized systems. Figure 31 shows a graphic representation of the arrangement of knife and kiss gating. In kiss gating, the casting must be completely in the cope or the drag with the runner overlapping the casting by 0.8 to 2.4 mm (1/32 to 3/32 in.). Actually, there is no gate in this system because

1044 / Casting of Nonferrous Alloys the metal goes directly from the runner into the casting. Knife gating is used when the casting is in both the cope and the drag and there is a contact at the parting line of 0.8 to 2.4 mm (1/32 to 3/32 in.) thickness just at the casting surface. Knife gating systems work well when the runner is in just the cope or just the drag or in both the cope and the drag. Maximizing Casting Quality. Excellent, clean, high-quality castings may be obtained for the copper-base alloy groups of narrow-, intermediate-, and wide-freezing-range alloys if the basic principles discussed for the pouring basin, sprue, sprue base, chokes, pouring rates, runners, and gates are applied. By following these recommendations, the maximum ease in molding, casting yield, and ease of removal of gates and runners may be obtained.

Feeding The objectives of feeding or risering are to eliminate surface sinks or draws and to reduce internal shrinkage porosity to acceptable levels (less than 1%). To minimize porosity, the feeding system must establish:

group III (long-freezing-range) alloys are described separately.

Feeding Group I and II Alloys The feeding technique for these alloys is similar to that used in the manufacture of steel castings. Gates and risers are positioned such that directional solidification is ensured, with due consideration being given to the feeding range of the alloy in question. To avoid hot spots, local chills may be applied to bosses, ribs, and to other sections having sudden changes in thickness. Solidification Contours. The first step in determining riser placement is to draw the solidification contours. This is illustrated by the hypothetical casting shown in Fig. 33, which consists essentially of a plate to which is attached a thinner section, C, and a boss, B. The thin end of the casting, C, would normally undergo rapid cooling after pouring as the result of edge-cooling effects. Thus, it is possible to place the riser at the heavy section, A,

and gate through the riser to provide favorable temperature gradients. The dotted lines in Fig. 34 show successive positions of the solidification front. As shown, porosity will develop in the boss unless a chill is placed on the boss or the riser is relocated there. A chill is a block of metal or other material with a higher heat conductivity and heat capacity than sand. Feeding Ranges. The number and location of the feeders to be used must be consistent with the feeding range of the alloy. The feeding range is the distance that can be fed by a feeder on a bar or plate. It is generally desirable to divide the casting into a number of sections to determine the number of risers to be used. Because all parts of a casting must be within the feeding range of at least one of the risers, it is important to have quantitative information regarding feeding ranges. The information regarding feeding ranges. The feeding range values for group I and II copper-base alloys have not been well documented. In the absence of specific data for particular alloys, satisfactory results can often be attained by applying

 Directional

solidification, as shown in Fig. 32, in which the solidification front is substantially V-shaped in a longitudinal cross section, with the large end of the V directed toward the riser  Steep temperature gradients along the casting toward the riser The feeding techniques for group I (shortfreezing-range) alloys and group II (mediumfreezing-range) alloys can be discussed together. The basic principles of risering of

Fig. 31

Basic kiss and knife gates

Fig. 32

Features of progressive and directional solidification. Source: Ref 32

Fig. 33

Hypothetical casting used to illustrate the principles of feeding technique. Source: Ref 33

Fig. 34

Mode of freezing the casting in Fig. 33 without special precaution to avoid shrinkage. Source: Ref 33

Casting of Copper and Copper Alloys / 1045

Fig. 35

Effect of chills in increasing feeding range of risers. Source: Ref 32

is obtained from a plot of VR/VC versus (L + W)/T, where VR and VC are the riser and casting volume, respectively. Work sponsored by the American Foundry Society has led to the development of a series of curves for aluminum bronzes, copper-nickel, and manganese bronzes (Fig. 36–39). Figures 36 to 39 indicate that the riser volumes necessary for sound castings can be reduced and the effectiveness of feeding improved by the use of exothermic sleeves. Insulating sleeves also can be used to increase the effectiveness of the risers. Exothermic sleeves, which were once popular, have now largely been discontinued in favor of insulating sleeves, which are more economical to use and cause fewer problems. In recent years, greater attention has been given to what has become known as the modulus method for calculating riser size, which includes the development of pertinent data and computer programs to check that riser volume is adequate. This method is now, and no doubt will continue to be, the most widely used technique in the industry. Chvorinov’s rule states that the freezing time, t, of a cast shape is given by the relationship:



t ¼ k ðV =AÞ2

(Eq 1)

where V and A are the volume and surface area, respectively, of the cast shape, and k is a constant proportionality whose value is dependent on the thermal properties of the metal and the mold. For convenience, the term (V/A) in Chvorinov’s equation is generally replaced by the symbol M, a value referred to as the modulus of the shape. Equation 1 can be rewritten more simply to read:



t ¼ k M2

Fig. 36

Naval Research Laboratories-type riser size curve for manganese bronze (alloy C86500). Source: Ref 34

values that have been developed for carbon steels. The following approximate values for feeding ranges have been quoted in the literature but should be used with caution: Alloy

Shape

Manganese bronze

Square bars Plates

Aluminum Square bars bronze Nickel-aluminum Square bars bronze Copper-nickel Square bars

Feeding distance, T

pffiffiffiffi 4 T to 10 T depending on thickness 5.5 to 8 T, depending on thickness pffiffiffiffi 8 T pffiffiffiffi <8 T pffiffiffiffi 5:5 T

Use of chills can further increase feeding range. Consequently, the spacing between risers may be increased to approximately ten times the section thickness if chills are located midway between each pair of risers (Fig. 35). Riser Size. From time to time, various methods have been proposed for the calculation of the optimal riser size to be used to feed a particular casting or casting section. One of the earlier methods was developed for steel castings at the Naval Research Laboratories. In this technique, an empirical “shape factor,” defined as the length (L) plus the width (W) of the casting divided by the thickness (T), that is, (L + W)/T, is first determined. The correct riser size

(Eq 2)

Because Chvorinov’s equation can be applied to any cast shape, it applies equally to that which is intended to be the casting itself and to the attached riser. With connected shapes, such as a riser and a casting, the surface area of each shape to be considered includes only those portions that contribute to the loss of heat during freezing. For the riser to be effective in feeding, its solidification time, tR, must be greater than the solidification time, tC, for the casting. This can be written:

 



tR k MR2 ¼ ¼ F 2 ; or MR2 ¼ F 2 MC2 tC k MC2

(Eq 3)

Further simplified, this becomes:



MR ¼ F MC

(Eq 4)

This means that the modulus for the riser, MR, must be greater than the modulus for the casting, MC, by some factor, F. Experience

1046 / Casting of Nonferrous Alloys it is attached. These values can be calculated if necessary; however, it is much easier (though less precise) to use data of the type shown in Table 11). The numbers listed in the table indicate the minimum values for the ratio between riser volume and casting volume (as percentages) to ensure that the riser can, indeed, supply the necessary amount of feed metal to the casting. Five general classes of castings are shown, ranging from “very chunky” to “rangy.” Notice that risers having a height-to-diameter ratio (H/ D) of 1 to 1 are more efficient than when the H/ D is 2 to 1. More important, it can be seen that insulated risers are far more efficient than those formed directly in the sand mold. Feeder Shape. One of the requirements of the riser is to remain liquid longer than the casting; that is, from Chvorinov’s rule: ðV =AÞR > ðV =AÞC

Fig. 37

Naval Research Laboratories-type riser curve for manganese bronze (alloy C86500) using different types of exothermic hot topping and top risers. Source: Ref 35

Fig. 38

Naval Research Laboratories-type riser curve for aluminum bronze (alloy C95300) using different types of exothermic hot topping and top risers. Source: Ref 35

has shown that the proper value of F depends on the metal being cast. A value of approximately 1.3 is preferred for the short-freezingrange copper-base alloys. As a practical working equation, therefore, one may say that with these alloys, the modulus of the riser should be approximately 1.3 times that of casting (or casting section) to be fed, or:

MR ¼ 1:3 MC

(Eq 5)

Equation 5 merely shows an empirical way of proportioning a riser so that it freezes more slowly than the casting. The other basic requirement of any riser is that it must have sufficient volume to provide the necessary amount of feed metal to the casting or casting section to which

(Eq 6)

The shape with the highest possible V/A ratio is the sphere. However; spherical risers are rarely used in industry because of molding considerations. The next-best shape for a riser is the cylinder. The H/D for cylindrical risers is in the range of 0.5 to 1.0. Riser Neck Dimensions. The ideal riser neck should be dimensioned such that it solidifies after the casting but slightly before the riser. With this arrangement, the shrinkage cavity is entirely within the riser, this being the last part of the casting-riser combination to solidify. Specific recommendations for the dimensions of riser necks are contained in the literature for ferrous alloys. These should apply to shortfreezing-range copper alloys and are given in Table 12. Hot Topping. Approximately 25 to 50% of the total heat from a copper-base alloy riser is lost from the exposed surface by radiation. In order to minimize this radiation loss and thereby increase the efficiency of the riser, some sort of cover should be used on the top surface. Any cover, even dry sand, is better than nothing at all. A reliable exothermic hot topping is one form of usable cover. Chills. The heat abstraction of the mold walls can be increased locally by the use of chills. Although expensive, metal chills are particularly effective because they reduce the solidification time by a factor of more than 55. As mentioned earlier, chills can be used to increase feeding distances and thereby reduce the number of feeders required. When it is impractical to attach feeders at certain locations, chills are particularly useful for initiating directional solidification, for example, at junctions and so on, which would otherwise be porous. Padding. The process of solidification can also be controlled by means of padding. Padding is the added section thickness (usually tapered) to promote directional solidification, and the bulk of it should be as close to the riser as possible.

Casting of Copper and Copper Alloys / 1047 Feeding Group III Alloys The “workhorse” alloys of the copper-base group are the leaded red brasses and tin bronzes, virtually all of which have wide freezing ranges. These alloys have practically no feeding range, and it is extremely difficult to get fully sound castings. The average run of castings in these alloys contains 1 to 2% porosity. Only small castings may exhibit porosity below 1%. Attempts to reduce it more by increasing the size of the risers are often disastrous and actually decrease the soundness of the casting rather than increase it. Experience has shown that success in achieving internal soundness depends on avoiding slow cooling rates. The foundryman has three possible means for doing this, within the limitations of casting design and available molding processes:  Minimize casting section thickness.  Reduce and/or evenly distribute the heat of

the metal entering the mold cavity.

 Use chills and mold materials of high chill-

ing power.

Fig. 39

Naval Research Laboratories-type riser curve for Cu-30Ni (alloy C96400) using exothermic sleeves and hot topping versus hot topping alone. Source: Ref 35

Table 11

 Gate into hot spot.  Riser into hot spot.  Ensure that riser freezes last (consider riser

Minimum volume requirements of risers Minimum VR/VC, % Insulated risers

Type of casting

Very chunky; cubes, and so on; dimensions in ratio 1:1.33:2(a) Chunky; dimensions in ratio 1:2:4(a) Average; dimensions in ratio 1:3:9(a) Fairly rangy; dimensions in ratio 1:10:10(a) Rangy; dimensions in ratio 1:15:30 or larger(a)

In order to produce relatively sound castings, the following points should be considered. Directional solidification, best used for relatively large, thick castings, can be promoted in various ways:

size, insulation, and chills).

Sand risers

H/D = 1:1

H/D = 2:1

H/D = 1.1

H/D = 2.1

32

40

140

198

26

32

106

140

19

22

58

75

13

15

30

38

8

9

12

14

(a) Ratio of thickness: width:length. Source: Ref 32

 Promote high thermal gradients by the use of

chills, preferably tapered chills unless casting section is light (less than 12.5 mm, or ½ in. thick).  Make sure risers are not so large that they unduly extend the solidification time of the casting, which would generate porosity beneath or behind the riser. Uniform solidification, best used for smaller, thin-wall castings, can be promoted in various ways:  Gate into cold spots, using several gates for

uniform temperature distribution.

Table 12 Type of riser

Riser neck dimensions Length, LN

General side Plate side

Short as feasible, not over D/2 Short as feasible, not over D/3

Top

Short as feasible, not over D/2

Cross section

Round, DN = 1.2 LN + 0.1D Rectangular, HN = 0.6 to 0.8T as neck length increases. WN = 2.5 LN + 0.18D Round, DN = LN + 0.2D

LN, DN, HN, WN: length, diameter, height, and width of riser neck, respectively, D, diameter of riser, T, thickness of plate casting. Source: Ref 36

Interaction of Gates and Risers. The effectiveness of side risers can be increased considerably by using a gating system that enters the mold cavity through the riser. The advantages of this arrangement are:

 Cleaner molten metal enters the mold cavity.  Because the metal in the riser remains liquid

for a longer time, steep thermal gradients are established to improve the soundness of the casting.

 Use no risers, except perhaps on gate areas.  Use chills on hot spots to ensure that they cool

at the same rate as the rest of the casting.

 Use chills on areas that must be machined,

thereby moving porosity to areas where the cat skin will be left unmachined; that is, maintain pressure tightness.  Gate into areas away from machined sections to maintain pressure tightness.  Use low pouring temperature (care should be taken to avoid misruns).  See whether increased gas content (no de-gassing, reduced deoxidation) or induced metalmold reaction increases pressure tightness.

1048 / Casting of Nonferrous Alloys  Make castings as thin as possible to increase

cooling rate and reduce machining. 9. ACKNOWLEDGMENTS

10.

This article was adapted from:  R.F. Schmidt, D.G. Schmidt, and M. Sahoo,

Copper and Copper Alloys, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 771–785  D.V. Neff, Nonferrous Molten Metal Processes, Casting, Vol 15, ASM Handbook, ASM International, 1988, p 445–496

11. 12. 13.

REFERENCES 1. Casting Copper-Base Alloys, 2nd ed., American Foundry Society, 2007 2. J.F. Wallace and R.J. Kissling, Fluxing of Copper Alloy Castings, Foundry, 1963 3. J.F. Wallace and R.J. Kissling, Gases in Copper Base Alloys, Foundry, Dec 1962, p 36–39; Jan 1963, p 64–68 4. L.V. Whiting and D.A. Brown, “Air/Oxygen Injection Refining of Secondary Copper Alloys,” Report MRP/PMRL 79-SO (J), Physical Metallurgy Research Laboratories, CANMET, 1979 5. J.E. Stalarcyzk, A. Cibula, P. Gregory, and R.W. Ruddle, The Removal of Lead from Copper in Fire Refining, J. Inst. Met., Vol 85, 1957, p 49 6. J.G. Peacey, G.P. Kubanek, and P. Tarasoff, “Arsenic and Antimony Removal from Copper by Blowing and Fluxing,” paper presented at the Annual Meeting of AIME (Las Vegas, NV), Nov 1980 7. A.V. Larsson, Influence of Aluminum on Properties of Cast Gun Metal and Removal of Aluminum by Slag, Trans. AFS, 1952, p 75 8. P.K. Trojan, T.R. Ostrom, and R.A. Flinn, Melt Control Variables in Copper Base Alloys, Trans. AFS, 1982, p 729; “High

14. 15.

16. 17. 18.

19. 20. 21. 22.

Conductivity Copper Alloys,” Bulletin 43, Foseco, Inc., 1972 R.J. Kissling and J.F. Wallace, Gases in Copper Base Alloys, Foundry, 1962, 1963 M.P. Renatv, C.M. Andres, G.P. Douglas, and R.A. Flinn, Solubility Relations in Liquid Copper-Nickel-Carbon-Oxygen Alloys, Trans. AFS, 1976, p 641 B.N. Ames and N.A. Kahn, Gas Absorption and Degasification of Cast Monel, Trans. AFS, 1947, p 558 T.R. Ostrom, P.K. Trojan, and R.A. Flinn, Gas Evolution from Copper Alloys—A Summary, Trans. AFS, 1981, p 731 T.R. Ostrom, P.K. Trojan, and R.A. Flinn, The Effects of Tin, Aluminum, Nickel, and Iron on Dissolved Gases in Molten Copper Alloys, Trans. AFS, 1980, p 437 Casting Copper Base Alloys, American Foundrymen’s Society, 1984 T.R. Ostrom, R.A. Flinn, and P.K. Trojan, Gas Content of Copper Alloy Melts—Test Equipment and Field Test Results, Trans. AFS, 1977, p 357 T.R. Ostrom, P.K. Trojan, and R.A. Flinn, Dissolved Gases in Commercial Copper Base Alloys, Trans. AFS, 1975, p 485 R.J. Cooksey and R.W. Ruddle, New Techniques in Degassing Copper Alloys, Trans. AFS, Vol 68, 1960 P.K. Trojan, S. Suga, and R.A. Flinn, Influence of Gas Porosity on Mechanical Properties of Aluminum Bronze, Trans. AFS, 1973, p 552 R.J. Kissling and J.F. Wallace, Fluxing of Copper Alloy Castings, Foundry, 1964 M.G. Neu and J.E. Gotheridge, Fluxing and Deoxidation Treatments for Copper, Trans. AFS, 1956, p 616 Y.T. Hsu and B.O. Reilly, Impurity Effects in High Conductivity Copper, J. Met., Dec 1977, p 21 “Lithium Cartridges for Treatment of Copper and Copper Alloys,” Product Bulletin 304, Lithium Corporation of America, 1986

23. R.C. Harris, Deoxidation Practice for Copper, Shell-Molded Castings, Trans. AFS, 1958, p 69 24. J.L. Dion, A. Couture, and J.O. Edwards, “Deoxidation of Copper for High Conductivity Castings,” Report MRP/PMRL-787(J), Physical Metallurgy Research Laboratories, CANMET, April 1978 25. A. Cibula, Grain Refining Additions for Cast Copper Alloys, J. Inst. Met., Vol 82, 1953, p 513 26. G.C. Gould, G.W. Form, and J.F. Wallace, Grain Refinement of Copper, Trans. AFS, 1960, p 258 27. R.J. Kissling and J.F. Wallace, Grain Refinement of Copper Alloy Castings, Foundry, June—July, 1963 28. A. Couture and J.O. Edwards, Grain Refinement of Sand Cast Bronzes and Its Influence on Their Properties, Trans. AFS, 1973, p 453 29. M. Sahoo, J.R. Barry, and K. Kleinschmidt, Use of Ceramic Foam Filters in the Brass and Bronze Foundry, Trans. AFS, 1981, p 611 30. “Copper Casting Alloys,” Publication 7014-0009, Copper Development Association, 1994 31. D.G. Schmidt, Gating of Copper Base Alloys, Trans. AFS, Vol 88, 1980, p 805– 816 32. Casting Copper-Base Alloys, American Foundrymen’s Society, 1964 33. R.W. Ruddle, Risering Copper Alloy Castings, Foundry, Vol 88, Jan 1960, p 78–83 34. R.A. Flinn, Copper, Brass and Bronze Castings—Their Structures, Properties and Applications, Non-Ferrous Founders’ Society, 1963 35. R.A. Flinn, R.E. Rote, and P.J. Guichelaar, Risering Design for Copper Alloys of Narrow and Extended Freezing Range, Trans. AFS, Vol 74, 1966, p 380–388 36. J.W. Wallace, Risering of Castings, Foundry, Vol 87, Nov 1959, p 74–81

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1049-1055 DOI: 10.1361/asmhba0005307

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Casting of Zinc Alloys ZINC CASTINGS are produced by several casting processes, but the vast majority of zinc castings are produced by hot chamber die casting. Other methods include sand casting, permanent mold casting, plaster casting, spin casting (silicone rubber molds), investment (lost wax) casting, squeeze casting, continuous or semicontinuous casting, and centrifugal casting. Zinc alloy castings have also been produced by semisolid casting methods. Zincmatrix metal-matrix composites have also been produced by various foundry methods. Zinc has a relatively low melting point of 420
C (788
F) and forms a eutectic with aluminum at 5% Al and 382
C (720
F). Thus, aluminum is the principal alloying element in the cast zinc alloys, providing good castability and increased cast strength. The addition of aluminum also greatly reduces the iron dissolution rate, permitting the use of ferrous die casting equipment. Because molten zinc does not dissolve hydrogen, there is no need to flux, and cast products are generally sound. Pure zinc is coarse grained, attacks iron when molten, and exhibits property anisotropy when cast and processed. Zinc casting alloys are described in more detail in the article “Zinc and Zinc Alloy Castings” in this Volume. The casting alloys can be divided into two principal classes based on aluminum content: the hypoeutectic alloys with approximately 4% Al for optimal mechanical properties, and the hypereutectic alloys with >5% Al for additional strength, hardness, and stiffness (but at the price of lower ductility). The die casting alloys are basically a family of Zn-4Al alloys. The gravity-fed castings (sand and permanent mold) are made of higher-aluminum-containing alloy.

Control of Alloy Composition Zinc alloys for die casting are sensitive to variations in composition and impurity levels—generally more so than aluminum alloys. However, limitations on the permissible amounts of added elements or impurities in zinc are liberal enough that a reasonable program of control and generally sound shop practice are sufficient to maintain adequate consistency in alloy composition. Agitation. Although agitation during melting will not affect alloy composition, it results in

melt oxidation and should therefore be minimized. Overheating often results in loss of aluminum and magnesium through oxidation and in an increase in iron due to a decrease in the scavenging action provided by aluminum. Use of Foundry Scrap. Clean scrap of acceptable composition returned from the foundry can be charged into the furnace in unlimited proportions, although use of 50% maximum of scrap per charge is recommended. Not surprisingly, the usual practice is to remelt zinc scrap runners, overflow wells, sprues, and castings. However, there are safeguards that should be employed to ensure that remelt does not disrupt the fine balance of additives in the melt. The ability to remelt zinc process scrap many times without losing properties is a significant advantage to the die caster. However, caution should be exercised to keep this material clean and free of unwanted substances. It should be stored separately away from other metals if it is accumulated in batches. If conveyed back to a central melter, conveyors and the furnaces should be covered when maintenance work is being done nearby or overhead. Floors and tables should be kept clean. If there is doubt as to the purity of the scrap, it should not be used in any proportion until it has been analyzed. Scrap returned for recycling must be free of oil and moisture. A safety hazard is created when oil or moisture is present on the metal being charged into the furnace. Some zinc scrap can be electroplated. This material should be remelted separately and added back in small quantities or, better still, sold outright. If this is not practical, the scrap should be fed back moderately. The electrodeposits will separate and float to the top of the bath, where they can be skimmed off. Agitation will increase copper, nickel, and chromium levels and should therefore be avoided. Castings that have had chemical surface treatments can generally be remelted as clean scrap. Under no circumstances should cadmium-plated, tin-plated, or soldered die castings be remelted. Caution should also be exercised when remelting returns from customer plants. These castings may contain lead plugs or other undesirable materials. It is usually recommended that the scrap not exceed 30 to 40% of the amount of newly prepared alloy. Melt Temperature and Fluxing. High temperatures and flux can change the percentage

of the alloying elements. No flux is needed when the melting stock is clean ingot, but 1.4 to 2.3 kg (3 to 5 lb) of a chloride or fluoride flux is added for each 450 kg (1000 lb) of metal when the melting stock is partly or totally comprised of trimmings, gates, and rejected castings. A few pounds of flux per ton of alloy will reduce the magnesium content, and greater flux additions can make the magnesium disappear completely. Consequently, it is necessary to control temperatures continuously, to flux properly, and to check the analysis.

Melt Processing Furnaces. In the past, the standard furnace at the hot chamber zinc die casting machine was a gas-fired unit that held a cast iron pot. Quite often, the furnace at the machine was also used for melting. Large installations generally had a central melting facility, usually gas fired and accommodating a cast iron pot. Hot metal was pumped or siphoned from the furnace into the transport crucible or ladle. The transport ladle was suspended from an overhead conveyor and filled and emptied by mechanical tipping mechanisms. Only one worker was required. However, the working conditions were far from satisfactory. Manual handling of the metal was heavy, hot, smoky, and dangerous. Furnaces are frequently an integral part of the die casting machine. Most furnaces for melting and alloying, as well as for holding, are fuel-fired open-pot, immersion tube heated, or induction heated. Most pots for melting and containing molten zinc alloys are cast from gray or ductile cast iron. Ladles, if used, are of cast iron or pressed steel. Regardless of its type, the furnace should be equipped with controls so that the temperature of the molten zinc can be maintained within 6
C (10
F). The furnace capacity required depends on the size of the casting machine, workpiece size, and production rate. Generally, a holding furnace should be able to hold at least four times the amount of metal required for 1 h of operation. The capacity range of melting furnaces is usually 450 to 9000 kg (1000 to 20,000 lb), although immersion tube furnaces can range up to 18 Mg (40,000 lb). Total furnace capacity is usually five to seven times the amount of metal required per hour.

1050 / Casting of Nonferrous Alloys The introduction of gas-fired immersion tube furnaces has eliminated the need for pots. These furnaces use metal tubes immersed in the metal bath. The results are excellent; furnace efficiencies are increased considerably by this almostdirect method of heating. Immersion tubes are used for melting as well as holding furnaces and for heating molten-metal launder systems. Many installations began using complete systems with immersion tubes to heat the melter, holders, and interconnecting launders. The launder system consists of three main components: a central furnace, a number of metal feed furnaces (one for each die casting machine), and a launder system connecting the furnaces. The central melting furnace is arranged to feed the main launder continuously with molten zinc. Each casting station is equipped with a holding furnace that has a branch launder connecting it to the main launder. The furnaces are fully enclosed, with tightsealing lids and extremely thick walls. The castable main refractory does not contaminate zinc alloys and is backed up with heavy insulating material. The objective is to achieve such efficient sealing and insulation that very little heat is felt when the furnace is touched. The heat source is either immersed in the melt or located underneath the furnace lid (or a combination of both). The openings for charging and discharging are narrow and have heat locks to keep heat loss to a minimum. As long as the melting furnace metal is maintained at a recommended level, molten zinc flows by gravity from the ladling chamber, through an exit cast into the side of the furnace, and into the main launder. Because the channels are located well below the surface of the bath, surface dross from the charge end does not enter the ladling end. This type of system—immersion zinc alloying, remelting, laundering, and holding—can be used to gravity feed a number of die casting machines. In this type of system, cold-charged metal has little, if any, effect on the temperature of the metal in the holding furnace of the die cast machine. The laundering and holding system requires skimming only once a month, which reduces metal loss due to dross formation/removal. The minimal effort required to maintain the immersion system increases efficiency, while also improving the working conditions. In a launder, the metal is not directly exposed to the atmosphere. The trough is lined with heavy insulation having extremely low thermal conductivity. This insulation is also nonwettable by molten zinc. The launders are also well insulated and sealed, and the cross-sectional area of the metal stream in the launder is small (26 cm2, or 4 in.2). The covers are heavily insulated and are hinged to permit inspection of any portion of the launder. Metal conveyance throughout the launder system into the machine-feed furnace is free from turbulence. The metal is moved smoothly and continuously under a protective layer of stationary metal oxides. Most important, metal

temperature fluctuations are virtually eliminated. In addition, almost all manual work is eliminated. Scrap Return. Directly below the trim press is the scrap conveyor, which goes to the melting furnace. Also leaving the trim press is the finished parts conveyor, which directs the parts for further processing. The transfer conveyors are typically steel belts or oscillating conveyors. An innovation is the overlapping steel belt. Each section of this belt is hinged on only one side, allowing the belt surface to swing free at the discharge point. This motion sheds the material being conveyed, minimizes carryover, and prevents the accumulation of material in the hinge points, which can be a primary cause of wear and jamming in the conventional hinged steel belt. The overlapping belt is designed to use standard belt components and can be used in conventional hinged steel belt frames. Inclusions in Zinc Alloys. In the zinc foundry and die casting alloys containing aluminum, nonmetallic oxide inclusions are usually less important than intermetallic inclusions. Iron-zinc and iron-aluminum intermetallic compounds may also be present, especially above 480
C (900
F) in melting in iron-base vessels. These particles may be small crystals less than 5 mm (200 min.) in size or may agglomerate into millimeter-sized aggregates. Other heavy metals present, such as lead, cadmium, nickel, manganese, and chromium, can also lead to intermetallic compound precipitation (for example, Mn2Al5, Ni2Al3, and CrAl4). These compounds (often called aluminides) can form when the concentration of impurity exceeds the liquid solubility level of the element, which is often quite low (<0.02% for manganese, silicon, chromium, and nickel in die casting alloys). The intermetallic particles can be lighter than the base alloy and therefore rise to the top dross. Fluxing salt inclusions and exogenous refractory particles can also be present, as with the other nonferrous alloy systems. Oxide inclusions in zinc are somewhat limited because of the high positive vapor pressure of zinc. However, alloys containing aluminum can be subject to greater oxidation of the aluminum constituent. Oxide inclusions can also result during melting or casting when excessive turbulence is encountered; thus, mechanical stirring or molten-metal pumping should be controlled. Other nonmetallic inclusions arise from refractory vessel erosion and include silica, alumina, and silicon carbide. There is an appreciable density difference between these oxides and the base melt, unlike in aluminum; for this reason, gravity separation and thorough removal are substantially easier. Fluxing of Zinc Alloys. In the melting of clean, pure zinc and zinc alloys, there is little need for cover fluxes to protect the melt because zinc does not oxidize appreciably or absorb hydrogen at normal melting temperatures. However, chloride-containing fluxes that form fluid slag covers can be used to minimize

melt loss if they are carefully skimmed from the melt before pouring. When melting dirty metal, scrap returns, and the like, both a cover flux and a reactive flux are advantageously used to separate entrapped metal from oxides. The principal alloying elements present in zinc die casting alloys include aluminum, magnesium, and copper. Manganese and silicon may be present as impurities in the remelt; chromium and nickel are often seen when plated scrap is remelted. In most cases, zinc die casting alloys can tolerate these impurity amounts up to the limit of their solubility during solidification (generally <0.02%). The quantities of these elements present in excess of solubility limits will form oxides and/or complex intermetallic compounds, especially when iron is present (from melting or casting vessels). It is possible to collect some of these impurity constituents, particularly oxides, in the dross phase; therefore, a fluid slag or flux cover is beneficial. At typical zinc melt operating temperatures, various mixtures of zinc chloride (ZnCl2), KCl, and NaCl form the basis of such fluxes. These fluxes will also agglomerate nonmetallic residues and contaminants from dirty scrap and will help cleanse charge materials during remelting. The zinc alloy drosses that form during melting, holding, turbulent transfer, and remelt operations can contain as much as 90% entrapped finely divided free metallic zinc, in addition to oxides, intermetallic compounds, and other debris. An exothermic drossing flux, usually containing nitrate salts and silicofluoride double salts, can be used to recover much of this entrapped zinc. The exothermic reaction, assisted by rabbling or raking the flux into intimate contact with the dross phase, serves to raise the local temperature, increase fluidity, decrease the surface tension of the oxide skin, and chemically reduce ZnO. Proper fluxing procedures will permit more than half of the entrapped zinc to be recovered, yielding only approximately 35 to 40% unrecoverable and oxidized fines. The amount of flux to be added is often determined empirically and is usually 1 to 4.5 kg (2 to 10 lb) per ton of metal being treated, but this may vary considerably. Suitable melt temperatures are usually above 480
C (900
F). As the flux is worked into the dross, the rich metallic appearance will give way to a drier, duller, and more powdery residue ready for final skimming. The dross reclamation process is often carried out directly on zinc remelt pot furnaces. Alternatively, a separate dross mixer consisting of a vessel and a rotating paddle is used in which hot drosses and flux are added, mixing begins, and the fluid metal being recovered runs out the bottom and is collected in an ingot mold beneath the dross mixer. Galvanizing Fluxes. Galvanizing is the practice of applying a protective zinc or zinc alloy coating on iron or steel surfaces. The zinc oxidizes or is attacked preferentially and thus protects the underlying metal from corrosion.

Casting of Zinc Alloys / 1051 In hot dip galvanizing, the articles to be coated are immersed in molten zinc in a galvanizing tank or kettle at a temperature ranging from 450 to 460
C (840 to 860
F). A molten salt flux is usually maintained on the surface of the kettle, serving a number of important functions. The flux serves as a preheating and drying medium to reduce sputtering and explosions, and it keeps the galvanizing zinc metal itself free from oxides. The flux also precleans the article being galvanized. The principal ingredients of galvanizing fluxes include ZnCl, ammonium chloride (NH4Cl), zinc ammonium chloride (ZnCl2 NH3), and occasionally other chloride salts used as extenders. Typical reactions that take place in the flux during the galvanizing process include:





2NH4 Cl þ Zn ! H2 þ NH3 þ ZnCl2 NH3

(Eq 1)

2NH4 Cl þ FeO ! H2 O þ 2NH3 þ FeCl2

(Eq 2)

FeCl2 þ Zn ! Fe þ ZnCl2

(Eq 3)



2NH4 Cl þ ZnO ! H2 O þ NH3 þ ZnCl2 NH3 (Eq 4)

Equation 1 explains the occurrence of zinc consumption, hydrogen fires in the flux, and ammonia odors. Equations 2 and 3 show how the flux dissolves iron oxide and how elemental iron is freed both to alloy with zinc (desirable) and to form dross (undesirable). When fresh flux containing the active ingredient NH4Cl is added to partially spent flux, the high-melting viscous basic zinc chlorides (shown as ZnO) are converted to the lower-melting zinc ammonium chlorides. Figure 1 shows the phase

diagram of the ZnCl2/NH4Cl system. At a normal operating temperature of 455
C (850
F), NH4Cl would vaporize. However, the liquidus boundary indicates that as the mixture becomes richer in ZnCl2, the mixture of chloride can contain 3 to 4% NH4Cl and remain a liquid rather than become a gas. Foaming agents (glycerin) are often added to slow the loss of NH4Cl due to vaporization. The working galvanizing flux is therefore metastable. Because NH4Cl is lost through vaporization and is used up in reactions, fresh flux containing NH4Cl must be added continuously to keep the flux active. The viscosity of the flux will also increase because of the loading or buildup of high-melting zinc and iron oxides and debris carried in with the steel articles. The addition of fresh flux reduces viscosity by converting these compounds to ammoniated zinc chloride. Aluminum is often added (usually <0.1%) to galvanizing baths to serve as a brightener for the zinc coating on the workpiece. Aluminum oxidizes readily and will also form AlCl3. The viscosity of the flux is increased with the presence of Al2O3, and AlCl3 formation depletes the flux of active NH4Cl.

Die Casting of Zinc Alloys Die casting of zinc alloys is usually the hot chamber type, in which the pressure chamber, or gooseneck, is submerged in the molten metal. Selection of materials for the component parts of a die casting machine is less of a problem in casting zinc alloys than in casting aluminum or copper alloys. As noted, zinc die alloys are cast at relatively low temperatures. Molten

zinc alloys also do not rapidly attack ferrous metals. Pots, goosenecks, and other components of the machine that come in contact with the molten metal are usually made of gray or ductile cast iron, although they can be made of cast steel. Sleeves and nozzle seats, because they receive high wear, have been made from nitrided alloy steel, hot work tool steel (such as H13), high- speed tool steel, and stainless steel. Frequently, the sleeve and nozzle seat of the gooseneck are replaceable to permit inexpensive repair. The injection cylinder can be either hydraulic or pneumatic. With the same stroke length, the amount of metal injected can be changed by increasing or decreasing the size of the cylinder bore. Injection pressures used for the die casting of zinc alloys usually range from 10.3 to 20.6 MPa (1500 to 3000 psi). The lower pressures are used for simpler castings; the higher pressures, for more complex ones. Good practice is to use the lowest pressure that will produce acceptable castings; however, a minimum pressure of 10.3 MPa (1500 psi) is essential for obtaining an acceptable combination of soundness, surface finish, and mechanical properties. Ideally, die casting installations that are to be automated would include the following features:  Die casting machine designed for automa-

  



Fig. 1

Binary phase diagram for the system ZnCl2/NH4CL

tion and equipped with closed-loop control of machine variables, reliable automatic central lubrication and hydraulic system, repeatable and recordable machine functions, interlocked cycle stop, and shutdown circuitry triggered by out-of-tolerance machine variables Constant-temperature molten-metal supply with interlocked shutdown circuitry monitoring temperature tolerances Optimum-designed and precision-built dies with closed-loop cooling and heating control Flexible, programmable multifunction die lubricator system to provide die release, surface cooling, and efficient lubrication of the moving die parts Integrated auxiliary equipment that matches the die casting machine in performance and physical properties and is designed to endure the die casting environment

Early machines were designed and built for operator-controlled die casting and are not suitable for automation. Automation is described further in the article “Automation in High-Pressure Die Casting” in this Volume. Dies. Zinc alloys can be cast in single-cavity, multiple-cavity, combination, or unit dies. Dies that are designed to produce zinc alloy castings can seldom be used to produce castings of aluminum alloys or other metals that are cast at higher temperatures, because zinc alloys can be cast to thinner sections, smaller radii, and closer tolerances than aluminum, magnesium, or copper alloys. However, a die designed for

1052 / Casting of Nonferrous Alloys casting the higher-melting alloys can be used for casting zinc alloys. The cover half of a die must be equipped with a nozzle seat that will provide a good seal with the gooseneck of the machine. A sprue hole, or bushing, and spreader must be incorporated into the die to ensure feeding of the molten metal to the runners and to permit removal of the solidified sprue with the casting. Die Materials. In the die casting of zinc alloys, die temperature is relatively low, usually ranging from 160 to 245
C (325 to 475
F). Therefore hot work tool steels are not generally required for dies. However, for extremely long runs and for high dimensional accuracy, hot worked tool steels such as H13 will provide optimum die life. Die hardness in the casting of zinc alloys is less critical than for alloys of higher casting temperature. Steels prehardened by the manufacturer to any maximum hardness consistent with reasonable machinability can generally be used. The typical hardness is 29 to 34 HRC (280 to 320 HB). Slides and Cores. Hardenable stainless steels such as type 440B are often used for cores. Hot work tool steels such as H13 can be used for both cores and slides. Because these tool steels respond readily to nitriding, they can be selectively hardened; a component made from one of them can be treated for different properties in different areas. For example, the main portion of cores and slides can be nitrided for wear resistance, and the end portions can be masked to resist nitriding for better resistance to heat checking and spalling. Lubricating the slides and cores with molybdenum disulfide or colloidal graphite in oil helps to ensure smooth action and to minimize wear. Clearance between the slides and the guides should be kept to a minimum to prevent the molten zinc from wedging between them. Ejector pins of nitrided H11 tool steel or 7140 alloy steel are available as stock items for insertion in the dies. Die Life. The service life of a die casting die is directly related to the temperature of the metal being cast, thermal gradients within the die, and frequency of exposure to high temperature. Because of the relatively low temperatures of the dies (see the section “Die Temperature” in this article) and of the molten metal, die life for casting zinc alloys is generally much longer than for casting aluminum, magnesium, or copper alloys; it is not unusual for dies to last for 1 million shots. Maximum die life depends largely on having welldesigned dies, trained operators, and a rigidly enforced program of machine and die maintenance. A deficiency in any of these areas will result in decreased die life. Die Temperature. The temperature at which a die will run (level out) during continuous operation depends on the weight of the shot, the surface area of the shot, the cycle speed, and the shape of the die. When dies are too cold, cold shuts, laminations, internal porosity,

incomplete filling, and poor finish with excessive flow marks are likely to result. Dies that are too hot cause shrinkage, heat sinks, excessive flash, spitting, poor ejection, soldering, and die erosion. For a new application, some experimentation is usually required to establish a satisfactory optimum die temperature. The optimum die temperature for zinc is usually between 160 and 245
C (325 and 475
F). The lower end of this range is used for thicksection castings, and the higher end is for thin-section castings. When hardware finish is required, higher die temperatures (near 245
C, or 475
F) are generally required, regardless of the casting thickness. Once established, die temperature should be maintained within 6
C (10
F). Some casting shapes require localized heating or cooling of the die above or below the established temperature. Metal overflows are often used to heat die areas surrounding the perimeters of castings that have thin sections far from the main runner. This method of local heating helps to fill thin sections and to improve casting finish. Conversely, water channels are frequently placed behind the runner area immediately adjacent to the sprue to provide localized cooling and to prevent soldering of the molten metal to the die. Control of Casting Temperature. Research has shown that the variation in die temperature is one of the most important parameters affecting production rates and casting quality. To eliminate defects caused by excessively low die temperature, the die caster should design the die with excess cooling capacity and then use a temperature control system to modulate the flow of coolant in the die to achieve the desired die temperature. This incorporates the use of thermocouples, solenoid valves, and controllers. If necessary, heat can be added to the die through the use of heating elements inserted in the die. A complete system can be purchased that will control both heating and cooling using oil as the heat-transfer medium. Some die casters have found the oil controller to be superior in overall performance and flexibility. Metal temperatures for casting the zinc alloys range from 400 to 440
C (755 to 825
F). Generally, the lower end of this range is used for castings with thick sections, and the higher end is for castings in which section thickness is near minimum. In practice, a temperature of 415
C (780
F) is used for a wide variety of casting sizes and shapes. For best results, including best casting finish, some experimentation is usually required to arrive at the optimum metal temperature for a given application. When the optimum temperature is established, it should be controlled within 6
C (10
F). Die Lubricants. Selection of the optimum lubricant (die release agent) for a specific application often requires some experimentation because of the various operating temperatures and die shapes. An optimum lubricant is one that carbonizes at the operating temperature.

A lubricant that carbonizes above the operating temperature will stain the casting, and one that carbonizes below the operating temperature will be used up on the first shot. The black oil and graphite lubricants have been replaced by water-based lubricants; this has reduced the fire hazard and smoky environments commonly found in die casting plants. Water-based lubricants are formulated to be an effective release agent and aid in cooling the dies. Die Lubrication System. Because of the long periods spent spraying the die and the variety of spray patterns the operators will use in a day’s production, a considerable amount of time can be saved by automating this operation. Installation of the spray unit also reduces the work load of the operator. A number of die spray reciprocators have appeared on the market in recent years. Automatic die spray has been the most cost-effective measure in the die casting process. It resulted in a production increase of up to 25% and has a significant reduction in rejects (primarily because of improved surface finish). The die spray serves to release the casting, to lubricate moving die parts, and to cool the die surface. Selection of a spray system should involve consideration of the following requirements:  Moving die parts should be lubricated by an



  

appropriate central lubricating system built into the die and coupled to an external lubricator. Die cooling should be accurately calculated and achieved by internal water channels. In marginal cases, additional cooling can be accomplished by the die spray. In such cases, the dilution is extended to avoid excessive use of the release agent and resultant buildup. The die sprayer must be easily programmable to perform its function with reasonable repetitive accuracy. Essential features of the sprayer include the ability to spray two different media and air blast. Lateral movement of the spray nozzles may be essential for reaching deep die cavities.

Considerations of only the basic requirements when selecting a spray system will result in dissatisfaction. Knowledge of die size, special spray patterns, and so on is also important. Once the system is operational, adjustments should require a minimum of time. The spray system should be selected to allow for ease in movement when installing dies and/ or making machine repairs. Each system has advantages and limitations. Most use air for spraying and blow-off. Some systems use air for movement, while others use hydraulic power either from a separate hydraulic power supply or from the hydraulic system of the die casting machine. Because of the air pressure fluctuation in most plants (anywhere from 550 to 825 kPa, or 80 to 120 psi), a hydraulic system gives more constant movement when

Casting of Zinc Alloys / 1053 spraying, and this in turn provides more control. Other areas that must be investigated before purchasing a spray system include the adequacy of the plant air supply, the need for a tank for die release, and the size of the area the spray pattern should cover (which depends on the die size to be run in the system). Casting Removal. Removal of the casting from the die cast machine is done with an unloader, extractor, robot, grabber, or drop system. Unloaders vary widely in function. Some units will unload the machine, spray the die, and trim the castings. Others will only unload. Some systems are programmable, and others require the adjustment of limit switches and timers. A drop-through system is the simplest means of unloading a die casting machine. However, there are several problems with the system. When a drop-through system is decided upon, a method of removing the dropped casting must be considered. Normally, a pit is dug under the machine, and a conveyor is installed to remove the castings. The pit is filled with water to be used as a quench. Quenching ensures the solidification of the casting when it reaches the conveyor, thus preventing bending. If the water is too shallow, the casting will hit the conveyor and will be bent. Sensing a falling part in one way is quite simple, using a limit switch, photo detector, infrared detector, or a radio wave device. The problem is to determine whether or not the complete shot has left the die. There are several methods of determining if the entire die is clear for the next cycle. One is to weigh each shot as it leaves the die. This method appears simple, but it must be accomplished under water or after the casting leaves the water. Another method is to use a radio wave device to scan both halves of the die in conjunction with a heat detector for the sprue bushing to determine whether the sprue has been removed. The equipment mentioned is not easily set up, and it requires frequent servicing. The only positive way to determine whether or not the die is clear for another cycle is to have an operator standing by, watching each cycle and clearing anything that sticks to the die. Grabbers. A slightly more complex method of casting removal is the grabber. This equipment, although not inexpensive, is relatively low in cost compared to other good casting removal units. Its ability to handle a wide variety of tasks, however, is limited by its lack of mobility. This type of equipment performs well, but the purchaser must take the limitations into consideration. One advantage of some grabbers is their ability to spray and sense, which reduces the total capital investment. The grabbing mechanism on most units can be modified to handle a variety of dies; this is sufficient for the average die caster. The grabber design is compatible with other die casting industry equipment; controls, both electrical and hydraulic, are standard, using basic design concepts to minimize complexity. These units can be

purchased with their own hydraulic power sources, or they can be connected to the hydraulic system of the die casting machine. Whatever system is used, the unit has been designed with ease of maintenance in mind. The unloader is not attached to the die casting machine but supports itself. This type of equipment is sometimes more sophisticated than the previously mentioned types and is generally self-sufficient, relying only on input and/ or output signals to control its operation. Some unloaders can be connected to the hydraulic system of the die casting machine, reducing the overall cost. Various accessories are available with unloaders. Some unloaders can be purchased with a trim press or spray system and sensing unit. Others are strictly unloaders that can be reprogrammed by the changing of limit switches and rotating cams. The unloader differs from the grabber in its versatility and its programmability. Each system has been developed with the die casting machine in mind. Except for one or two specific items, these machines are designed using the standard relay logic and hydraulic systems of the die casting machine industry. For the most part, the unloader is a more sophisticated and versatile grabber. Robots, or programmable unloaders, are versatile. The robot can be moved through a series of operations and can store each operation in memory. It is the most expensive of all unloaders and the most sophisticated. Robots can usually be serviced by trained maintenance personnel because most of the work needed to correct a malfunction consists of removal and replacement. However, with all of its versatility, the robot can handle only 70 to 80% of the average dies. With additional equipment, the robot can handle approximately 90% of the average dies. The most important aspects of the robot are the reduced time it takes to program and the speed at which it operates. Some robots can be programmed to operate several pieces of equipment at the same time. The robot can also be programmed to spray the die with a very precise pattern. The choice between a robot and an extractor is subject to individual consideration. Generally, a single-purpose extractor is more reliable, more cost effective, and simpler to program and operate than a robot. Although the repetitive accuracy of robots and extractors is well known, placing any casting accurately in a trim press is not an easy task. As with any type of casting removal, there must be a method of detecting the casting once it has been removed from the die. This detecting can be accomplished with limit switches, photocells, infrared probes, or tactile sensors. Infrared sensors can sometimes be triggered by the surrounding environment instead of the casting and therefore may not be completely reliable. Tactile sensors consist of multiple probes or antennas connected to a low-voltage detection

system. When the probes make contact with the casting, the detection circuit is complete, and the next step of the casting cycle is initiated. Detection systems having as many as 12 probes are common; therefore, multiple-cavity gates and different portions of the same casting can be sensed. For example, deep bosses or similar features that tend to stick in the die can be individually monitored. The aforementioned sensing systems (infrared or tactile) are possible only when a robot is used, because it is necessary to bring the complete shot accurately to the probes. Newer sensing systems have memory capability in that all the multiple probes do not need to contact the casting simultaneously. This feature allows a “fly through” of the shot to permit quicker sensing with no loss in reliability. In the fastrunning zinc machines, such a capability is economically significant. Any of these detection devices must be mounted in some predetermined position near the casting removal location. Trimming. In finish trimming (not mere breaking of the casting from the gate), the part must almost always be quenched in water-soluble coolant/lubricant similar to that used in machining zinc castings. Failure to include soluble oil usually causes the trim die plates to solder, with consequent tearing of the trimmed part. Quenching several hundred pounds of shots per hour in a small tank requires either a heat exchanger for the quenching fluid or connection to a larger remote reservoir so that the heat can be dissipated naturally. Another method would be to use a spray and to recycle from the tank. If the robot or extractor dips the shot in a quench tank and then places it in a trim press, a drag-out conveyor can be fitted into the tank to remove the flash that settles to the bottom. The primary reason for mating a robot to the die casting machine is that the die casting process is a single-position oriented process; that is, the process begins with molten metal and ends with a part always in the same location and identically oriented with respect to fixed points in space. This sequence can be compared to the operation of a trimming press, which is a two-position process:  Random-oriented positions in containers

ahead of the press

 Fixed-oriented position on the location of

the trimming die Present-day robots, on the other hand, can only be effective in transferring parts that have fixed or identical orientation relative to some reference point. Therefore, although a robot can operate or service a die casting machine, it cannot operate a trim press in the same manner as a trim operator. However, if advantage is taken of the fact that the robot preserves position utility (part orientation) after it has unloaded a die casting machine, then a robot can be used as the connecting link between the tandem processes of

1054 / Casting of Nonferrous Alloys casting and trimming. To remove the trimmed part, gates, runners and overflows, conveyors, or chutes are required. An alternative to the cast-trim operation using a robot is the removal of the part from the casting machine to pipe- or screw-type conveyors that feature a buffer storage capacity, because the robot has the capability of placing a shot on such a conveyor, which is usually at least 2 to 2.5 m (6 to 8 ft) above the floor level. With a buffer storage conveyor, the entire system is not shut down for trim press problems. Automation of the process requires some method of trimming the casting. Three basic choices can be considered:  Die casting and in-die degating  Die casting machine and automatically

loaded trim press

 Die casting automatically with separate

trimming department Separate trimming was a preferred method in the past; this method suited the manual casting shop. The current range of sophisticated machinery makes this choice less attractive. The first and second choices listed previously result in decidedly higher productivity. Because of the limited production use of in-die degating, the second choice is preferred at this time. The production line concept also favors the integrated or interconnected trimming, where the orientation of the cast component is retained; thus, the casting can be easily transferred to a following machine for further machining operation. The frequently heard objection regarding the breakdown of one unit stopping the entire line is less valid with the currently offered machinery than it was in the past. Solid-state technology, improved limit switches, and state-of-the-art hydraulics possess proven track records. The statistics of the early 1970s are no longer valid. One can state that the single largest cause of stoppages is die failures. Hydraulically operated presses are used for cast-trim operations, and various designs (vertical, inclined, and horizontal) have been used to help overcome the problem of part, flash, overflow, and gate removal from the trimming location. The robot can accurately place the shot of castings on a location and can remove the part, gate, and flash off the location, but such a procedure almost always causes a delay in the casting cycle, forcing some economic trade-offs. The use of a shuttle press can overcome this problem. Another advantage of the robot in a cast-trim operation is in meeting the requirements of the Occupational Safety and Health Administration Power Press Standard concerning no hands at the point of operation by elimination of the press operator. Conveyors. The last item required for any installation is some method of removing castings from the trim press. A conveyor may not seem very important, but it can be the weak link in a trouble-free system. The conveyor must give workers time to perform their jobs

and must supply adequate storage capacity. The most widely used conveyor is the belt type, which can withstand the most adverse conditions. However, pipe- or screw-type conveyors, which feature a buffer storage capacity, should not be ruled out. With a buffer storage conveyor, it is not necessary to have personnel always stationed at the end of the conveyor. This feature not only allows ease of administration but also permits secondary operations linked to the conveyor to be done at an incentive pace. Other conveyors would probably be flat-top or roller conveyors, depending on the specifics of the particular job.

Other Casting Processes for Zinc Alloys Sand Casting. The ZA alloys, especially ZA12 and ZA-27, are being increasingly used in gravity sand casting operations. The wide freezing range of the ZA-27 alloy (109
C, or 230
F) means that control of solidification is especially important for this alloy. The use of chills or patterns that promote directional solidification is recommended. Permanent mold casting is done using both metallic and machined graphite molds. Cast

Fig. 2

The Gircast process

iron or steel is most commonly used for metallic permanent molds. The use of graphite molds permits as-cast tolerances similar to those obtained in die casting. Machining time is reduced or eliminated, making the graphite process attractive for intermediate production volumes (500 to 20,000 parts per year). Plaster Mold Casting. Zinc alloys are frequently cast in plaster molds, most often for prototype castings. The die casting alloys AG40A and AC41A are often used, but a proprietary alloy whose coefficient of thermal expansion is very close to that of aluminum alloys is frequently cast. Master patterns appropriate for this zinc alloy or for aluminum can be made according to a single shrinkage rule. Nominal compositions of the three zinc alloys mentioned are: Composition, % Alloy

AG40A AC41A Proprietary

Al

Cu

Mg

Zn

4.10 4.10 12.00

0.10 max 1.00 0.87

0.035 0.055 0.02

bal bal bal

Except for allowances for shrinkage, the technique for making the molds is the same as that for aluminum alloys.

Casting of Zinc Alloys / 1055 Squeeze casting is a process in which the liquid metal solidifies under pressure in closed dies held together by a hydraulic press. Essentially, the metal is forged to near-net or net shape while it solidifies. Metal-matrix composites (MMCs) are manufactured by squeeze casting by infiltrating a porous ceramic preform with the liquid metal under pressure. This process has been employed to cast MMCs with ZA alloy matrices and silicon carbide or alumina chopped-fiber reinforcements. Semisolid Casting. Zinc casting alloys are also being processed by semisolid casting. In this process, alloys are poured with negative superheat (that is, the pouring temperature is between the liquidus and the solidus). Vigorous mechanical agitation of the cooling metal melt prevents the formation of normal dendrites

and maintains the solid fraction of the melt in the form of rounded, primary particles. One semisolid metal processing method that has been applied to ZA alloys is the Gircast process. As shown in Fig. 2, this stir casting process involves three major steps:

(2) the paddles are raised, and (3) the induction heating means are retracted downward to release the crucible. The thermocouple is retracted upward. As soon as operations 1 to 3 have been executed, the centrifuged casting motor is started.

 The alloy temperature is elevated to TC1,

which is higher than the liquidus temperature TL (TC1 > TL).  The agitator is lowered, and the paddles are rotated to stir the metal at a temperature TC1. The crucible is then cooled to a temperature TC2 intermediate between the solidus and the liquidus temperature (TS < TC2 < TL).  The following operations are then performed simultaneously: (1) The agitator is stopped,

ACKNOWLEDGMENT Adopted from:  D.C.H. Nevison, Zinc and Zinc Alloys,

Casting, Vol 15, ASM Handbook, ASM International, 1988, p 786–797  D. Neff, Nonferrous Molten Metal Processes, Casting, Vol 15, Metals Handbook, 9th ed., ASM International, 1988, p 445

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1059-1084 DOI: 10.1361/asmhba0005331

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Aluminum and Aluminum Alloy Castings J. Fred Major, Rio Tinto Alcan

ALUMINUM CASTINGS have played an integral role in the growth of the aluminum industry since its inception in the late 19th century. The first commercial aluminum products were castings, such as cooking utensils and decorative parts, which exploited the novelty and utility of the new metal. Those early applications rapidly expanded to address the requirements of a wide range of engineering specifications. Alloy development and characterization of physical and mechanical characteristics provided the basis for new product development through the decades that followed. Casting processes were developed to extend the capabilities of foundries in new commercial and technical applications. The technology of molten metal processing, solidification, and property development has been advanced to assist the foundryman with the means of economical and reliable production of parts that consistently meet specified requirements. Today, aluminum alloy castings are produced in hundreds of compositions by all commercial casting processes, including green sand, dry sand, composite mold, plaster mold, investment casting, permanent mold, counter gravity lowpressure casting, and pressure die casting. Casting processes are normally divided into two categories: expendable mold processes and those processes in which castings are produced repetitively in tooling of extended life. Examples are green and dry sand molding for the former, and die and permanent mold casting for the latter. Alloys can also be divided into two groups: those most suitable for casting by any gravity or low-pressure process and those used in pressure die casting. A finer distinction is made between alloys suitable for permanent mold application and those for other gravity processes. In general, the most alloy-versatile processes (that is, those in which the largest number of alloys can be used) are those in which mold collapse accompanies pouring and solidification. The least forgiving processes, which require special consideration in alloy selection, are more rigid or permanent mold processes. The least alloy-versatile casting process based on process requirements is conventional high-pressure die casting, in which more than the usual measures of castability apply. The process demands a high level of fluidity, hot

strength, hot tear resistance, and die soldering resistance. Material constraints that formerly limited the design engineer’s alloy choice once a casting process had been selected are increasingly being blurred by advances in foundry technique. In the same way, process selection is also less restricted today. For example, many alloys thought to be unusable in permanent molds because of casting characteristics are in production by that process. Similarly specialized high-pressure die casting processes are in use that are compatible with new and existing alloys, some of which would have been thought uncastable by that process.

Chemical Compositions Systems used to designate casting compositions are not internationally standardized. In the United States, comprehensive listings are maintained by general procurement specifications issued through government agencies (federal and military, for example) and by technical societies such as the American Society for Testing and Materials and the Society of Automotive Engineers. Alloy registrations by the Aluminum Association are in broadest use; its nomenclature is decimalized to define foundry alloy composition variations. Designations in the form xxx.1 and xxx.2 include the composition of specific alloys in remelt ingot form suitable for foundry use. Designations in the form xxx.0 in all cases define composition limits applicable to castings. Prefix letters used primarily to define differences in impurity limits denote further variations in specified compositions. Accordingly, one of the most common gravity-cast alloys, 356, is shown in variations A356, B356, and C356; each of these alloys has identical major alloy contents but has decreasing limits applicable to impurities, especially iron content. Aluminum Association composition limits for registered aluminum foundry alloys used to cast shapes are given in Table 1. Table 1 does not include wrought alloys that are cast into ingots or billets intended for fabrication by mechanical deformation.

Although the nomenclature and designations for various casting alloys are standardized in North America, many important alloys have been developed for engineered casting production worldwide. For the most part, each nation (and in many cases the individual firm) has developed its own alloy nomenclature. Excellent references are available that correlate, cross reference, or otherwise define significant compositions in international use (Ref 2–4). Because differences in process capabilities exist, not all alloys can be cast by all methods. Alloys are usually separated into families in which alloy characteristics are considered a function or process requirements. Table 1 lists common compositions. The Aluminum Association designation system attempts alloy family recognition by the scheme:
1xx.x: Controlled unalloyed compositions
2xx.x: Aluminum alloys containing copper

as the major alloying element


3xx.x: Aluminum-silicon alloys also contain-

ing magnesium and/or copper


4xx.x: Binary aluminum-silicon alloys
5xx.x: Aluminum alloys containing magne-

sium as the major alloying element


6xx.x: Currently unused
7xx.x: Aluminum alloys containing zinc as the

major alloying element, usually also containing additions of either copper, magnesium, chromium, manganese, or combinations of these elements
8xx.x: Aluminum alloys containing tin as the major alloying element
9xx.x: Currently unused A separate four-digit Aluminum Association designation system exists for wrought aluminum compositions.

Effects of Alloying and Impurity Elements Antimony. At concentration levels equal to or greater than 0.05%, antimony refines the eutectic aluminum-silicon phase to a lamellar form in hypoeutectic compositions. The effectiveness of antimony in altering the eutectic structure

1060 / Nonferrous Alloy Castings Table 1

Chemical composition limits for registered aluminum alloys in the form of xxx.0 castings, xxx.1 ingot and xxx.2 ingot

Only composition limits that are Identical to those listed herein or are registered with The Aluminum Association should be designated as “AA” alloys. Composition(b), wt% AA No.

201.0 A201.0 B201.0 203.0 204.0 206.0 B206.0 A206.0 240.0 242.0 A242.0 295.0 296.0 301.0 302.0 303.0 308.0 318.0 319.0 A319.0 B319.0 320.0 328.0(f)(u) 332.0 333.0 A333.0 336.0 339.0 354.0 355.0 A355.0 C355.0 356.0 A356.0 B356.0 C356.0 F356.0 357.0 A357.0 B357.0 C357.0 D357.0 E357.0 F357.0 358.0 359.0 A359.0 360.0(k) A360.0(k) 361.0 363.0 364.0 365.0 366.0 369.0 A380.0(k) B380.0 C380.0 D380.0 381.0 383.0 A383.0 384.0 A384.0 B384.0 C384.0 390.0 A390.0 B390.0 391.0

Products(a)

S S S S S, P S, P S, P S, P S S, P S S P ... ... ... S, P S, P S, P S, P S, P S, P S P P P P P P S, P S, P S, P S, P S, P S, P S, P S, P S, P S, P S, P S, P S S, P, I S, P, I S, P S, P ... D D D S, P D D D D D D D D D D D D D D D S, P D D

Si

0.10 0.05 0.05 0.30 0.20 0.10 0.05 0.05 0.50 0.7 0.6 0.7–1.5 2.0–3.0 9.5–10.5 9.5–10.5 9.5–10.5 5.0–6.0 5.5–6.5 5.5–6.5 5.5–6.5 5.5–6.5 5.0–8.0 7.5–8.5 8.5–10.5 8.0–10.0 8.0–10.0 11.0–13.0 11.0–13.0 8.6–9.4 4.5–5.5 4.5–5.5 4.5–5.5 6.5–7.5 6.5–7.5 6.5–7.5 6.5–7.5 6.5–7.5 6.5–7.5 6.5–7.5 6.5–7.5 6.5–7.5 6.5–7.5 6.5–7.5 6.5–7.5 7.6–8.6 8.5–9.5 8.5–9.5 9.0–10.0 9.0–10.0 9.5–10.5 4.5–6.0 7.5–9.5 9.5–11.5 6.5–7.5 11.0–12.0 7.5–9.5 7.5–9.5 7.5–9.5 7.5–9.5 9.0–10.0 9.5–11.5 9.5–11.5 10.5–12.0 10.5–12.0 10.5–12.0 10.5–12.0 16.0–18.0 16.0–18.0 16.0–18.0 18.0–20.0

Fe

0.15 0.10 0.05 0.50 0.35 0.15 0.10 0.10 0.50 1.0 0.8 1.0 1.2 0.8–1.5 0.25 0.8–1.5 1.0 1.0 1.0 1.0 1.2 1.2 1.0 1.2 1.0 1.0 1.2 1.2 0.20 0.6(g) 0.09 0.20 0.6(g) 0.20 0.09 0.07 0.20 0.15 0.20 0.09 0.09 0.20 0.10 0.10 0.30 0.20 0.25 2.0 1.3 1.1 1.1 1.5 0.15 0.15 1.3 1.3 1.3 1.3 1.3 1.3 1.3 1.3 1.3 1.3 1.3 1.3 1.3 0.50 1.3 1.2

Cu

4.0–5.2 4.0–5.0 4.5–5.0 4.5–5.5 4.2–5.0 4.2–5.0 4.2–5.0 4.2–5.0 7.0–9.0 3.5–4.5 3.7–4.5 4.0–5.0 4.0–5.0 3.0–3.5 2.8–3.2 0.20 4.0–5.0 3.0–4.0 3.0–4.0 3.0–4.0 3.0–4.0 2.0–4.0 1.0–2.0 2.0–4.0 3.0–4.0 3.0–4.0 0.50–1.5 1.5–3.0 1.6–2.0 1.0–1.5 1.0–1.5 1.0–1.5 0.25 0.20 0.05 0.05 0.20 0.05 0.20 0.05 0.05 ... ... 0.20 0.20 0.20 0.20 0.6 0.6 0.50 2.5–3.5 0.20 0.03 0.05 0.50 3.0–4.0 3.0–4.0 3.0–4.0 3.0–4.0 3.0–4.0 2.0–3.0 2.0–3.0 3.0–4.5 3.0–4.5 3.0–4.5 3.0–4.5 4.0–5.0 4.0–5.0 4.0–5.0 0.20

Mn

0.20–0.50 0.20–0.40 0.20–0.50 0.20–0.30 0.10 0.20–0.50 0.20–0.50 0.20–0.50 0.30–0.7 0.35 0.10 0.35 0.35 0.50–0.8 ... 0.50–0.8 0.50 0.50 0.50 0.50 0.8 0.8 0.20–0.6 0.50 0.50 0.50 0.35 0.50 0.10 0.50(g) 0.05 0.10 0.35(g) 0.10 0.05 0.05 0.10 0.03 0.10 0.05 0.05 0.10 0.10 0.10 0.20 0.10 0.10 0.35 0.35 0.25 (l) 0.10 0.50–0.8 0.03 0.35 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.10 0.10 0.50 0.30

Mg

0.15–0.55 0.15–0.35 0.25–0.35 0.10 0.15–0.35 0.15–0.35 0.15–0.35 0.15–0.35 5.5–6.5 1.2–1.8 1.2–1.7 0.03 0.05 0.25–0.50 0.7–1.2 0.45–0.7 0.10 0.10–0.6 0.10 0.10 0.10–0.50 0.05–0.6 0.20–0.6 0.50–1.5 0.05–0.50 0.05–0.50 0.7–1.3 0.50–1.5 0.40–0.6 0.40–0.6 0.45–0.6 0.40–0.6 0.20–0.45 0.25–0.45 0.25–0.45 0.25–0.45 0.17–0.25 0.45–0.6 0.40–0.7 0.40–0.6 0.45–0.7 0.55–0.6 0.55–0.6 0.40–0.7 0.40–0.6 0.50–0.7 0.40–0.6 0.40–0.6 0.40–0.6 0.40–0.6 0.15–0.40 0.20–0.40 0.10–0.60 0.50–1.2 0.25–0.45 0.10 0.10 0.10–0.30 0.10–0.30 0.13 0.10 0.10–0.30 0.10 0.10 0.10–0.30 0.10–0.30 0.45–0.65 0.45–0.85 0.45–0.65 0.40–0.7

Cr

... ... ... ... ... ... ... ... ... 0.25 0.15–0.25 ... ... ... ... ... ... ... ... ... ... ... 0.35 ... ... ... ... ... ... 0.25 ... ... ... ... ... ... ... ... ... ... ... ... ... ... 0.20 ... ... ... ... 0.20–0.30 (l) 0.25–0.50 ... ... 0.30–0.40 ... ... ... ... 0.15 ... ... ... ... ... ... ... ... ... ...

Others Ni

... ... ... 1.3–1.7 0.05 0.05 0.05 0.05 0.30–0.7 1.7–2.3 1.8–2.3 ... 0.35 1.0–1.5 1.0–1.5 ... ... 0.35 0.35 0.35 0.50 0.35 0.25 0.50 0.50 0.50 2.0–3.0 0.50–1.5 ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... 0.50 0.50 0.20–0.30 0.25 0.15 ... ... 0.05 0.50 0.50 0.50 0.50 0.50 0.30 0.30 0.50 0.50 0.50 0.50 ... ... 0.10 ...

Zn

... ... ... 0.10 0.10 0.10 0.10 0.10 0.10 0.35 0.10 0.35 0.50 0.05 0.05 0.05 1.0 1.0 1.0 3.0 1.0 3.0 1.5 1.0 1.0 3.0 0.35 1.0 0.10 0.35 0.05 0.10 0.35 0.10 0.05 0.05 0.10 0.05 0.10 0.05 0.05 ... ... 0.10 0.20 0.10 0.05 0.50 0.50 0.50 3.0–4.5 0.15 0.07 0.05 1.0 3.0 1.0 3.0 1.0 3.0 3.0 3.0 3.0 1.0 1.0 3.0 0.10 0.10 1.5 0.10

Ti

0.15–0.35 0.15–0.35 0.15–0.35 0.15–0.25 0.15–0.30 0.15–0.30 0.10 0.15–0.30 0.20 0.25 0.07–0.20 0.25 0.25 0.20 0.20 0.20 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.20 0.25 0.04–0.20 0.20 0.25 0.20 0.04–0.20 0.04–0.20 0.04–0.20 0.20 0.04–0.20 0.04–0.20 0.04–0.20 0.10–0.20 0.10–0.20 0.04–0.20 0.10–0.20 0.20 0.20 ... ... 0.20 0.20 ... 0.04–0.15 0.20 ... ... ... ... ... 0.20 ... ... ... ... ... ... 0.20 0.20 0.20 0.20

Sn

... ... ... ... 0.05 0.05 0.05 0.05 ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... 0.15 0.15 0.10 0.25 0.15 ... ... 0.10 0.35 0.35 0.35 0.35 0.15 0.15 0.15 0.35 0.35 0.35 0.35 ... ... ... ...

... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...

... ... ... ... ...

... ... ... ... ... ... ... ... ... ... ... ... ... ... ...

Each

Total

0.05(c) 0.03(c) 0.05(d) 0.05(e) 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 ... 0.03 0.03 0.03 ... ... ... ... ... ... ... ... ... ... 0.05 ... 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05(h) 0.05 0.05(h) 0.05(h) 0.05(i) 0.05(i) 0.05(j) 0.05 0.03 ... ... 0.05 (m) 0.05(n) 0.03(o) 0.05 0.05 ... ... ... ... (p) ... ... ... ... ... ... 0.10 0.10 0.10 0.10

0.10 0.10 0.15 0.20 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.35 0.10 0.10 0.10 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 ... 0.50 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.10 0.25 0.25 0.15 0.30 0.15 0.10 0.15 0.15 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.20 0.20 0.20 0.20

(continued) (a) D, die castion; P, permanent mold; S, sand; I, investment casting. Other products may pertain to the composition shown even though not listed. (b) Compositions are maximum unless shown as a range or a minimum. All compositions contain balance of aluminum. (c) 0.40–1.0 Ag. (d) 0.50–1.0 Ag. (e) 0.025 max Mn + Cr + Ti + V. (f) Used for semisolid formel products. (g) If iron exceeds 0.45, manganese content shall not be less than one-half iron content. (h) 0.04–0.07 Be. (i) 0.002 max Be. (j) 0.10–0.30 Be. (k) A360.1, A380.1, and A413.1 ingot is used to produce 360.0 and A360.0, 380.0 and A380.0, 413.0 and A413.0 castings, respectively. (l) 0.8 max Mn + Cr. (m) 0.25 max Pb. (n) 0.02–0.04 Be. (o) Casting: 0.001 max P, 0.005–0.015 Sr. Ingot: 0.001 max P, 0.010–0.020 Sr. (p) 0.15 max Sb, 0.15 max Pb. (q) 0.08–0.15 V. (r) 0.10 max Pb. (s) 0.003–0.007 Be, 0.005 max B. (t) Used for centrifugally cast products. (u) Reactivation date of alloy designation. Data extracted from Ref 1; see Ref 1 for more details.

Aluminum and Aluminum Alloy Castings / 1061 Table 1

(continued) Composition(b), wt%

AA No.

A391.0 B391.0 392.0 393.0 413.0(k) A413.0(k) B413.0 443.0 A443.0 B443.0 C443.0 444.0 A444.0 505.0(t) 511.0 512.0 513.0 514.0 515.0 516.0 518.0 520.0 535.0 A535.0 B535.0 705.0 707.0 709.0(t) 710.0 711.0 712.0 713.0 771.0 772.0 850.0 851.0 852.0 853.0

Products(a)

P S D S, D D S, S, S S, D S, P

P, D

P P P P

S S P S D D D S S S S S, P S, P S P S S, S S S, S, S, S,

P

P P P P

Si

18.0–20.0 18.0–20.0 18.0–20.0 21.0–23.0 11.0–13.0 11.0–13.0 11.0–13.0 4.5–6.0 4.5–6.0 4.5–6.0 4.5–6.0 6.5–7.5 6.5–7.5 0.40–0.8 0.30–0.7 1.4–2.2 0.30 0.35 0.50–1.0 0.30–1.5 0.35 0.25 0.15 0.20 0.15 0.20 0.20 0.40 0.15 0.30 0.30 0.25 0.15 0.15 0.7 2.0–3.0 0.40 5.5–6.5

Fe

0.6 0.20 1.5 1.3 2.0 1.3 0.50 0.8 0.8 0.8 2.0 0.6 0.20 0.7 0.50 0.6 0.40 0.50 1.3 0.35–1.0 1.8 0.30 0.15 0.20 0.15 0.8 0.8 0.50 0.50 0.7–1.4 0.50 1.1 0.15 0.15 0.7 0.7 0.7 0.7

Cu

Mn

0.20 0.20 0.40–0.6 0.7–1.1 1.0 1.0 0.10 0.6 0.30 0.16 0.6 0.25 0.10 0.15–0.40 0.15 0.35 0.10 0.15 0.20 0.30 0.25 0.25 0.05 0.10 0.10 0.20 0.20 1.2–20 0.35–0.6 0.35–0.6 0.25 0.40–1.0 0.10 0.10 0.7–1.3 0.7–1.3 1.7–2.3 3.0–4.0

0.30(g) 0.30 0.20–0.6 0.10 0.35 0.35 0.35 0.50 0.50 0.35 0.35 0.35 0.10 0.15 0.35 0.8 0.30 0.35 0.40–0.6 0.15–0.40 0.35 0.15 0.10–0.25 0.10–0.25 0.05 0.40–0.6 0.40–0.6 0.30 0.05 0.05 0.10 0.6 0.10 0.10 0.10 0.10 0.10 0.60

Mg

0.40–0.7 0.40–0.7 0.8–1.2 0.7–1.3 0.10 0.10 0.05 0.05 0.05 0.05 0.10 0.10 0.05 0.8–1.2 3.5–4.5 3.5–4.5 3.5–4.5 3.5–4.5 2.5–4.0 2.5–4.5 7.5–8.5 9.5–10.6 6.2–7.5 6.5–7.5 6.5–7.5 1.4–1.8 1.8–2.4 2.1–2.9 0.6–0.8 0.25–0.45 0.50–0.65 0.20–0.50 0.8–1.0 0.6–0.8 0.10 0.10 0.6–0.9 ...

Cr

... ... ... ... ... ... ... 0.25 0.25 ... ... ... ... 0.04–0.35 ... 0.25 ... ... ... ... ... ... ... ... ... 0.20–0.40 0.20–0.40 0.18–0.28 ... ... 0.40–0.6 0.35 0.06–0.20 0.06–0.20 ... ... ... ...

Others Ni

... ... 0.50 2.0–2.5 0.60 0.50 0.05 ... ... ... 0.50 ... ... ... ... ... ... ... ... 0.25–0.40 0.15 ... ... ... ... ... ... ... ... ... ... 0.15 ... ... 0.7–1.3 0.30–0.7 0.9–1.5 ...

Zn

0.10 0.10 0.50 0.10 0.50 0.50 0.10 0.50 0.50 0.35 0.50 0.35 0.10 0.25 0.15 0.35 1.4–2.2 0.15 0.10 0.20 0.15 0.15 ... ... ... 2.7–3.3 4.0–4.5 5.1–6.1 6.0–7.0 6.0–7.0 5.0–8.5 7.0–8.0 6.5–7.5 6.0–7.0 ... ... ... ...

Ti

0.20 0.20 0.20 0.10–0.20 ... ... 0.25 0.25 0.25 0.25 ... 0.25 0.20 0.15 0.25 0.25 0.20 0.25 ... 0.10–0.20 ... 0.25 0.10–0.25 0.25 0.10–0.25 0.25 0.25 0.20 0.25 0.20 0.15–0.25 0.25 0.10–0.20 0.10–0.20 0.20 0.20 0.20 0.20

Sn

... ... 0.30 ... 0.15 0.15 ... ... ... ... 0.15 ... ... ... ... ... ... ... ... 0.10 0.15 ... ... ... ... ... ... ... ... ... ... ... ... ... 5.5–7.0 5.5–7.0 5.5–7.0 5.5–7.0

Each

... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...

0.10 0.10 0.15 0.05(q) ... ... 0.05 ... ... 0.05 ... 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05(r) ... 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.10 0.05 0.05 ... ... ... ...

Total

0.20 0.20 0.50 0.15 0.25 0.25 0.20 0.35 0.35 0.15 0.25 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 ... 0.25 0.15 0.15(s) 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.20 0.25 0.15 0.15 0.30 0.30 0.30 0.30

(a) D, die castion; P, permanent mold; S, sand; I, investment casting. Other products may pertain to the composition shown even though not listed. (b) Compositions are maximum unless shown as a range or a minimum. All compositions contain balance of aluminum. (g) If iron exceeds 0.45, manganese content shall not be less than one-half iron content. (k) A360.1, A380.1, and A413.1 ingot is used to produce 360.0 and A360.0, 380.0 and A380.0, 413.0 and A413.0 castings, respectively. (p) 0.15 max Sb, 0.15 max Pb. (q) 0.08–0.15 V. (r) 0.10 max Pb. (s) 0.003–0.007 Be, 0.005 max B. (t) Used for centrifugally cast products. (u) Reactivation date of alloy designation. Data extracted from Ref 1; see Ref 1 for more details.

depends on an absence of phosphorus and on an adequately rapid rate of solidification. Antimony also reacts with either sodium or strontium to form coarse intermetallics with adverse effects on castability and eutectic structure. Antimony will chemically combine with and precipitate the common modifiers sodium and strontium; therefore, antimony has the potential to act as a modification poisoner if it enters the scrap recycling stream. Chemical analysis of antimony content is very difficult, even with modern optical emission spectrometers, because of the overlap between its spectrum and that of iron. Antimony is classified as a heavy metal with potential toxicity and hygiene implications, especially as associated with the possibility of stibine gas formation and the effects of human exposure to other antimony compounds. Beryllium additions of as low as a few parts per million may be effective in reducing oxidation losses and associated inclusions in magnesium-containing compositions. Studies have shown that proportionally increased beryllium concentrations are required for oxidation suppression as magnesium content increases. At higher concentrations (>0.04%), beryllium affects the form and composition of iron-

containing intermetallics, markedly improving strength and ductility. In addition to beneficially changing the morphology of the insoluble phase, beryllium changes its composition, rejecting magnesium from the Al-Fe-Si complex and thus permitting its full use for hardening purposes. Beryllium-containing compounds are, however, numbered among the known carcinogens that require specific precautions in melting, molten metal handling, dross handling and disposition, and blasting, grinding, and welding. Standards define the maximum beryllium in welding rod and weld base metal as 0.008 and 0.010%, respectively. Both acute and chronic berylliosis are possible upon exposure to beryllium fumes. The chronic form can appear years after the exposure has ended. For these reasons, there has been a movement to eliminate this element from aluminum alloys in general. Bismuth improves the machinability of cast aluminum alloys at concentrations greater than 0.1%. It is also known to increase shrinkage tendencies in aluminum-silicon alloys at concentrations greater than 30ppm. Boron combines with other metals to form borides present as solid particles in liquid aluminum, such as AlB2 and TiB2. The titanium

diboride particles then act as nucleation sites for the active grain-refining phase TiAl3. Once nucleated from the molten aluminum, this last phase serves as the nucleation site for new grains. Intermetallic borides reduce tool life in machining operations because they are coarse particles and form objectionable inclusions with detrimental effects on mechanical properties and ductility. At high boron concentrations, borides contribute to furnace sludging, particle agglomeration, and increased risk of casting inclusions. However, boron treatment of aluminum-containing peritectic elements is practiced to improve purity and electrical conductivity in rotor casting. Higher rotor alloy grades may specify boron to exceed titanium and vanadium contents to ensure either the complexing or precipitation of these elements for improved electrical performance (see the section “Rotor Castings” in this article). Cadmium in concentrations exceeding 0.1% improves machinability. Precautions that acknowledge volatilization at 767  C (1413  F) are essential for toxicity reasons. Calcium is a weak aluminum-silicon eutectic modifier. It increases hydrogen solubility and is often responsible for casting porosity at trace concentration levels. Calcium concentrations

1062 / Nonferrous Alloy Castings greater than approximately 0.005% also adversely affect ductility in aluminum-magnesium alloys. Chromium additions are commonly made in low concentrations to room-temperature aging and thermally unstable compositions in which germination and grain growth are known to occur. Chromium typically forms the compound CrAl7, which displays extremely limited solid-state solubility and is therefore useful in suppressing grain growth. Sludge that contains iron, manganese, and chromium is sometimes encountered in die-casting compositions, but it is rarely encountered in gravity-casting alloys. Chromium improves corrosion resistance in certain alloys and increases quench sensitivity at higher concentrations. Copper. The first and most widely used aluminum alloys were those containing 4 to 10% Cu. Copper substantially improves strength and hardness in the as-cast and heat treated conditions. Alloys containing 4 to 6% Cu respond most strongly to thermal treatment. Copper generally reduces resistance to general corrosion and, in specific compositions and material conditions, stresscorrosion susceptibility. Additions of copper also reduce hot tear resistance and decrease castability. Hydrogen is the only gaseous species with appreciable solubility in liquid aluminum. The solubility of hydrogen is considerably lower in solid aluminum than in the liquid so there is the propensity for hydrogen to cause porosity through precipitation from the liquid in the form of bubbles as the metal solidifies. Hydrogen bubbles, commonly called gas porosity, are avoided by degassing the metal with lowhydrogen inert or active gases using either a lance, porous media, or rotary impeller degasser in increasing order of cost and effectiveness. The most common source of hydrogen is moisture from the atmosphere over the molten metal, as aluminum will reduce water vapor to form surface layer of alumina (Al2O3) and dissolved hydrogen in the melt. For this reason, air used to pump low-pressure furnaces is normally dried to a low dew point. Inert gas covers may be used over the metal in critical applications. Indium is added, together with zinc, in the casting of sacrificial anodes. Indium helps to disrupt the normally tough, coherent aluminum oxide surface coating typical of aluminum alloys and hence enhances its ability to corrode sacrificially. Iron improves hot tear resistance and decreases the tendency for die sticking or soldering in die casting. Increases in iron content are however, accompanied by substantially decreased ductility. Iron reacts to form a myriad of insoluble phases in aluminum alloy melts, the most common of which are FeAl3, FeMnAl6, and aAlFeSi. In aluminum-silicon alloys, the two most common intermetallic precipitates are b-needles (FeSiAl5) and the Mn-corrected ascript ((Fe,Mn)3Si2Al15). These essentially insoluble phases are responsible for improvements in strength, especially at elevated temperature.

They do, however, reduce the ductility. As the fraction of insoluble phase increases with increased iron content, casting considerations such as flowability and feeding characteristics are adversely affected. Iron participates in the formation of sludging phases with manganese, chromium, and other elements, particularly at the concentrations and low holding temperatures normally associated with high-pressure die casting. In fact the sludging factor, a simple expression combining the concentrations of iron, manganese, and chromium, is given as: SF ¼ wt%Fe þ 2ðwt%MnÞ þ 3ðwt%CrÞ

(Eq 1)

Plots of this factor versus temperature delineate the minimum temperature below which sludging can commonly be expected in highpressure die casting. Lead is commonly used in aluminum casting alloys at greater than 0.1% for improved machinability. Magnesium is the basis for strength and hardness development in heat treated aluminum-silicon alloys and is commonly used in more complex aluminum-silicon alloys containing copper, nickel, and other elements for the same purpose. The Mg2Si-based precipitationhardening system displays a useful solubility limit corresponding to approximately 0.70% Mg, beyond which either no further strengthening occurs or matrix softening takes place. Common premium-strength compositions in the aluminum-silicon family employ magnesium in the range of 0.40 to 0.070% (see the section “Premium Engineered Castings” in this article). Binary aluminum-magnesium alloys are widely used in applications requiring a bright surface finish and corrosion resistance, as well as attractive combinations of strength and ductility. Common compositions range from 4 to 10% Mg, and compositions containing more than 7% Mg are heat treatable. Instability and room-temperature aging characteristics at higher magnesium concentrations encourage heat treatment. Manganese is used for two main purposes in shape casting. The first is to correct the iron phases that form in secondary foundry alloy compositions from b-needles to a-script, thereby improving ductility. The second use is in advanced die-casting processes where manganese can be substituted for iron to reduce die soldering. The higher solubility of manganese in aluminum combined with its reduced tendency to form brittle phases yields an alloy with better properties than those in which iron is used as the exclusive remedy for die soldering. Some evidence also exists that a high volume fraction of MnAl6 in alloys containing more than 0.5% Mn may beneficially influence internal casting soundness. Mercury. Compositions containing mercury were developed as sacrificial anode materials for cathodic protection systems, especially in marine environments. The use of these optimally electronegative alloys, which do not passivate in seawater, was severely restricted for environmental reasons.

Nickel is usually employed with copper to enhance elevated-temperature properties. It also reduces the coefficient of thermal expansion. Nickel will accelerate filliform corrosion in applications such as coated automotive wheels if present in high enough trace quantities. Values up to roughly 100 ppm seem to be acceptable, however. Phosphorus reacts with aluminum to form AlP, which nucleates and refines the primary silicon phase in hypereutectic aluminum-silicon alloys. At parts per million concentrations, phosphorus coarsens the eutectic structure in hypoeutectic aluminum-silicon alloys. Phosphorus diminishes the effectiveness of the common eutectic modifiers sodium and strontium as well as the eutectic refiner, antimony. Silicon. The outstanding effect of silicon in aluminum alloys is the improvement of casting characteristics. Addition of silicon to pure aluminum dramatically improves fluidity, hot tear resistance, and feeding characteristics. The most prominently used compositions in all casting processes are those of the aluminum-silicon family. Commercial alloys span the hypoeutectic and hypereutectic ranges up to about 25% Si. In general, an optimum range of silicon content can be assigned to casting processes. For slow cooling rate processes (such as plaster, investment, and sand casting), the range is 5 to 7%, for permanent mold casting 7 to 9%, and for die casting 8 to 12%. The bases for these recommendations are the relationship between cooling rate and fluidity and the effect of percentage of eutectic on feeding. Silicon additions are also accompanied by a reduction in specific gravity and coefficient of thermal expansion. Silver is used in only a limited range of aluminum-copper premium-strength alloys at concentrations of 0.5 to 1.0%. Silver contributes to precipitation hardening and stress-corrosion resistance. Sodium modifies the aluminum-silicon eutectic. Its presence is embrittling in aluminum-magnesium alloys. Sodium reacts with phosphorus, which reduces its own effectiveness in modifying the eutectic in hypoeutectic alloys as well as reducing the refining effect of phosphorus in hypereutectic alloys. The effects of sodium and phosphorous are mutually antagonistic. Strontium is used to modify the aluminumsilicon eutectic. Effective modification can be achieved at very low addition levels, but a range of recovered strontium of 0.0040 to 0.02% is commonly used. Higher-addition levels frequently are associated with casting porosity, especially in processes or in thick-section parts in which solidification occurs more slowly. This increase in porosity can be used to compensate for shrinkage in castings where increased distributed porosity is tolerable. Degassing efficiency as determined by vacuum solidification tests may also be adversely affected at higher strontium levels. The effects of strontium and phosphorous are mutually

Aluminum and Aluminum Alloy Castings / 1063 antagonistic in the same manner as discussed in the section on sodium. Tin is effective in improving antifriction characteristics and is therefore useful in bearing applications. Casting alloys may contain up to 25% Sn. Additions can also be made to improve machinability. Tin may influence precipitation-hardening response in some alloy systems at levels in the hundreds of ppm. Titanium is extensively used to refine the grain structure of aluminum casting alloys, often in combination with smaller amounts of boron. Titanium in excess of the TiB2 stoichiometry is necessary for effective grain refinement of wrought alloys. Grain refinement of foundry alloys is effectively accomplished with ratios of Ti to B of 5 to 1. Titanium is often employed at concentrations greater than those required for grain refinement to reduce cracking tendencies in hot short compositions. Zinc. No significant benefits are obtained by the addition of zinc alone to aluminum. Accompanied by the addition of copper and/or magnesium, however, zinc results in attractive heat treatable or naturally aging compositions. A number of such compositions are in common use. Zinc is also commonly found in secondary-gravity and die-casting compositions.

Structure Control A number of factors define the metallurgical microstructure in aluminum castings. Of primary importance are dendrite cell size or dendrite arm spacing, the form and distribution of microstructural phases, and grain size. The foundryman can control the fineness of dendrite structure by controlling the solidification rate. Microstructural characteristics such as the size and distribution of primary and intermetallic phases are considerably more complex. However, chemistry control (particularly control of impurity element concentrations), control of element ratios based on the stoichiometry of intermetallic phases, and control of solidification conditions to ensure uniform size and distribution of intermetallics are all means to the end of structural control. Effective grain refinement shares and strongly influences the same objectives. The use of modifiers and refiners to influence eutectic and hypereutectic structures in aluminum-silicon alloys is also an example of the manner in which microstructures and macrostructures can be optimized through alloy chemistry in combination with control of solidification conditions in foundry operations.

Secondary Dendrite Arm Spacing In all conventional processes, solidification takes place through the formation of dendrites in the liquid solution. The cells contained within the dendrite structure correspond to the dimensions separating the secondary arms of primary dendrites and are controlled for a given composition exclusively by solidification rate.

In fact, the secondary dendrite arm spacing (DAS) is a simple inverse cube root function of the solidification time. Figure 1 displays the microstructures typical of conventional casting processes as well as those encountered in semisolid casting. The solidification rate increases, and hence the DAS decreases, in the order shown for the processes of sand, low-pressure permanent mold, squeeze casting, and die casting. Through microstructural examination, it is possible to define the rate at which a given region of a casting has solidified by reference to data obtained from unidirectionally solidified samples spanning solidification rates represented by the full range of various casting processes. Figure 2 illustrates the improvement in mechanical properties achievable by the change in dendrite formation controlled by solidification rate. In premium engineered castings and in many other casting applications, careful attention is given to obtaining solidification rates corresponding to optimum mechanical property development. It should be kept in mind that solidification rate affects more than just the dendrite arm spacing. The scale of the aluminum-silicon eutectic structure and the size and distribution of intermetallics are also strongly impacted by the solidification rate. It is the dendrite arm spacing, however, that is the easily measured indication of the solidification time.

benefits of faster cycle times as well as reduced thermal shock to the dies with a resultant increased tooling life. In addition, the high chill rates associated with this process may allow T5 properties close to those of T6 properties in sand or permanent mold to be achieved if the parts are quenched straight out of the dies. Alloys used for SSM parts today are primarily A356/357 alloys, but applications using 2xx and wrought 6xxx compositions are also being investigated.

Nondendritic Microstructures

Under normal solidification conditions spanning the full range of commercial casting processes, aluminum alloys without grain refiners exhibit coarse columnar and/or coarse equiaxed structures. The coarse columnar grain structure (Fig. 3a) is less resistant to cracking during solidification and postsolidification cooling than the well-refined grain structure of the same alloy shown in Fig. 3(b). The mechanical tension imposed on areas of the casting during and after solidification is distributed over grain boundaries. The more grain boundaries, the greater the area of grain-boundary surfaces over which these stresses are distributed, thus lowering stress concentrations. Grain refinement also increases the volume fraction of solid at which grains impinge and lock together. This is known as the coherency, and the later this occurs in the solidification process, the less the tendency for hot tearing to occur. A fine grain structure also minimizes the effects on castability and properties associated with the size and distribution of normally occurring intermetallics. Large, insoluble intermetallic particles that are present or form in the temperature range between liquidus and solidus reduce feeding. A finer grain size promotes the formation of finer, more evenly distributed intermetallic particles with corresponding improvements in feeding characteristics. Because most of these more brittle phases precipitate late in the solidification process, their preferential formation at grain boundaries also

Emerging casting processes include thixocasting and rheocasting. Both are variants of semisolid metal casting and have evolved from work at the Massachusetts Institute of Technology in the 1970s. Both variants are capable of creating a nondendritic microstructure from a wide range of alloys. Hypoeutectic aluminum-silicon alloys will typically display a microstructure in which the a-Al primary phase is present in the form of noninterconnected matrix surrounded by the aluminum-silicon eutectic. Figure 1 shows semisolid metal (SSM) microstructures typical of these processes. Evolution in this area has proceeded from thixocasting (electromagnetically or mechanically stirred direct-chill-cast (DC-cast) billet that is reheated until the eutectic is liquefied) to slurry-ondemand processes that create SSM slugs directly from liquid in the case of rheocasting. The resultant semisolid metal slugs serve as feedstock to either vertical or horizontal highpressure die-casting machines. The semisolid slurry flows thixotropically; that is, the viscosity decreases as the applied shear rate increases. This is different from liquid metal, and it helps to prevent many of the turbulence and entrained air-related defects inherent in the conventional die-casting process. Since a SSM melt is already typically half solid, only half as much energy needs to be removed to complete the solidification process. This gives process

Grain Structure A fine, equiaxed grain structure is normally desired in aluminum castings. The type and size of grains formed are determined by alloy composition, solidification rate, and the addition of master alloys (grain refiners) containing intermetallic phase particles, which provide sites for heterogeneous grain nucleation. Grain-Refinement Effects. A finer grain size promotes improved casting soundness by minimizing shrinkage, hot cracking, and hydrogen porosity. The advantages of effective grain refinement are:







Improved feeding characteristics Increased tear resistance Improved mechanical properties Increased pressure tightness Improved response to thermal treatment Improved appearance following chemical, electrochemical, and mechanical finishing

1064 / Nonferrous Alloy Castings Characteristic Comparison Among Casting Process Rheocasting

Conventional casting process

Thixocasting 1

1

1 Liquid

Liquid

Liquid

Nucleation 2

2 Cooling and holding

Semisolid

Semisolid

Semisolid

Sand casting 3

3

Reheating

Solid Billet in room temperature

Casting 1 Temperature

Time

1 Temperature

2 Melting point

2 Melting point

3 Solidification point

3 Solidification point

Low pressure casting

Solid

Casting Time

Die casting 1 Temperature

Solid Squeeze casting Time

Diecasting

Squeeze casting

200 mm Rheocasting

Thixocasting

Low pressure casting

Fig. 1

Casting processes can be characterized according to their solidification rate as well as their solidification profile. Conventional liquid-based processes display a dendritic microstructure whose scale is dependent on the time to solidify. The faster the freezing rate, the smaller the dendrite arm spacing. Thixocasting and rheocasting display nondendritic microstructures in which the aAl phase is present as nonconnected spheroids. Source: Ref 5

profoundly affects tear resistance and mechanical properties in coarse grain structures. By reducing the magnitude of grain-boundary effects through grain refinement, the hot-cracking tendencies of some predominantly solidsolution alloys, such as those of the 2xx and 5xx families, can be substantially reduced. Porosity, if present, is of smaller discrete void size in fine-grain parts. The size of intergranular shrinkage voids is directly influenced by grain size. The previously mentioned effects of structural refinement on feeding characteristics minimize the potential formation of larger shrinkage cavities, and when hydrogen porosity is present, larger pore size with more damaging impact on properties will be

experienced in coarse-grain rather than finegrain castings. With respect to this point, it is important to differentiate between grain refinement with dissolved titanium alone and with particulate refiners containing borides as boride particles that can nucleate hydrogen bubbles, resulting in more numerous, finer, smaller pores. The finer distribution of soluble intermetallics throughout grain-refined castings results in faster and more complete response to thermal treatment. More consistent mechanical properties can be expected following thermal treatment when the structure comprises finer grains. Large grains frequently emphasize different reflectances based on random crystal orientation, resulting in an orange peel or spangled

appearance following chemical finishing, anodizing, or machining. Tearing also becomes more pronounced in the machining of coarsegrained structures. Grain Refinement. All aluminum alloys can be made to solidify with a fully equiaxed, fine grain structure through the use of suitable grain-refining additions. The most widely used grain refiners are master alloys of titanium, or of titanium and boron, in aluminum. Aluminum-titanium refiners generally contain from 3 to 10% Ti. The same range of titanium concentrations is used in Al-Ti-B refiners with boron contents from 0.2 to 1% and titanium-to-boron ratios ranging from about 1 to 50. It should be noted that 1-to-1 ratio titanium-to-boron master

Aluminum and Aluminum Alloy Castings / 1065 Dendrite cell size, in. x 10−4 303

5

10

15

20

25

30

35

40

276

45 44

40

248

36

221

Stress, ksi

Stress, MPa

Ultimate tensile strength

32 Yield

193

28

16

Elongation, %

12

8

4

0 13

Fig. 2

25

38

51 64 76 Dendrite cell size, µm

89

102

114

Tensile properties versus dendrite cell size for four heats of aluminum alloy A356-T62 plaster cast plates

operations such as direct chill casting, in which there is a short residence time between their addition in a trough and billet solidification as they will eventually convert to aluminum carbides if left in holding furnaces for long periods. Although grain refiners of these types can be considered conventional hardeners or master alloys, they differ from master alloys added to the melt for alloying purposes alone. To be effective, grain refiners must introduce controlled, predictable, and operative quantities of aluminides (and/or borides) in the correct form, size, and distribution for grain nucleation. Wrought refiner in rod form, developed for the continuous treatment of aluminum in primary operations, is available in sheared lengths for foundry use. The same grain-refining compositions are furnished in waffle form. In addition to grain-refining master alloys, salts (usually in compacted form) that react with molten aluminum to form combinations of TiAl3 and TiB2 are also available, although seldom used commercially today. Grain-Refinement Mechanisms. Despite much progress in understanding the fundamentals of grain refinement, no universally accepted theory or mechanism exists to satisfy laboratory and industrial experience. It is known that TiAl3 is an active phase in the nucleation of aluminum crystals, ostensibly because of similarities in crystallographic lattice spacing. Nucleation may occur on TiAl3 substrates that are undissolved or precipitate at sufficiently high titanium concentrations by peritectic reaction. Grain refinement can be achieved at much lower titanium concentrations than those predicted by the binary aluminum-titanium peritectic point of 0.15%. For this reason, other theories, such as conucleation of the aluminide by TiB2 or carbides and constitutional effects on the peritectic reaction, are presumed to be influential. Recent findings also suggest the active role of more complex borides of the Ti-Al-B type in grain nucleation. Dissolved titanium, and other elements, can influence the grain size via the growth restriction factor (GRF). Essentially, each dissolved element is treated as a constitutional undercooling agent and the effects are summed according to: GRF ¼

(a)

Fig. 3

(b) As-cast Al-7Si ingots showing the effects of grain refinement. (a) No grain refiner. (b) Grain refined. Both etched using Poulton’s etch; both 2. Courtesy of W.G. Lidman, KB Alloys, Inc.

alloys, and even pure boron, have been shown to be effective grain refiners of aluminum-silicon alloys. They are less effective on foundry or wrought compositions that do not contain silicon. Recent developments also include master

alloys that include both grain-refining elements and compounds as well as strontium for modification. Titanium and aluminum carbides are also effective in the nucleation of grains. Titanium carbide based grain refiners are used in

X

ðki  1Þmi Co

(Eq 2)

where ki and mi are the distribution coefficient between liquid and solid and the slope of the liquidus line for the ith element, respectively. The concentration of the additive element is represented by Co. Titanium makes a strong contribution to the growth restriction factor. The impact of titanium is strongly influenced by the silicon content, however, since, at low silicon levels characteristic of most common wrought alloys, the grains are growing in a plane front or cellular manner. Growth restriction will lead to greater undercooling and, hence, increased

1066 / Nonferrous Alloy Castings use a dry sand mold of fixed design or routine samples, such as those poured in vacuum solidification or for mechanical property tests. The grain size of cast structures obtained by sectioning the casting itself, polishing, and etching is also determined under magnification. Determinations are usually comparative and judgmental, but quantitative grain size measurement by intercept methods is also practiced. Such procedures are outlined in ASTM E 112. Thermal analysis equipment is also commercially available that will give a grain size number based on the liquidus undercooling measured in a standardized instrumented sample.

1000

Grain size, mm

750

500

Al-Si-Fe-Ti (>3wt% Si) Al-Si-Fe-Ti (<3wt% Si) Al-Si-Ti (>3wt% Si) Al-Si-Ti (<3wt% Si) Al-Fe-Ti Al-Ti

250

Modification and Refinement of Aluminum-Silicon Alloys

0 0

20

40

60

80

S(ki⫺1)miCo

Fig. 4

The effect of elements dissolved in liquid aluminum, as opposed to solid particulate compounds, on the grain size is shown by means of the growth restriction factor. Wrought aluminum compositions are generally to the left of the minimum in the curve and exhibit a decrease in grain size with the addition of elements such as titanium. The high silicon foundry alloys show the opposite effect, and the grain size can actually increase with the addition of elements such as titanium. Source: Ref 6

nucleation. High-silicon alloys, on the other hand, are solute rich and grow in a completely dendritic manner. As solute is efficiently directed into the interdendritic channels, there is less undercooling. The combination of factors leads to the result shown in Fig. 4 in which grain size decreases with increasing growth restriction factor for dilute compositions typical of most wrought aluminum alloys and then switches over and increases as the growth restriction factor reaches the higher values typical of the more highly alloyed aluminum-silicon and other common foundry alloys. Therefore, one observes that silicon reduces grain-refiner effectiveness or that aluminumsilicon alloys are more difficult to grain refine than are wrought compositions. It should be noted that these effects are active on growth of the solid/liquid interface and do not take into account heterogeneous nucleation. Elemental effects on grain size caused by the growth restriction factor can be overpowered by the far greater impact on nucleation of the metallurgical phases present in or formed from commercial grain refiners. Additions of titanium in the form of master alloys to aluminum casting compositions normally result in significantly finer and equiaxed grain structure. The period of effectiveness following grain-refiner addition and the potency of grain-refining action are enhanced by the presence of TiB2. In the testing of some compositions, notably those of the aluminum-silicon family, aluminum borides and titanium boride in the absence of excess titanium have been found to provide effective grain refinement. These may also exhibit less density segregation than grain refiners that rely on undissolved TiB2.

The role of borides in enhancing grain-refinement effectiveness and extending its useful duration is observed in both casting and wrought alloys, forming the basis for its use. However, when the boride is present in the form of large, agglomerated particles, it assumes the character of a highly objectionable inclusion with especially damaging effects in machining. Particle agglomeration is found in master alloys of poor quality, or it may occur as a result of long, quiescent holding periods. For the latter reason, it is essential that holding furnaces be routinely and thoroughly cleaned when boron-containing master alloys are used. It is also good practice to add particulate-based grain refiners a few minutes before casting, if practical, to limit settling of the particles as much as possible. The objective in every case in which master alloys or other grain refiners are added to the melt is the release of constituent particles capable of nucleating grain formation to ensure uniform, fine, equiaxed grain structure. The selection of an appropriate grain refiner, practices for grain-refiner addition, and practices covering holding and pouring of castings following grain-refiner addition are usually developed by the foundry after considering casting and product requirements and after referral to the performance characteristics of commercial grain refiners furnished by the supplier. Grain-Size Tests. Various tests have been devised to sample molten aluminum for the purpose of determining the effectiveness of grain refinement. These tests employ principles of controlled solidification to ensure accurate, meaningful, and reproducible grain size determination following polishing and etching. For many foundries, a useful standardized test can

Hypoeutectic aluminum-silicon alloys can be improved by inducing structural modification of the normally occurring eutectic. In general, the greatest benefits are achieved in alloys containing from 5% Si to the eutectic concentration; this range includes most common gravity-cast compositions.

Chemical Modifiers The addition of certain elements, such as calcium, sodium, strontium, and antimony, to hypoeutectic aluminum-silicon alloys results in a finer lamellar or fibrous eutectic network. It is also understood that increased solidification rates are useful in providing similar structures. There is, however, no agreement on the mechanisms involved. The most popular explanations suggest that modifying additions suppress the growth of silicon crystals within the eutectic, providing a finer distribution of lamellae relative to the growth of the eutectic. Various degrees of eutectic modification are shown in Fig. 5. The results of modification by strontium, sodium, and calcium are similar. Sodium is the most powerful modifier, followed by strontium and calcium, respectively. Each of these elements is mutually compatible so that combinations of modification additions can be made without adverse effects. Eutectic modification is, however, transient when artificially promoted by additions of these elements. Figure 6 illustrates the relative effectiveness of various modifiers as a function of time at temperature. Antimony has been advocated as a permanent means of achieving structural modification. In this case, the modified structure differs; a more refined acicular or lamellar eutectic is obtained compared to the uniform fibrous structures of sodium-, calcium-, or strontium-modified metal. As a result, although permanent, the end result obtained via antimony is not as good. In fact, particularly at low freezing rates where very large lamellar structures can arise, antimony may give worse results than even an unmodified alloy.

Aluminum and Aluminum Alloy Castings / 1067

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 5

Varying degrees of aluminum-silicon eutectic modification ranging from unmodified (a) to well modified (f). See Fig. 6 for the effectiveness of various modifiers.

Antimony is not compatible with other modifying elements. In cases in which antimony and other modifiers are present, coarse antimonycontaining intermetallics are formed that preclude the attainment of an effectively modified structure and adversely affect casting results. The much higher quantities of antimony usually employed means that this element can easily overpower the other modifiers by simply reacting with them to form intermetallics that precipitate, essentially removing the modifiers from the melt. Confusion can arise under these conditions since the modifiers will still be detected in spectrographic analysis while the antimony, because its emission spectra overlaps with

iron lines, may not. Thermal analysis of eutectic undercooling is the sure way to settle this question if such a problem is suspected. Modifier additions are usually accompanied by an increase in porosity. In the case of sodium and calcium, the reactions involved in element solution are invariably turbulent or are accompanied by compound reactions that by their nature increase dissolved hydrogen levels. In the case of strontium, the mode of eutectic solidification is changed, making the dendritic area more tortuous by the nucleation and growth of eutectic colonies throughout the dendrite network. This makes feeding more difficult toward the end of solidification and

shortens feeding distances. This effect is particularly noticeable when solidification takes place under low-temperature gradients. For sodium, calcium, and strontium modifiers, the removal of dirt and impurities by reactive gas (chlorine) fluxing also results in the removal of the modifying element. Recommended practices are to obtain modification through additions of modifying elements added to well-processed (cleaned) melts, followed by inert gas fluxing to acceptable hydrogen levels. No such disadvantages accompany antimony use. Calcium and sodium can be added to molten aluminum in metallic or salt flux form. Vacuum-prepackaged sodium metal is still used. A sodium-cored aluminum wire form is also available in Europe. Strontium is currently available in many forms, including aluminumstrontium master alloys containing either 10% or 90% Sr, respectively. The former is inert and easily handled, and its recovery increases with melt temperature. The latter is reactive and must be stored in sealed cans or packets for use. It undergoes reactive dissolution, and its recovery has been shown to decrease with increasing melt temperature. Very low sodium concentrations (0.001%) are required for effective modification. More typically, additions are made to obtain a sodium content in the melt of 0.005 to 0.015% to obtain practical casting times between remodification additions. A much wider range of strontium concentrations is in use. In general, addition rates far exceed those required for effective sodium modification. An operating range 50 ppm wide between 0.004 and 0.020% is modern standard industry practice. Normally, good modification is achievable in the range of 0.004 to 0.015% Sr. Remodification through strontium additions may be required, although retreatment is less frequent than for sodium, particularly at the lower strontium levels where fade rates can be quite low. To be effective in modification, antimony must be alloyed to approximately 0.06%. In practice, antimony is used in the much higher range of 0.10 to 0.50%. It is possible to achieve a state of overmodification, in which eutectic coarsening occurs, when sodium and/or strontium are used in excessive amounts. The corollary effects of reduced fluidity and susceptibility to porosityrelated problems are usually encountered well before overmodification may be experienced. The Importance of Phosphorus. It has been well established that phosphorus interferes with the modification mechanism. Phosphorus reacts with sodium and probably with strontium and calcium to form phosphides that nullify the modification additions. It is therefore desirable to use low-phosphorus metal when modification is a process objective and/or to make larger modifier additions to compensate for phosphorus-related losses. Primary producers control phosphorus contents in smelting and processing through the judicious choice of feed materials to provide less than 0.001% P in alloyed ingot. Some older

1068 / Nonferrous Alloy Castings Grain size, in.

(F) 0

0.016

0.032

0.048

0.064

8 Sodium modified alloy Unmodified alloy

Elongation, %

7 6 5 4 3 2

Alloy 356 + sodium Alloy 356 + strontium Alloy 356 + sodium/strontium Alloy 356 + sodium/strontium (C)

(B)

304 44

294 284

41

274 38

264 254

Hardness, HRB

90

(A) 0

Fig. 6

314

Ultimate tensile strength, ksi

1

(D) Ultimate tensile strength, MPa

Degree of modification (see Fig. 5)

(E)

1

2

3 Time after modification, h

4

5

6

89 88 87 86 85

0

0.2

0.6 1.0 Grain size, mm

1.4

1.8

Effectiveness of sodium and strontium modifiers as a function of time. See Fig. 5 for degrees of modification.

Fig. 7

smelters, although less efficient at aluminum metal recovery, are capable of even lower levels. At these levels, normal additions of modification agents are effective in achieving modified structures. However, phosphorus contamination may occur in the foundry through contamination by phosphate-bonded refractories and mortars, crucible glazes, and by phosphorus contained in other melt additions, such as master alloys and alloying elements including silicon. Effects of Modification. Typically, modified structures display somewhat higher tensile properties and appreciably improved ductility when compared to similar but unmodified structures. Figure 7 illustrates the desirable effects on mechanical properties that can be achieved by modification. Improved performance in casting is characterized by superior resistance to elevatedtemperature cracking. In addition, modification has been shown to reduce the time required in the solution heat treatment of aluminum-silicon alloys. This results from two distinct objectives of the solutionizing procedure. The primary objective is to maximize the solid solution of hardening elements before quenching, and the second is to round, and in some cases spheroidize, silicon particles found in the aluminum-silicon eutectic by diffusion. The second objective can take from minutes to hours to days depending on the coarseness of the microstructure and the degree of eutectic modification achieved in the

casting process. A high degree of eutectic modification significantly reduces the time required for this purpose.

Refinement of Hypereutectic Aluminum-Silicon Alloys The elimination of large, coarse primary silicon crystals that are harmful in the casting and machining of hypereutectic silicon alloy compositions is a function of primary silicon refinement. Phosphorus added to molten alloys containing more than the eutectic concentration of silicon, made in the form of metallic phosphorus or phosphorus-containing compounds such as phosphor-copper and phosphorus pentachloride, has a marked effect on the distribution and form of the primary silicon phase. Investigations have shown that retained trace concentrations as low as 0.0015 through 0.03% P are effective in achieving the refined structure. Disagreements on recommended phosphorus ranges and addition rates have been caused by the extreme difficulty of accurately sampling and analyzing for phosphorus. More recent developments using vacuum-stage spectrographic or quantometric analysis now provide rapid and accurate phosphorus measurements. Following melt treatment by phosphorus-containing compounds, refinement can be expected to be less transient than the effects of

Mechanical properties of as-cast A356 alloy tensile specimens as a function of modification and grain size

conventional modifiers on hypoeutectic modification. Furthermore, the solidification of phosphorus-treated melts, cooling to room temperature, reheating, remelting, and resampling in repetitive tests have shown that refinement is not lost; however, primary silicon particle size increases gradually, responding to a loss in phosphorus concentration. Common degassing methods accelerate phosphorus loss, especially when chlorine or freon is used. In fact, brief inert gas fluxing is frequently used to reactivate aluminum phosphide nuclei, presumably by resuspension. Practices that are recommended for melt refinement are:
The alloy should be thoroughly chlorine or

salt fluxed before refining to remove phosphorus-scavenging impurities such as strontium, calcium, and sodium.
Brief fluxing after the addition of phosphorus is recommended to remove the hydrogen introduced during the addition and to distribute the aluminum phosphide nuclei uniformly in the melt. Figure 8 illustrates the microstructural differences between refined and unrefined structures. Effects of Refinement. Refinement substantially improves mechanical properties and castability. In some cases, especially at higher

Aluminum and Aluminum Alloy Castings / 1069

Foundry Alloys for Specific Casting Applications Die-Casting Compositions

(a)

(b)

(c)

Fig. 8

Effect of phosphorus refinement on the microstructure of Al-22Si-1Ni-1Cu alloy. (a) Unrefined. (b) Phosphorus-refined. (c) Refined and fluxed. All 100

silicon concentrations, refinement forms the basis for acceptable foundry results. In wearresistant applications, refinement prevents patchy or silicon-devoid areas and hence improves the consistency of behavior under wear conditions. The appearance of machined surfaces is also generally improved. Modification and Refinement. No elements are known that beneficially affect both eutectic and hypereutectic phases. The potential negative consequences of employing modifying and refining additions in the melt are characterized by the interaction of phosphorus with calcium, sodium, and strontium. Phosphorous is mutually antagonistic to the effects of all of these elements and vice versa.

The most important compositions in use for high-pressure die casting are the highly castable and forgiving alloys of the aluminum-silicon family. Of these, alloys A380, 383, and their variations predominate. Magnesium content is usually controlled at low levels to minimize oxidation and the generation of oxides in the casting process. Nevertheless, alloys containing appreciable magnesium concentrations are routinely produced to achieve increased hardness as well as to avoid the costs associated with magnesium removal in secondary refineries. Iron content of 0.7% or greater is preferred in most die-casting operations to maximize elevated-temperature strength, to facilitate ejection, and to minimize soldering to the die face. Improved ductility through reduced iron content has been an incentive resulting in widespread efforts to develop a tolerance for iron as low as approximately 0.25%. These efforts focus on process refinements and improved die lubrication. Hypereutectic aluminum-silicon alloys are growing in importance as their valuable characteristics and excellent die-casting properties are exploited in automotive and other applications. Recent developments in die-casting alloys include those developed for advanced die-casting processes designed to produce heat treatable, weldable, structural parts. These processes make use of specialized die and shot sleeve lubricants to prevent metal contamination. Careful solidification modeling of the filling process is also carried out to reduce the turbulence normally associated with high-pressure, high-velocity injection die-casting processes to an absolute minimum. The main feature of these processes, however, is the use of a high vacuum on the casting cavity and shot sleeve to eliminate the entrainment of air. Aluminum-silicon alloys designed for these processes generally use manganese in place of iron to combat die soldering and are thus primary as opposed to secondary alloy compositions.

Rotor Castings Most squirrel-cage induction-type electric motors employ an integrally cast aluminum rotor. There are many economic and manufacturing process advantages that constitute an improvement of this type of rotor construction over wire-wound assemblies. The rotors produced by the casting process range in diameter from 25 to approximately 760 mm (1 to 30 in.). Several casting processes are used to produce cast aluminum rotors. The principal processes used are vertical and horizontal cold chamber die casting and, to a lesser extent, centrifugal and permanent mold casting. Aluminum alloys display a wide range of electrical characteristics, and the choice of

a specific alloy for a cast motor rotor depends on the specified operating characteristics of the motor. Standardized alloys of high, intermediate, and low conductivity have been developed for this industry. Conductivity. Most cast aluminum motor rotors produced are in carefully controlled, more pure compositions 100.0, 130.0, 150.0, 160.0, and 170.0 (99.0, 99.3, 99.5, 99.6, and 99.7% Al, respectively). Impurities in these alloys are controlled to minimize variations in electrical performance based on conductivity and to minimize the occurrence of microshrinkage and cracks during casting. Rotor alloy 100.0 contains a significantly larger amount of iron and other impurities, and this generally improves castability. With higher iron content, crack resistance is improved, and a lower tendency toward shrinkage formation will be observed. This alloy is recommended when the maximum dimension of the part is greater than 127 mm (5 in.). For the same reasons, alloy 150.0 is preferred over 170.0 in casting performance, for example. Minimum conductivities for each rotor alloy grade are: Alloy

Minimum conductivity, %IACS

100.1 130.1 150.1 160.1 170.1

54 56 57 58 59

IACS, International Annealed Copper Standard

For motor rotors requiring high resistivity (for example, motors with high starting torque) more highly alloyed die casting compositions are commonly used. The most popular are alloys 443.2 and A380.2. By choosing alloys such as these, conductivities from 25 to 35% IACS can be obtained; in fact, highly experimental alloys with even higher resistivities have been developed for motor rotor applications. Although gross casting defects may adversely influence electrical performance, the conductivity of alloys employed in rotor manufacture is more exclusively controlled by composition. Table 2 lists the effects of various elements in and out of solution on the resistivity of aluminum. Simple calculation using these values accurately predicts total resistivity and its reciprocal conductivity for any composition. A more general and easy-to-use formula for conductivity calculation that offers sufficient accuracy for most purposes is: Conductivity ¼ 63:50  6:9x  83y

(Eq 3)

where electrical conductivity is in percent IACS, x = iron + silicon (in weight percent), and y = titanium + vanadium + manganese + chromium (in weight percent). Reference to specified composition limits for rotor alloys shows the use of composition controls that reflect electrical considerations.

1070 / Nonferrous Alloy Castings Table 2 Effect of elements in and out of solution on the electrical resistivity of aluminum Average increase in resistivity per weight percent, mV  cm(a) Element

Chromium Copper Iron Lithium Magnesium Manganese Nickel Silicon Titanium Vanadium Zinc Zirconium

Maximum solubility in aluminum, %

In solution

Out of solution(b)

0.77 5.65 0.052 4.0 14.9 1.82 0.05 1.65 1.0 0.5 82.8 0.28

4.00 0.344 2.56 3.31 0.54(c) 2.94 0.81 0.65 2.88 3.58 0.094(d) 1.74

0.18 0.030 0.058 0.68 0.22(c) 0.34 0.061 0.059 0.12 0.28 0.023(d) 0.044

(a) Add indicated increase to the base resistivity for high-purity aluminum (2.65 mO  cm at 20  C, or 68  F, or 2 71 mO  cm at 25  C, or 77  F). (b) Limited to about twice the concentration given for the maximum solid solubility except as noted. (c) Limited to approximately 10%. (d) Limited to approximately 20%. Source: Ref 5.

The peritectic elements are limited because their presence is harmful to electrical conductivity. Prealloyed ingot produced to these specifications may be produced by boron addition, which precipitates these elements out as solids that settle out before casting if the initial impurity levels warrant it. In addition, iron and silicon contents are subject to control by ratio with the objective of promoting the aAl-Fe-Si phase intermetallics least harmful to castability. Ignoring these important relationships results in variable electrical performance and, of equal importance, variable casting results. As in all casting operations, but especially in die casting, establishment of reproducible casting conditions depends heavily on the rhythm of the process itself. Variable chemical composition (for example, when unfurnaced* and thus untreated unalloyed primary aluminum grades are used for cost advantage over rotor alloys) will always introduce a degree of unpredictable casting results that will adversely affect both process and product performance. In contrast, the use of rotor alloys ensures optimum casting and controlled, consistent product results. Casting Processes. The horizontal cold chamber die-casting method is recommended and is the most widely used for the highvolume production of fractional-horsepower motor rotors. Multiple-cavity dies are usually employed to cast several rotors of the same or different design simultaneously. The vertical-pressure die-casting method has been successfully used for many years to cast both fractional and integral horsepower units. In this process, the lower press platen contains a sump or depression into which molten metal is automatically or manually poured. The sump is typically lined with a refractory material, such as Fiberfrax, mica, or other insulating paper. Sprayed refractory coatings can also be used to insulate and to protect the basin from attack. The mold and indexed mounting for the laminate stack are located in the upper platen. Molten *That is, tapped from primary potroom transfer ladles in a smelter and not subject to alloying and control in an alloying furnace.

metal is forced into the die through a network of small gates separating the lower and upper platens. These gates are normally tapered in the direction of metal flow through the base plate and into the collector ring of the part. For either horizontal or vertical casting processes, preheating of the laminate pack is recommended, but in practice preheating is rarely performed. The preheat temperature for the laminate assembly is from 205 to 540  C (400 to 1000  F). When preheating is not performed, individual ferrous laminations must be free of surface contamination, especially oil and moisture. Preheating also oxidizes the laminate and its sheared outer surface to inhibit metallurgical bonding during the casting process. A postcasting thermal treatment of heating to a temperature in excess of 260  C (500  F) will effectively shear the brittle intermetallics formed when bonding does occur. Advantages are attributed to more severe postcasting thermal treatment to ensure bond separation and to oxidize the interface surfaces. Annealing at 425 to 510  C (800 to 950  F) followed by still-air cooling is conventional, with cooling practices dictated by the electrical advantages of rejecting solute phases from solid solution. The foundryman should be certain that the laminations are clean and free of excessive burrs that might result in current leakage. The preheating of poorly sheared laminate disks at higher temperature is recommended. Good die venting is essential in rotor casting. Failure to provide and to maintain vents will always result in excessive end-ring porosity and unsoundness in the conductor bars. Another defect encountered in rotor casting that is related to venting is backfilled conductor bars. This condition is induced by preferential flow in one region of the cavity and simultaneous molten metal entry to a single bar from both end rings. Accurate laminate assembly and placement in the mold cavity, clean laminates, minimum die lubrication, effective gate design, and adequate die venting are essential in the production of quality rotors. Gating scrap is often directly recharged to the furnace in rotor casting. This results in the

potentially heavy oxide contamination in the metal source that is a major cause of casting defects. Iron cores inadvertently charged with the gating scrap are a potential source of serious electrical conductivity degradation. The most common defects, apart from misruns or nonfills, are broken bars resulting from different rates of thermal contraction between the core and the conductor bars and from massive oxide inclusions, which significantly influence flowability and casting recovery and obstruct current flow in the conductor bars. The direct recharging of scrap may include the inadvertent or intentional addition of iron, particularly in the form of steel laminates readily available to operators for the purpose of improving casting results at the expense of electrical characteristics. The direct recharging of scrap also influences temperature control with obvious harmful effects on process variability. For best results, a temperature control range of 10  C (18  F) is recommended.

Premium Engineered Castings Premium engineered castings provide higher levels of quality and reliability than are found in conventionally produced parts. These castings may display optimum performance in one or more of the following characteristics: mechanical properties (often determined by prolongations or test coupons machined from representative parts), soundness (determined radiographically), dimensional accuracy, and finish. However, castings of this classification are notable primarily for mechanical property attainment that reflects extreme soundness, fine dendrite arm spacing, and well-refined grain structure. These technical objectives require the use of chemical compositions competent to display premium engineering properties. Alloys considered to be premium engineered compositions appear in separately negotiated specifications or in specifications such as military specification SAE-AMS-A-21180, which is extensively used in the United States for premium engineered casting procurement. Alloys commonly considered premium by definition and specification are 201.0, C355.0, A206.0, B206.0, A356.0, 224.0, A357.0, 249.0, 358.0, and 354.0. All alloys employed in premium engineered casting work are characterized by optimum concentrations of hardening elements and restrictively controlled impurities. Although any alloy can be produced in cast form with properties and soundness conforming to a general description of premium values relative to corresponding commercial limits, only those alloys demonstrating yield strength, tensile strength, and especially elongation in a premium range belong in this discussion. They fall into two categories: highstrength aluminum-silicon compositions and those alloys of the 2xx series, which, by restricting impurity element concentrations, provide

Aluminum and Aluminum Alloy Castings / 1071 outstanding ductility, toughness, and tensile properties with notably poorer castability. In all premium casting alloys, impurities are strictly limited for the purposes of improving ductility. In aluminum-silicon alloys, this translates to control of iron at or below 0.010% with measurable advantages in further reductions to the range of 0.03 to 0.05%, the practical limit of commercial smelting capability. Beryllium is present in A357.0, C357.0, and D357.0 alloys not to inhibit oxidation (although that is a corollary benefit) but to alter the form of the insoluble Al-Mg-Fe-Si phase to a more nodular form least detrimental to ductility. Beryllium also alters the chemistry of the insoluble phase to exclude magnesium, which then becomes fully available for hardening purposes. Magnesium is normally controlled in the upper specification range to maximize Mg2Si formation for strength development at some loss in ductility. The usefulness of beryllium declines at iron levels below 0.07%, and the variants E357.0 and F357.0 have been registered to provide an exit strategy from the use of this element. This move is being driven by health and safety considerations associated with the recognized toxicity of beryllium and its ability to cause both acute and chronic beryllium disease (also known as berylliosis, an allergic response by the human body, particularly in the lungs), and the carcinogenicity of certain beryllium compounds. The presence of iron and silicon in alloys such as 295.0 contributes to reasonably good castability. For the development of premium properties in the 2xx family of alloys, these impurities are severely restricted. As a result of these composition restrictions, all premium engineered casting alloys of the 2xx type can be characterized as extremely sensitive to crack formation. They are also highly susceptible to shrinkage, requiring unusual foundry expertise in gating and risering. Design engineers and the producing foundry usually must collaborate to develop configurations offering maximum compatibility with casting requirements. The development of hot isostatic pressing for aluminum alloys is pertinent to the broad range of premium castings, but is especially relevant for the more difficult-to-cast aluminum-copper series (see the article “Hot Isostatic Pressing of Castings” in this Volume). Table 3 defines the minimum mechanical property limits normally applied to premium engineered castings. It must be emphasized that the negotiation of limits for specific parts is usual practice, with higher as well as lower specific limits based on part design criteria and foundry capabilities. Melt Preparation Metal treatment is performed for this class of casting to achieve the highest possible quality. Fluxing to hydrogen levels lower than 0.10 mL/100 g and effective removal of oxides and other nonmetallics by active gas fluxing, salt injection fluxing, and/or filtration are essential in meeting specified soundness levels and mechanical property limits.

Extreme care is also exercised in all phases of metal handling and pouring to preserve metal quality. Proprietary pouring methods have been developed for this purpose, including mechanical pumping, displacement, and nonturbulent transfer from metal source to the casting cavity. Casting Processes. Molds are usually dry sand and other materials comprising composite mold construction, such as plaster and metallic sections. Elaborate planning and control of solidification through mold design and controlled chilling are normal. Solidification modeling is frequently employed in this process, particularly for high-volume castings. Many premium engineered castings are produced by other gravity-casting processes, notably, permanent mold and plaster molding. In every case, however, the same exacting approach is necessary for control of solidification to achieve the structures and soundness needed to furnish castings within specification. Dimensional Control. Dimensional variations depend on part size, complexity, and alloy, but the use of highly accurate mold and core patterns and the dimensional control of molding media provide for extreme dimensional accuracy. Tolerances applicable to typical casting dimensions can be specified at 0.254 mm (0.010 in.), and in special cases, tolerances as small as 0.127 mm (0.005 in.) may apply. Even the inspection methods used in these cases represent an unusual capability on the part of premium engineered casting foundries. Datum plane measuring concepts are required that are agreed upon by both customer and vendor. The datum plane system is a cost-effective means of establishing dimensions, and it is highly adaptable to automated coordinate measurement. Another advantage of premium engineered castings is the attainment of structural integrity in thin walls. This ranges from 1.52 to 2.03 mm (0.060 to 0.080 in.) for large plane areas and is as low as 0.51 to 0.76 mm (0.020 to 0.030 in.) for more limited casting sections. Casting surface finish is always a function of mold surface and characteristics. The selection of mold materials and the accuracy of mold finish maintained in premium casting operations ensure that specified requirements are met.

Heat Treatment The metallurgy of aluminum and its alloys fortunately offers a wide range of opportunities for employing thermal treatment practices to obtain desirable combinations of mechanical and physical properties. Through alloying and temper selection, it is possible to achieve an impressive array of features that are largely responsible for the current use of aluminum alloy castings in virtually every field of application. Although the term heat treatment is often used to describe the procedures required to achieve maximum strength in any suitable composition through the sequence of solution

Table 3 Mechanical property specifications for premium engineered castings Ultimate tensile strength (min) Alloy

Class

MPa

ksi

0 2% offset yield strength (min) MPa

ksi

Elongation in 50 mm (2 in), %

Specimens cut from designated casting areas 249.0 10 379 55 310 45 11 345 50 276 40 354.0 10(a) 324 47 248 36 11(a) 296 43 228 33 355.0 10(a) 283 41 214 31 11(a) 255 37 207 30 12(a) 241 35 193 28 A356.0 10(a) 262 38 193 28 11(a) 228 33 186 27 12(a) 221 32 152 22 A357.0 10(a) 262 38 193 28 11(a) 283 41 214 31 224.0 10 310 45 241 35 11 345 50 255 37 XA201.0 10 386 56 331 48 11 386 56 331 48

3 2 3 2 3 1 1 5 3 2 5 3 2 3 3 15

Specimens cut from any 249.0 1 345 2 379 3 414 354.0 1(a)(b) 324 2(a)(b) 345 C355.0 1(b) 283 2(a)(b) 303 3(a)(b) 345 A356.0 1(b) 262 2(a)(b) 276 3(a)(b) 310 A357.0 1(a)(b) 310 2(a)(b) 345 224.0 1 345 2 379 XA201.0 1 414 2 414

2 3 5 3 2 3 3 2 5 3 3 3 5 3 5 5 3

casting area 50 276 55 310 60 345 47 248 50 290 41 214 44 228 50 276 38 193 40 207 45 234 45 241 50 276 50 255 55 255 60 345 60 345

40 45 50 36 42 31 33 40 28 30 34 35 40 37 37 50 50

(a) Values from specification SAE-AMS-A-21180. (b) This class is obtainable in favorable casting configurations and must be negotiated with the foundry for the particular configuration desired.

heat treatment, quenching, and precipitation hardening, in its broadest meaning, heat treatment comprises all thermal practices intended to modify the metallurgical structure of products in such a way that physical and mechanical characteristics are controllably altered to meet specific engineering criteria. In all cases, one or more of the following objectives form the basis for temper selection:
Increase hardness for improved machinability
Increase strength and/or produce the

mechanical properties specified for a particular material condition
Stabilize mechanical and physical properties
Ensure dimensional stability as a function of time under service conditions
Relieve residual stresses induced by casting, quenching, machining, welding, or other operations To achieve any of these objectives, parts can be annealed, solution heat treated, quenched, precipitation hardened, overaged, or treated with combinations of these practices. In some simple shapes (for example, bearings), thermal treatment can also include plastic deformation in the form of cold work.

1072 / Nonferrous Alloy Castings Temper Designations and Practices The Aluminum Association has standardized the definitions and nomenclature applicable to thermal practice and maintains a registry of standard heat treatment practices and designations for industry use. Standardized temper designations applicable to castings are:
F: as-fabricated temper (in the case of cast







ings this is the as-cast condition) O: (formerly T2, T2x): annealed (thermally stress relieved) T4: solution heat treated and quenched T5: artificially aged from the F temper T6: solution heat treated, quenched, and artificially aged T7: solution heat treated, quenched, and overaged T8: cold reduced before aging to improve compressive yield strength (bearings only)

Variations in thermal treatment practice are shown as the second and third digits in the standard designations; for example, T61, T62, T572, and so on. There is no consistent convention for the assignment of temper designation variations except that for solution heat treated alloys general practice has been to assign tempers T6, T61, and T62 to an ascending order in strength development up to full hardness. Recommended thermal treatment practices for aluminum casting alloys are listed in Table 4.

Table 4

Solution heat treatment(b)

Aging treatment

Temperature(c) Alloy

201.0

204.0 A/B206

242.0

295.0

296.0

319.0

328.0 332.0 333.0

336.0 354.0 355.0

Principles of Heat Treatment The heat treatment of aluminum alloys is based on the varying solubilities of metallurgical phases in a crystallographically isotropic system. Because the solubility of the many elements increases with increasing temperature to the solidus, as in the binary eutectic system shown in Fig. 9, the variations in degree of solution and the formation and distribution of precipitated phases can be used to influence material properties. In addition to the phase and morphology changes associated with soluble elements and compounds, other (sometimes desirable) effects accompany elevated-temperature treatment. The microsegregation characteristic of all cast structures is minimized or eliminated. Residual stresses caused by solidification or by prior quenching are reduced, insoluble phases may be physically altered, and susceptibility to corrosion, especially in certain compositions, may be affected. Solution Heat Treatment. Exposure to temperatures corresponding to maximum safe limits relative to the lowest melting temperature for a specific composition results in the dissolution of soluble phases formed during and after solidification. The rate of heating to solution temperature is technically unimportant in casting compositions except when more than one soluble phase is present, as in the Al-Cu-Mn-Mg, Al-Si-Cu-Mg, and Al-Zn-Cu-Mg systems. In these cases, stepped heat treatment is sometimes essential to avoid the melting of lower-melting

Typical heat treatments for aluminum alloy sand and permanent mold castings

Temper

T6

S

T7

S

T4 T4

S or P S or P

T43

S or P

T7

S or P

O(c) T571

S S P S S or P S S S S P P P S S P S P P P P P P (g) S or P S P P S P S P S P

T77 T61 T4 T6 T62 T7 T4 T6 T7 T5 T6 T6 T5 T5 T6 T7 T551 T65 T51 T6 T62 T7 T71

C355.0

T6 T61

356.0

T51 T6 T7 T71

A356.0

T6 T6 T61

T7 T71

357.0 A357.0 359.0 A444.0

T6 T61

T4



Type of casting(a)

S or P S P S P S P S P S P S P S P P S (g) (g) P



C

F

Temperature(c) Time, h

510–515 525–530 510–515 525–530 520 510 Then 530 510 Then 530 510 Then 530 ... ... ... 515 515 515 515 515 515 510 510 510 ... 505 505 515 ... ... 505 505 ... 515 525–535 ... 525 525 525 525 525 525 525 525 525 ... ... 540 540 540 540 540 540 540 540 540 540

950–960 980–990 950–960 980–990 970 950 990 950 990 950 990 ... ... ... 960 960 960 960 960 960 950 950 950 ... 940 940 960 ... ... 950 940 ... 960 980–995 ... 980 980 980 980 980 980 980 980 980 ... ... 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000

2 14–20 2 14–20 10 2 14–20 2 14–20 2 14–20 ... ... ... 5(f) 2–6(f) 12 12 12 12 8 8 8 ... 6–12 4–12 12 ... ... 6–12 6–12 ... 8 10–12 ... 12 4–12 4–12 12 4–12 12 4–12 12 6–12 ... ... 6–12 4–12 6–12 4–12 6–12 4–12 6–12 4–12 6–12 4–12

540 540 540 540 ... 540 540 540 540 540

1000 1000 1000 1000 ... 1000 1000 1000 1000 1000

6–12 4–12 6–12 4–12 ... 8 10–12 8–12 10–14 8–12



C



F

Time, h

Room temperature Then 155 310 Room temperature Then 190 370

12–24 20 12–24 5

Room temperature Room temperature Then 160 320 Room temperature Then 190 370 345 650 205 400 165–170 330–340 330–355 625–675 205–230 400–450 ... ... 155 310 155 310 260 500 ... ... 155 310 260 500 205 400 155 310 155 310 155 310 205 400 205 400 155 310 260 500 205 400 205 400 (h) (h) 225 440 155 310 155 310 170 340 225 440 225 440 245 475 245 475 155 310 Room temperature 155 310 225 440 155 310 155 310 205 400 225 440 245 475 245 475 155 310 155 310 165 330 Room temperature Then 155 310 225 440 225 440 245 475 245 475 155 310 165 330 155 310 (h) (h) (h) (h)

5 days 12–24 0.5–1.0 12–24 4–5 3 8 22–26 2 (min) 1.3 ... 3–6 12–24 4–6 ... 2–8 4–6 8 2–5 2–5 2–5 7–9 7–9 2–5 4–6 7–9 7–9 (h) 7–9 3–5 2–5 14–18 3–5 3–9 4–6 3–6 3–5 8 (min) 10–12 7–9 3–5 2–5 3–5 7–9 2–4 7–9 2–5 2–5 6–12 8 (min) 6–12 8 8 3–6 3–6 6–12 6–12 10–12 (h) (h)

(continued) (a) S. sand: P. permanent mold. (b) Unless otherwise indicated, solution treating is followed by quenching in water at 65–100  C (150–212  F). (c) Except where ranges are given listed. Listed temperature are 6  C or 10  F. (d) Stress relieve for dimensional stability as follows: hold 5 h at 413 = 14  C (775 = 25  F); furnace cool to 345  C (650  F) over a period of 2 h or more; furnace cool to 230  C (450  F) over a period of notmore than ½ h; furnace cool to 120  C (250  F) over a period of approximately 2 h; cool to room temperature in still air outside the furnace. (e) No quench required: cool in still air outside furnace. (f) Air-blast quench from solution-treating temperature. (g) Casting process varies (sand, permanent mold, or composite) depending on desired mechanical properties. (h) Solution heat treat as indicated, then artificially age by heating uniformly at the temperature and for the time necessary to develop the desired mechanical properties. (i) Quench in water at 65–100  C (150–212  F) for 10–20 s only. (j) Cool to room temperature in still air outside furnace.

Aluminum and Aluminum Alloy Castings / 1073 Table 4

(continued) Solution heat treatment(b) Temperature(c) 

Type of casting(a)



C

F

Alloy

Temper

520.0 535.0 705.0

T4 T5(d) T5

S S S

430 400 ...

810 750 ...

18(i) 5 ...

P 707.0

T5

S P

T7

... ... ... ... ... 530 530 ... ... ...

... ... ... ... ... 990 990 ... ... ...

... ... ... ... ... 8–16 4–8 ... ... ...

Time, h

710.0 711.0 712.0

T5 T1 T5

S P S P S

713.0

T5

S or P

...

...

...

771.0

T53(d) T5 T51 T52 T6 T71 T5 T5 T6 T5

S S S S S S S or P S or P P S or P

415(j) ... ... ... 590(j) 590(e) ... ... 480 ...

775(j) ... ... ... 1090(j) 1090(e) ... ... 900 ...

5(j) ... ... ... 6(j) 6(e) ... ... 6 ...

850.0 851.0 852.0

Aging treatment Temperature(c) 

C



F

Room temperature 100 210 Room temperature 100 210 155 310 Room temperature 100 210 175 350 175 350 Room temperature Room temperature Room temperature or 155 315 Room temperature or 120 250 180(j) 360(j) 180(j) 360(j) 205 405 (d) (d) 130 265 140 285 220 430 220 430 220 430 220 430

Time, h

21 days 8 21 days 10 3–5 21 days 8 4–10 4–10 21 days 21 days 21 days 6–8 21 days 16 4(j) 3–5(j) 6 (d) 3 15 7–9 7–9 4 7–9

(a) S. sand: P. permanent mold. (b) Unless otherwise indicated, solution treating is followed by quenching in water at 65–100  C (150–212  F). (c) Except where ranges are given listed. Listed temperature are 6  C or 10  F. (d) Stress relieve for dimensional stability as follows: hold 5 h at 413 = 14  C (775 = 25  F); furnace cool to 345  C (650  F) over a period of 2 h or more; furnace cool to 230  C (450  F) over a period of notmore than ½ h; furnace cool to 120  C (250  F) over a period of approximately 2 h; cool to room temperature in still air outside the furnace. (e) No quench required: cool in still air outside furnace. (f) Air-blast quench from solution-treating temperature. (g) Casting process varies (sand, permanent mold, or composite) depending on desired mechanical properties. (h) Solution heat treat as indicated, then artificially age by heating uniformly at the temperature and for the time necessary to develop the desired mechanical properties. (i) Quench in water at 65–100  C (150–212  F) for 10–20 s only. (j) Cool to room temperature in still air outside furnace.

Fig. 9

Portion of aluminum-copper binary phase diagram. Temperature ranges for annealing, precipitation heat treating, and solution heat treating are indicated.

phases. Very rapid heating can result in nonequilibrium melting in highly segregated structures, but the conditions required for such occurrences are most unlikely in engineered castings. In all cases, the most complete degree of solution is desirable for subsequent hardening, and

a number of factors are involved. Different casting processes and foundry practices result in microstructural differences with relevance to heat treatment practice. The coarser microstructures associated with slow solidification rate processes require a longer solution heat treatment exposure.

The time required at temperature to achieve solution is progressively shorter for investment, sand, and permanent mold castings, but thin-wall sand castings produced with extensive use of chills can often display finer microstructures than heavy-section permanent mold parts produced in such a way that process advantages are not exploited. For these reasons, solution heat treatment practices can be optimized for any specific part to achieve solution with the shortest reasonable cycle once production practice is finalized, even though most foundries and heat treaters tend to standardize to a practice with a large margin of safety. Because the slope of the characteristic solvus illustrated in Fig. 9 changes as temperature approaches the eutectic melting point, it is apparent that temperature is critical in determining the degree of solution that can be attained. In addition, temperature affects diffusion rates, and this directly influences the degree of solution as a function of time at temperature. Within the temperature ranges defined for solution heat treatment by applicable specifications lies a significant corresponding range of solution times. The knowledgeable heat treater/foundry seeking to obtain superior properties based on solution heat treatment will bias temperature selection within specification limits to obtain the highest degree of solution. Superior furnaces, thermocouples, and furnace controls, representing the state of the art, along with recognition of the normal resistance of cast structures to melting based on diffusion considerations, offer the potential of superior property development compared to historically more conservative temperature selection. However, although temperatures just below the eutectic melting point are desirable for optimum property development, it is critical that eutectic melting resulting in brittle intergranular eutectic networks be avoided. Another important factor in the heat treatment of aluminum-silicon foundry alloys is the effect of solutionizing time on the morphology of the silicon phase. This is especially important for castings that are produced without chemical modification. Many old heat treatment specifications call out solutionizing times much greater than those needed for the maximum dissolution of the hardening elements to accomplish thermal modification of the silicon phase. Recent realization of this has allowed many foundries that do modify their castings to reduce solutionizing times substantially. Figure 10 shows the effect of solutionizing on unmodified microstructures. The three micrographs of chemically unmodified A356 in this figure are presented at the same magnification to emphasize the difference in scale of the microstructures. The micrographs would be representative of, in decreasing order of DAS, a thick sand casting, a fairly thick permanent mold casting, and a very thin permanent mold casting. The degree to which the acicular Si phase can be broken up and rounded into an almost modified morphology is strongly dependent on the original scale of the microstructure. Coarse structures such as seen in Fig. 10(a) may not achieve any degree of modification thermally even after long solution

1074 / Nonferrous Alloy Castings

100 µm

(a)

Fig. 10

(b)

(c)

(a) A DAS of 70 mm and 12 h at 540  C (1010  F). (b) A DAS of 38 mm and 4 h at 540  C. (c) A DAS of 25 mm and 90 min at 540  C. Source: Ref 8

50 µm

50 µm

(a)

Fig. 11

100 µm

100 µm

(b) Strontium-modified A356 alloy solidified with a DAS of 24 mm. (a) F temper (as-cast), (b) T6 temper. Source: Ref 8

times. The progressively finer structures seen in Fig. 10(b) and (c) can achieve a nearly modified structure in relatively short time frames, however. The solution heat treatment of alloy A444, which contains no soluble phase, is justified solely by this phenomenon and its effect on ductility. Limited solubility also results in similar boundary physical changes in other insoluble intermetallics. The silicon phase in modified aluminum-silicon alloys is also impacted by solutionizing. Figure 11 shows the effect of the solutionizing step of the T6 temper on an A356 part. The fine fibrous microstructure is broken up and undergoes ripening. The silicon fibers are too fine to easily resolve before solutionizing, but coarsen to levels close to those seen in thermally modified samples after heat treatment, although they are more spherical and do not show residual small platelets as thermally modified castings are apt to. Note that chemical modification is largely insensitive to the thickness of the part and resultant scale of the microstructure. Coarsening is a function of time at temperature and is another good reason, besides economics, to find the optimum solutionizing time as opposed to being overly conservative.

Quenching. This discussion separates quenching as a distinct step in thermal practice leading to the metastable solution heat treated (T4) condition. Specific parameters can be associated with the heating of parts to achieve solution, and separate parameters apply to the steps required to achieve the highest postquench degree of retained solution. Rapid cooling from solution temperature to room temperature is essential, and it is possible to describe quenching as the most difficult and often least controlled step in thermal treatment. For optimum results, quench delay must be minimized. Although specifications often define quench delay limits, in practice, the shortest possible delay is observed. This may require specialized equipment, such as bottom-drop or continuous furnaces. Excessive delays result in temperature drop and the rapid formation of coarse precipitate in a temperature range in which the effects of precipitation are lost for hardening purposes. Even though castings are characteristically more tolerant of quench delay than wrought products based on structure and diffusion, excessive quench delays result in less than optimum strengthening potential.

Water is the quench medium of choice for aluminum alloys, and its temperature has a major effect on results. Most commercial quenching is accomplished in water near the boiling point, but room temperature, 65  C (150  F), and 80  C (180  F) have also become common standardized alternatives. Because higher potential strength is generally thought to be associated with the most rapid quenching and because corrosion and stress corrosion performance are usually enhanced by rapid quenching, it would appear that room-temperature water should routinely be employed. However, the selection of quench temperature becomes less obvious when the objective of quenching is to retain the highest possible degree of solution with the least warpage or distortion and the lowest level of induced residual stresses consistent with commercial or specified requirements. The key to the compromise between goals involving property development and the physical consequences of quenching is uniformity of heat extraction, which is in turn a complex function of the operable heat-extraction mechanism. Nucleate, vapor film, and convective boiling occur with dramatically different heatextraction rates at different intervals. Differences in section thickness, load density, positioning, and casting geometry also influence the results. As section thicknesses increase, the metallurgical advantages of quench rates obtained with water temperatures less than 65  C (150  F) diminish, but the cooling-rate advantage of the 65 versus the 100  C (150 versus 212  F) quench temperature is retained independently of section thickness. In addition to developing racking and loading methods that space and orient parts for the most uniform quenching, quenchant additions can be made:
To promote stable vapor film boiling by the

deposition of compounds on the surface of parts as they are submerged in the quench solution
To suppress variations in heat flux by increasing vapor film boiling stability

Aluminum and Aluminum Alloy Castings / 1075 through chemically decreased quench solution surface tension
To moderate quench rate for a given water temperature Quenching rates are also affected by the surface condition of the parts. More rapid quenching occurs with oxidized, stained, and rough surfaces, while bright, freshly machined, and etched surfaces quench more slowly. The common use of water as a quenching medium is largely based on its superiority in terms of quenching characteristics relative to other materials. Nevertheless, quenching has been accomplished in oil, salt baths, and organic solutions. For many compositions, fan or mist quenching is feasible as a means of obtaining dramatic reductions in residual stress levels with considerable sacrifices in hardening potential. Natural Aging. In general, two routes can be followed after the quenching operation. In the first case, the castings may be artificially aged immediately without hesitation. The second route involves a natural aging step at room temperature before being artificially aged. Most aluminum alloys age harden naturally to some extent after quenching; that is, properties change as a function of time at room temperature solely as a result of Guinier-Preston (GP) zone formation within the solid solution. The extent of change is highly alloy dependent. For example, room-temperature aging in alloys such as A356 and C355 occurs within 48 h, with insignificant changes taking place thereafter. Alloy 520, normally used in the T4 condition, age hardens over a period of years, and a number of Al-Zn-Mg alloys that are used without heat treatment exhibit rapid changes in properties over 3 or 4 weeks and harden at progressively reduced rates thereafter. Alloys such as those of the 206 family age naturally over a period of a few days to 2 weeks and are thereafter stable indefinitely. Parts that are to undergo subsequent artificial aging may undergo natural aging for a number of reasons, one of these being a desire for consistency in properties. It may be impossible, for a number of practical reasons, to ensure that all of the castings are fed into the artificial aging ovens with no delay. The natural aging stage provides a buffer to absorb any delays in the heat treat process so that all castings are subjected to the same thermal cycle. This is important when data from the castings are used to establish the foundry’s process capability in terms of property outcomes since variation in the thermal cycle will increase the variation of mechanical properties, leading to a reduction in the reported process capability. Another reason to include a natural aging step arises when the castings must undergo a postquench straightening operation to eliminate quench distortion. The best time to do this is with the castings in the T4 temper, the most ductile condition possible. The effects of natural aging with subsequent artificial aging are quite complex. Natural aging has been shown, for A356 alloy, to reduce the

yield stress slightly while improving the elongation for natural aging times up to 8 h followed by aging at 155  C (310  F). Natural aging for longer times can lead to recovery of yield stress at the expense of ductility. Aging at 170  C (340  F) has also been found to cancel the reduction in strength. Precipitation Heat Treating/Aging. In general, parts that have been solution heat treated and quenched display tensile properties and elongation superior to those of the as-cast (F) condition. However, the T4 condition is rarely used. Instead, the advantages of aging or precipitation hardening are obtained by thermal treatment following quenching. Among these advantages are increased strength and hardness with a corresponding loss in ductility, improved machinability, the development of more stable mechanical properties, and reduced residual stresses. It is through precipitation treatment that cast aluminum products may be most powerfully differentiated. The process of hardening is accelerated by artificial aging at temperatures ranging from approximately 95 to 260  C (200 to 500  F), depending on the alloy and the properties desired. In artificial aging, supersaturation, which characterized the room-temperature solution condition, is relieved by the precipitation of solute, which proceeds in stages with specific structural effects. At low aging temperatures or during transition at higher temperatures, the principal change is the diffusion of solute atoms to high-energy sites within the lattice, producing distortion of the lattice planes and forming concentrations of subcritical crystal nuclei. With continued exposure to aging temperature, these sites reach or fail to reach critical nucleation size, a stage leading to the formation of discrete particles displaying the identifiable crystallographic character of the precipitated phase. These transitional phase particles grow with an increase in coherency strains until, with sufficient time and temperature, interfacial bond strength is exceeded. Coherence is lost and with it the strengthening effects associated with precipitate formation and growth. Continued growth of the now equilibrium phase occurs without strength benefit, corresponding to the overaged condition. The practice to be employed in artificial aging depends entirely on the desired level of property development. Aging curves have been developed to facilitate process selection. The results of a large number of aging-curve studies are reflected in specification recommendations and industry references. The heat treater can reasonably predict the results of aging by reference to these curves. It should be noted that longer times at lower aging temperatures generally result in higher peakstrength values. The behavior of aging response or rate of property change as a function of time at peak strength points is also of interest. The flatter curves associated with lower aging temperatures allow greater tolerance in the effects of time-temperature variations. Unlike solution heat treatment, the time required to reach the aging

temperature may be significant, but is seldom included in age-cycle control. The energy imparted during heating to the precipitation-hardening temperature can be integrated into the control sequence to control results more accurately with minimum cycle time. The overaged T7 condition is less common than the T6 temper, but there are good reasons for its use. Precipitation hardening as practiced for the T6 condition results in reductions in residual stress levels imposed by quenching of 10 to 35%. By definition, overaging consists of carrying the aging cycle to a point beyond peak hardness, but it is most often conducted at higher temperatures than those used for the fully hardened condition. A substantial further decrease in residual stresses is associated with the highertemperature aging treatment. Furthermore, parts become more dimensionally stable as a result of more complete degrowthing, and increased stability in properties and performance is ensured when service involves exposure at elevated temperatures. One area where T7 heat treatments are common is for automotive powertrain components. Some of the secondary alloys used in the manufacture of cylinder heads and engine blocks will become very hard but also brittle if heat treated to a fully hard T6 condition. Underaged to maintain ductility, they may be prone to growth in service at engine temperatures. T7 tempers are an elegant solution to both problems as they yield a dimensionally stable component with acceptable ductility and sufficient strength. Annealing was originally assigned the designation T2, but is now known as the O temper. It is rarely employed, but is useful in providing parts with extreme dimensional and physical stability and the lowest practical level of residual stresses. The annealed condition is also characterized by extremely low strength and a correspondingly poor level of machinability. Typical annealing practices are for relatively short (2–4 h) exposures at a minimum temperature of 345  C (650  F). Higher-temperature practices are employed for more complete relaxation of residual stresses. The cooling rate from annealing temperatures must be controlled such that residual stresses are not reinduced and resolutionizing of elements with subsequent instability are avoided. Heat Treating Problems. Under less than optimum conditions, the measurable results of thermal practices may be other than those anticipated. When specification limits are not met, analytical procedures and judgments are applied to establish a corrective course of action based on evidence or assumptions concerning the cause of failure. Mechanical property limits statistically define normalcy for a given composition heat treated by specific practices. Chemical composition is a major variable in mechanical property development, and when mechanical properties have failed specification limits, chemistry is the logical starting point for investigation. The role of trace elements in mechanical property development is important because these

1076 / Nonferrous Alloy Castings elements are often not separately specified in alloy specifications except as “others each” or “others total.” Sodium and calcium are embrittling in 5xx alloys. Low-melting elements such as lead, tin, and bismuth may under some conditions form embrittling intergranular networks with similar effects. Elements such as tin can change the precipitation heat treating response when present in excessive quantities in coppercontaining alloys. Insoluble impurity elements are generally responsible for decreases in elongation. Low concentrations of soluble elements in heat treatable compositions naturally result in the more frequent distribution of mechanical property values in the lower specification range. Element relationships such as Cu-Mg, Si-Mg, Fe-Si, Fe-Mn, and Zn-Cu-Mg are also important considerations in defining the causes of abnormal mechanical property response to thermal treatment. A second important consideration is that of casting soundness. Unsound castings will not consistently meet specified mechanical property or other limits sensitive to structural integrity. All defects adversely affect strength and elongation. Shrinkage, hydrogen porosity, cracks, inclusions, and other casting-related defects influence mechanical properties adversely, and their effects should be considered before addressing the possibility of heat treatment problems. The quality of solution heat treatment can be assessed in several ways. There is, of course, the rounding effect on insoluble phases, which serves as evidence of elevated-temperature exposure. The elimination of coring in many alloys is another indication of elevated-temperature treatment. The effective solution of soluble phases can be determined microscopically. Undissolved solute can be distinguished from the appearance of precipitate that forms at high temperatures and results from quench delay or an inadequate or incomplete quench. There is also a tendency for the precipitates formed as a result of quench delay or inadequate quench to concentrate at grain boundaries as opposed to more normal distribution through the microstructure for properly solution heat treated and aged material. Although the overaged condition is microscopically apparent, underaging is difficult to assess because of the submicroscopic nature of transitional precipitation. Evidence of acceptable aging practice is best obtained from aging-furnace records, which might indicate errors in the age cycle. Underaging can be corrected by additional aging, but for all other heat treatment aberrations except those associated with objectionable conditions such as high-temperature oxidation or eutectic melting, resolution heat treatment is an acceptable corrective action despite myths to the contrary. Recent evidence indicates that hypoeutectic aluminum-silicon should be resolutionized no more than twice, however. Eutectic melting occurs when the eutectic melting temperature is exceeded, resulting in the characteristic rosettes

of resolidified eutectic and/or intergranular eutectic. High-temperature oxidation is a misnamed condition of hydrogen diffusion that affects surface layers during elevated-temperature treatment. This condition can result from grease, oil, or moisture contamination on casting surfaces or in the furnace atmosphere and is sometimes aggravated by sulfur (as in heat treatment furnaces also used for magnesium alloy castings) or other furnace refractory contamination. In the case of resolutionizing aluminum-copper alloys, it is essential that re-solution heat treatment be conducted at a temperature equivalent to or higher than the original solutionizing temperature to ensure effective re-solution. However, when the results of repeated heat treatment prove unsatisfactory, it should be apparent that other conditions are responsible for mechanical property failure.

Quality Control The most effective method of determining the combined consequences of chemistry, material condition, and heat treatment is the determination of tensile strength, yield strength, and elongation. This is often done by testing separate cast tensile specimens that accompany the parts and bars through the heat treat cycle. These values most conclusively indicate the acceptability of the alloy being used and heat treatment process being applied. Caution should be exercised though since the inevitable DAS difference between parts and bars, together with any other process differences, will make the properties obtained from separately cast bars differ from those obtained from production bars. Hardness can be established as an acceptance criterion through negotiation, but it is less adequate in aluminum alloys than in other metal systems for control purposes. Hardness in aluminum corresponds only approximately to yield strength, and although hardness can be viewed as an easily measurable indicator of material condition, it is not normally accurate enough to serve as a guaranteed limit. Electrical conductivity only approximates material condition in cast or cast and heat treated structures, and it remains excessively variable for most control purposes.

Stability Because thermal treatment affects the stability of mechanical properties and directly influences residual stress levels, additional discussion of these two considerations is warranted. Stability is defined as the condition of unchanging structural and physical characteristics as a function of time under specific conditions of application. As previously mentioned, the metastable T4 condition is subject to hardening (extensive in some alloys and limited in others) at room and higher temperatures. Under service conditions involving temperatures

greater than room temperature, significant physical and mechanical changes are to be expected. Although these changes may be acceptable for a given application, changes in susceptibility to corrosion and stress corrosion, which can be associated with transitional states in some alloy systems, must be carefully considered. The most stable conditions obtainable are associated with the (in order of decreasing stability) annealed, overaged, and as-cast conditions. Underaged and solution heat treated parts are least stable. Residual stresses are caused by differing rates of cooling, especially postsolidification cooling, quenching from solution heat treatment temperature, and drastic changes in temperature at any intermediate step. Residual stress development depends on the differential rate of cooling, section thickness, and material strength. Stresses imposed by quenching from the solution heat treatment temperature are much more important than casting stresses or stresses normally obtained in conventional processing. Decreasing the severity of the quench from solution heat treatment results in a lower level of residual stresses but with correspondingly decreased material strength. Air quenching may provide a useful compromise in applications requiring unusual dimensional stability. Residual stresses can also be relaxed by exposure to elevated temperature followed by slow cooling, or by plastic deformation. Plastic deformation that is routinely practiced for stress relief in wrought products has little application in the complex designs of engineered products such as castings; therefore, stress relief becomes more exclusively a function of postquench thermal treatment. Overaging results in significant reductions in residual stresses, and annealing provides a practical minimum in residual stress levels.

Natural versus Artificial Property Development When foundrymen work closely with design engineers, alloys and tempers can be selected that are capable of meeting both the manufacturing objectives of the foundry and the engineering criteria at the lowest possible cost. A frequent point of discussion is whether a superior selection might be a naturally hardening alloy (a composition that hardens naturally at room temperature after casting) or an alloy requiring heat treatment for mechanical property development. Unfortunately, room-temperature hardening alloys, which offer a wide range of attractive properties, are also characterized by relatively inferior casting characteristics so that final material selection may become an issue of foundry versus application considerations. The selection of a heat treatable composition with good casting characteristics in combination with effective heat treatment often offers significant advantages in cost, performance, uniformity, and reliability.

Aluminum and Aluminum Alloy Castings / 1077

Welding Foundry welding is employed to salvage scrap castings primarily by correcting surface and dimensional defects or irregularities. The inert gas shielded arc welding processes are normally used, rather than the processes using flux mixtures or flux-coated filler rods. The welding techniques used for castings are similar to those for wrought products. However, special consideration must be given to the thicker-surface aluminum oxide film and the void content of castings. The quality of weldments is closely related to the soundness of the casting in the weld area. The quality of conventional welds is generally very poor in castings displaying hydrogen porosity and in most die castings. Welding Method. Casting flaws or areas requiring correction are usually small; therefore, gas tungsten arc welding (GTAW) is normally used. Gas metal arc welding can be used if economics and operating conditions indicate that satisfactory results can be achieved. Equipment for GTAW. Alternating-current welding transformers that deliver a balanced wave and incorporate high-frequency stabilization minimize the production of tungsten inclusions and produce the highest-quality welds. In general, the alternating-current/direct-current (ac/dc) and unbalanced wave types of power supply tend to spit tungsten. Tungsten inclusions in a weld are considered as detrimental as other casting defects of similar dimension. Pure tungsten is preferred for facilitating a stable arc and minimizing tungsten spitting. The second choice is zirconiated tungsten. Shielding Gas. Argon and mixtures of argon and helium gases are used to gas tungsten arc weld aluminum castings. When a gas mixture is used, helium usually constitutes 25 to 50% of the total mixture. The helium helps produce a hotter arc at a particular current setting, clean up the weld puddle if casting quality is a problem, and reduce the size of gas porosity voids in the weld. A Y-tube connector can be used to blend the appropriate volumes of gases directly from the individual gas bottles and flowraters rather than maintaining a supply of a variety of premixed gas ratios. Preparation for Welding. Flaws should be removed by pneumatic chipping or hand-tool milling. This operation should be performed without the use of a lubricant, if possible. The leading edge of the chipping tool should be U-shaped rather than V-shaped to facilitate welding. The milling or deburring tool should have a coarse cutting bit to prevent it from becoming loaded with aluminum. Cavity Preparation. The cavity that is to be repaired should be smooth, that is, without exaggerated ridges, and the sides of the cavity should be chamfered rather than being a steep-sided hole. Chamfering promotes good fusion in the casting repair. It also provides a means for any gas in the casting that is released during welding to become uniformly distributed throughout the

weld rather than becoming concentrated as linear porosity around the edge of the weld. Cleaning. Oil and grease, if present, should be removed with a locally applied solvent or by a vapor degreasing operation. The type of solvent used (toluene, for example) should not leave a residue that may gasify the weld or produce harmful fumes during the welding operation. Etchants should not be used to clean the casting before welding unless absolutely necessary. Etchants are usually corrosive and tend to be retained in the surface pores of the casting. They can also contribute to gas porosity in the weld if the casting surface containing these trapped liquids becomes involved in the welding operation. Impregnated Castings. Castings should not be impregnated for pressure tightness before welding. If an impregnated casting must be welded, the impregnant can be partially removed by prolonged heating at a temperature of 150 to 205  C (300 to 400  F). Surface Preparation. After cleaning with a solvent and just before welding, the area to be welded and the immediately adjacent area (a 13 mm, or 1/2 in., band around the weld area) should be hand brushed with a stainless steel bristle brush to reduce the thickness of the aluminum oxide on the casting surface and to remove any sand or foreign material. In cases in which the as-cast surface is to be welded, the surface should be scraped or scarfed to a depth of 1.5 mm (0.06 in.) or more to overcome the effects of the oxidized, rough casting surface. The filler metal is often the same alloy as the casting being welded. Filler material can be cast in rod-shaped molds when production castings are being poured. If the castings are to be heat treated after welding, the filler material selected may be different from that used if ease of welding is the only consideration. The size of the filler rod used for welding is usually the largest one that will not freeze the weld puddle. To minimize hot short weld cracking, weld composition should be maintained as closely as possible to a ratio of 70% filler alloy to 30% base alloy. Preheating. Although it can be beneficial in reducing the size of gas porosity in the weld, preheating of the casting is not necessary before making a weld repair. The ability to maintain a puddle is the primary consideration. In some cases, however, preheating may be necessary to overcome cracking, distortion, or welding speed problems. Each case is different, and the need for preheating depends on the alloy, section thickness, casting size and geometry, residual stress, and potential thermal gradients between hot and cool areas. When it is required, preheating can be accomplished by placing the entire casting in a furnace, or if the size of the casting does not permit this method, the casting can be preheated with a gas torch. The casting is usually heated to 150 to 205  C (300 to 400  F). Welding Technique. Weld quality is directly related to casting quality. The general problems encountered in making good welds

with castings are the same as those encountered with wrought alloys except that the aluminum oxide on casting surfaces is thicker than on wrought alloys and must be reduced to diminish welding problems. A good general rule of welding technique is to “get in and then get out quickly.” To accomplish fast melting and weld puddle cleanup, helium shielding gas in a mixture with argon gas should be used. Cracks cannot be melted out, however, because the melting point of the aluminum oxide that lines the cracks is prohibitively high. The cracks must be chipped out and the cavity filled with weld metal. Postweld Heat Treatment. Castings that require heat treatment should be heat treated after welding. Castings that have been heat treated and then welded will suffer localized loss of tensile properties and an increase in localized residual stresses . However, postweld heat treatment will restore their properties if the proper filler alloy has been used in welding.

Properties of Aluminum Casting Alloys The physical and mechanical properties of aluminum casting alloys are well documented, and example data are listed in Tables 5 to 9 in this article (see also the Selected References). The discussion in this article is not intended as a comprehensive discussion of all the commonly tested properties of aluminum casting alloys (for example, tensile strength). Rather, certain properties, such as fatigue resistance and corrosion resistance, are discussed in terms of testing, problems in testing, and how the properties of cast aluminum may differ from those of wrought alloys. Melt fluidity, shrinkage during cooling, and hot cracking are also discussed. It must be noted that the data given in Tables 5 to 9 are useful for comparing alloys, but they should not be used for design purposes. Properties for design must be obtained from pertinent specifications or design standards and/or by negotiation with the producer.

Fatigue When a part is subjected to cyclical loads, cracks may develop in highly stressed areas and propagate with time. Such cracks may occur at stress levels below the characteristic ultimate strength of the material. Failure by such a mechanism is termed fatigue failure. Fatigue cracks frequently initiate at stress raisers such as abrupt changes in section, holes, notches, machining gouges, stamped emblems, the edges of a weld bead, porosity, surface oxide concentrations, dross inclusions, or other surface discontinuities. Fatigue Testing. Separately cast test bars are now seldom used, although they do prevent the destruction of production parts and, hence, help to hold down costs. Unfortunately, the fatigue

1078 / Nonferrous Alloy Castings Table 5

Typical physical properties of aluminum casting alloys Approximate melting range

Density(b) Alloy

201.0 206.0 A206.0 224.0 238.0 240.0 242.0

295.0 296.0

308.0 319.0 324.0 332.0 333.0

336.0 354.0 355.0

C355.0 356.0

A356.0 357.0 A357.0 358.0 359.0 360.0 A360.0 364.0 380.0 A380.0 384.0 390.0 392.0 413.0 A413.0 443.0

A444.0 511.0 512.0 513.0 514.0 518.0 520.0 535.0 A535.0 B535.0 705.0 707.0 710.0 711.0 712.0 713.0

Temper and product form(a)

T6 (S) T7 (P)

T62 (S) F (P) F (S) O (S) T77 (S) T571 (P) T61 (P) T4 (S) T62 (S) T4 (P) T6 (P) T62 (S) F (P) F (S) F (P) F (P) T5 (P) F (P) T5 (P) T6 (P) T7 (P) T551 (P) F (P) T51 (S) T6 (S) T61 (S) T7 (S) T6 (P) T61 (S) T51 (S) T6 (S) T7 (S) T6 (P) T6 (S) T6 (S) T6 (S) T6 (S) T6 (S) F (D) F (D) F (D) F (D) F (D) F (D) F (D) T5 (D) F (P) F (D) F (D) F (S) O (S) F (D) F (P) F (S) F (S) F (P) F (S) F (D) T4 (S) F (S) F (D) F (S) F (S) F (S) F (S) F (P) F (S) F (S)

Specific gravity(b)

kg/m3

lb/ln3

2.80 2.80 2.8 2.8 2.81 2.95 2.78 2.81 2.81 2.81 2.81 2.81 2.81 2.80 2.80 2.80 2.79 2.79 2.79 2.67 2.76 2.77 2.77 2.77 2.77 2.72 2.71 2.71 2.71 2.71 2.71 2.71 2.71 2.68 2.68 2.68 2.68 2.69 2.68 2.68 2.68 2.67 2.68 2.68 2.63 2.76 2.76 2.70 2.73 2.73 2.64 2.66 2.66 2.69 2.69 2.69 2.68 2.66 2.65 2.68 2.65 2.53 2.57 2.62 2.54 2.62 2.76 2.77 2.81 2.84 2.82 2.84

2796 2796 2796 2796 2824 1938 2768 2823 2823 2823 2823 2823 2823 2796 2796 2796 2796 2796 2796 2658 2768 2768 2768 2768 2768 2713 2713 2713 2713 2713 2713 2713 2713 2685 2685 2685 2685 2713 2713 2713 2658 2685 2685 2685 2630 2740 2740 2713 2740 2740 2630 2657 2657 2685 2685 2685 2685 2657 2657 2685 2657 2519 2574 2519 2547 2630 2768 2768 2823 2851 2823 2879

0.101 0.101 0.101 0.101 0.102 0.107 0.100 0.102 0.102 0.102 0.102 0.102 0.102 0.101 0.101 0.101 0.101 0.101 0.101 0.096 0.100 0.100 0.100 0.100 0.100 0.098 0.098 0.098 0.098 0.098 0.098 0.098 0.098 0.097 0.097 0.097 0.097 0.098 0.098 0.098 0.096 0.097 0.097 0.097 0.095 0.099 0.099 0.098 0.099 0.099 0.095 0.096 0.096 0.097 0.097 0.097 0.097 0.096 0.096 0.097 0.096 0.091 0.093 0.091 0.092 0.095 0.100 0.100 0.102 0.103 0.102 0.104



C

570–650 570–650 570–650 570–650 550–645 510–600 515–605 530–635 525–635 525–635 525–635 520–645 520–645 520–630 520–630 520–630 520–615 520–605 520–605 545–605 520–580 520–585 520–585 520–585 520–585 540–570 540–600 550–620 550–620 550–620 550–620 550–620 550–620 560–615 560–615 560–615 560–615 560–610 560–615 555–610 560–600 565–600 570–590 570–590 560–600 520–590 520–590 480–580 510–650 510–650 550–670 575–585 575–585 575–630 575–630 575–630 575–630 590–640 590–630 580–640 600–640 540–620 450–600 550–630 550–620 550–630 600–640 585–630 600–650 600–645 600–640 595–630



F

1060–1200 1060–1200 1060–1200 1060–1200 1020–1190 950–1110 960–1120 990–1180 980–1180 980–1180 980–1180 970–1190 970–1190 970–1170 970–1170 970–1170 970–1140 970–1120 970–1120 1010–1120 970–1080 970–1090 970–1090 970–1090 970–1090 1000–1060 1000–1110 1020–1150 1020–1150 1020–1150 1020–1150 1020–1150 1020–1150 1040–1140 1040–1140 1040–1140 1040–1140 1040–1130 1040–1140 1030–1130 1040–1110 1050–1110 1060–1090 1060–1090 1040–1110 970–1090 970–1090 900–1080 950–1200 950–1200 1020–1240 1070–1090 1070–1090 1070–1170 1070–1170 1070–1170 1070–1170 1090–1180 1090–1170 1080–1180 1110–1180 1000–1150 840–1110 1020–1170 1020–1150 1020–1170 1110–1180 1090–1170 1110–1200 1110–1190 1110–1180 1100–1170

Electrical conductivity, % IACS

27–32 32–34 ... ... 30 25 23 44 38 34 33 35 35 33 33 33 37 27 28 34 26 26 29 29 35 29 32 43 36 37 42 39 39 43 39 40 41 40 39 40 39 35 37 37 30 27 27 23 25 24 22 39 39 37 42 37 41 36 38 34 35 24 21 23 23 24 25 25 35 40 40 37

Thermal conductivity at 25  C (77  F), W/m K

0.29 0.29 0.29 0.29 0.28 0.25 0.23 0.40 0.36 0.32 0.32 0.33 0.34 0.32 0.32 0.32 0.34 0.27 0.28 0.37 0.25 0.25 0.29 0.28 0.34 0.28 0.30 0.40 0.34 0.35 0.39 0.36 0.35 0.40 0.36 0.37 0.37 0.36 0.36 0.38 0.36 0.33 0.35 0.35 0.29 0.26 0.26 0.23 0.32 0.32 0.22 0.37 0.37 0.35 0.39 0.34 0.38 0.34 0.35 0.32 0.33 0.24 0.21 0.24 0.24 0.23 0.25 0.25 0.33 0.38 0.38 0.37



Coefficient of thermal expansion, per C  106 (per  F  106)

20–100  C (68–212  F)

20–300  C (68–570  F)

34.7 (19.3) 44.5 (24.7) 34.7 (19.3) 44.5 (24.7) ... ... ... ... 21.4 (11.9) 22.1 (12.3) ... 22.1 (12.3) 22.5 (12.5) 22.5 (12.5) 22.9 (12.7) 22.9 (12.7) 22.0 (12.2) 22.0 (12.2) ... 21.4 (11.9) 21.6 (12.0) 21.6 (12.0) 21.4 (11.9) 20.7 (11.5) 20.7 (11.5) 20.7 (11.5) 20.7 (11.5) 20.7 (11.5) 18.9 (10.5) 20.9 (11.6) 22.3 (12.4) 22.3 (12.4) 22.3 (12.4) 22.3 (12.4) 22.3 (12.4) 22.3 (12.4) 21.4 (11.9) 21.4 (11.9) 21.4 (11.9) 21.4 (11.9) 21.4 (11.9) 21.4 (11.9) 21.4 (11.9) 21.4 (11.9) 20.9 (11.6) 20.9 (11.6) 21.1 (11.7) 20.9 (11.6) 21.2 (11.8) 21.1 (11.7) 20.3 (11.3) 18.5 (10.3) 18.0 (10.0) 18.5 (10.3) 20.5 (11.4) ... 22.1 (12.3) ... ... 21.8 (12.1) 23.6 (13.1) 22.9 (12.7) 23.9 (13.3) 23.9 (13.3) 24.1 (13.4) 25.2 (14.0) 23.6 (13.1) 24.1 (13.4) 24.5 (13.6) 23.6 (13.1) 23.8 (13.2) 24.1 (13.4) 23.6 (13.1) 23.6 (13.1) 23.9 (13.3)

22.9 (12.7) 24.3 (13.5) ... 23.6 (13.1) 24.5 (13.6) 24 5 (13.6) 24.8 (13.8) 24.8 (13.8) 23.9 (13.3) 23.9 (13.3) ... 22.9 (12.7) 24.1 (13.4) 24.1 (13.4) 23.2 (12.9) 22.3 (12.4) 22.7 (12.6) 22.7 (12.6) 22.7 (12.6) 22.7 (12.6) 20.9 (11.6) 22.9 (12.7) 24.7 (13.7) 24.7 (13.7) 24.7 (13.7) 24.7 (13.7) 24.7 (13.7) 24.7 (13.7) 23.4 (13.0) 23.4 (13.0) 23.4 (13.0) 23.4 (13.0) 23.4 (13.0) 23.4 (13.0) 23.6 (13.1) 23.4 (13.0) 22.9 (12.7) 22.9 (12.7) 22.9 (12.7) 22.9 (12.7) 22.5 (12.5) 22.7 (12.6) 22.1 (12.3) ... ... 20.2 (11.2) 22.5 (12.5) ... 24.1 (13.4) ... ... 23.8 (13.2) 25.7 (14.3) 24.8 (13.8) 25.9 (14.4) 25.9 (14.4) 26.1 (14.5) 27.0 (15.0) 26.5 (14.7) 26.1 (14.5) 26.5 (14.7) 25.7 (14.3) 25.9 (14.4) 26.3 (14.6) 25.6 (14.2) 25.6 (14.2) 25.9 (14.4)

(continued) (a) S, sand cast; P. permanent mold: D. die cast (b) The specific gravity and weight data in this table assume solid (void-free) metal Because some porosity cannot be avoided in commercial castings, their specific gravity or weight is slightly less than the theoretical value.

Aluminum and Aluminum Alloy Castings / 1079 Table 5

(continued) Approximate melting range

Density(b) Alloy

Temper and product form(a)

850.0 851.0 852.0

T5 (S) T5 (S) T5 (S)

Specific gravity(b)

kg/m3

lb/ln3

2.87 2.83 2.88

2851 2823 2879

0.103 0.102 0.104



C

225–650 230–630 210–635

F

Electrical conductivity, % IACS

Thermal conductivity at 25  C (77  F), W/m K

440–1200 450–1170 410–1180

47 43 45

0.44 0.40 0.42





Coefficient of thermal expansion, per C  106 (per  F  106)

20–100  C (68–212  F)

20–300  C (68–570  F)

... 22.7 (12.6) 23.2 (12.9)

... ... ...

(a) S, sand cast; P. permanent mold: D. die cast. (b) The specific gravity and weight data in this table assume solid (void-free) metal Because some porosity cannot be avoided in commercial castings, their specific gravity or weight is slightly less than the theoretical value.

Table 6 Ratings of castability, corrosion resistance, machinabilty, and weldability for aluminum casting alloys I, best; 5, worst. Individual alloys may have different ratings for other casting processes. Alloy

Resistance to hot cracking(a)

Sand casting alloys 201.0 4 208.0 2 213.0 3 222.0 4 240.0 4 242.0 4 A242.0 4 295.0 4 319.0 2 354.0 1 355.0 1 A356.0 1 357.0 1 359.0 1 A390.0 3 A443.0 1 444.0 1 511.0 4 512.0 3 514.0 4 520.0 2 535.0 4 A535.0 4 B535.0 4 705.0 5 707.0 5 710.0 5 711.0 5 712.0 4 713.0 4 771.0 4 772.0 4 850.0 4 851.0 4 852.0 4

Pressure tightness

Fluidity (b)

Shrinkage tendency

Corrosion resistance (c)

3 2 3 4 4 3 4 4 2 1 1 1 1 1 3 1 1 5 4 5 5 5 5 5 4 4 3 4 4 4 4 4 4 4 4

3 2 2 3 3 4 3 4 2 1 1 1 1 1 3 1 1 4 4 4 4 4 4 4 4 4 4 5 3 3 3 3 4 4 4

4 2 3 4 4 4 4 3 2 1 1 1 1 1 3 1 1 5 4 5 5 5 4 4 4 4 4 4 3 4 3 3 4 4 4

4 4 4 4 4 4 4 3 3 3 3 2 2 2 2 2 2 1 1 1 1 1 1 1 2 2 2 3 3 2 2 2 3 3 3

1 3 2 1 3 2 2 2 3 3 3 3 3 3 4 4 4 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

2 3 2 3 4 3 3 2 2 2 2 2 2 1 2 4 1 4 4 4 5 3 4 4 4 4 4 3 4 3 ... ... 4 4 4

3 2 3

4 3 4

4 4 4

1 2 1

2 2 3

Permanent mold casting alloys 201.0 4 3 213.0 3 3 222.0 4 4

Machinability Weldability (d) (e)

Alloy

238.0 240.0 296.0 308.0 319.0 332.0 333.0 336.0 354.0 355.0 C355.0 356.0 A356.0 357.0 A357.0 359.0 A390.0 443.0 A444.0 512.0 513.0 711.0 771.0 772.0 850.0 851.0 852.0

Resistance to hot cracking(a)

Pressure tightness

Fluidity (b)

Shrinkage tendency

Corrosion resistance (c)

2 4 4 2 2 1 1 1 1 1 1 1 1 1 1 1 2 1 1 3 4 5 4 4 4 4 4

3 4 3 2 2 2 1 2 1 1 1 1 1 1 1 1 2 1 1 4 5 4 4 4 4 4 4

2 3 4 2 2 1 2 2 1 1 1 1 1 1 1 1 2 2 1 4 4 5 3 3 4 4 4

2 4 3 2 2 2 2 3 1 2 2 1 1 1 1 1 3 1 1 4 4 4 3 3 4 4 4

4 4 4 4 3 3 3 3 3 3 3 2 2 2 2 2 2 2 2 1 1 3 2 2 3 3 3

2 3 3 3 3 4 3 4 3 3 3 3 3 3 3 3 4 5 3 2 1 1 1 1 1 1 1

3 4 4 3 2 2 3 2 2 2 2 2 2 2 2 1 2 1 1 4 5 3 ... ... 4 4 4

1 1 2 1 2 2 2 2 3 5 5

2 2 1 2 2 1 2 1 3 5 5

2 2 3 5 4 3 2 2 2 1 1

3 3 4 3 3 3 4 4 5 2 1

4 4 3 4 4 4 2 4 4 4 4

... ... ... ... ... ... ... ... ... ... ...

Die casting alloys 360.0 1 A360.0 1 364.0 2 380.0 2 A380.0 2 384.0 2 390.0 2 413.0 1 C443.0 2 515.0 4 518.0 5

Machinability Weldability (d) (e)

(a) Ability of alloy to withstand stresses from contraction while cooling through hot short or brittle temperature range. (b) Ability of liquid alloy to flow readily in mold and to fill thin sections. (c) Based on resistance of alloy in standard salt spray test. (d) Composite rating based on ease of cutting, chip characteristics, quality of finish, and tool life. (e) Based on ability of material to be fusion welded with filler rod of same alloy.

resistance of such specimens is usually higher than that of the cast parts they represent because of factors such as relative soundness and the presence after heat treatment of beneficial residual surface compressive stresses. It may even be impossible to produce these by the same process used to produce the parts in the case of processes requiring expensive tooling. It is therefore generally considered appropriate to machine specimens from critical casting areas when fatigue properties are important. Cast-on coupons or extensions may be another option. In the case of high-volume parts

such as cast automotive wheels and suspension parts, regular full part testing on various fixtures and bogies is frequently considered a normal part of the ongoing quality-control plan with set failure mode and effect analysis responses called for if a part fails unexpectedly or prematurely. The frequency of testing may be reduced as a history of consistently positive test results is accumulated. Recent work on crack propagation during fatigue of foundry alloys has found that residual stresses significantly impact the fatigue crack advance per cycle, or da/dn, curves, and can

result in erroneous values if not compensated for either physically or numerically. The fatigue strength of castings is normally lower than that of wrought materials. However, discontinuities that considerably reduce the fatigue strength of a wrought material (for example, holes) have much less effect on the fatigue performance of a cast part (that is, castings are less notch sensitive than wrought materials). Porosity is more commonly encountered in castings than in wrought forms. As well, the microstructures of castings are generally more heterogeneous than wrought forms as has been

1080 / Nonferrous Alloy Castings Table 7 Typical mechanical properties of aluminum casting alloys Ultimate tensile strength Alloy

Temper

0.2% offset yield strength

Elongation in 50 mm (2 in.), %

Hardness, HB(a)

MPa

ksi

MPa

ksi

414 448 467 354 145 165 186 283 421 380 235 214 186 221 207 214 221 250 283 186 207 250 159 193 241 269 264 241 269 164 172 228 234 193 159 179 278 207 172 179 345 278 317 179 179 278 250 131 145 159 145 138 172 331 250 241 241 241 138 138 186

60 65 68 51 21 24 27 41 61 55 34 31 27 32 30 31 32 36 41 27 30 36 23 28 35 39 38 35 39 24 25 33 34 28 23 26 40 30 25 26 50 40 46 26 26 40 36 19 21 23 21 20 25 48 36 35 35 35 20 20 27

255 379 414 250 97 103 138 276 331 276 200 217 124 207 159 ... 110 165 221 124 179 164 83 159 172 241 250 200 200 124 138 164 207 145 83 124 207 138 90 117 296 234 248 179 179 278 250 55 62 62 83 90 83 179 124 172 172 172 76 76 152

37 55 60 36 14 15 20 40 48 40 28 30 18 30 23 ... 16 24 32 18 26 24 12 23 25 35 26 29 29 18 20 24 30 21 12 18 30 20 13 17 43 34 36 26 26 40 36 8 9 9 12 13 12 26 18 25 25 25 11 11 22

17.0 8.0 5.5 7.0 2.5 1.5 1.0 <0.5 4.0 10.0 1.0 0.5 1.0 0.5 2.0 2.0 8.5 5.0 2.0 2.0 1.5 2.0 3.0 1.5 3.0 1.0 0.5 1.5 5.0 6.0 2.0 3.5 2.0 3.5 6.0 3.0 6.0 3.0 5.0 3.0 2.0 3.0 3.0 <1.0 <1.0 <1.0 <1.0 8.0 9.0 12.0 3.0 2.0 9.0 16.0 9.0 5.0 5.0 5.0 8.0 5.0 2.0

... 130 ... ... 55 70 80 115 ... 123 90 ... 70 85 75 ... 60 75 90 70 80 80 ... 65 80 90 85 75 85 ... 60 70 75 60 ... ... 75 ... ... ... 90 60 85 100 100 140 115 40 ... ... 50 50 50 75 65 75 75 74 45 45 65

Permanent moid casting alloys 201.0 T43 414 T6 448 T7 469

60 65 68

255 379 414

37 55 60

17.0 8.0 5.0

... 130 ...

Sand casting alloys 201.0 T43 T6 T7 A206.0 T4 208.0 F 213.0 F 222.0 O T61 T62 224.0 T72 240.0 F 242.0 F O T571 T77 A242.0 T75 295.0 T4 T6 T62 319.0 F T5 T6 355.0 F T51 T6 T61 T7 T71 C355.0 T6 356.0 F T51 T6 T7 T71 A356.0 F T51 T6 T71 357.0 F T51 T6 T7 A357.0 T6 A390.0 F T5 T6 T7 443.0 F A444.0 F T4 511.0 F 512.0 F 514.0 F 520.0 T4 A535.0 F 710.0 F 712.0(h) F 713.0(h) F 850.0 T5 851.0 T5 852.0 T5

Ultimate tensile strength

0.2% offset yield strength

Elongation in 50 mm (2 in.), %

Hardness, HB(a)

Alloy

Temper

MPa

ksi

MPa

ksi

A206.0

T4 T7 F T52 T551 T65 F T571 T61 T63 T7 T4 T6 T7 F F T6 F T5 T62 T5 F T5 T6 T7 T551 T65 F T51 T6 T7 T61 F T51 T6 T6 T62 F T5 T6 T7 F F T4 F F T5 T5 T5

431 436 207 241 255 331 207 276 324 476 278 255 276 270 193 234 276 207 248 310 248 234 234 290 255 248 324 179 186 262 221 283 193 200 359 359 345 200 200 310 262 159 165 159 186 241 159 138 221

62 63 30 35 37 48 30 40 47 69 62 37 40 39 28 34 40 30 36 45 36 34 34 42 37 36 47 26 27 38 32 41 28 29 52 52 50 29 29 45 38 23 24 23 27 35 23 20 32

264 347 165 214 241 248 165 234 290 414 359 131 179 138 110 131 186 110 179 269 193 131 172 207 193 193 296 124 138 186 165 207 103 145 296 290 290 200 200 310 262 62 76 69 110 124 76 76 159

38 50 24 31 35 36 24 34 42 60 52 19 26 20 16 19 27 16 26 39 28 19 25 30 28 28 43 18 20 27 24 30 15 21 43 42 42 29 29 45 38 9 11 10 16 18 11 11 23

17.0 11.7 1.5 1.0 <0.5 <0.5 1.5 1.0 0.5 6.0 9.0 9.0 5.0 4.5 2.0 2.5 3.0 4.0 3.0 3.0 1.0 2.0 1.0 1.5 2.0 0.5 0.5 5.0 2.0 5.0 6.0 10.0 6.0 4.0 5.0 5.0 5.5 <1.0 <1.0 <1.0 <1.0 10.0 13.0 21.0 7.0 8.0 12.0 5.0 5.0

... ... 85 100 115 140 100 105 110 ... ... 75 90 80 70 85 95 70 90 105 105 90 100 105 90 105 125 ... ... 80 70 90 ... ... 100 100 ... 110 110 145 120 45 44 45 60 70 45 45 70

F F F F F F F T5 F F F F F F F

324 317 296 331 324 324 279 296 290 296 241 228 276 283 310

47 46 43 48 47 47 40.5 43 42 43 35 33 40 41 45

172 165 159 165 159 172 241 265 262 145 110 110 152 ... 186

25 24 23 24 23 25 35 38.5 38 21 16 16 22 ... 27

3.0 5.0 7.5 3.0 4.0 1.0 1.0 1.0 <0.5 2.5 3.5 9.0 10.0 10.0 8.0

75 75 ... 80 75 ... 120 ... ... 80 80 50 ... ... 80

213.0 222.0

238.0 242.0 249.0 296.0

308.0 319.0 324.0

332.0 333.0

336.0 356.0

A356.0 357.0

A357.0 359.0 A390.0

443.0 A444.0 513.0 711.0 850.0 851.0 852.0 Die casting alloys 360.0 A360.0 364.0 380.0 A380.0 384.0 390.0 392.0 413.0 A413.0 443.0 513.0 515.0 518.0

(a) 500 kg (1100 lb) load on 10 mm (0.4 in.) ball.

seen earlier in the DAS/elongation relationship. Figure 12 gives an example of the impact of changes in both pore size and secondary dendrite arm spacing on fatigue life in A356 alloy. Modern automotive requirements frequently call for more than R.R. Moore S/N curves or

set fatigue lives at a given stress or strain amplitude. Table 9 shows two examples of full data sets, both monotonic and cyclical, collected for safety critical suspension parts. Premium engineered castings combine a number of conventional casting procedures

in a selective manner to provide premium strength, soundness, dimensional tolerances, surface finish, or a combination of these characteristics. As a result, premium engineered castings typically display higher fatigue resistance than conventional castings. Improved surface

Aluminum and Aluminum Alloy Castings / 1081 Table 8

Typical values for squeeze cast and semisolid metal cast (SSM) alloys Yield Stress MPa

Tensile Stress ksi

MPa

ksi

Elongation, %

Hardness

Squeeze Cast A206-T4 356-T61 A356-T61 B356-T61 A356-T6 357-T6

248–269 231.7  5.5 246.8  11.7 222.7  7.6 145–165 241–262

36–39 33.6  0.8 35.8  1.7 32.3  1.1 21–24 35–38

400–421 371.2  1.4 299.9  6.9 302.7  6.2 255–276 324–338

58–61 46  0.2 43.5  1.0 43.9  0.9 37–40 47–49

21–26 11.9  1.9 6.6  2.0 17.6  1.8 13–17 8–10

63–75 HRB 113  2 HV 117  4 HV 102  HV 40–56 HRB 52–68 HRB

SSM 319-F 319-T4 319-T5 319-T6 356-T5 356-T6 357-T5 357-T6 A357-T61 390-T5 390-T6

132–146 172–185 227–248 284–296 152–221 152–245 124–219 138–303 243  11.7 225–270 341–385

19.1–21.1 25–26.9 32.9–36 41.2–42.9 22.0–32.0 22.0–35.5 18.0–31.8 20.0–44.0 35.3  1.7 32.6–39.2 49.5–55.9

244–267 290–343 285–327 345–387 214–295 262–317 241–289 255–339 316.5  9.0 225–270 341–385

35.4–38.7 42–49.7 41.3–47.4 50.0–56.1 31.0–42.8 38.0–46.0 35.0–41.9 37.0–49.2 45.9  1.3 32.6–39.2 49.5–55.9

3.1–6.2 5.5–14.0 1.4–5.3 3.0–7.3 3.0–14.0 4.5–20.0 4.7–11.0 4.0–13.2 8.4  2.0 < 0.2 < 1.0

... 50–57 HRB 70–73 HRB 74–79 HRB 40–49 HRB 40–51 HRB ... ... 112  5 HV 72–86 HRB 86–92 HRB

(a) Adapted from Ref 9.

Table 9 Example mechanical property data sets A356.0-T6 cast automotive suspension components Mechanical property

Example casting 1

Yield strength, MPa (ksi) Tensile strength, MPa (ksi) Elongation on 5D, % (Lowest / Mean / Highest) Cyclic 0.2% yield strength, MPa (ksi) Cyclic strain hardening exponent Cyclic strength coefficient, MPa (ksi) Fatigue strength coefficient, MPa (ksi) Fatigue strength exponent Fatigue ductility coefficient Fatigue ductility exponent pffiffiffiffi pffiffiffiffi Fracture toughness (Kq), MPa m/(ksi in.) Compressive yield strength Precision modulus, GPa/(msi)pffiffiffiffi pffiffiffiffi △K threshold (R = 0.1), MPapm ffiffiffiffi.) ffiffiffiffi/(ksipin △K threshold (R = 0.5), MPa m/(ksi in.)

Example casting 2

234 (33.9) 309 (44.8) 7.2 / 9.3 / 11.2 314 (45.5) 0.044 422 (61.2) 831 (120.6) 0.131 1.637 1.008 25.6 (23.3) 244 (35.4) 76.4 (11.1) 3.75 (3.41) 2.58 (2.35)

251 (36.5) 328 (47.6) 4.8 / 9.6 / 12.8 302 (43.8) 0.071 468 / 67.9 573 (83) 0.097 0.124 0.656 21.8 (19.9) 255 (37.0) 74.2 (10.8) 4.22 (3.84) 2.52 (2.29)

Source: Ref 10. Note that these data are for illustrative purposes and were taken from samples excised from cast parts. ASM, USAMP-AMD/DPO, associate members, and Westmoreland Mechanical Testing and Research assume no responsibility or liability for its use. Process capability data like these are foundry-specific, and designers should consult with the specific foundry for appropriate design data.

Cycles to failure

106

105

104 DAS: 25 mm, pore size: 30 mm DAS: 75 mm, pore size: 30 mm DAS: 25 mm, pore size: 1.9 mm DAS: 75 mm, pore size: 1.9 mm

103

102 80

Fig. 12

100

120 140 Stress amplitude, MPa

160

180

A plot from a statistical study of the effect of largest pore size and DAS on the fully reversed axial fatigue life of A356-T61. Source: Ref. 11

finish, better soundness at the surface, finer grain size, and adjustments in heat treated condition may all contribute to this enhanced fatigue resistance. Preventing Fatigue Cracking. Because castings are less notch-sensitive than wrought materials, it is not so critical to avoid discontinuities with castings as it is with wrought products. The following general considerations, however, should be kept in mind for castings that are to be subject to repeated loading:
Use gradual changes in section and symme-

try of design where practical


Minimize scratches and the presence and

magnitude of surface defects or discontinuities of any kind in highly stressed areas
Alleviate possible fretting between contacting surfaces
Provide proper maintenance and suitable protection against corrosion
Institute proper monitoring and control over process factors such as strontium content and hydrogen level that will influence the size and amount of porosity present in the part

Corrosion The corrosion resistance of aluminum casting alloys in service depends on such factors as alloy composition and the environment to which the alloy is exposed. The mechanical and physical conditions under which the alloy is exposed are also important. Detailed information on the corrosion resistance of wrought aluminum alloys is available in the article “Corrosion of Aluminum and Aluminum Alloys” in Corrosion, Volume 13B (2005) of the ASM Handbook. Minimizing Corrosion Problems. Corrosion can be minimized by designing the application to include protective measures and by selecting the alloy that will economically meet all design requirements, including those pertaining to corrosion. In many applications, the corrosion resistance of various casting alloys is similar enough that alloy selection can be based on other factors. Only in severe environments, such as marine atmospheres or heavily polluted urban industrial atmospheres, do differences in the general-corrosion resistances of various aluminum casting alloys become apparent. Special corrosion problems may arise when aluminum castings are painted or coated. Filiform corrosion is an example of common concern in aluminum alloy wheels that have been clear coated. Impurity levels in the alloy combined with the physical condition of the coating can affect how strongly this effect is seen when the clear coat is penetrated by a stone chip or scratch.

Mechanical Properties Mechanical Properties and Casting Design. Any casting design is based on two expectations:

1082 / Nonferrous Alloy Castings
The casting will meet certain minimum

requirements for mechanical properties that are characteristic of the alloy and temper selected.
The combination of design and properties will provide reliable service performance. The mechanical properties of castings produced by a particular process in a specific alloy and temper are critical considerations in alloy and temper selection. In high-volume automotive applications this is normally defined as the process capability, expressed in terms of statistical measures, and is considered to be a characteristic of each individual foundry/process/alloy combination. As a result, modern foundries must now do more of their own testing to be able to present their own individual process capability data when bidding for parts in this industry. Examination of Tables 5 to 8 will quickly show that microstructural features, such as the DAS, are commonly handled indirectly by means of reference to the process. Sand castings solidify more slowly than permanent mold castings resulting in a generally larger DAS and hence, lower achievable ductility. Squeeze-cast or semi-solid formed (SSF) parts are yet another new and emerging category. The ability to predict DAS values via computer simulation is being used increasingly to define the casting in terms of maximum values in specific locations, however. Mechanical Properties and Quality Assurance. Some type of quality-assurance program is required to ascertain that metal composition, molten metal quality, foundry practices, and heat treatment are consistently under control, and that parts with the expected properties are being consistently produced. Specimens for mechanical property testing may be cast separately, integrally cast, or cut from representative castings. Minimum Values for Mechanical Tests. Minimum mechanical property limits are usually defined by the terms of general procurement specifications, such as those developed by government agencies and technical societies. These documents often specify testing frequency, tensile bar type and design, lot definitions, testing procedures, and the limits applicable to test results. By references to general-process specifications, these documents also invoke standards and limits for many additional supplier obligations, such as specific practices and controls in melt preparation, heat treatment, radiographic and liquid penetrant inspection, and test procedures and interpretation. Some specifications for sand, permanent mold, plaster, and investment castings have defined the correlation between test results from specimens cut from the casting and separately cast specimens. A frequent error is the assumption that test values determined from these sources should agree. Rather, the properties of separately cast specimens should be expected to be superior to those of specimens machined from the casting. In the absence of more specific guidelines, a generally accepted relationship

defines the average tensile and yield strengths of machined specimens as not less than 75% of the minimum requirements for separately cast specimens, and elongation as not less than 25% of the minimum requirement. These relations may be useful in establishing the commercial acceptability of parts in dispute. Premium Engineered Castings. Quality assurance for premium engineered castings frequently involves tensile testing of specimens machined from castings or cast integrally with the casting. Premium engineered castings also frequently make use of chills, in sand casting processes, which give permanent-mold or better properties despite the process. For these reasons, property requirements are established by negotiation and agreement between customer and supplier and may include guarantees for critical areas of the casting that differ from those specified for the entirety of the part.

Molten Metal Fluidity The Nature of Fluidity. Fluidity is the characteristic of a molten alloy that enables mold filling. The foundry definition of fluidity is simply the distance that the metal will flow in a set configuration before it solidifies. This differs from the physics definition, which usually is given as the inverse of viscosity. Different alloys have different fluidities and therefore different abilities to provide definition of detail and integrity in thin sections. Factors Affecting Fluidity. Fluidity depends on two major factors: the intrinsic fluid properties of the molten metal and casting conditions. The properties usually thought to influence fluidity are viscosity, surface tension, the character of the surface oxide film, inclusion content, and the manner in which the alloy solidifies. Casting conditions that influence fluidity include part configuration; physical measures of the fluid dynamics of the system, such as metallostatic pressure drops, casting head, and velocities; mold material; mold surface characteristics; heat flux; rate of pouring; and degree of superheat. Viscosity. The measured viscosities of molten aluminum alloys are quite low and fall within a relatively narrow range. Kinematic viscosity (viscosity/specific gravity) is less than that of water. It is evident on this basis that viscosity is not strongly influential in determining casting behavior and therefore is an unlikely source of variability in casting results. Surface Tension and Oxide Film. A high surface tension has the effect of increasing the pressure required for liquid metal flow. A number of elements influence surface tension, primarily through their effects on the surface tension of the oxide. Figure 13 illustrates the effect of selected elements on surface tension. In aluminum alloys, the true effect of surface tension is overpowered by the influence of surface oxide film characteristics. The oxide film on pure aluminum, for example, triples apparent surface tension.

Inclusions in the form of suspended insoluble nonmetallic particles dramatically reduce the fluidity of molten aluminum. This is true even of intentional insoluble intermetallic additions such as the borides typical of commercial grain refiners. Solidification. It has been shown that fluidity is inversely proportional to freezing range (that is, fluidity is highest for pure metals and eutectics, and lowest for solid-solution alloys). The manner in which solidification occurs may also influence fluidity. Fluidity Testing. Correlation between fluidity test results (usually of the spiral or variable-dimension type) and foundry experience has been poor. No accepted method for the determination or prediction of fluidity, with relevance to the casting process, has been developed.

Shrinkage For most metals, the transformation from the liquid to the solid state is accompanied by a decrease in volume. In aluminum alloys, volumetric solidification shrinkage can range from 3.5 to 8.5%. The tendency for formation of shrinkage porosity is related to both the liquid/ solid volume fraction and the solidification temperature range of the alloy. Riser requirements relative to the casting weight can be expected to increase with increasing solidification temperature range. Requirements for the establishment of more severe thermal gradients, such as by the use of chills or insulated riser sleeves, or in permanent mold by selection and graduation of mold coatings or selectively cooling or heating mold sections, also increase with increasing solidification temperature range.

Hot Cracking The Source of Hot Cracking. As previously mentioned, aluminum alloys shrink from 3.5 to 8.5% during solidification. A further, much smaller contraction occurs in postsolidification cooling to room temperature. Tearing or hot cracking occurs during solidification, when the greatest amount of shrinkage occurs and when the casting is least resistant to stresses imposed by the geometrical constraints of the mold. This type of cracking, sometimes termed hot shortness, is always intergranular and is a feature to some extent of all casting alloys. Hot strength (resistance to cracking at solidification temperature), however, is alloy dependent. Minimizing Hot Cracking. Hot-cracking problems can be minimized by proper selection of casting process, alloy selection, part and mold design, solidification control, and grain refinement. Aluminum casting alloys demonstrate varying degrees of resistance to cracking; therefore, the crack sensitivity of alloys capable of meeting part requirements should be an important criterion in final alloy selection.

Aluminum and Aluminum Alloy Castings / 1083 No. 403, 3rd ed., 2003, North American Die Casting Association, Rosemount, IL, p. 3–6 to 3–25 10. “Design and Product Optimization for Cast Light Metals,” USAMP-LMD 110 Project, AFS, Des Plaines IL, 2001 11. J.F. Major, Porosity Control and Fatigue Behaviour in A356-T61 Aluminum Alloy, AFS Trans., Vol 97–94, 1997, p. 901–906 SELECTED REFERENCES General References
Aluminum Casting Technology, American

Foundrymen’s Society, 1986


J.E. Hatch, Ed., Aluminum: Properties and

Physical Metallurgy, American Society for Metals, 1984
K.R. Van Horn, Ed., Aluminum, Vol 1–3, American Society for Metals, 1967 Alloys
Aluminum Standards and Data, The Alumi-

num Association, 1984 Casting Processes
R. Bruner, The Metallurgy of Die Casting


Fig. 13

Effect of various elements on surface tension of 99.99% Al in argon at 700 to 740  C (1290 to 1365  F).

Mold Design. Improper mold design is a frequent cause of hot cracking. It is important to avoid abrupt changes in cross-sectional area and inadequately radiused or filleted angles or corners. These are design features that can sharply increase and concentrate stresses resulting from restrained contraction during and following solidification. The intelligent use of chills and/or insulation can also be beneficial in preventing hot cracks. Effective grain refinement to ensure a fine, equiaxed grain structure is especially important in hot-short alloys or in casting designs with features that aggravate cracking tendencies. Fine grain size reduces local stress concentrations by minimizing grain-boundary effects. ACKNOWLEDGMENT This article has been updated and revised from E.L. Rooy, Aluminum and Aluminum Alloys, Casting, Vol 15, Metals Handbook, 9th ed., ASM International, 1988, p 743–770. REFERENCES 1. Registration Record of Aluminum Association Alloy Designations and Composition Limits for Aluminum Alloys in the Form of Castings and Ingot, The Aluminum Association, April 2002

2. J.P. Frick, Ed., Woldman’s Engineering Alloys, 9th ed., ASM International, 2000 3. Handbook of International Alloy Compositions and Designations, Metals and Ceramics Information Center, Battelle Memorial Institute, 1976 4. J. Datta, Aluminum-Schlu¨ssel—Key to Aluminum Alloys, 5th ed., Aluminium Verlag, 1997 5. An.-de Figueredo, Ed., Science and Technology of Semi-Solid Metal Processing, North American Die Casting Association, Rosemount, IL, 2001 6. L. Backerud and M. Johnsson, The Relative Importance of Nucleation and Growth Mechanisms to Control Grain Size in Various Aluminum Alloys, Light Metals 1996, W. Hale, Ed., The Minerals, Metals, and Materials Society, 1996, p. 679–685 7. K.R. Van Horn, Ed., Properties, Physical Metallurgy, and Phase Diagrams, Vol 1, Aluminum, American Society for Metals, 1967, p 174 8. F. Major and D. Apelian, A Microstructural Atlas of Common Commercial Al-Si-X Structural Castings, Proc. AFS International Conference on Structural Aluminum Casting, Nov. 2–4, 2003 (Orlando, FL), American Foundrymen’s Society, p. 267–285 9. Product Specification, “Standards for Die Castings Produced by the Semi-Solid and Squeeze Casting Process,” Publication










Alloys, Society of Die Casting Engineers, 1986 Core and Mold Process Control, American Foundrymen’s Society, 1977 Foundry Core Practice, American Foundrymen’s Society, 1966 Fundamental Molding Sand Technology, American Foundrymen’s Society, 1973 E.A. Herman, Die Casting Handbook, Society of Die Casting Engineers, 1982 Plaster Mold Handbook, American Foundrymen’s Society, 1984 Precision Ceramic Castings, Paper 107, Trans. AFS, 1976 A.C. Street, The Die Casting Book, Portcullis Press, 1977 B. Upton, Pressure Die Casting, Society of Die Casting Engineers, 1982

Solidification
J. Burke, M. Flemings, and A. Gorum, Solid








ification Technology, Brook Hill Publishing, 1974 J.E. Gruzleski, Microstructure Development During Metalcasting, American Foundrymen’s Society, Inc., Des Plaines, IL, 2000 W. Kurz and E. Fisher, Fundamentals of Solidification, TransTech Publications, 1986 A. Montgomery, Metallography of Aluminum Casting Alloys, Am. Foundryman, April 1949 S. Shankar, Y. Riddle, and M. Makhlouf, Nucleation Mechanism of the Eutectic Phases in Aluminum-Silicon Hypoeutectic Alloys, Acta Mater., Vol 52, 2004, p. 4447–4460 Solidification, American Society for Metals, 1971

1084 / Nonferrous Alloy Castings Dendrite Arm Spacing
K. Oswalt and M. Misra, Dendrite Arm Spac-

ing, Paper 51, Trans. AFS, 1980


O. Atasoy, F. Yilmaz, and R. Elliot, Growth



K. Radhakrishna, S. Seshan, and M. Seshadri,

Dendrite Arm Spacing and Mechanical Properties of Aluminum Alloy Castings, Metallurgical Society, Indian Institute of Science, 1979
R. Spear and G. Gardner, Dendrite Cell Size, Paper 65, Trans. AFS, 1963







Nondendritic Microstructures
NADCA Product Specification: Standards for

Die Castings Produced by the Semi-Solid and Squeeze Casting Processes, NADCA Publication No. 403, 3rd ed., North American Die Casting Association, Rosemount, IL, 2003







Grain Refinement
L. Backerud and M. Johnsson, The Relative












Importance of Nucleation and Growth Mechanisms to Control Grain Size in Various Aluminum Alloys, Light Metals 1996, Wayne Hale, Ed., The Minerals, Metals, and Materials Society, 1996, p. 679–685 P. Cooper, et al., Characterisation of a New Generation of Grain Refiners for the Foundry Industry, Light Metals 2003, Paul Crepeau, Ed., The Minerals, Metals, and Materials Society, 2003, p. 923–928 M. Guzowski and G. Sigworth, Grain Refining of Hypoeutectic Al-Si Alloys, Paper 172, Trans. AFS, 1985, p 85–172 M. Guzowski, G. Sigworth, and D. Sentner, The Role of Boron in the Grain Refinement of Aluminum with Titanium, Metall. Trans. A, Vol 10A, April 1987 R. Kotschi and C. Loper, Grain Refinement in Cast Aluminum Alloys, Paper 113, Trans. AFS, 1977 Solidification Characteristics of Aluminum Alloys, Skan Aluminum, 1986

Modification and Refinement
D. Apelian, G. Sigworth, and K. Whaler,

Assessment of Modification and Grain Refining of Aluminum Silicon Casting Alloys by Thermal Analysis, Paper 161, Trans. AFS, 1984













Structures in Aluminum-Silicon Alloys, J. Cryst. Growth, Jan–Feb 1984 S. Bercovici, Solidification, Structure, and Properties of Aluminum Silicon Alloys, Giesserei, Vol 67 (No. 17), 1980 J. Charbonnier and A. Kearney, French Experience and Developments in Thermal Analysis of Aluminum Casting Alloys, Paper 133, Trans. AFS, 1984 J. Charbonnier, J. Perrier, and R. Portalier, Recent Developments in Aluminum-Silicon Alloys Having Guaranteed Structures or Properties, Cast Met. J., Dec 1978 N. Handiak, J. Gruzleski, and D. Argo, Sodium, Strontium, and Antimony Interactions During the Modification of AS7G03 (A356) Alloys, Paper 20, Trans. AFS, 1987 J.F. Major and J.W. Rutter, The Effect of Sr and P on the Solid/Liquid Interface of the Al-Si Eutectic, Mater. Sci. Technol., Vol 5, 1989, p. 645–656 E. Rooy, Summary of Technical Information on Hypereutectic Al-Si Alloys, Paper 44, Trans. AFS, 1972 M. Shamsuzzoha and L. Hogan, The Crystal Morphology of Fibrous Silicon in Strontium Modified Al-Si Eutectic, Philos. Mag., Vol 54 (No. 4), 1986 G. Sigworth, Observations on the Refinement of Hypereutectic Silicon Alloys, Paper 82, Trans. AFS, 1982 O. Vorrent, J. Evensen, and T. Pedersen, Microstructure and Mechanical Properties of Al-Si (Mg) Casting Alloys, Paper 162, Trans. AFS, 1984 J. Weiss and C. Loper, Primary Silicon in Hypereutectic Aluminum-Silicon Casting Alloys, Paper 32, Trans. AFS, 1987 C. Zheng, L. Yao, and Q. Zhang, Effects of Cooling Rate and Modifier Concentration on Modification of Al-Si Eutectic Alloys, Acta Metall., Vol 18 (No. 6), Dec 1982

Heat Treatment

(Toronto, Ontario, Canada), American Institute of Mining, Metallurgical, and Petroleum Engineers, Oct 1985
S. Shivkumar, C. Keller, and D. Apelian, Aging Behaviour in Cast Al-Si-Mg Alloys, Trans. AFS, 1990, p. 905–911
S. Shivkumar, S. Ricci Jr., B. Steenhoff, and D. Apelian, An Experimental Study to Optimize the Heat Treatment of A356 Alloy, Trans. AFS, 1989, p. 791–810 Properties
G. Bouse and M. Behrendt, Metallurgical and
















J.F. Hernandez-Paz, F. Paray, and J.E. Gru-

zleski, Natural Aging and Heat Treatment of A356 Aluminum Alloy, Trans. AFS, Paper No. 04-009(2), 2004
E. Rooy, “Practical Aspects of Heat Treatment,” Paper presented at the Fall Meeting




Mechanical Property Characterization of Premium Quality Vacuum Investment Cast 200 and 300 Series Aluminum Alloys, Adv. Cast. Technol., Nov 1986 G.J. Courval, B. Chamberlain, and D. Pattemore, Effects of Alloy Composition and Condition on Filiform Corrosion Performance of Cast Aluminum Wheels, in Applications for Aluminum in Vehicle Design, Publication No. SP-1251, Technical Paper No. 970021, SAE International, 1997, p. 81–89 Directionally Solidified Aluminum Foundry Alloys, Paper 69, Trans. AFS, 1987 M. Holt and K. Bogardus, The “Hot” Aluminum Alloys, Prod. Eng., Aug 1965 J. Gilbert Kaufman and Elwin L. Rooy, Aluminum Alloy Castings: Properties Processes and Applications, ASM International, 2004 D.A. Lados and D. Apelian, The Effect of Residual Stress on the Fatigue Crack Growth Behavior of Al-Si-Mg Cast Alloys—Mechanisms and Corrective Mathematical Models, Metall. Mater. Trans. A, Vol 37 (No. 1), 2006, p 133–145 F. Mollard, Understanding Fluidity, Paper 33, Trans. AFS, 1987 E. Rooy, Improved Casting Properties and Integrity with Hot Isostatic Processing, Mod. Cast., Dec 1983 G. Scott, D. Granger, and B. Cheney, Fracture Toughness and Tensile Properties of Directionally Solidified Aluminum Foundry Alloys, Trans. AFS, 1987, p 69 J. Tirpak, “Elevated Temperature Properties of Cast Aluminum Alloys A201-T7 and A357-T6,” AFWAL-TR-85-4114, Air Force Wright Aeronautical Laboratories, 1985

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1085-1094 DOI: 10.1361/asmhba0005332

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Copper and Copper Alloy Castings Kumar Sadayappan, Laurence Whiting, and Mahi Sahoo, CANMET—Materials Technology Laboratory Harold T. Michels, Copper Development Association

COPPER and its alloys were the first metals discovered by mankind some 60 centuries ago. Copper smelting and casting by artisans is known to have occurred as early as 3000 B.C. in many parts of the world. This long history of successful employment of copper was due to the ease of recognition of ores, ease of production, and availability. Copper and its alloys are readily processed by casting, forming, machining, brazing, soldering, polishing, and plating. The properties of copper alloys occur in unique combinations found in no other alloy system. Historically, key properties were relative strength, ductility, and toughness, which have been surpassed in contemporary use by high thermal and electrical conductivity, a good range of mechanical properties, excellent ductility and toughness, as well as superior corrosion resistance in many environments.

Alloying Additions Copper is generally alloyed with other elements to improve its castability and properties. Pure copper is extremely difficult to cast as well as being prone to cracking, gas porosity, and internal cavities. The casting characteristics and mechanical strength of copper are improved by the addition of alloying elements such as zinc, tin, aluminum, chromium, silver, beryllium, silicon, and nickel. However, in applications requiring the electrical conductivity of pure copper, alloying is avoided. The major and minor alloying additions and their impact on the properties of copper are briefly discussed in this section.

Zinc Zinc is a relatively inexpensive, yet potent, alloying element in copper, imparting strength and hardness through solid-solution strengthening. Zinc is the major alloying element in brasses. It is either added alone or in combination with other elements. Copper alloys may contain between 1 and 45% Zn. The solid solubility of zinc in copper is approximately 36%, which results in a single-phase microstructure.

Higher levels of zinc promote mixed microstructure in alloys known as alpha-beta brass.

Tin Tin is the oldest known alloying element in copper and is a key alloying element in many bronzes. Tin is also a potent solid-solution strengthener in copper, even more so than zinc, because less tin is required for an equivalent increase in strength. Unlike zinc, tin also improves corrosion resistance. Tin reduces the melting temperature of copper and increases the fluidity, thereby making these alloys easier to melt and cast. However, tin increases the solidification range of copper alloys. The level of tin and zinc in thin-section green sand casting alloys should not exceed 11.5% and even less in thicker-section castings; otherwise, porosity becomes an issue. Strength can be regained by addition of nickel.

Lead Although lead is insoluble in copper, many commercial brasses and bronzes contain significant amounts of lead. Lead solidifies last, so it is found at the grain boundaries or in interdendritic areas. The addition of lead helps to seal the normal shrinkage porosity formed toward the end of solidification. Leaded copper alloys have improved pressure tightness as a result of this sealing of the interdendritic spaces. However, lead is not required for pressure tightness in Cu-Sn-Ni alloys. High-lead alloys are used for bearings, where the lead on machined surfaces eliminates the tendency of copper parts to gall or bond to each other. Lead also improves machinability by providing a natural break, so that the removed metal forms chips rather than metal ribbons that foul the tool bits.

aluminum. Small amounts of aluminum improve the fluidity of copper alloys, such as yellow brasses and silicon brasses for permanent mold casting.

Silicon Copper-silicon alloys contain less than 5% Si. Like aluminum, silicon forms a solid solution with copper. The excellent fluidity of copper-silicon alloys enables them to be used for art castings and permanent mold casting.

Nickel Copper and nickel are completely miscible in both the liquid and solid phases. Copper alloys with higher nickel contents are known as nickel silvers due to their white color. Copper-nickel alloys are highly resistant to corrosion, so the alloys are used in the chemical, petroleum, food, and dairy industries. Nickel improves the quality, strength, and creep resistance of tin bronze and semired brass castings and is more effective than lead in improving pressure tightness.

Beryllium Beryllium is added to copper in small quantities to increase strength by forming an interdendritic precipitate, without compromising the electrical and thermal conductivity or ductility of high-copper alloys.

Chromium Chromium is added to high-copper alloys to increase strength, with only a minor loss in the electrical conductivity.

Aluminum

Iron

The solubility of aluminum in copper is limited to 9%. Aluminum improves the strength of copper, so high-strength copper alloys, such as aluminum bronzes and high-strength yellow brasses, contain significant quantities of

Iron is not soluble in copper. However, iron is used as a grain refiner in yellow brasses, high-strength yellow brasses, and aluminum bronzes. Although the mechanism of the refinement is not well understood, iron increases the

1086 / Nonferrous Alloy Castings melting temperature of the copper alloys, and iron additions to the liquid alloys at higher temperatures result in precipitates in the liquid that persist in the casting. Excessive iron levels can cause segregation and hard spots in these alloys. In tin bronzes and semired brass, iron increases the strength and hardness but at great cost to ductility. For these alloys, iron and tin should be less than 0.01 and 4%, respectively, if minimum magnetic susceptibility is required.

Minor Additions Antimony is used in small quantities to reduce the specific corrosion problem, known as dezincification, that occurs in high-zinc brasses. Antimony (up to 0.25%) reduces the strength and ductility of tin bronzes. Bismuth, when present in pure copper and some high-strength alloys such as aluminum bronzes, is considered an impurity. Bismuth embrittles these alloys even when present in amounts as low as 0.01%. However, bismuth is being used as a substitute for lead in red brasses and yellow brasses in potable water component systems to enhance machinability. Selenium is added to red brasses and coppersilicon alloys to improve the subsequent machinability of the castings. Selenium can also be used along with bismuth. Manganese is used to neutralize the deleterious effects of iron in red brasses. It can increase the strength of yellow brasses and aluminum bronzes to a certain extent but is deleterious to 88-10-2 Cu-Sn-Zn and 85-5-5-5 leaded red brass. Phosphorus is used to deoxidize copper, bronzes, and red brasses. Although phosphorus reduces electrical conductivity, it is used to lower the oxygen level in high-copper alloys, followed by the addition of other costlier elements, such as boron, to eliminate the last traces of oxygen. Phosphorus is added to red brasses to reduce oxygen levels, but residual levels are limited to 0.02% when the metal is to be cast in green sand because of potential reactions with water in the sand. The structure as well as the fluidity and castability of copper alloys are dramatically improved when oxides are eliminated. Although 0.05% P improves the machinability of red brasses and copper-silicon alloys, it is only used in copper-silicon alloys when cast in permanent molds. Sulfur and arsenic levels are no longer issues for alloys made from primary copper, since the advent of electrolytic purification of copper. Sulfur, if present, can react with oxygen at higher pouring temperatures and cause gas porosity.

bismuth, and sulfur are examples of impurities from the original ores, while iron, silicon, and aluminum chiefly occur from recycling and processing. Nevertheless, different impurity elements affect each alloy family. Lead and bismuth are considered impurities in aluminum bronzes, while aluminum and silicon are harmful in red brasses. The tolerance of impurities depends on the alloy and the intended use. Most critical levels (if known) are included in the specifications. Gases. Dissolved gases, or reactions that produce gases, are a major problem in many nonferrous melts, because on freezing they can increase the level of porosity and thus decrease the strength of the material. Porosity cannot be avoided in many copper alloys. Gas porosity is revealed on polishing as round holes, while shrinkage porosity occurs as irregular, rough cavities. Interconnected porosity is undesirable in castings intended for plumbing, compressed gases, or vacuum systems, and leaking often occurs when these cavities are opened during

machining (castings should be designed so as to avoid machining both as-cast surfaces). The most common gases present in copper and copper alloys are oxygen and hydrogen from the reactions between moisture in the air and green sand. However, copper melt cannot have both hydrogen and oxygen in high quantities because there is an inverse relation between the two gases. As shown in Fig. 1, when the melt is rich in hydrogen, it will have very low levels of oxygen, and vice versa. Typical air melting causes dissolved hydrogen to be reduced to low levels. The oxygen is then removed, typically by phosphorus. In gas-fired melting furnaces, the liquid metal surface should not be exposed to the flame; otherwise, both oxygen and hydrogen will dissolve. During melting in channel furnaces, the liquid metal is covered with graphite to reduce oxygen levels, because oxides damage the refractories. Hydrogen in these melts is removed by flushing with dry nitrogen or argon. Since there is usually a small amount of hydrogen and oxygen in the last metal to freeze, some steam

Impurities Impurities in copper alloys occur as a result of processing of the original ores and from processing and recycling. Arsenic, antimony,

Fig. 1 pressures.

Hydrogen-oxygen equilibrium in liquid copper. The millimeter indicators on the curves are the water vapor pressures above the copper melt; the pressure scales under the x-axis are the equivalent hydrogen partial

Copper and Copper Alloy Castings / 1087 porosity will result. Sulfur, when present, can react with hydrogen or oxygen to produce gas pockets on freezing. Alloys containing more than 15% Zn are usually low in oxygen and hydrogen when melted. The oxygen is reduced by the formation of zinc oxide, while the hydrogen is removed by the unavoidable volatilization of zinc from the liquid metal. Aluminum bronzes are also low in oxygen but need to be degassed by blowing dry nitrogen through the melt. High superheats exacerbate gas problems, particularly in copper-nickel alloys.

Alloying Systems and Specifications Copper alloys are identified by the Unified Numbering System (UNS). This system was developed during the 1970s to give a precise description of the composition range for each alloy and is now an internationally recognized designation system. The UNS is managed jointly by ASTM International and the Society of Automotive Engineers. The Copper Development Association administers the registration, numbering, compilation, and compositions of specific alloys. Cast alloys are numbered from C80000 through C99999. The numbers for families of copper alloys are grouped by composition (Table 1). Because the designations are based on precise compositional limits for the final castings, the chemical analyses of commercial ingots may differ slightly from the designated composition. Such variances anticipate processing conditions in which some elements are lost by oxidation or volatilization and

Table 1 Unified Numbering System (UNS) classification of copper casting alloys Family

Coppers High copper Copper-tin-zinc (red brasses) Copper-tin-zinc (semired brasses) Copper-zinc (yellow brasses) Manganese bronzes Copper-silicon Copper-bismuth Copper-tin (tin bronzes) Copper-tin-lead (leaded tin bronzes) Copper-tin-lead (highleaded tin bronzes) Copper-tin-nickel (nickel-tin bronzes) Copper-aluminum-iron Copper-aluminumnickel-iron (aluminum bronzes) Copper-nickel-iron (nickel coppers) Copper-nickel-zinc (nickel silvers) Copper-lead Special alloys

UNS No.

C80001–C81399 C81400–C83299 C83300–C83999 C84xxx C85xxx C86xxx C87xxx C88xxx, C89xxx C90xxx, C91xxx C92xxx

impurities gained when the ingots and rejects are remelted to produce satisfactory castings.

Coppers These numbers are reserved for alloys with a designated minimum copper content of 99.3% or higher, which usually contain traces of silver or phosphorus. The presence of silver imparts annealing resistance, while phosphorus, a deoxidizer, aids in welding. These alloys are used in applications where the high thermal and electrical conductivity are desirable, such as electrical connectors and in hot metal handling. These coppers have relatively low strength.

High-Copper Alloys The cast high-copper alloys have designated copper contents in excess of 94%. Copper is alloyed with other elements along with silver to obtain properties not achievable using pure copper. High-copper alloys are unique in that they combine high strength with high thermal and electrical conductivity, two properties that are seldom found together in the same material. These high coppers are alloyed with small amounts of elements, including beryllium, silicon, nickel, tin, zinc, and chromium, and are constituted to have improved strength properties compared to high-purity copper while maintaining a minimum of 85% electrical conductivity. They are widely used for cast electrical conducting members. Chromium coppers are alloys containing up to 1.5% Cr. The strength of chromium copper is approximately twice that of pure copper, while the electrical conductivity is only 20% lower than that of pure copper. Other potential alloying agents are less effective and cause greater loss of conductivity. Typical applications for chromium copper are welding clamps and high-strength electrical connectors. Chromiumcopper castings can be heat treated for precipitation strengthening. Beryllium coppers contain up to 2.25% Be along with some nickel and cobalt. These alloys can be precipitation hardened to obtain high strength and high conductivity. These alloys are to be handled carefully due to the toxic nature of beryllium, with particular attention to any fumes and machining dusts.

Brasses

C93xxx–C94500 C94600–C94900 C95xxx

C96xxx C97xxx C98xxx C99xxx

These alloys contain zinc as the principal alloying element, with or without other designated alloying elements such as tin, lead, iron, aluminum, nickel, and silicon. The cast alloys comprise three major families of brasses: CuSn-Zn alloys (red, semired, and yellow brasses), manganese bronze alloys (high-strength yellow brasses), and Cu-Zn-Si alloys (silicon brasses and bronzes). Alloys with more than 20% Zn are prone to selective corrosion known as dezincification.

Red brasses are alloys with up to 8% Zn and may contain varying amounts of tin. The color of red brass is attributable to its relatively low zinc content. The highest-volume cast red brass alloy, commonly known as 85-5-5-5 (i.e., 85% Cu, 5% Sn, 5% Pb and 5% Zn), has been used commercially for several hundred years and accounts for more tonnage than any other cast copper alloy. In foundry alloys, lead may be present in various amounts to promote pressure tightness in castings in service and to facilitate free machining during the manufacturing process. Most commercial red and semired brasses invariably contain more than 1% Pb, although newer leadfree versions of these alloys are now available. Early versions of lead-free alloys contained both bismuth and selenium. These elements act synergistically to improve machinability. However, selenium has become expensive, and because of the low ductility of Cu-Bi-Se alloys, other alloys are being promoted for potable water (plumbing) applications. Semired brasses contain more than 8% Zn and contain tin and lead as other alloying elements. Yellow brasses are even lower in cost than the red brasses, because their zinc content is higher, at 20 to 39%. Other than zinc, these alloys contain low levels of tin, lead, and aluminum. Yellow brass has a pleasant yellow color that can be polished to a high luster. This accounts in part for its selection for decorative hardware. High-strength yellow brasses, also known imprecisely as manganese bronzes, contain high zinc contents along with significant quantities of manganese, aluminum, and iron. Iron in these alloys causes fine grains that enhance the properties. Tin and lead may be present in these alloys as well. Ship propellers are the most important application for these alloys. Silicon bronzes and silicon brasses are essentially alloys with up to 20% Zn and up to 5% Si. They have low melting points and high fluidity, which are desirable for art castings, permanent mold, and pressure die casting applications. Bismuth Brasses and Bronzes. In recent times, in response to the restrictions on the lead content in drinking water, brasses with very low levels of lead were developed. These alloys often contain bismuth to achieve the same machinability and casting characteristics as the leaded alloys. Sometimes, selenium can be added to these alloys to further enhance machinability. Other alloying additions, such as tin, zinc, iron, and nickel, remain the same as in red or yellow brasses.

Bronzes Bronze is now an imprecise classification, but originally bronze referred to alloys in which tin was the major alloying element. Today, the term bronze applies to a broader class of alloys

1088 / Nonferrous Alloy Castings in which the principal alloying element is neither zinc nor nickel. Nevertheless, the name bronze is in common use for a number of alloys that contain little, if any, tin. The term bronze is generally used with a modifier rather than alone. The foundry alloys include four main families of bronzes: copper-tin alloys (tin bronzes), copper-tin-lead alloys (leaded and high-leaded tin bronzes), copper-tin-nickel alloys (nickel-tin bronzes), and copper-aluminum alloys (aluminum bronzes). The alloys known as manganese bronzes, in which zinc is the major alloying element, are in fact highstrength brasses and are included in the previous brass category. Tin Bronzes. Modern cast tin bronzes are very similar to the alloys found in the many relics from the Bronze Age, over 3500 years ago. They are basically alloys of copper and tin, where the tin content can be as high as 20%. The good corrosion resistance of tin bronzes to aqueous solutions accounts, in part, for the survival of these Bronze Age relics to this day. Additional attributes of tin bronzes include reasonably high strength, good wear resistance, and a lower coefficient of friction when compared to steel, making these alloys very useful for bearings, piston rings, and gears. Leaded Tin Bronzes. High-leaded tin bronzes containing more than 7% Sn and 7% Pb are used for bearings and bushings as well as pump and impeller components with high corrosion resistance and strength. Aluminum bronzes have complex metallurgical structures. Alloying elements always include aluminum and varying amounts of manganese, iron, and, in some versions, nickel. Aluminum imparts both strength and oxidization resistance by virtue of the formation of alumina (Al2O3)-rich protective films. These alloys are very wear resistant and exhibit good casting and welding characteristics. They are resistant to corrosion in many environments, including seawater, chlorides, and dilute acids. Applications are varied and include propellers, valves, pickling hooks, pickling baskets, and wear rings. Aluminum bronzes and particularly nickel-aluminum bronzes are useful in fluidmoving applications, such as pump impellers, because these alloys have superior resistance to erosion, corrosion, and cavitation. Aluminum bronzes are used in permanent mold casting in addition to sand casting, because aluminum improves the fluidity of the liquid and enables complex shapes to be produced.

be limited to 0.5% to facilitate welding. Their excellent corrosion resistance in seawater, combined with their high strengths, account for their wide use in pipes, heat exchangers, valves, ship tail-shaft sleeves, and other marine applications. They are also used in coins.

Copper-Nickel-Zinc Alloys Commonly known as nickel silvers, these alloys contain 11 to 27% Ni and 1 to 25% Zn. The presence of nickel primarily accounts for their pleasant silver luster, but, in spite of their name, nickel silvers do not contain silver. Although nickel silvers are highly alloyed, they are simple solid solutions. They offer good corrosion resistance and are easily cast and machined. Major uses include hardware for food processing, seals, architectural trim, musical instrument valves, and door keys.

Leaded Coppers These comprise a series of cast alloys of copper with 20% or more lead, usually with a small amount of silver present but without tin or zinc.

Special Alloys Alloys whose chemical compositions do not fall into any of the previous categories are categorized as special alloys. For example, Inco recently announced some new alloys containing chromium and silicon.

Physical Metallurgy Copper-Tin-Lead-Zinc Alloys As a group, the Cu-Sn-Pb-Zn alloys exhibit very wide freezing ranges, varying from approximately

Copper-Nickels The copper-nickel alloys are simple solid solutions of nickel in copper with or without other designated alloying elements. The nickel content varies from 9.0 to 33.0%. Small amounts of manganese (0.05 to 1.5%) and iron (0.4 to 1.8%) are also present. Silicon with either niobium or chromium improves the strength of the 30% Ni alloy, but silicon should

Fig. 2

Copper-zinc phase diagram

90 to 170  C (195 to 340  F). Consequently, all of these alloys solidify in a mushy, dendritic manner with the attendant effects of pronounced segregation, undercooling, and nonequilibrium, which are caused by inadequate diffusion and exacerbated by blocked feeding channels. These alloys do not lend themselves to heat treatment or melt processes, such as inoculation or grain refinement, that are intended to enhance properties by structural modification. Hence, these alloys are always used in the as-cast condition, and their microstructures depend essentially on alloy content, purity, and rate of solidification. A familiarity with typical microstructural features of the various alloys, together with an understanding of possible abnormalities that can occur, can assist the metallurgist in evaluating and controlling alloy quality. Technically, the alloys of this group are either ternary or quaternary. However, lead is nearly or entirely insoluble in the solid alloys and thus has little or no effect on alloy phase relationships within the Cu-Sn-Zn system. Therefore, from a practical point of view, the alloys of this group can be considered as binary or ternary, with or without lead being present as an insoluble additive. Even though the cast microstructures of these alloys represent nonequilibrium conditions, they can best be interpreted and understood by consideration of phase-equilibrium diagrams. Likewise, the ternary Cu-Sn-Zn system is easier to understand if first the solubility limits and general phase relationships are considered in the binary copper-zinc and copper-tin systems. Copper Zinc. Zinc has a relatively wide range of solubility in copper in the solid state, whereby compositions containing up to 32.5% Zn solidify as single-phase alpha (a) solid solutions (Fig. 2). This is to be expected for all the alloys in brasses that contain a maximum of 15% Zn. Note that in amounts up to 15%, zinc

Copper and Copper Alloy Castings / 1089

Fig. 4

Fig. 3

Copper-tin phase diagram

produces only a slight lowering of liquidus temperature, and the binary alloys have very narrow freezing ranges. At higher zinc levels, due to segregation effects, a second, intermediate solid solution (b) begins to form. This beta phase is the base for most high-strength yellow brasses. The presence of other alloying elements, such as aluminum, silicon, and manganese, shifts the alpha-beta transition composition. Copper Tin. Tin produces a rather drastic effect on both the melting temperature (liquidus) (Fig. 3) and freezing range of copper when added in amounts of up to 10%, which is the limit in commercial alloys. Within this composition range, the phase-equilibrium diagram indicates that alloys should solidify as single-phase, alpha, solid solutions—which are analogous to those of the copper-zinc system described. For practical purposes, the line on the copper-tin diagram showing the limiting tin solubility in the alpha phase can be drawn vertically downward at the maximum value of 15.8%, from 520  C (968  F) to room temperature. The epsilon (e) phase, which should form at 350  C (662  F), is rarely encountered because of the sluggishness of the transformation. Hence, the alpha-plus-delta field may be considered as extending to room temperature. However, because of the very wide freezing ranges of the copper-tin alloys, together with pronounced undercooling and slow diffusion, considerable segregation occurs during freezing. This causes high composition gradients to exist within the crystals, their central portions being rich in copper, surrounded by zones increasingly rich in tin. This phenomenon, known as coring, commonly occurs in the as-cast structure. Another important consequence of this nonequilibrium freezing is the fact that the last liquid present during solidification contains much more tin than the nominal composition of the alloy. Thus, the final metal

to solidify may actually follow the peritectic reaction. This results in the eventual formation of a hard, brittle delta (d) phase, occurring as bluish lakes of alpha-delta eutectoid at the alloy grain boundaries. Single-phase or alpha-phase materials tend to be uniform. If the tin content exceeds 11%, then as the metal is cooled below 400  C (750  F), some of the alpha is transformed to delta. These finely dispersed particles dramatically affect slip planes by locking them. This structure has greater hardness and wear resistance than the single-phase material. Copper-Tin-Zinc. The 500  C (932  F) isothermal section of the ternary Cu-Sn-Zn phase-equilibrium diagram is shown in Fig. 4 and represents the effective phase relationships existing at room temperature. The phases present, within the range of compositions of commercial interest, can be understood in terms of what has already been discussed in connection with the various binary copper alloy phase diagrams. The principal phase occurring in all the commercial alloys is that designated as alpha (a) in Fig. 4. The field represents a continuous series of solid solutions analogous to the alpha phases in the binary copper-zinc and copper-tin systems. When both alloying elements are present simultaneously, each reduces the solubility of the other element to some extent, and the solid curved line that bounds the alpha region shows the effective solubility limits. The wide freezing ranges of the ternary alloys result in as-cast structures wherein the alpha matrix shows pronounced coring. The second phase that occurs consistently in all the leaded alloys is, of course, the lead itself. The quantity and distribution of the lead depends on the amount present in the alloy, the freezing range of the alloy, and its actual rate of solidification. Chill castings, for a given lead content, tend to exhibit finer grain size

Copper-tin-zinc ternary phase diagram

and, consequently, finer distribution of the lead particles than do sand castings. The lead inclusions are of a dark-gray color, and their shapes tend to conform to the dendritic grain boundaries of the matrix. Since the lead phase is virtually insoluble in the solid condition, there is no need to consider the constitution of these alloys in terms of a quaternary Cu-Sn-Pb-Zn equilibrium diagram. Instead, the lead content is ignored, and the percentage composition of the alloy is recalculated in terms of the copper/tin/zinc ratio. For example, alloy C83600 (85-5-5-5) would be expected to solidify in the same manner as its unleaded equivalent— 89½ Cu, 5¼ Sn, 5¼ Zn. Again in the ternary diagram, the alpha field is bounded by a two-phased region of compositions consisting of alpha plus gamma (g). The gamma phase in the ternary diagram, like the alpha phase, also consists of a continuous series of solid solutions of tin and zinc in copper. Metallurgically, this phase is said to be isomorphous with the gamma phase of the copper-zinc system and the delta phase of the copper-tin system. When, due to nonequilibrium freezing conditions, the gamma phase appears in the solidified structures of the ternary alloys, it has the identical appearance of the delta phase that constitutes the alpha-delta eutectoid in copper-tin alloys. For this reason, it is commonly referred to as the delta phase. In appearance, this phase forms a part of an alpha-delta eutectoid that occurs as bluish inclusions, often elongated, found at the grain boundaries of the matrix. The alpha-delta eutectoid occurs most often in the tin bronzes with 8 to 10% Sn and is more prevalent during rapid freezing. It is frequently detected in leaded red brass compositions but less often in the semired brasses. Excessive amounts of alpha-delta eutectoid are considered undesirable because they reduce the elongation of the alloy. The occurrence of this phase in commercial alloys is regulated by restricting the sum of the tin and zinc contents.

1090 / Nonferrous Alloy Castings Aluminum Bronzes Aluminum bronzes containing up to approximately 9.4% Al should exhibit a single-phase alpha (a) structure at room temperature. However, alloys containing more than 8% Al at the beginning of solidification always solidify as a two-phase material consisting of alpha + beta (a + b). On slow cooling, the high-temperature b phase should transform to a, but because the transformation rate is too slow, alloys containing 8 to 9.4% Al eventually remain as a two-phase material. In alloys containing more than 9.4% Al, the b phase undergoes a eutectoid transformation, if cooled slowly in the region of 900 to 565  C (1650 to 1050  F), and forms a mixture of alpha + gamma2 (a + g2). The most common alloying element added to copper-aluminum alloys is iron, which is added as a grain refiner to improve the toughness. Iron also reduces the formation of the g2 phase during solid transformation. The room-temperature microstructure of an aluminum bronze with 5% Fe consists of a and b phases with precipitates of iron (d) based on Fe3Al. The other common alloying element, although not added alone in aluminum bronzes, is nickel, which causes precipitation of NiAl particles when the alloy is cooled slowly. Nickel-aluminum bronzes contain nickel, iron, and manganese as alloying elements. Their microstructure always consists of various nickel- or iron-rich kappa (k) phases in an alpha matrix. In addition, because of nonequilibrium solidification and cooling, which is characteristic of the casting process, the alloys with more than approximately 8.5% Al contain substantial b phase at higher temperatures. This phase further transforms to a + g2 as the casting cools to room temperature, as explained earlier, and additional k phases also form in this reaction. The phase diagrams of the aluminum bronze alloy systems are shown in Fig. 5. The k phases, depending on the morphology and compositions, are identified as kI, kII, kIII, and kIV. The schematic diagram indicating these various phases in nickel-aluminum bronzes and a typical microstructure are shown in Fig. 6(a and b). The g2 phase is detrimental to both ductility and corrosion resistance. The amount of this phase in the microstructure increases with the aluminum content; therefore, the higherstrength, higher-aluminum-content aluminum bronzes C95300, C95400, and C95500 generally rely on heat treatment and/or alloying with nickel and manganese to prevent formation of excessive g2 phase in the microstructure. Both nickel and iron combine with the aluminum to form complex k phases and thus increase the amount of aluminum necessary to introduce the g2 phase. The addition of manganese to the aluminum bronze is also reported to slow down the transformation of b phase to the a + g2 eutectoid. Hence, alloys such as Mn-Ni-Al bronze C95700 solidify in an all-b structure, although

Fig. 5

Phase diagrams of Cu-Al, Cu-Al-Fe, Cu-Al-Ni, and Cu-Al-Fe-Ni systems

Fig. 6

Different phases in nickel-aluminum bronzes.

Copper and Copper Alloy Castings / 1091 during slow cooling some alpha is formed. While aluminum has the greatest effect on properties (increasing strength and reducing ductility), manganese acts similarly but is less effective. C95700 is often selected because heat treatment is not required for excellent properties. Slow cooling following stress relief increases magnetic permeability and reduces ductility and impact resistance due to low-temperature precipitation of manganese-rich compounds. Silicon, like manganese, has a positive aluminum equivalent (1% Si is equivalent to 1.6% Al) and has the advantage of reducing the difficulties associated with the breakdown of the b phase to the a + g2 eutectoid at slow cooling rates.

Types of Copper Castings Copper alloys can be cast by many processes, including sand casting, permanent mold casting, precision casting (plaster and investment molds), high-pressure die casting, and low-pressure die casting. Green sand molds are the most common practice for copper alloys, although chemically bonded sand molds are used for very large castings such as ship propellers. Very small, intricate parts are made using investment or plaster molds. The permanent mold casting process is widely used for copper alloys with a narrow freezing range, such as aluminum bronzes and high coppers. These permanent metal molds can be used for 50,000 shots. The copper alloys are poured with solely the assistance of gravity (gravity casting) or forced upward into the die cavity by pressure (low-pressure permanent mold) or partial vacuum (vacuum casting). The advantages are near-net shape, fine surface finish, close tolerance control, thin section size, higher mechanical properties, and greater uniformity of casting variables. Copper alloy rolls and tubes are produced by the centrifugal casting process. Additional information on these casting processes can be found in the articles dedicated to the specific casting processes in this Volume.

Properties The nominal compositions of some sand-cast copper alloys are presented in Table 2, along with their typical strengths. Specified properties can be obtained from ASTM specifications B 148, B 369, B 584, B 763, and B 770. Tables 3 and 4 present the minimum mechanical properties specified for die-cast (B 176) and permanent mold cast bars (B 806), respectively.

Heat Treatment Few copper alloys are amenable to heat treatment. Nevertheless, many large castings can be improved by annealing and stress-relieving treatments. High coppers containing beryllium

or chromium are heat treated to obtain high strength and good conductivity. The other family of heat treatable copper alloys is aluminum bronzes containing more than 10% Al.

Beryllium Copper These alloys can be heat treated by solution treating and aging. Solution-treating temperature limits must be observed if optimal properties are to be obtained from the precipitation-hardening treatment. Exceeding the upper limit causes grain growth, which results in a brittle metal that does not fully respond to precipitation hardening. Solution heating below the specified minimum temperature results in insufficient solution of the beryllium-rich phase, a cause of lower hardness after precipitation hardening. After castings are solution treated, they are quenched in water. All castings, except those of alloy C82000, may be solution heated in air and water quenched immediately after removal from the furnace. Alloy C82000 must be solution treated in a protective atmosphere, such as cracked ammonia or natural gas. The duration of the solution treatment depends on section thickness, with the treatment time increased by 1 h for each additional inch of thickness for castings thicker than 25 mm (1 in.). Following solution heating, the castings are precipitation hardened. Table 5 shows the heat treating cycles for beryllium-copper alloys.

Chromium Copper Alloy C81500 (Cu-1Cr) can be heat treated in the same manner as beryllium copper. Here, the solution treatment is 1 h at 980 to 1015  C (1800 to 1860  F), followed by a water quench. Next, the castings are precipitation hardened at 500  C (930  F) for 2 h. Because chromium is sensitive to oxidation, a protective atmosphere should be used to avoid an oxidized zone of approximately 3 mm (0.1 in.) on the casting surface. If heat treating is done in an air furnace, the castings must be machined after treatment to remove this oxide in order to obtain accurate conductivity and hardness measurements.

Aluminum Bronzes These common high-strength casting alloys containing more than 10% Al are heat treatable. These are alloys whose normal microstructures contain more than one phase, to the extent that beneficial quench and temper treatments are possible. The copper-aluminum alloys normally containing iron are heat treated by procedures somewhat similar to those used for heat treatment of steel and have isothermal transformation diagrams that resemble those of carbon steels. For these alloys, the quench-hardening treatment is essentially a high-temperature soak intended to dissolve all of the beta phase into the alpha phase. Quenching results in a hard,

room-temperature, martensitic phase that can be subsequently tempered to promote the formation of fine precipitates in the form of needles, known as tempered martensite. Table 6 shows typical heat treatments for three principal aluminum bronze alloys. Protective atmospheres should be used to avoid an oxidized zone of approximately 3 mm (0.12 in.) on the casting surface. Castings heat treated in an air furnace must be machined after treatment to remove this oxide in order to obtain accurate conductivity and hardness measurements.

Welding Copper alloys, such as cupronickels, aluminum bronzes, and manganese bronzes, are readily weldable by a variety of methods. Other alloys are weldable only by the use of special techniques.

Cupronickels Within the recommended chemistry ranges, the cupronickel alloys are freely weldable by all methods. These alloys may be welded by shielded metal arc, gas metal arc, or gas tungsten arc methods. The shielded metal arc process is most often used, mainly due to the more convenient interpass temperature control and the ready availability of suitable filler metal. American Welding Society class 5.6 ECuNi filler metals are available for welding these alloys. It is desirable to limit the silicon content in the chromium-modified 70/30 alloy between 0.3 and 0.5% to enhance weldability. The presence of certain impurities, such as lead, sulfur, phosphorus, bismuth, selenium, antimony, and boron, can contribute to poor weldability and have a detrimental effect on weld properties. Interpass temperatures should not exceed 120  C (250  F).

Aluminum Bronzes Aluminum bronzes, such as C95200, C95300, C95400, and C95800, can be joined by most welding processes using commercially available electrodes (Ampcotrode 10 for 952 and 953, Ampcotrode 150 for 954, and E10 for 958). Metal inert gas (MIG) shielded arc or tungsten inert gas (TIG) arc welding produce the best results. Metal arc processes using the correct grade of coated electrodes give satisfactory results. Carbon arc welding has been reported to produce good results. However, oxyacetylene welding is seldom successful. Weld repairing to rectify small defects and building up and hardfacing are being performed on cast components. Interpass temperatures can vary between 38 and 177  C (100 and 350  F) for C95800 and between 205 and 425  C (400 and 800  F) for C95200, C95300, and C95400.

1092 / Nonferrous Alloy Castings Table 2

Nominal chemical composition and typical mechanical properties of sand-cast alloys Nominal composition, %

Alloy type

UNS No.

Cu

Sn

Pb

Zn

Ni

Fe

Al

Yield strength, 0.5%

Mn

Si

Other

MPa

ksi

Tensile strength MPa

ksi

Elongation, %

Freezing range,  C

Copper

C81100

100

...

...

...

...

...

...

...

...

...

62

9

172

25

40

19

Chrome copper

C81500 as-cast Heat treated

99

...

...

...

...

...

...

...

...

1 Cr

83 290

12 42

214 379

31 55

25 18

10 27

Beryllium copper

C81400 as-cast Heat treated C82000 as-cast Heat treated C82200 as-cast Heat treated C82400 as-cast Heat treated C82500 as-cast Heat treated C82600 as-cast Heat treated C82800 as-cast Heat treated

99.1

...

...

...

...

...

...

...

...

0.06 Be, 0.8 Cr

...

...

...

...

...

96.8

...

...

...

...

...

...

...

...

0.6 Be, 2.6 Co

...

...

...

2.0

...

...

...

...

0.6 Be, 0.3 Co

97.8

...

...

...

...

...

...

...

...

1.7 Be, 0.43 Co

97.2

...

...

...

...

...

...

...

0.3

2.0 Be, 0.5 Co

96.8

...

...

...

...

...

...

...

0.3

2.4 Be, 0.5 Co

96.6

...

...

...

...

...

...

...

0.3

2.6 Be, 0.5 Co

20 75 (HT) 25 75 (HT) 40 145 45 150 50 155 50 155

345 662 (HT) 380 655 (HT) 489 1069 552 1103 552 1138 552 1138

50 96 (HT) 55 95 (HT) 70 155 80 155 80 165 80 165

20 6 (HT) 20 7 (HT) 15 1 20 1 10 1 10 1

117

97.9

138 517 (HT) 206 517 (HT) 276 1000 310 1035 345 1069 345 1069

C83450

88

2.5

2

6.5

1

...

...

...

...

...

103

15

255

37

31

155

C83600 C83800

85 83

5 4

5 6

5 7

... ...

... ...

... ...

... ...

... ...

... ...

117 110

17 16

255 241

37 35

30 25

156 161

Leaded semired brass

C84400

81

3

7

9

...

...

...

...

...

...

103

15

234

34

26

161

C84800

76

2.5

6.5

15

...

...

...

...

...

...

97

14

255

37

35

122

Leaded yellow brass

C85200

72

1

3

24

...

...

...

...

...

...

90

13

262

38

35

14

C85400 C85700

67 61

1 1

3 1

29 37

... ...

... ...

... ...

... ...

... ...

... ...

83 124

12 18

234 345

34 50

35 40

14 28

C86200

63

...

...

27

...

3

4

3

...

...

331

48

655

95

20

43

C86300 C86400 C86500 C86700

61 58 58 58

... 1 ... 1

... 1 ... 1

27 38 39 34

... ... ... ...

3 1 1 2

6 0.5 1 2

3 0.5 1 2

... ... ... ...

... ... ... ...

462 172 200 290

67 25 29 42

821 448 490 586

119 65 71 85

18 20 30 20

38 18 18 18

Silicon brass

C87400

82

...

...

14

...

...

...

...

3.5

0.5 Pb

165

24

379

55

30

95

C87500

82

...

...

14

...

...

...

...

4

...

207

30

462

67

21

95

Silicon bronze

C87300

95

...

...

...

...

...

...

1

4

...

...

...

...

...

...

...

C87600 C87610 C87800

91 92 82

... ... ...

... ... ...

5 4 14

... ... ...

... ... ...

... ... ...

... ... ...

4 4 4

... ... ...

221 170 345

32 25 50

455 400 586

66 58 85

20 35 25

111 111 ...

C89510

87

5

...

5

...

...

...

...

...

1 Bi, 0.5 Se

124

18

241

35

20

205

Leaded red brass

Manganese bronze

Bismuthselenium brass

88 97 125 97 47

C89520

86

5.5

...

5

...

...

...

...

...

1.9 Bi, 0.9 Se

124

18

214

31

15

196

Bismuth brass Tin bronze

C89844

84.5

4

...

8

...

...

...

...

...

3 Bi

103

15

234

34

30

167

C90300 C90500

88 88

8 10

... ...

4 2

... ...

... ...

... ...

... ...

... ...

... ...

145 152

21 22

310 310

45 45

30 25

146 145

Leaded tin bronze

C92200

88

6

1.5

4.5

...

...

...

...

...

...

138

20

276

40

30

162

C92300 C92600

87 87

8 10

1 1

4 2

... ...

... ...

... ...

... ...

... ...

... ...

138 138

20 20

276 303

40 44

25 30

145 139

C93200

83

7

7

3

...

...

...

...

...

...

124

18

241

35

20

123

C93500 C93700 C93800 C94300

85 80 78 71

5 10 7 5

9 10 15 24

1 ... ... ...

... ... ... ...

... ... ... ...

... ... ... ...

... ... ... ...

... ... ... ...

... ... ... ...

110 124 110 90

16 18 16 13

221 241 207 186

32 35 30 27

20 20 18 15

145 167 89 ...

C95200

88

...

...

...

...

3

9

...

...

...

186

27

552

80

35

3

C95300 as-cast Heat treated C95400 as-cast Heat treated C95410 as-cast Heat treated C95500 as-cast Heat treated C95600

89

...

...

...

...

1

10

...

...

...

...

...

...

...

4

11

...

...

...

84

...

...

...

2

4

10

...

...

...

81

...

...

...

4

4

11

...

...

...

91

...

...

...

...

...

7

...

2

...

27 42 35 54 35 54 44 68 34

517 586 586 724 586 724 689 827 517

75 85 85 105 85 105 100 120 75

25 15 18 8 18 8 12 10 18

5

85

186 290 241 372 241 372 303 469 234

High-leaded tin bronze

Aluminum bronze

(continued) HT, heat treated

11 11 16 22

Copper and Copper Alloy Castings / 1093 Table 2

(continued) Nominal composition, %

Alloy type

UNS No.

Cu

Sn

Pb

Zn

Ni

Yield strength, 0.5%

Fe

Al

Mn

Si

Other

MPa

ksi

Tensile strength MPa

ksi

Elongation, %

Freezing range,  C

C95700 C95800

75 81

... ...

... ...

... ...

2 4.5

3 4

8 9

12 1.2

... ...

... ...

310 262

45 38

655 655

95 95

26 25

40 17

Copper nickel

C96200

87.5

...

...

...

10

1.5

...

0.9

0.1

...

...

...

...

...

...

50

C96400

67

...

...

...

30

0.7

...

0.8

0.5

1.0 Nb

255

37

469

68

28

67

Coppernickelberyllium Leaded nickel bronze

C96700 Heat treated

67.2

0.3

...

...

31.0

0.6

...

0.7

...

1.2 Be, 0.3 Zr, 0.3 Ti

552 (HT)

80 (HT)

862 (HT)

125 (HT)

10 (HT)

...

C97300

57

2

9

20

12

...

...

...

...

...

117

17

241

35

20

30

C97600 C97800

64 66

4 5

4 2

6 2

20 25

... ...

... ...

... ...

... ...

... ...

165 207

24 30

310 379

45 55

20 15

35 40

HT, heat treated

Table 6 Heat treatment of aluminum bronze alloys

Table 3 Nominal chemical composition and typical mechanical properties of die-cast alloys Yield strength, 0.5%

Nominal composition, %

Tensile strength

Elongation, %

ASTM specification

Alloy type

UNS No.

Cu

Sn

Pb

Zn

ksi

MPa ksi

Yellow brass

C85700

61

1

1

37 . . . . . . . . . . . . . . .

...

124

18

345

50

40

B 176

C85800 C86500

62 1 1 36 . . . . . . . . . . . . . . . 58 . . . . . . 39 . . . 1 1 1 ...

... ...

207 200

30 29

379 490

55 71

15 30

B 176 B 176

C87800

82 . . . . . . 14 . . . . . . . . . . . .

4

...

345

50

586

85

25

B 176

C99700

58 . . .

2

22

...

1

12

...

...

186

27

448

65

15

B 176

C99750

58 . . .

1

20 . . . . . .

1

20 . . .

...

221

32

448

65

30

B 176

Manganese bronze Silicon bronze White brass

Ni

5

Fe

Al

Mn

Si

Other MPa

Alloy type

Aluminum bronze

Silicon bronze White brass Silicon bronze

Table 5

UNS No.

Yield strength, 0.5%

Tensile strength

ksi MPa

ksi

Elongation, %

ASTM specification

235

34

565

82

20

B 806

... ... ...

320 275 254

46 40 37

725 105 690 100 575 83

11 10 29

B 806 B 806 B 806

C99700 58 . . . 2 22 5 . . . 1 12 . . . C87500 82 . . . . . . 14 . . . . . . . . . . . . 4

... ...

275 249

40 36

620 566

90 82

15 15

B 806 B 806

C87800 82 . . . . . . 14 . . . . . . . . . . . .

...

254

37

575

83

29

B 806

Cu

Sn

Pb

Zn

Ni

Fe

Al

Mn

Si

C95300 89 . . . . . . . . . . . .

1

10 . . . . . .

...

C95400 84 . . . . . . . . . 1 4 11 . . . . . . C95410 83 . . . . . . . . . 2 4 11 1 . . . C87800 82 . . . . . . 14 . . . . . . . . . . . . 4

Other MPa

4

Heat treatment of beryllium coppers Solution treatment

Aging treatment

Temperature Alloy

C81400 C82000 C82200 C82400 C82500 C82600 C82800



C

980–1010 915–925 925 800–815 800–815 800–815 800–815

F

Time, h

1800–1850 1675–1700 1700 1475–1500 1475–1500 1475–1500 1475–1500

1 1 1 1 1 1 1

C95400 and C95410 C95500

>1 h at 800–890  C (1470–1635  F) >1 h at 870–910  C (1600–1675  F)

>2 h at 620–660  C (1150–1225  F) >2 h at 620–660  C (1150–1225  F)

2 h at 870–925  C (1600–1700  F)

>2 h at 495–540  C (925–1000  F)



C

480 480 455 345 345 345 345



Oxide films that are always present in aluminum bronzes should be removed by using a flux or by grinding or scratch-brushing before welding. Degreasing should also be done to remove oil films that may cause porosity. Postweld stress relief or heat treatment is not performed on large components such as propellers, unless a customer’s specification requires it. However, castings for marine applications should be postweld heat treated at approximately 700  C (1290  F) for 6 h. The Cu-Mn-Al bronzes (alloy C95700) can be easily welded because of their low thermal conductivity and the absence of a brittle temperature range. The procedure is similar to aluminum bronzes. Flux-coated electrodes of matching composition are available for MIG and TIG welding. Preheating is not necessary but may be advantageous to minimize distortion. Postweld heat treatment at 650  C (1200  F) is recommended to achieve maximum corrosion resistance.

Manganese Bronzes

Temperature 

C95300

Annealing treatment (followed by air quench)

Source: Ref 1

Table 4 Nominal chemical composition and minimum specified mechanical properties of permanent mold cast alloys Nominal composition, %

Alloy

Solution treatment (followed by water quench)

F

Time, h

900 900 850 650 650 650 650

2 3 3 3 3 3 3

The C86500 manganese bronze is welded using TIG or MIG. The parent metal must have an alpha-beta microstructure to be weldable. The surfaces are prepared as for aluminum bronzes but preheated to 150  C (300  F) before welding. Interpass temperatures are kept between 150 and 427  C (300 and 800  F).

1094 / Nonferrous Alloy Castings A postweld stress relief is performed, heating at 10  C (18  F) or less per hour to 315 to 427  C (600 to 800  F), holding for 6 h, and cooling at a rate of 10  C (18  F) or less per hour.

Tin Bronzes and Leaded Tin Bronzes These bronzes have a long solidification range and are prone to hot shortness, which gives rise to cracking on cooling. Hence, they are difficult to weld or weld repair. Another welding problem is related to electrodes. Although flux-coated electrodes are available for metal arc welding, the slag formed during welding is very strong and adherent. Unsound welds will result if such adherent slag is not removed. Hence, it may be necessary to per-

form interpass cleaning by grinding between each pass. Use of an inert gas shielding process can minimize the formation of slag. Leaded tin bronzes can be welded using alternating welding current with a thoriated tungsten electrode for the tungsten arc process and direct current electrode positive for the metal arc and inert gas shielded metal arc process. It is advisable to avoid sputtering during welding because it may cause porosity.

SELECTED REFERENCES  Casting    

REFERENCE 1. “Standard Specification for AluminumBronze Sand Castings,” B 148–97, Annual Book of ASTM Standards, ASTM International, 2003



Copper-Base Alloys, American Foundry Society, 2006 Cast Products Alloy Data, Standards Handbook, Part 7, Copper Development Association, Inc., 1978 Heat Treating, Vol 4, Metals Handbook, 9th ed., American Society for Metals, 1981 F. Hudson, Gunmetal Castings, Hart Publishing Co, Inc., New York, 1968 P.J. Macken and A.A. Smith, The Aluminum Bronzes, Copper Development Association, 1966 H. Meigh, Cast and Wrought Aluminum Bronzes—Properties, Processes and Structure, Copper Development Association, U.K., 2000

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1095-1099 DOI: 10.1361/asmhba0005333

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Zinc and Zinc Alloy Castings Frank E. Goodwin, International Zinc Association

DIE CASTING is the process most often used for shaping zinc alloys. Sand casting, permanent mold casting, and continuous casting of zinc alloys are also practiced, but account for less than 10% of zinc casting tonnage. All processes are based on the same family of alloys. The most commonly used zinc casting alloys are UNS Z33520 (alloy 3) and a modification of this alloy UNS Z33523 distinguished by the commercial designation alloy 7. The compositions of these alloys are given in Table 1, while mechanical properties of cast zinc alloys are shown in Table 2. Although alloy 3 is more frequently specified, the properties of the two alloys are generally similar. The second and fourth alloys listed in Table 1, UNS Z35531 and UNS Z35541 are used when

Table 1

higher tensile strength or hardness is required. The former is commonly termed alloy 5 and the latter alloy 2. Alloy 2 is more common in Europe than North America. Alloys 2, 3, 5, and 7 are used exclusively for pressure die casting. Higher-aluminum zinc-base casting alloys are also available. These alloys were originally developed for gravity casting where they have substituted for aluminum, brass, bronze, and cast malleable iron in uses ranging from plumbing fixtures, pumps, and impellers to automotive vehicle parts and, recently, bronze-bearing substitutes. The three members of this family of alloys are generically identified industrywide as ZA-8, ZA-12, and ZA-27. As with the loweraluminum conventional zinc die casting alloys,

Compositions of zinc casting alloys

Alloy

Applicable Common standards designation UNS No. (ASTM)

No. 3 No. 5 No. 7 No. 2 ZA-8 ZA-12 ZA-27 ACuZinc 5

Z33530 Z35531 Z33523 Z35541 Z35636 Z35631 Z35841 Z46541

B 86 B 86 B 86 B 86 B 86 B 86 B 86 B 894

Alloying Additions

Composition(a), wt% Al

3.5–4.3 3.5–4.3 3.5–4.3 3.5–4.3 8.0–8.8 10.5–11.5 25.0–28.0 2.9

Cu

Mg

0.25 0.75–1.25 0.25 2.5–3.0 0.8–1.3 0.5–1.25 2.0–2.5 5.5

0.02–0.03 0.03–0.08 0.005–0.02 0.020–0.050 0.015–0.03 0.015–0.03 0.01–0.02 0.04

Fe

0.100 0.075 0.10 0.100 0.075 0.10 ... 0.075

Pb

Cd

Sn

Ni

Zn

0.005 0.005 0.003 0.005 0.005 0.004 0.004 0.005

0.004 0.004 0.002 0.004 0.004 0.003 0.003 0.004

0.003 0.003 0.001 0.003 0.003 0.002 0.002 0.003

... ... 0.005–0.02 ... ... ... ... ...

bal bal bal bal bal bal bal bal

(a) Maximum unless range is given or otherwise indicated

Table 2 Alloy and product form(s)

they contain zinc plus aluminum, with small amounts of copper and magnesium. The numerical components of the ZA alloys indicate the approximate aluminum content. Compositions and properties of these alloys are also shown in Tables 1 and 2, respectively. All three ZA alloys show hardness superior or equivalent to that of aluminum, brass, or bronze, allowing improved wear resistance together with resistance to galling. Alloys 2, 3, 5, 7, ZA-8, and ZA-12 are considered nonincendiary and sparkproof. This means that the zinc alloys will not ignite hazardous fuel air mixtures, vapors, or particulate matter when struck by rusted ferrous materials. The nonmagnetic properties of zinc make it suitable for use in electronics and for delicate moving parts that might otherwise be adversely affected by magnetic disturbances.

Comparison of typical mechanical properties of zinc casting alloys Ultimate tensile strength MPa

No. 2 (D) 358 No. 3 (D) 283 No. 5 (D) 331 No. 7 (D) 283 ZA-8 (S) 248–276 ZA-8 (P) 221–255 ZA-8 (D) 372 ZA-12 (S) 276–317 ZA-12 (P) 310–345 ZA-12 (D) 400 ZA-27 (Sb) 400–440 ZA-27 (P) 421–427 ZA-27 (D) 421

ksi

52 41 48 41 36–40 32–37 54 40–46 45–50 58 58–64 61–62 61

0.2% offset yield Elongation strength in 50 mm Hardness, MPa ksi (2 in.), % HB

... ... ... ... 200 207 290 207 207 317 365 365 365

... ... ... ... 29 30 42 30 30 46 53 53 53

7 10 7 13 1–2 1–2 6–10 1–3 1–3 4–7 3–6 1 1–3

100 82 91 80 80–90 85–90 95–110 90–105 90–105 95–115 110–120 110–120 105–125

Impact strength




Fatigue strength

Young’s modulus

J

ft lbf

MPa

ksi

GPa

106 psi

48(c) 58(c) 65(c) 58(c) 20 ... 42(c) 25(c) ... 28(c) 47(c) ... ...

35 43 48 43 15 ... 31 19 ... 21 35 ... ...

58 47.6 56.5 ... ... 51.8 ... 103.5 ... ... 172.5 ... ...

8.5 6.9 8.2 ... ... 7.5 ... 15 ... ... 25 ... ...

... ... ... ... 85.5 85.5 ... 83.0 83.0 ... 75.2 75.2 ...

... ... ... ... 12.4 12.4 ... 12.0 12.0 ... 10.9 10.9 ...

(a) D, die cast; S, sand cast; P, permanent cast. (b) As-cast. (c) Unnotched Charpy. (d) Notched Charpy. (f) Izod

Strict control of chemical composition, including limiting of impurities, is absolutely essential to prevent any chance of intergranular (intercrystalline) corrosion, dimensional changes, or loss of mechanical properties. Aluminum is added to zinc to strengthen the alloy, reduce grain size, and minimize the attack of the molten metal on the iron and steel in the casting and handling equipment. Aluminum increases the fluidity of the molten alloy and improves its castability. As indicated in Table 1, aluminum content ranges from 3.5 to 4.3% for alloys 3 and 5. An aluminum content lower than 3.5% requires higher than normal metal temperatures for satisfactory castability. The higher temperatures result in accelerated attack on the dies. Other disadvantages of a low aluminum content are low strength and less dimensional stability than alloys containing aluminum in the specified range. When aluminum content is higher than 4.3%, the impact strength of the castings is reduced. The extremely brittle zinc-aluminum eutectic forms at about 5% Al and must be avoided. Magnesium content must also be carefully maintained within the ranges shown in Table 1. Magnesium is added primarily to minimize susceptibility to intergranular corrosion caused

1096 / Nonferrous Alloy Castings by the presence of impurities. Excessive amounts of magnesium lower the fluidity of the melt, promote hot tear cracking, increase hardness, and decrease elongation. Cracking is confined to castings of complex form that are free to shrink in the die. Copper, like magnesium, minimizes the undesirable effects of impurities and, to a smaller extent, increases the hardness and strength of the castings. Castings containing more than about 1.25% Cu are less dimensionally stable than those with less copper. The copper range for alloys 5 is 0.6 to 1.25. Keeping copper content above the lower limit in each alloy in Table 1 ensures the high tensile and hardness values shown, while keeping the copper content below the upper limit indicated for each alloy ensures that aging changes in the castings at room temperature will be minimized. Iron in amounts up to 0.10% has little detrimental effect, but may contribute to problems in buffing or machining. Iron will be in solid solution in zinc when the amount is less than 0.02 wt%. Higher iron content, that is, more than 0.02 wt% in zinc casting alloys, can result in the formation of hard iron-aluminum compounds, which can produce comet tails during buffing and can dull tools during machining. Nickel, chromium, silicon, and manganese are not harmful in amounts up to the solubility limit of each in zinc (0.02% Ni, 0.02% Cr, 0.035% Si, and 0.5% Mn). When these metals exceed their solubility limits, they form intermetallic compounds with aluminum that are less dense than the zinc melt and therefore can be skimmed off its surface. Lead, cadmium, and tin at levels exceeding the limits shown in Table 1 can cause die cast parts to swell, crack, or distort. The defects can occur within 1 year. The maximum limit

for lead, which can promote the occurrence of a subsurface network of corrosion, is 0.006%. Cadmium is detrimental in its effect at some concentrations and neutral at others. As such, the maximum limit for cadmium is at 0.005%. Tin, like lead, can promote subsurface intergranular corrosion and is more potent than lead. For this reason, its impurity limit is set at 0.003% for the conventional casting alloys.

Physical Metallurgy The conventional zinc casting alloys are hypoeutectic, and under equilibrium solidification the zinc-rich eta phase should form first, followed by eutectic. However, the rapidity of solidification during the zinc die casting process results in metastable structures. In alloys 2, 3, 5, and 7 a characteristic five-layer microstructure is seen. This consists of:  Layer 1—a surface chill layer of fine pri   

mary eta in a dark etching eutectic matrix Layer 2—a thin often multiple layer of cellular eutectic Layer 3—a thick layer of “normal” structure, that is, dendrites of eta within a eutectic matrix Layer 4—an irregular layer richer in eutectic than normal Layer 5—a thick central region of increased primary dendrite content

Fig. 2, which shows a very weak layer 2 formation and a very similar structure in layers 1 and 2. This casting was made with a gate velocity of 35 m/s (115 ft/s) and a die temperature of 235  C (455  F) followed by air cooling. Figure 3 shows very poor development of the surface layers. This was cast with the same conditions as the alloy in Fig. 2 except that a 1.5 mm (0.06 in.) thickness was cast rather than 0.75 mm (0.03 in.). Finally, very well formed surface layers are shown in Fig. 4. This casting was made with a gate velocity of 50 m/s (164 ft/s), a die temperature of 235  C (455  F), a wall thickness of 1.5 mm (0.06 in.), and was water quenched after ejection. The degree of development of the layer structures is slightly greater in alloy 3 than in alloy 5. Alloys ZA-8 and 2 also have five distinguishable layers of microstructure; however,

Layer 1 Layer 2

Layer 3

50 µm

Fig. 2

Similar layers 1 and 3 and vestigal layer 2, alloy 3

Depending on process conditions, each of these layers can be present in greater or lesser quantity, or even absent. In alloy 2, for example, a multiple layer 2 formation can be seen in Fig. 1, a 0.75 mm (0.03 in.) thick casting with a gate velocity of 50 m/s (164 ft/s) and a die temperature of 180  C (355  F). This can be contrasted with

500 µm

Fig. 3

Short sides of alloy 3 casting, showing poor development of surface layers

Layer 2 Layer 1

500 µm

500 µm

Fig. 1

Multiple layer 2 formation at the surface of alloy 2

Fig. 4

Well-formed surface layers and side fissure, alloy 3

Zinc and Zinc Alloy Castings / 1097 in the case of ZA-8, a hypereutectic alloy, an aluminum-rich beta phase is observed rather than the primary zinc-rich eta phase. The primary beta phase that forms in ZA-8 decomposes below 280  C (535  F). The aluminum-rich phase is etched dark, while the eutectic etch is white, in contrast with the lower aluminum content alloys. The development of layers and other features is similar to alloys with lower aluminum content; however, surface layers overall are less well developed, and the central layer more strongly developed, in ZA-8. The higher copper content of alloy 2 does not produce any significant changes in microstructure with process variable changes. The degree of development of the five-layer structure is less than that observed in alloys 3 and 5.

Die Casting and Aging The lower aluminum alloys, Alloys 2, 3, 5, and 7, are always cast in hot-chamber zinc die casting machines. ZA-8 can be hot-chamber die cast and is also permanent mold cast. As a pressure die casting, it has excellent properties, especially creep, compared with other zinc alloys. ZA-12 and ZA-27 are cold-chamber die cast and can be sand cast, too. ZA-12 is also graphite permanent mold cast. In all conditions ZA-12 shows excellent pressure tightness and can be easily machined. Because of its wide freezing range ZA-27 can require careful control during solidification, especially with regard to the use of chills to obtain homogeneous castings. All three ZA alloys are also available in continuous cast solid and hollow shapes that are most often used for bearing or prototyping applications. A key advantage of the conventional lowaluminum zinc die castings alloys is their ability to be cast to very precise dimensions, often eliminating the need for machining and other secondary operations. Related to this is their ability to be cast with zero or negligible draft, allowing shafts to run in cored holes, or gear assemblies to run together, without further machining. These alloys are cast by the highly automated hot-chamber die casting process

and increasingly use small high-speed machines, thus improving the economics of production. The casting fluidity of these alloys is excellent and castings of section thicknesses of 0.75 mm (0.03 in.) is commonplace, offsetting the higher density of zinc alloys compared with the less fluid aluminum and magnesium die casting alloy compositions. In many cases, the bearing properties of these zinc alloys are sufficient to preclude the need for pressed-in copper bearings, further reducing cost. The excellent castability and mechanical properties of these alloys has allowed previous multipart assemblies to be cast as a single piece, greatly reducing costs. The tensile properties of these alloys are influenced by casting wall thickness with a decrease in casting thickness resulting in an increase in engineering stress. Low die temperatures and water quenching of parts after casting also increase strength. Ductility increases with increasing thickness and die temperature and also with air cooling rather than water quenching. Gate velocity appears to have little effect on tensile properties but, because of interactions with other casting parameters, a relatively low gate velocity is preferred if maximum properties are to be obtained. The effects of process variables and section thickness on the mechanical properties of Alloy 3 are shown in Fig. 5. The thicker castings have reduced ultimate tensile strength and yield strength compared to thinner castings. As die temperature increases, the strength reduction between the thinner and thicker castings increases, demonstrating the beneficial effect of lower die temperature. Also, as die temperature increases, the Young’s modulus and elongation of thick castings is greater than for thin castings. As cooling rate increases (by direct quenching of the casting as it ejected from the die rather than air cooling), thick castings have greater ultimate tensile strength and yield strength, but ductility is reduced. Finally, the interaction between thickness and cooling rate shows that the beneficial effect of cooling rate on strength is reduced by increasing thickness, but ductility is unaffected. This indicates that the favorable effect on strength of a low casting

thickness is stronger if the specimens are water quenched after ejection. The opposite effect is also true; the negative influence of a high wall thickness is increased if the specimens are air cooled. Similar trends are also seen with alloy 2, 5, and ZA-8 with increased gate velocity exerting a stronger effect on increasing strength for these two alloys. During the rapid solidification of die casting, copper in the primary phase is overly enriched and, with time, precipitates as the equilibrium phase. This results in a volume change, and stabilizing heat treatments are routinely applied if rigid dimensional tolerances are to be met. The higher the heat treatment temperature the shorter the stabilizing time required. A practical limit to heat treatment temperature to prevent blistering of the casting or other problems is 100  C (212  F). A common heat treatment consists of 3 to 6 h at 100  C (212  F) followed by air cooling. The time required for heat treatment extends to 10 to 20 h for a treatment temperature of 70  C (158  F). In laboratory work, longer treatment times (95  C (203  F) for 10 days) were used, but this longer treatment only removes a small portion of dimensional changes beyond the shorter treatment times given previously. Aging at 95  C (203  F) for 10 days reduces the ultimate tensile strength of the low-aluminum alloys by about 40 MPa (6 ksi) and that of the higher aluminum ZA alloys by about 80 MPa (12 ksi). The elongation of alloy 2 was reduced from 7% to 2% by this aging treatment while the elongations of all other alloys are increased. Impact strength of alloy 2 is also adversely affected by this aging treatment, reducing from 48 to 7 J (35 to 5 ft lbf). Impact properties of the other low-aluminum alloys are minimally affected. For ZA-8, impact strength is reduced from 42 to 17 J (31 to 13 ft lbf) after aging, whereas for ZA-12 the reduction is from 29 to 19 J (21 to 14 ft lbf). For ZA-27, the reduction in impact strength is 5 to 2.2 J (4 to 1.6 ft lbf). Consistent with metallographic observations, the minimal effect of aging on alloy 3, 5, and 7 is related to their lower copper content compared with the other alloys.













Specialty Alloys Thickness Die temperture

E (GPa) EL (%) UTS (MPa)

Cooling rate

YS (MPa)

Tickness/ cooling rate −30

Fig. 5

−20

−10

0

10

20

Variation of mechanical properties with casting parameters between 0.75 mm (0.03 in.) and 1.5 mm (0.06 in.) castings of alloy 3. Tensile testing conducted at 20  C (68  F)

The gravity-cast version of alloy 2 is known as “Kirksite” and finds application as a forming die alloy. In sand cast form, it is used to make two-piece dies for forming of sheet metal parts such as components for use in transportation and aerospace industries. In Europe, alloys of comparable composition, designated Kayem 1 and Kayem 2, are used in similar applications. Cast-to-size molds made of Kirksite are used for plastics injection molding for both shortterm prototyping and production operations. Two die-forming alloy compositions are included in ASTM B 793 (Table 3). Slush casting alloys are used extensively for the production of hollow castings such as table

1098 / Nonferrous Alloy Castings Table 3 Nominal compositions of zinc casting alloys used for sheet metal forming dies and for slush casting alloys in ingot form Alloy Common designation

Composition, % UNS No.

Al

Cd max

Cu

Fe max

Pb max

Mg

Sn max

Zn

Forming die alloys (ASTM B 793) Alloy A Z25543 3.5–4.3 Alloy B Z35542 3.9–4.3

0.005 0.003

2.5–3.5 2.5–2.9

0.100 0.075

0.007 0.003

0.02–0.10 0.02–0.05

0.005 0.001

bal bal

Slush casting alloys (ASTM B 792) Alloy A Z34510 4.50–5.00 Alloy B Z30500 5.25–5.75

0.005 0.005

0.2–0.3 0.1 max

0.100 0.100

0.007 0.007

... ...

0.005 0.005

bal bal

lamp bases. The molten alloy is poured into the mold unit until it is full or nearly full, and then the mold is inverted allowing the unsolidified metal (slush) to run out. The solidified shell that is left is then removed. The thickness of the shell depends on the time interval between pouring and inverting the mold, the melt and mold temperatures, and the mold material. Two slush casting alloys are included in ASTM B 792. A higher-copper die casting alloy is also used in commercial production and is listed in ASTM B 894. This alloy, commonly referred to as ACuZinc 5, has 5 to 6% Cu, 2.5 to 3.3% Al, 0.025 to 0.05% Mg, and limitations on impurity similar to the conventional low-aluminum casting alloys (see Table 1). The tensile strength of this alloy ranges between 310 and 355 MPa (45 and 51 ksi) with a yield stress in the range of 240 to 284 MPa (35 to 41 ksi). Elongations between 4.6 and 9.4% are measured. The Brinell hardness is 105 to 115 (500 kg, 10 Mn).

Finishing of Die Castings For many applications, surface finish is not important, but where necessary, a wide range of applied finishes can be used. Finishes for die castings can be functional, decorative, or both. Almost any desired texture, such as simulated cloth, leather, or wood grain, can be cast simply and economically. Zinc-base alloys have good corrosion resistance in normal atmospheric conditions, in aqueous solutions, and when used with petroleum products. The corrosion resistance is enhanced by painting, plating, or chromate or phosphate treatment and is substantially improved by anodizing. The ZA alloys readily accept a wide variety of decorative and corrosion-resistant surface finishes. Low-cost components are painted to match adjacent parts, chromium plated to offer a durable luster (except ZA-27), and brush finished to take on the rich appearance of brass, bronze, or stainless steel at a fraction of the cost. Anodizing produces a thin, ceramiclike, abrasion-resistant film exhibiting excellent resistance to most natural and industrial corrosive materials. Chromium plating is the most widely used finish for zinc alloy die castings. Normally, to obtain a high-quality bright chromium finish

on castings, they must be buffed and polished before plating. However, a combination of modern zinc die casting and plating techniques substantially reduces or, in some cases, eliminates the need for mechanical polishing. If a finish is decorative rather than protective, this does not imply that it requires only casual consideration; electroplating must be carried out to an exacting specification if it is to be satisfactory. For example, an unplated die cast door handle on an automobile would suffer little corrosion (apart from discoloration) in service, while a poorly plated one would soon exhibit blistering and pitting. Therefore, plating is normally required only for appearance, but when it is used, it must be of sufficient quality to provide long-term durability. Polishing. Where polishing is necessary, the following points must be kept in mind. Sharp corners and edges are difficult to polish without damaging their outlines; a radius of 0.8 mm (0.03 in.) will help polishing and plating. Large flat surfaces are difficult to polish evenly, but undulations will be less noticeable if the surface has a slight crown (1.5 mm/100 mm, or 0.015 in./in.). Small recesses and acute angles are impossible to reach with a polishing wheel. Chromate Finishes. Chemical treatments prevent the growth of white corrosion products on the surface when the castings are exposed to stagnant moisture, such as condensation. This treatment results in a dull green-yellow finish. Painting. Most types of paint can be successfully applied, but surface preparation and pretreatment are very important. Clear Lacquers. Many attempts have been made to formulate clear lacquers to preserve the bright finish of polished zinc. The recently developed nonyellowing polyurethane lacquers containing rubeanic acid show promise for this application. Acrylic lacquers are also being used, but they are less resistant to mechanical damage. Electropainting (Electrophoretic Painting.) The electropainting of die castings is well established. The process requires specially formulated water-base soluble resin paints, and this imposes some limitations on the colored pigments that can be used. The advantage of electropainting, in which the parts to be painted are made cathodic relative to the steel tank containing the paint, is that a very dense and uniform paint film can be applied to a complex surface, so that one coat of electropaint can

replace two coats of conventional paint. It is possible to achieve a high gloss with electrodeposited paints, but color matching with other paints still presents problems, especially over long production runs. Only one coat can be electrodeposited because the dry paint is an insulator. Some paint manufacturers recommend giving zinc die castings a chromate treatment before electropainting. Plastic Coatings. Epoxy powder coatings are being increasingly used for zinc die castings as an alternative to heat curing paints. The process is economical because the only labor required is loading and unloading the conveyor that carries the castings past the spray gun and through the curing oven. Matte or glossy coatings are formed about 0.04 mm (1.5 mils) thick. Plated Finishes. Where zinc alloy die castings are correctly designed, properly produced and carefully prepared plated finishes are very satisfactory. They have excellent decorative value and, given a coating of sufficient thickness and good adherence, a very long life. Copper and nickel can both be plated directly onto zinc alloy, but it is nickel applied over a preliminary deposit of copper that is most commonly used to enhance the corrosion resistance of the casting, coupled with a chromium finish. For special effects, die castings plated with copper can be treated to simulate antique bronze. Nickel can also be wire brushed or polished with a coarse abrasive to give various attractive satin finishes before chromium plating. Black chromium is another finish with decorative possibilities. Vacuum Metallizing. Die castings that need a bright finish but are unlikely to be damaged by rubbing or knocks can be vacuum metallized. Color effects, such as those of brass, can be created by immersing parts in a dye solution that tints the protective lacquer very evenly. Anodizing. A process for anodizing zinc and zinc alloy die castings has been developed that provides excellent resistance to salt water and detergent solutions and good resistance to abrasion. The finish is available only in matte, light green, gray, and brownish tints.

Metalworking and Machining Bending and Forming. The ductility of zinc makes it possible to incorporate integral rivets, to shape integral flanges to curving contours, to bend hollow arms, to spin out undercuts, to form projections, or to twist parts of the casting through 90 or more. It is possible to use a flat parting that provides parallel bosses, cored holes, or studs at right angles to the parting and then to form the casting so that the axes of these elements are no longer parallel. Thin plates with cast bosses or holes at right angles to the surface require much less expensive dies than if cast to a curved shape. Inserts are generally used for one or both of the following reasons:

Zinc and Zinc Alloy Castings / 1099  To provide greater strength, hardness, wear

resistance, ductility, flexibility, or some other property not possessed, at least in the same degree, by the die casting alloy  To provide shapes of parts or passages that cannot be cored or cast or are less expensive or better as inserts Inserts can usually be cast in place, but there are many cases in which they are applied after casting in holes cored for the purpose. The object of casting the insert in place is either to anchor it securely or to locate it in a position where it could not be placed after casting. Machining. Zinc die castings are manufactured to very fine tolerances as-cast, and if any machining is necessary, the cuts required are usually light. A minimum machining allowance of 0.25 mm (0.01 in.) is recommended, along with a maximum of 0.5 to 1 mm (0.02 to 0.04 in.). Machining operations are made easier by the good machining qualities of zinc die casting alloys. Design drawings should show where machining is to be done and should indicate how much metal is to be removed in machining, unless this is left to the judgment of the die caster. Surfaces to be machined should be of minimum area, consistent with other requirements, and when possible should be positioned to simplify machining. If possible, location should be from the fixed die half. For example, flats can often be trued by simple grinding if the surfaces to be ground are accessible. Placing flats (such as boss faces) all in one plane expedites grinding. Such surfaces should be slightly above surrounding areas that do not require machining. When the number of castings to be produced from a die is small, the die cost must be kept low. In such cases, it is sometimes preferable to avoid expensive machining on the die and to perform additional machining on the castings. Small holes (3 mm, or 0.12 in.) in thin sections are often drilled or punched in preference to coring because the flash from cored holes must normally be removed by such operations. It is almost as quick to drill or punch the full depth of the hole as to remove flash. The drilling operation can be simplified by casting a start for the drill.

Joining Soldering. Zinc alloy die castings are not easy to solder because of the aluminum

content of the alloy. Furthermore, the soldered joint may be subject to intergranular corrosion arising from the lead and tin of the solder diffusing into the alloy. These disadvantages can be overcome by plating the casting with a metal such as copper and soldering onto the plated surface. Care must be taken to ensure that no soldered die castings (plated or nonplated) are remelted with other scrap die castings. A very small amount of solder can spoil a large batch of alloy when the two are melted together. Repair welding of zinc alloy die castings is recommended only when the damaged casting cannot be replaced. In such cases, and for emergency repairs, a procedure has been developed for building up the damaged part by welding, using an alloy rod of the same composition as the casting with the gentle heat of a slightly reducing oxyacetylene flame. Stud welding has been used, but it is not satisfactory, because the alloy adjacent to the stud is weakened and may break. Adhesive Bonding. When appropriate, zinc die castings can be bonded with a range of modern high-strength adhesives, particularly those based on epoxy and phenolic resins. The final choice of adhesive for any application should be discussed with the adhesive supplier. Mechanical Fastening. Studs formed as an integral part of the casting usually cost less than inserted studs and in general constitute a highly economical means of fastening a casting to a mating part. Production rates are seriously impaired when separate inserts must be placed in hot dies before each shot. For integral studs that are at least 6 mm (0.24 in.) in diameter on large or medium-size castings, little trouble during handling is experienced. With small, light castings, proportionately smaller studs can be used safely. Threads. The ability to cast threads is a major advantage of the die casting process. Cast threads should be specified wherever their use reduces cost over that for cut threads. Most external threads can be cast. It is common practice to cast coarse external threads over 20 mm (0.8 in.) in diameter if they are located at a die parting. The weight of a component with a cast screw thread is important. If it is heavy, mechanical bruising may damage fine threads. Pitch errors are likely to be greater than with cut or rolled threads. It is advisable to limit the length of engagement to one-half a diameter. Tooling costs will be higher than

for unthreaded components because of the accuracy necessary in die sinking and the increased die maintenance. Most internal threads are more expensive cast than cut. Cast internal threads are occasionally useful for very steep pitches, and whatever the pitch, the thread can be carried down to a shoulder or to the bottom of a blind hole. All holes requiring fine threads are tapped, and cast interior threads less than 20 mm (0.8 in.) in diameter are rarely economical.

Applications for Zinc Die Castings The automotive industry is the largest user of zinc die castings. Some of the important mechanical components made as zinc alloy die castings are carburetor bodies, bodies for fuel pumps, windshield wiper parts, control panels, grilles, horns, and parts for hydraulic brakes. Structural and decorative zinc alloy castings include grilles for radios and radiators, lamp and instrument bezels, steering wheel hubs, interior and exterior hardware, instrument panels, and body moldings. Other applications include the electrical, electronic, and appliance industries, business machines and other light machines of all types such as beverage vending machines and tools. Building hardware, padlocks, and toys and novelties are major areas of application for zinc die castings. ACKNOWLEDGMENT This article has been updated and revised from Dale C.H. Nevison, Zinc and Zinc Alloys, Casting, Vol 15, Metals Handbook, ASM International, 1988, p 786–797. SELECTED REFERENCES  NADCA Product Specification Standards

for Die Castings, North American Die Casting Association, Wheeling, IL, www. diecasting.org  F. Porter, Zinc Handbook: Properties, Processing and Use in Design, Marcel Dekker, 1991  F.C. Porter, Corrosion Resistance of Zinc and Zinc Alloys, Marcel Dekker, 1994  G.O. Cowie, Stress Calculations for Zinc Die Castings, International Lead Zinc Research Organization, Durham, NC, 1988

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1100-1113 DOI: 10.1361/asmhba0005334

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Magnesium and Magnesium Alloys Ken Savage, Magnesium Elektron

MAGNESIUM ALLOY CASTINGS can be produced by nearly all of the conventional casting methods, namely, sand, permanent, and semipermanent mold and shell, investment, and die casting. The choice of a casting method for a particular part depends on factors such as the configuration of the proposed design, the application, the properties required, the total number of castings required, and the properties of the alloy. The discussion here focuses on the variety of alloys, furnaces, and associated melting equipment, and on the casting methods available for manufacturing magnesium castings.

Magnesium Alloys A large range of magnesium-base alloys is available for the production of castings. However, not all alloys are suitable for production by all casting methods. For example, alloys normally cast by the permold process are somewhat limited in number, and those used in the die-casting process are even more restricted. The method of codification used in North America to designate magnesium alloy castings is taken from ASTM Standard Practice B 275 (Table 1). It gives an immediate, approximate idea of the chemical composition of an alloy, with letters representing the main constituents and figures representing the percentages of these constituents. The nominal compositions of the alloys used for sand, investment, and permold castings are shown in Table 2, and those for die castings are shown in Table 3. For example, consider the three alloys AZ91A, AZ91B, and AZ91C. In these designations:
“A”





represents aluminum, the alloying element specified in the greatest amount. “Z” represents zinc, the alloying element specified in the second-greatest amount. “9” indicates that the rounded mean aluminum percentage lies between 8.6 and 9.4. “1” signifies that the rounded mean of the zinc lies between 0.6 and 1.4. “A” as the final letter in the first example indicates that this is the first alloy whose composition qualified assignment of the designation AZ91.


The final serial letters “B” and “C” in the

second and third examples signify alloys subsequently developed whose specified compositions differ slightly from the first and from one another, but do not differ sufficiently to effect a change in the basic designation. Although alloys used for the die-casting process are somewhat limited in number, more of the Al-Zn-Mn alloys (for example, the AZ91 type, particularly the high-purity grade) are now being used. A large, growing application for die castings is the automotive market. Magnesium castings of all types have found use in many commercial applications, especially where their lightness and rigidity are a

major advantage, such as for chain saw bodies, computer components, camera bodies, and certain portable tools and equipment. Magnesium alloy sand castings are used extensively in aerospace components. Sand Casting. Magnesium alloy sand castings are used in aerospace applications because they offer a clear weight advantage over aluminum and other materials. A considerable amount of research and development on these alloys has resulted in some spectacular improvements in general properties compared with the earlier AZ types (Ref 1). Although there has been, and still is, a large volume of castings for aerospace applications being produced in the older, conventional AZ-type alloys, the trend is toward the production of a

Table 1 Standard three-part ASTM system of alloy designations for magnesium alloys First part

Second part

Indicates the two principal alloying elements

Indicates the amounts of the two principal elements

Consists of two code letters representing the two main alloying elements arranged in order of decreasing percentage (or alphabetically if percentages are equal) A—Aluminum

Whole numbers

B—Bismuth C—Copper D—Cadmium E—Rare earth F—Iron G—Magnesium H—Thorium J—Strontium K—Zirconium L—Lithium M—Manganese N—Nickel P—Lead Q—Silver R—Chromium S—Silicon T—Tin V—Gadolinium W—Yttrium Y—Antimony Z—Zinc

Third part

Distinguishes between different alloys with the same percentages of the two principal alloying elements Consists of two numbers corresponding to Consists of a letter of the alphabet assigned in order as compositions rounded-off percentages of the two main become standard alloying elements and arranged in same order as alloy designations in first part A—First compositions, registered ASTM B—Second compositions, registered ASTM C—Third compositions, registered ASTM D—High-purity, registered ASTM E—High corrosion resistant, registered ASTM X1—Not registered with ASTM

Magnesium and Magnesium Alloys / 1101 Table 2 Nominal compositions of magnesium casting alloys for sand, investment, and permanent mold castings Composition, % Alloy

AM100A AZ63A AZ81A AZ91C AZ91E AZ92A EV31A EZ33A QE22A(b) EQ21A(b)(c) K1A ZE41A ZE63A ZK51A ZK61A WE43B WE54A

Al

Zn

Mn

10.0 6.0 8.0 9.0 9.0 9.0 ... ... ... ... ... ... ... ... ... ... ...

... 3.0 0.7 0.7 2.0 2.0 0.4 2.7 ... ... ... 4.2 5.7 4.6 6.0 ... ...

0.1 min 0.15 0.13 0.13 0.10 0.10 ... ... ... ... ... ... ... ... ... ... ...

Rare earths

Gd

Y

Zr

... ... ... ... ... ... 3.3(a) 3.3 2.0 2.0 ... 1.2 2.5 ... ... 3.2(d) 3.5(e)

... ... ... ... ... ... 1.4 ... ... ... ... ... ... ... ... ... ...

... ... ... ... ... ... ... ... ... ... ... ... ... ... ... 4.0 5.25

... ... ... ... ... ... 0.6 0.6 0.6 0.6 0.6 0.7 0.7 0.7 0.7 0.5 0.5

(a) Comprising up to 0.4% other rare earths in addition to the 2.9% Nd present. (b) These alloys also contain silver, that is, 2.5% in QE22A and 1.5% in EQ21A. (c) EQ21A also contains 0.10% Cu. (d) Comprising 1.0.% other heavy rare earths in addition to the 2.25% Nd present. (e) Comprising 1.75% other heavy rare earths in addition to the 1.75% Nd present.

Table 3

Nominal compositions of magnesium casting alloys for die castings Alloying element

Alloy

AJ52A AJ62A AM50A AM60A AM60B AS21A AS21B AS41A AS41B AZ91A AZ91B AZ91D AE44

Al

5.0 6.0 5.0 6.0 6.0 2.25 2.25 4.25 4.25 9.0 9.0 9.0 4.0

Mn

0.4(a) 0.4(a) 0.35(a) 0.30 0.35(a) 0.35 0.10 0.35 0.50(a) 0.13 min 0.13 min 0.30(a) 0.25

Si

Sr

Zn

Re

Mg

... ... ... ... ... 1.0 1.0 1.0 1.0 ... ... ... ...

2.0 2.4 ... ... ... ... ... ... ... ... ... ... ...

... ... ... ... ... ... ... ... ... 0.7 0.7 0.7

... ... ... ... ... ... ... ... ... ... ... ... 4.0

bal bal bal bal bal bal bal bal bal bal bal bal bal

(a) Manganese content to be dependent on iron content.

greater proportion of aerospace castings in the relatively new zirconium types. Although the Mg-Al and Mg-Al-Zn alloys are generally easy to cast, they are limited in certain respects. They exhibit microshrinkage when sand cast, and they are not suitable for applications in which temperatures of over 95  C (200  F) are experienced. The magnesium rare-earth zirconium alloys were developed to overcome these limitations. Sand castings in the EZ33A alloy do in fact show excellent pressure tightness. The greater tendency of the zirconium-containing alloys to oxidize is overcome by the use of specially developed melting processes. The two Mg-ZnZr alloys originally developed, ZK51A and ZK61A, exhibit high mechanical properties, but suffer from hot-shortness cracking and are nonweldable. For normal, fairly moderate temperature applications (up to 160  C, or 320  F), the two alloys ZE41A and EZ33A are finding the greatest use. They are very castable and can be used to make very satisfactory castings of considerable complexity. In addition, they have

the advantage of requiring only a T5 heat treatment (that is, precipitation treatment). When a demand arose in some aerospace engine applications for the retention of high mechanical properties at higher elevated temperatures (up to 205  C, or 400  F), thorium was substituted for the rare earth metal content in alloys of the ZE and EZ type, giving rise to the alloys of the type ZH62A and HZ32. Not only were there substantial improvements in mechanical properties at elevated temperatures in these alloys, but good castability and welding characteristics also were retained. The thorium-containing alloys, however, exhibited a greater tendency for oxidation, requiring greater care in meltdown and pouring. The mildly radioactive nature of the dross and sludges from processing these alloys and the disposal of these by-products are associated problems with this alloy group. The thoriumcontaining alloys are no longer used for new applications and have been removed from most International Standards. A further development aimed at improving both room-temperature and elevated-temperature

mechanical properties produced an alloy designated QE22A. In this alloy, silver replaced some of the zinc, and the high mechanical properties were obtained by grain refinement with zirconium and by a heat treatment to the full T6 condition (that is, solution heat treated, quenched H2O, and precipitation aged). However, problems were experienced with both of these alloys. The use of thorium has become increasingly unpopular environmentally, and the high price of silver together with the need for good corrosion resistance has led to a considerable amount of research and development work on alternative alloy types. The more recent alloys emerging from this research contain yttrium in combination with other rare earth metals (that is, WE43A, WE43B, and WE54A), replacing both thorium and silver (Ref 2). These alloys have superior elevated-temperature properties and a corrosion resistance almost as good as the high-purity Mg-Al-Zn types (AZ91D). The latest alloy is Elektron 21 (coded EV31A), which has good elevated temperature performance, good corrosion resistance, and improved ease of casting. The alloys used for investment casting are very similar to those used for the sand-casting process. Permanent Mold Casting. In general, the alloys that are normally sand cast are also suitable for permanent mold casting (see the article “Permanent Mold Casting” in this Volume). The exception to this are the alloys of the MgZn-Zr type (for example, ZK51 and ZK61A), which exhibit strong hot- shortness tendencies and are consequently unsuitable for processing by this method. Die Casting. The alloys from which die castings are normally made are mainly of the MgAl-Zn type, for example, AZ91. Two versions of this alloy from which die castings have been made for many years are AZ91A and AZ91B. The only difference between these two versions is the higher allowable copper impurity in AZ91B. Further work has produced a high-purity version of the alloy in which the nickel, iron, and copper impurity levels are very low and the iron-to-manganese ratio in the alloy is strictly controlled. This high-purity alloy shows a much higher corrosion resistance than the earlier grades. More recent development work has produced alloys such as AJ52A, AJ62A containing strontium, and AS41B containing silicon, with the aim of improving the high-temperature properties while retaining good corrosion resistance. AE44 alloy, containing rare earths, has also been introduced. Use of this type of magnesium alloy has increased in the automotive industry. Thixomolding involves the casting of a semisolid alloy into a mold by an injectionmolding process. The equipment is specifically designed to process coarse particles of a magnesium alloy. The particles are carefully heated and subjected to a mechanical shear to produce

1102 / Nonferrous Alloy Castings the slurry, which is then injected into a mold. Argon gas is used to prevent burning. Modification of the alloy, process temperature, and shear rate can significantly affect the microstructure of the components made by this production method. The mechanical properties are reported to show improvements, mostly because of a reduction in porosity. Pressure tightness of components is also improved when compared to die-cast parts. Thixomolding of magnesium alloys is a process that is developing and applications are continually expanding. Further information on thixomolding can be found in Ref 3.

General Applications The most important feature of magnesium castings, which gives rise to their preferred use compared with other metals and materials, is their light weight. Because of this, magnesium castings have found considerable use since World War II in aircraft and aerospace applications, both military and commercial. More recently, as a result of a general requirement for lighter weight automobiles to conserve energy, there has been a growing use of magnesium in the automotive field, principally as die castings. Magnesium, however, has other important casting advantages over other metals:
It is an abundantly available metal.
It is easier to machine than aluminum.
It can be machined much faster than alumi-

num, preferably dry. In the die-casting process, it can be cast up to four times faster than aluminum because of the lower heat content. Die lives are considerably longer than with the aluminum alloys, because much less soldering onto the die surfaces takes place. When protected correctly, particularly against galvanic effects, it behaves in a very satisfactory manner. Modern casting methods and the application of protective coatings currently available ensure long life for welldesigned components. Support ring (cast iron)

Crucible (cast low-carbon steel)

Exhaust stack

Clay firebrick Cleanout door

Burner opening

Fig. 1

Cast refractory

Cross section of a stationary fuel-fired furnace for open-crucible melting of magnesium alloys

Current technology makes it possible to produce parts of considerable complexity having thin-wall sections. The end product has a high degree of stability as well as being light in weight. New die-cast automotive components include instrument panel beams, engine blocks, and radiator supports. Many cell phones incorporate die-cast magnesium alloy components.

Melting Furnaces and Auxiliary Pouring Equipment Furnaces for melting and holding molten magnesium casting alloys are generally the indirectly heated crucible type, of a design similar to those used for the aluminum casting alloys. The different chemical and physical properties of the magnesium alloys compared to aluminum alloys, however, necessitate the use of different crucible materials and refractory linings and the modification of process equipment design. Molten magnesium does not attack iron in the same way as molten aluminum. Magnesium can therefore be melted and held at temperature in crucibles fabricated from ferrous materials. It is common practice, therefore, especially with larger castings, to melt and process the molten magnesium alloy and to pour the casting from the same steel crucible. Figure 1 shows the cross-sectional design of a typical fuel-fired stationary crucible furnace of the bale-out type, from which metal for small castings can be hand poured using ladles. This allows the crucible to be supported from the top by a flange, leaving a space below the crucible. Not only is this a distinct advantage in the transfer of heat to the crucible charge, it also ensures that there is room for easy removal of any detached scale that might form on the outer surface of the crucible during the melting operation. The furnace chamber has a base that slopes toward a cleanout door. A progressive thinning of the crucible walls can occur, which may tend to be localized in fuel-fired furnaces because of flame impingement. There is the possibility of molten-metal leakage if the wall thickness is not checked regularly. With scale, there is also the possibility of reaction between the iron oxide and the molten magnesium, which can be explosive in character. Furnace bottoms must therefore be kept free from scale buildup. It is important also to have a runout pan capable of taking the full crucible content in the event of leakage. More recently, and especially in cases where it is difficult to check on scale formation, it is possible to use steel crucibles that are clad with a nickel chrome alloy on the outside heating surface to eliminate scaling and still not detract from the heating efficiency of the furnace. The furnace lining refractories are important, because molten magnesium reacts violently with some refractories. High-alumina refractories and high-density “superduty” firebrick of a typical 57% Si and 43% Al

composition have been found to give satisfactory results. The cleanout door on fuel-fired furnaces can be easily opened for its intended purpose. With electric-resistance crucible furnaces it is common practice to seal the cleanout door with a sheet of low melting point material, such as zinc, which will not act as a barrier to molten magnesium alloy in the event of a runout, but will prevent the “chimney” effect, which can accelerate oxidation of the crucible. The type and size of furnace used largely depends on the type of casting operation. A small jobbing foundry, carrying out a batchtype operation in a wide range of different alloys, normally uses the lift-out crucible technique. Larger-scale operations, typically operating on a more restricted range of casting alloys, may employ a larger bulk melting unit from which the molten alloy for the casting operation is distributed to holding crucible furnaces. Metal treatments are carried out, and the metal is either poured directly from the crucible or hand ladled to the casting molds. In cold-chamber die-casting operations, the supply of molten alloy to the machine is maintained by hand ladling or by a pump. A coldchamber machine differs from a hot-chamber machine in that the shot sleeve and injection plunger are not submerged in the molten metal. However, the size of parts made by hot-chamber die casting is limited. Because the injection mechanism in a hot-chamber machine is submerged in the furnace bath, the operation is faster and the plunger pressure can be adjusted to force the molten metal into the finest die detail. Electromagnetic pumps can also be used for this purpose. The overriding consideration in metal transfer is that the metal must be transferred in as nonturbulent a manner as possible to prevent oxidation, which can give rise to oxide skins and inclusions in the final casting. Excessive oxidation has ruled against the use of direct-fired reverberatory furnaces similar to those used quite satisfactorily for aluminum alloys. When magnesium becomes molten, it tends to oxidize and burn, unless care is taken to protect the molten metal surface against oxidation. Molten magnesium alloys behave differently from aluminum alloys, which tend to form a continuous, impervious oxide skin on the molten bath, limiting further oxidation. Magnesium alloys, on the other hand, form a loose, permeable oxide coating on the molten metal surface. This allows oxygen to pass through and support burning below the oxide at the surface. Protection of the molten alloy using either a flux or a protective gas cover to exclude oxygen is therefore necessary. Burning, which can occur at or above the melting point of the alloy, is prevented by the use of either a flux sprinkled on the molten metal or a suitable fluxless technique using a gas mix containing a mix of dry air, carbon dioxide, and SF6. Both procedures are described in detail later in this article.

Magnesium and Magnesium Alloys / 1103 Increasingly stringent environmental controls in the foundry have rendered the older sulfurdioxide domed bale-out furnace unacceptable. Crucibles. The indirect heating crucible method of melting is of comparatively low thermal efficiency. Electric coreless induction furnaces, although much higher in initial capital cost, have lower running costs and occupy less floor space than fuel-fired units. Crucibles range in size from about 30 to 910 kg (60 to 2000 lb). In the lower range, they may be constructed as steel weldments. The steels normally used are of low carbon content, that is, less than 0.12% C. Because of the extremely adverse effect of nickel and copper on the corrosion resistance of magnesium alloys, these two elements in the steel must be restricted to less than 0.10% each. Magnesium melting operations, particularly if a flux melting procedure is used, normally result in the formation of a sludge with a comparatively low thermal conductivity at the bottom of the crucible. If this material is not periodically removed, overheating of the crucible can result in this area, accompanied by excessive scaling of the crucible. Excessive oxide buildup on the crucible walls can have the same effect. Records of the number of charges being melted should be kept for each crucible as a routine safety measure. Less buildup is normally experienced with the fluxless method of melting. It is a very desirable procedure, however, to periodically withdraw the crucibles from use and allow them to soak, filled with water, to remove all buildup. Pouring Ladles. For smaller castings, hand ladling can be conventionally used, with pouring ladles taking molten alloy from a bale-out-

Fig. 2

type furnace. Pouring ladles can be bucket shaped for the slightly larger range of castings and hemispheric shaped for the smaller ones. Both types of ladles however, should be constructed from low-carbon, low-nickel steel and be about 12 gage (2.67 mm, or 0.105 in., thick). A typical design of a bucket-type ladle is shown in Fig. 2. Essential design features include an overflow guard and a bottom-pour spout to avoid the possibility of the pour being flux contaminated. Other essential items of metal handling equipment include sludge-removal ladles, sludge pans to contain this material, stirrers, puddling tools, and skimmers. All of these pieces of equipment should have the same steel composition as the crucible. Thermocouples. Accurate temperature control is critical to the processing of magnesium alloys. Iron-constantan- or chromel-alumel-type thermocouples are recommended. There should be a permanent type of installation such that temperature determinations can be made at appropriate stages in the melting and metal treatment process. Light-gage thermocouples in thermocouple protection tubes of either mild steel or nickel-free stainless steel should be used.

Melting Procedures and Process Parameters There are basically two main systems, flux and fluxless, for the melting and pouring of magnesium alloys. Each of them, if done correctly, is perfectly capable of producing good metal for the casting process. Many of the precautions taken apply equally to both methods.

Construction details of a typical ladle used for pouring magnesium alloys. Dimensions given in inches.

The Flux Process. The basic requirement in melting magnesium is to prevent contact with oxygen as it melts. Early attempts to develop a system using gaseous protection were not completely satisfactory. Successful melting procedures became possible only when flux methods were developed. In the course of time, suitable fluxes for handling both the Mg-Al-ZnMn and the Mg-Zr-containing alloys emerged, and corresponding techniques for producing clean, flux-free castings were developed. These procedures have been successfully used for since the 1970s (Ref 4). A typical flux-melting procedure would be for the crucible with a small quantity of flux (about 1.5% of charge weight) placed in the bottom, to be preheated to dull red heat. The metal charge to be loaded must be clean, dry, and free from oil, oxide, sand, and corrosion. There should be no foreign metals present. Contamination by even tiny pieces of sand or debris from other alloys must be prevented by close control of melt parameters. No oxide-contaminated material should be allowed to enter the melt charge. All materials that contain dirt or oxide should be separately refined and ingotted before being recycled into production melts. “Bridging” of the metal charge in the crucible should be avoided, the objective being to feed the charge progressively into the crucible and maintain an advancing level of liquid alloy. During this procedure, additional flux is lightly sprinkled onto the melt surface. There are separate proprietary fluxes for each type of magnesium alloy (that is, AZ types and magnesium-zirconium types). Supplier instructions for these fluxes must be precisely followed, and their use must be restricted to the type of alloy for which they were developed. During the melting process, localized overheating of the charge must be prevented. The process of chlorination of the melt for refining purposes is no longer considered to be an acceptable practice unless effective steps are taken to collect chlorine fumes. The Fluxless Process. With the flux technique, particularly in the field of die castings, the presence of flux gave some operational difficulties even with hot-chamber die-casting processes. Flux inclusions in the castings were not uncommon, creating a major hindrance to the greater use of magnesium. A significant breakthrough in this area resulted in the development of a fluxless process for use in the melting, holding, and pouring of magnesium alloys (Ref 5–10). This involved the use of air/sulfur-hexafluoride gas or air/carbon-dioxide/sulfur-hexafluoride gas mixes of low (2%) SF6 concentration. The protection provided by this cover gas was very effective, with the added advantage that the mix was nontoxic and odorless. This process became immediately accepted by both the ingot producers and the die-casting sections of the foundry industry, because it addressed this objection to the flux melting method.

1104 / Nonferrous Alloy Castings The new melting process was next extended to the sand-casting process (Ref 11). There were two aspects of the sand casting that had to be allowed for. The temperatures for pouring the magnesium alloys, particularly the zirconium alloy (see Fig. 3), were appreciably higher than for the die-casting alloys. Also, the sand process is generally a more open method, in that the molten alloy cannot be enclosed to the same degree as with die castings. For these reasons the gas mixture used by the sand caster is generally richer in sulfur-hexafluoride content, and sometimes, particularly with the magnesium-zirconium alloys, the parent gas is carbon dioxide. It is expected that the use of sulfur hexafluoride (SF6) will be prohibited in the near future because of environmental reasons. Alternative gases for the protection of magnesium alloys are being sought by the industry. Options for alternative gases include:





CO2/SO2 Air/SO2 CO2/HFC-134a Novec 612

Note that some of the above systems may only be used under a license from the developer. The fluxless methods proved to be much more acceptable to the die-casting process, and in practice melting losses were considerably reduced (by about half in some cases). The main reasons for the higher melting losses with flux melting are the entrapment of small globules of metal in the sludge at the bottom of the crucible and the difficulty of recovering such metal. In the fluxless method, the amount of sludge at the bottom of the crucibles is greatly reduced. Grain Refinement. The earlier procedure of superheating the melt to 870 to 925  C (1600 to 1700  F) followed by a rapid cooling to process temperature is no longer popular or acceptable because it considerably shortens crucible life and can increase the iron content of the alloy melted. With AZ-type alloys tablets of hexachlorethane or hexachlorbenzine are still used for grain refinement and degassing. However, these tablets are no longer environmentally

acceptable. Environmentally safe degassing can be accomplished by bubbling pure argon or nitrogen gas into the melt. Alternative proprietary materials for grain refinement are available from Foseco (Foseco Nucleant 5000 tablets) and Magnesium Elektron (Elektron GR). With the magnesium-zirconium alloys, very effective grain refinement is achieved by the zirconium addition. To achieve the optimum fineness of grain, it is necessary for the melt to be supersaturated with soluble zirconium. Insoluble zirconium complexes can also be present in the melt, resulting from contaminations of different types. Aluminum and silicon are very undesirable for this reason. Thus, it is necessary for an excess of zirconium to be added to the melt, over and above that which is theoretically required, to provide the required soluble zirconium level. It is also necessary, for the same reason, to maintain the heel of zirconium-bearing material at the bottom of the crucible into which the insoluble zirconium complexes settle. To prevent any of this liquid heel from being poured off into the castings, an adequate amount of molten alloy (that is, about 15% of the charge weight) is left behind after the casting molds have been poured. Suitable zirconium hardener products are available on the market. Undue disturbance of the melt during pouring must be avoided. Great care must be taken not to overpour, and the melting procedure must allow adequate settling time. The normal control check carried out on the AZ alloy is to fracture a small test sample cast in sand for visual examination. For the magnesium-zirconium alloys, a small chill cast bar is fractured and visually examined. Comparison is made with norms on which the

Table 4

Summary of grain-refining methods for magnesium alloys

Alloy

Grain-refining element or treatment

Pouring of a permanent mold for an Elektron 21 alloy casting using the fluxless technique. Courtesy of Magnesium Elektron

Degree of refinement

Magnesium Al, Zn, rare earths, Mild Th, Si, Ca, Zr, others Zirconium Extreme

Mg-Al, Mg-AlZn-Mn

Carbon inoculation Marked Superheating Marked

Mg-AlMn-(Zn)

Ferric chloride

Marked

Mg-Zn(Re-Mn)

Ferric chloride, iron-zinc alloy

Very marked

Ammonia

Very marked

Mg-Mn

Fig. 3

grain size has been checked metallographically. The grain-size value regarded as satisfactory is 0.030 mm (0.0012 in.). A very important factor, which normally requires continuous surveillance, is the possibility of cross-contamination by alloy mixing. Rare-earth alloying elements, which are strong carbide formers, suppress gram refining by limiting the amount of carbon available for Al4C3 nuclei formation. Zinc-iron master alloys can be used for grain refining rare earth containing alloys. Methods for grain refining magnesium alloys are summarized in Table 4. Alloying. Most foundries purchase prealloyed ingot, which is subsequently charged into the melting furnace with a proportion of the process scrap. In the case of some die-casting operations, the amount of process scrap generated is low, and it is economically feasible to have this scrap recycled. With the AZ alloys used in the sand-casting and die-casting operations, little correction to the composition is necessary. However, the magnesium-zirconium alloys contain alloying constituents that tend to be lost during each remelt operation and need to be added each time the material is remelted. Such corrections can be made by adding the pure metals themselves (such as zinc, misch metal, and so forth), or hardener alloys with a fairly high content of the alloying element. Zirconium added as a master alloy of about 30% in magnesium and cerium rare earths added as a master alloy with a content of 20% rare earths in magnesium are examples. Composition control must allow for the fact that the addition of a master alloy to correct one element may lead to a dilution of the melt, causing the content of other elements to be

Calcium + nitrogen Mild Zirconium Increases with decreasing manganese

Probable mechanism

Interfering elements(a)

Remarks

...

...

Al, Si, Fe, H, Sn, Sb, Co, Ni, (Mn), (Be), (Li), others ... Be, Zr, Ti, (excess Mn)

...

Concentration gradients Nucleation by zirconium or Zr-enriched Mg Carbide nucleation Nucleation by Al-Mn(Fe) compounds or carbon Nucleation by Fe-Al-Mn compounds and carbon Nucleation by iron compounds Nucleation by hydrogen ... Nucleation by zirconium or Zrenriched Mg

Zr, Be

... ...

Requires manganese

Al, Si, Th

Iron-zinc requires manganese content of  1% Al, Si, Th Gassy metal; Manganese can exceed 1% ... ... Al, Si, Fe, H, Sn, Sb, Co, ... Ni, Mn, others

(a) Elements in parentheses interfere to a lesser extent. Source: E.F. Emley, Principles of Magnesium Technology, Pergamon Press, 1966

Magnesium and Magnesium Alloys / 1105 reduced. Normally, these master alloys are added into the melt by placing them into a preheated steel basket, from which they readily dissolve into the melt. A “puddling” technique involving either manual or mechanical stirring, followed by a settling procedure to allow insoluble zirconium complexes to settle, produces the required degree of supersaturation of the melt with zirconium. Care must be taken not to hold the melt too long or to allow the melt temperature to fall, however, because losses of zirconium will result. Melt Treatments. In the melt-down operation, oxidation of the metal must be prevented by the sprinkling of flux on the melting metal or, in the case of fluxless techniques, by the efficient use of a protective gas atmosphere. Depending largely on the type of alloy and the melting and casting process used, the gas may consist of sulfur hexafluoride/air, or sulfur hexafluoride/carbon dioxide/air mixtures. For example, with the die-casting process, in which comparatively low casting temperatures are used and the molten metals can be effectively enclosed, sulfur hexafluoride/air mixture with a low sulfur hexafluoride content (typically <0.25%) provides adequate protection. With sand castings, particularly with the magnesium-zirconium alloys, higher melt temperatures are used. To provide adequate protection, it is normal to employ a carbon dioxide/ sulfur hexafluoride or carbon dioxide/air/sulfur hexafluoride mixture. It is also normal for the sulfur hexafluoride content of the mixture to be increased up to a maximum of 2% although, as mentioned previously, the use of sulfur hexafluoride gas will be withdrawn and other effective systems used in its place. Gas Content and Grain Size. Currently, the process of chlorination with the AZ-type alloys to clean and degas the melt is rarely acceptable on environmental and safety grounds. Also, the earlier process of grain refinement by superheating these alloys to a high temperature (that is, 850  C, or 1560  F), followed by a rapid cool to casting temperature is unacceptable because of the increased iron pickup from the crucible and because of the reduced of crucible life caused by the high furnace chamber temperatures. These earlier techniques were replaced by the use of carbon inoculation, using hexachlorethane or hexachlorbenzine. These materials, added in the form of compressed tablets, simultaneously provide degassing and grainrefining effects. Alternative, more environmentally friendly materials are now available for this purpose. With the zirconium alloys, the situation is different in that a separate degassing or grainrefinement treatment is no longer necessary, both of these features being taken care of by the addition of zirconium. With flux melting of the Al-Zn-Mn-type alloys, fluxes high in magnesium chloride are employed, but there are inherent dangers if these fluxes are used with the zirconium alloys. However, special proprietary fluxes have been

developed for the zirconium alloy to avoid these problems. Foundry Control Procedures. When sand casting an AZ-type alloy, apart from conducting standard spectrographic control of composition, it is normal to pour a sand-cast round bar approximately 19 mm (0.75 in.) in diameter and then fracture it for visual examination. Examination of this fractured surface and comparison with sand-cast standards serve as a useful check of the degree of grain refinement achieved. With the zirconium alloys, a grain size check is usually carried out on a standard chill cast bar that is fractured and examined in a similar way. This test satisfactorily determines whether the zirconium content is at an acceptable level for a prescribed degree of grain refinement. Comparison standards are set up to ensure that the grain size meets a 0.030 mm (0.0012 in.) size limit. Chemical composition control is normally carried out spectrometrically using cast slugs poured into a steel mold. A large proportion of the sand castings made in magnesium alloys are for aircraft or military requirements, for which very stringent checks and inspections are required. Normally, samples for tensile testing must be provided from each melt. These are either separately cast test bars made in a sand mold or coupons attached to the parent casting. Frequently, complete destructive testing of sample castings is called for to further demonstrate the quality level of the workpiece. For the die castings, the standard control procedures involve not only visual and dimensional checks but also inspection of the castings by fluoroscopic screening. Pouring Methods. The method of pouring depends on the casting process. Thus, pouring can vary from dip ladling from a bale-out crucible, to more automated systems using metered shots of molten alloy, to a die-casting machine. The automated hot-chamber machine is another possibility, and large sand castings are currently being poured directly from the crucible in single or multiple crucible pours. In the flux-melting method, the dried flux on the top of the molten metal must be completely removed by careful skimming. Just as important is the complete removal with a steel wire brush of any loose flux on the rim of the crucible or the pouring lip. During this operation, and until pouring is complete, oxidation is controlled by a suitable protective gas or by dusting the surface of the molten metal with a mixture of equal parts of coarse sulfur and fine boric acid. Proprietary material is available for this purpose. The surface of the metal should be left as undisturbed as possible until the melt is ready to pour. At this stage, the skin formed by the protective agent is pushed back to ensure that none of it enters the molten metal stream. During the actual pour, the metal stream can be adequately protected by use of a protective gas system or by dusting with sulfur. When pouring, it is undesirable for the molten alloy to enter the sprue directly from the

pouring ladle, because this tends to cause any oxide on the metal stream to be carried directly into the mold. If a properly designed pouring box that allows for a degree of separation is used, this will not occur. Because the flux melting process results in a quantity of residual flux-bearing material at the bottom of the crucible, the pouring crucible should never be completely emptied to avoid pouring some of this material off into the molds. With the zirconium alloys, this is even more important, and about 15% of the charged weight is retained as a heel. With these alloys, the heel will contain residual zirconium and zirconium complexes in addition to the flux and some entrapped magnesium particles. In the fluxless melting method, there is still some surface oxide, which should be carefully skimmed from the melt surface, while maintaining the flow of the protective gas. The gas flow can be temporarily discontinued while the crucible is removed from the furnace, but the gas flow must be continued until the pouring is completed. The same protective gas mix used for melting is used on the metal stream during pouring and is used to flush the molds to supplement the action of the inhibitors. The heel left behind can be smaller in volume than the sludge from the flux-melting procedure because there is no residual flux. The clean, sludge-free portion of the heel can be recycled into subsequent melts. Pouring. Small sand, permold, or die castings can be poured from hand ladles. For this purpose, clean, preheated bottom-pour ladles can be used to dip the molten alloy from open crucibles. The same precautions against oxidation must be observed as with direct pouring from the crucible. Oxidation and burning of the metal in the hand ladle can be minimized by use of a protective gas or by sprinkling powdered sulfur on the metal surface. For hand ladling a special flux pot can be used into which the ladle is immersed between pours to dissolve nonmetallics and also keep the ladle hot. The ladle must be drained until it is free of flux. “Double-filling” the ladle, avoiding undue agitation, and taking more metal than is actually needed for the pour are measures that help prevent the carryover of flux into the castings. The possibility of flux contamination and associated corrosion problems and the tedium of the precautions listed previously have spawned the development of automated pouring and fluxless methods in both the low- and high-pressure die-casting fields. Pressure Sand Casting. Several magnesium foundries now use a pressure system when making certain castings. The sand mold is placed on a steel plate located on top of the crucible containing the prepared alloy. A vertical pipe is lowered through a hole in the steel plate and immersed into the liquid metal and the top of the pipe sealed into the sand mold running system. The high-pressure system requires that the crucible has a sealed lid incorporating the pipe.

1106 / Nonferrous Alloy Castings The atmosphere above the melt is pressurized, and liquid metal is forced up the pipe and into the mold cavity. Once the metal has solidified sufficiently, then the pressure is released and the pipe assembly removed. In the low-pressure, or vacuum, system a pipe is connected to the bottom of the mold running system. A steel cover called a bell jar is fitted around the mold and sealed to the steel plate supporting the mold. The pipe is immersed into the liquid metal, and a partial vacuum applied to the bell jar to evacuate the atmosphere within the mold cavity. The alloy rises up the pipe, through the filter and running system, and enters the mold through the ingates. The advantages of the pressure system are:
The liquid metal is not exposed to air and so

the effect of oxidation is reduced.


The number of changes in flow direction is

reduced, thus minimizing turbulence.


The metal flow rate within the mold can be

controlled instead of relying on gravity. Safety Precautions. The usual precautions for handling other molten metals are even more exacting with magnesium. These include the use of face shields and fireproof clothing by plant personnel. With magnesium there is a greater hazard in that moisture, from whatever source, increases the danger of explosion and fire: when moisture comes in contact with molten magnesium, it generates hydrogen, a potential source of explosion. Certain minimum precautions must therefore be observed:
All scrap must be clean and dry; corroded

material should be precleaned.


Any fluxes, which tend to be hygroscopic,

must be kept dry and stored in airtight containers.
Ladles, tools, and anything that will come into contact with molten magnesium must be completely dry and preheated.
The danger of molten magnesium coming into contact with iron oxide scale must be avoided.

Sand Casting Sand castings are produced in a wide range of alloys, weighing from a few ounces to as much as 1400 kg (3000 lb). Successful production of sand castings became possible only when the means for preventing metal/mold reactions from occurring were developed. This was achieved by the addition of suitable inhibitors to the sand mix used in the manufacture of the mold and the cores. These inhibitors include, used singly or in combination: sulfur, boric acid, potassium fluoborate, and ammonium fluosilicate. The amount of inhibitor that must be added to the sand to prevent metal/mold reaction depends on the moisture content of the sand. The original process was called the green sand

method, in which the bond is produced by the natural clay present in the sand being activated by the addition of water or by the use of the clay products southern or western bentonite. Diethylene glycol is frequently added to the latter type of mix to minimize the amount of water necessary and to prevent the sand from drying out. In these green sand mixes, the water content can be in the range 2.0 to 4.0%. A correspondingly high inhibitor content is therefore required. The amount of inhibitor is also dependent on the pouring temperature, the type of alloy being cast, and the casting section thickness. The higher the pouring temperature, the higher the reactivity and the need for larger inhibitor additions. The heavier the casting section, the slower the cooling rate. The inhibitors, particularly the more volatile ones, are more easily lost from the mold surface and need to be replenished from the mold areas at a distance from the heavy section. The green sand molding process is still used extensively for the manufacture of a wide range of sand castings. It has certain limitations, but is economical because the sand can be reused repeatedly. To reuse the sand, it is necessary only to restore the moisture, inhibitor, and glycol contents and then to suitably mull the sand mixture. There is no loss of expensive binder materials used in the modern chemically bonded sands. Automated sand-mixing equipment, with delivery of the sand mix to the point of use, is commonly employed. Molding Sands. It is essential to carry out various sand control tests, as described elsewhere, to maintain a consistent sand mix. In most magnesium foundries making a wide range of sizes of castings, the molding sand and inhibitor type and amount are normally adjusted to suit the heaviest section being cast. Molding sands must have high permeability to allow free passage of the generated mold gases away from the metal/mold interface, that is, 60 to 90 AFS. For this reason, the basic sand used should be comparatively coarse grained. This is especially important because most additives tend to reduce permeability. Unfortunately, these relatively coarse sands tend to produce rough casting surfaces. The choice is therefore between a good casting surface on the one hand and the venting of generated mold gases on the other. For this reason, it

is of the utmost importance to assist the venting of mold or cores and any other reduction in gas generation. The natural venting through the body of the core must be augmented by additional vent channels drilled into the core. These channels must be capable of rapidly taking gases outside of the mold by means of core prints. Assisted evacuation of these gases is sometimes used. Natural-bonded sand can and has been used quite satisfactorily for small molds, but very careful control is needed to maintain satisfactory results because of its relatively high clay content and poor uniformity. Better and more consistent results are obtained by the use of a synthetic sand mix. A washed and graded silica sand is used, with the bond being provided by carefully controlled additions of western or southern bentonite. The essential properties of these two types of bentonite are quite different, and by joint use and controlled blending an optimum balance of desired properties can be obtained. Western bentonite imparts toughness (a combination of high green strength and low deformation). Because these green sand mixes are repeatedly reused, losses of moisture, inhibitors, and so forth can occur, and these materials must be restored at each remulling. Frequent and regular testing of the properties and composition of the sand is imperative. Also, mulling must be carried out on a controlled basis. The composition and typical properties of some molding sands are shown in Table 5 (Ref 12). A variant of these molding sand mixes, which has also been used successfully, employs a modified type of bentonite using an oil as a mixing agent for bonding. Control over the properties and additives is as important as it is with the conventional green sand process. More efficient air extraction is necessary to remove burnt oil fumes and render the process environmentally acceptable for present-day standards. Essentially, the type of sand mix chosen is largely dependent on the molding process used and the sand-cast product being made. Advantages and Disadvantages of Green Sand Molding. Green sand molding is the least expensive of the sand-casting processes. This is because only small amounts of additives are needed each time the sand is mulled, and it is not necessary to use expensive reclamation systems to recycle the sand. However, the process

Table 5 Typical compositions and properties of magnesium molding sands British Ingredient or property

Clay Bentonite Moisture Sulfur Boric acid Potassium fluoborate Diethylene glycol Compressive strength (green, psi) Permeability (green)

Synthetic

Natural

... 4 23/4 4 ½ ... 0.1 8–10 100

8 ... 5–6 6 0.3 ... ... 10–13 30–40

American Synthetic

... 3 2½–3½ 4½ 0.3 ... ¼ 11–15 80–100

... 3–7 1½–2½ 1¼–2¼ ¼ ... ½–1¼ 7–11 120–180

... 4 2¼–3¼ 2 2 ...

/

34

8–12 50

Semisynthetic

... 3 2½ 3 ... 1½ 2½ 8–10 90–150

... 3 1¼–2¼ 2 1½ ... 1/3 9–12 80–90

1 3 3 1½ 1½ ½ 1½ 10–12 80–100

1 5 2½–3 3 ... 1½

/

34

8–10 100–140

Magnesium and Magnesium Alloys / 1107

Fig. 4

Main transmission housing for a heavy lift helicopter that was sand cast in WE43B magnesium alloy having a T6 temper. Courtesy of Fansteel Wellman Dynamics. Casting weight = 206 lb (93 kg).

Fig. 5

Typical engine gearbox for a private or commuter turboprop aircraft cast from QE22A magnesium alloy with T6 temper. Courtesy of Haley Industries Ltd.

does not lend itself to the production of cast components of high complexity. Also, it is not capable of producing castings that satisfy the level of dimensional accuracy currently being demanded for many applications. Fortunately, technology in the mold- and core-making fields has advanced sufficiently to keep pace with these challenges. The first breakthrough, which occurred during the World War II era, was the development of the shell molding process (also known as the Croning process) for the manufacture of shell cores

Fig. 6

Gearbox housing for a military fighter aircraft composed of ZE41A magnesium alloy of T5 temper. Courtesy of Haley Industries Ltd.

and molds. Almost concurrently the carbon dioxide/silicate process came into general use. Both of these processes improved the quality and dimensional accuracy of castings. Also, the core breakdown characteristics were much improved from those of the earlier oil sand cores. The carbon dioxide/silicate process, originally developed for cores, was adapted to the manufacture of molds of much improved dimensional accuracy and good breakdown. The reclamation of these, to enable sand to be recycled, proved to be difficult. With the carbon dioxide/silicate process, there was appreciably less moisture present after the molds were gassed than there was with green sand. It was therefore possible to considerably reduce the amount of inhibitor addition. In the carbon dioxide/silicate process, the sand is fully hardened by passing carbon dioxide through the core in situ in the core box. This eliminates most of the distortions that occurred with the oil sand core and affords a significant energy saving. It is also possible to use the cores in a mold very soon after they are produced. Following this development there were a number of further advances involving various types of phenolic, furan, and epoxy resins of the chemical self-set or gas-hardening type. The latter group involves hardening by the use of air, carbon dioxide, sulfur dioxide, methyl formate, or organic amines. Preference for the different proprietary systems vary from foundry to foundry, but in general they all produce molds and cores much superior to those obtained from the green sand molds or oil-sand/baked cores. It is largely these improvements in materials and processes that have enabled the foundry industry to produce complex castings such as those illustrated in Fig. 4 to 6 (Ref 13). The aircraft gearbox housing shown in Fig. 6 is a sand casting in ZE41A-T5 alloy made by the furan chemical-set process. It contains 56

cores within its configuration. A number of these are small-diameter cores forming complex oil passageway configurations. The accurate positioning of these cores and the thinwall sections in this casting are made possible only by the use of this type of process. Casting solidification rates can be increased locally by the use of metallic chills or by the use of zircon sand to help in producing an optimum solidification pattern in the casting. Proprietary sprays, which are compatible with molten magnesium alloy, are often used to improve the surface hardness of the mold and thereby reduce the effects of molten metal impingement. The gap between the cope and drag may be sealed by the use of a proprietary core paste material to prevent metal leakage or flashing in the casting. Light torching of the mold, using a nonoxidizing acetylene flame, which deposits carbon on the metal contact surface, improves fluidity. Coremaking Procedures. Major cores, other than small pipe cores, are normally produced in a silica sand mix and should have approximately the same grain fineness as the molds themselves. Frequently, smaller cores, particularly those of small diameter and long length, may be of a different type of sand mix than that used for the molds. This factor becomes particularly important if the process allows a substantial amount of core sand to enter the mold sand system during shakeout. Subsequent recycling of the main sand burden may therefore change in grain size distribution. In fact, the introduction of new sand used for the cores may have a desirable “sweetening” effect on the main sand system. It would be normal for the molds to be made from a sand mix with a high content of recycled/reclaimed sand. The sand mixes used for core production must also contain an inhibitor addition, usually of a type similar to that used for the mold. The binders used for cores are similar to the types used in the mold. However, because magnesium alloys have a comparatively low heat content, not as much heat is available for the breakdown of the core as for other metals. Careful choice of binder type and of the percentage of binder is important. Other desirable features of cores are:
They must be capable of being handled and

drilled for venting without deterioration.


They must have good shelf life and not

be affected by atmospheric conditions of storage.
They must be entirely stable and not distort in storage.
The larger cores must be capable of incorporating chills, when necessary, in the core to produce directional solidification within the casting. For these reasons, it is desirable to choose a core binder and core system that produce a degree of strength and dimensional stability to

1108 / Nonferrous Alloy Castings prevent sagging without the necessity for core dryers or bottom boards. The carbon dioxide/silicate process has been used extensively for cores for magnesium alloys and is a good example of the earlier gas-hardening process. Details of the carbon dioxide/silicate and shell molding methods can be found elsewhere in this Volume. Modern resin binder systems in which the core sand mix is hardened by an acidic activating agent or by the passage of acid gases such as sulfur dioxide in situ in the core box have gained considerable popularity. Cores can be made by automated equipment that can be very closely controlled to perform all of the necessary functions of coremaking. Hot-box methods of core production, also used extensively for magnesium castings, are essentially the same as for other metals. Magnesium is widely used for aerospace and race car engine applications, where the particular attributes of the metal and its weight-saving potential are required. The ability to incorporate complex small-diameter oil passageways by coring is an important requirement. Using some of the improved core binders and coremaking processes described, those requirements can be met. If much higher production costs can be tolerated, it is possible to produce even smaller-diameter cores by the use of special processes described in Ref 14. These include the production of ceramic cores, salt cores, wire cores with wash coatings, quartz glass cores, fiberglass cores with ceramic coatings, and wire helix cores with wash coatings. The latter process can be used to manually produce cores with a diameter as small as 2 mm (0.079 in.). For many casting designs, it would be quite impossible to produce these passageways by machining operations. Shrinkage Allowances. Nominal shrinkage allowance data for the various alloys are available and can serve as a general guide. It is generally much safer, however, to establish factors for shrinkage under a particular set of foundry conditions. Gating practices for magnesium have been largely developed to accommodate some of the specific chemical and physical characteristics of the metal and its alloys. Probably the most important of these is the pronounced tendency of the alloys to oxidize. This is particularly the case with those alloying constituents such as rare earth metals and thorium. The second factor is the very low density of these alloys. Magnesium has only two-thirds the weight of aluminum at a comparable volume. Because of these two factors, minimizing turbulent effects in a mold must be a major objective in any casting method. Turbulence causes the formation of oxide skins and films (dross) being folded into the flowing metal and appearing later in the casting as inclusions or surface pits. The technique for introducing the molten alloy into the mold cavity must therefore be one designed to minimize turbulence. For example, the metal must be introduced at the level of the lowest part of the casting.

Gravity-pour systems are normally used, with the metal passing from a pouring cup down a sprue and into a runner system situated at the bottom of the mold. One or more sprues may be used, these normally being tapered so that the metal is choked at their bases. This prevents the aspiration of air. Particularly long sprues should be broken by an intermediate step. During the pour, the pouring cup or box should be kept full. From the bottom of the sprue, the design is such that the metal is allowed to take only a horizontal or upward flow. The normal procedure is for the metal to pass through a runner channel and then through gates into the bottom-most region of the casting. It is standard practice to position screens or filters to interrupt the metal flow before it enters the mold cavity to remove any oxide resulting from pour turbulence. The local cross section of runner bar is enlarged locally so that the flow of the molten alloy is not impeded. The screens may consist of a piece of perforated steel sheet, often with coarse steel wool as a backup. There are also proprietary woven screens of stainless steel wire. These are preformed into a wedge shape to fit easily into a matching cavity formed in the running system. The use of these filters ensures a degree of consistency and is normally a preferred technique. Apart from their essential function of removing oxide, filters also regulate the metal flow into the mold cavity. The design of the runner bar itself can also help to remove dross (slag). Thus, if the crosssectional area in the runner bar is increased (that is, compared to the sprue), the flow rate decreases, and oxide can be skimmed from the metal stream before it enters the mold cavity. To enhance this effect, the cross section of the runner channel is made fairly large and long and is taken beyond the last gate into the casting. The design of the runner system should be such that the runner bar is completely filled before any metal can flow through the gates into the casting. It is desirable, therefore, for as much of the casting as is possible to be formed in the cope, or to use a three-part mold with the runner system in the lowest section. At all times, metal spurting must be prevented. This can be achieved by making the runner bar cross section larger than the sprue cross section, and the total cross section of the gates larger than the runner channel cross section. It is customary to use a ratio of dimensions such as 1:2:4 or 1:4:8, according to the experience of the particular foundry. Molten metal should never be allowed to fall within the mold from one level to a lower level, because entrapment of gas/air and oxide inclusions will result. To avoid misruns, particularly in thin-cast sections, it is accepted practice to have fast pouring rates, high pouring temperatures, and more numerous gates that are closely clustered around the casting.

A particular problem to avoid if possible is a design mold in which a thick section and a thin wall are at the same level as the metal being poured. The metal flow will be preferential to the thick section, which may result in misruns in the thin section. Proper design of the gating system can overcome this problem. Risers. The function of risers in magnesium castings is similar to that in other metals. Magnesium alloy sand castings have a high propensity for shrinkage. For this reason, a larger amount of risering is required than with most other cast metals. Because of this and the relatively high cost of some alloys, it is necessary to increase the feeding efficiency of risers, while still reducing their total volume. This can be achieved by the use of insulating sleeves that encase the risers. Another important advantage of this technique, resulting from the improved casting yield, is the reduction in the amount of recycled process scrap. Chills, which may consist of cast-to-form iron castings, zircon sand, or cores containing steel shot, are used extensively to promote progressive or directional solidification of the casting from the bottom to the top in the gravity-pouring system. Heavy sections or bosses in the casting that are remote from direct riser feed should be chilled so that they conform to the general pattern of directional solidification within the casting. The very fine grain of the magnesium-zirconium alloys is mainly derived from the very marked grain-refining effect of the zirconium content. However, with the nonzirconium alloys, chilling, in combination with a good grain-refinement technique, produces a fine grain and superior mechanical properties. Shakeout. Small castings, up to about 90 kg (200 lb) cast weight, should be allowed to stand after casting until they have cooled to below 260  C (500  F). This is because the magnesium alloys tend to exhibit hot-shortness cracking tendencies; if the castings are removed from the molds too soon, they may show cracks. The larger the casting, the longer they should be allowed to cool before shakeout. Traditional methods for removal of the casting from the mold use vibratory screens equipped to separate chills by the use of electromagnets, a noisy and dusty process, however. Modern methods include placing the mold on a revolving table in an airtight chamber and bombarding it with steel shot to break down the mold. The process must be controlled to avoid abrasion of casting surfaces. It is very important, subsequently in the process, to acid etch the castings to remove any possible surface contamination, thereby precluding reduced corrosion resistance. To preserve optimum corrosion resistance, the use of a nonmetallic abrasive medium, such as alumina, instead of steel shot for cleaning the casting is desirable. With the resin-bonded sand, thermal reclamation is becoming popular because the product obtained is very clean and is equivalent to new sand. It can therefore be used at a high percentage for mold production.

Magnesium and Magnesium Alloys / 1109 Current Applications. The production of sand castings to the extremely high level of dimensional quality currently demanded by the aircraft designer is dependent on the quality of the pattern and core box equipment. These normally consist of precision equipment manufactured in metal and plastic. Wooden pattern equipment, although appreciably less expensive, is rarely used for this type of application. It is, however, quite suitable for castings where the production numbers are low and the dimensional quality requirement is not as stringent. One particular feature of modern mold and core production that has encouraged its widespread use for the manufacture of complex gearbox-type castings for aerospace application is the capability to include complex coring systems within these designs. It is possible, for example, to incorporate very complex systems of small-diameter oil passageway cores and, often surrounding these, some very thin wall sections. In some cases, complex designs that previously could only be made by the investment-casting method can now be made by these refined sand-casting methods. The examples shown in Fig. 4 to 6 show applications of these techniques.

Investment Casting Throughout the history of magnesium alloy castings, designs involving cast components that were very complex, embraced very thin wall sections, had fine surface finish, and attained tight dimensional tolerances became the market for investment casting. As already mentioned, the gap between modern sand casting and investment casting has been considerably narrowed. However, the investment-casting process is still capable of producing castings of tighter tolerances and thinner wall sections than those of sand castings. The disadvantages of the investment process are the cost of the casting in terms of both capital equipment and cost per casting. There is also a much greater restriction on the size of the casting that can be produced. The same alloys can be cast by this process as with sand casting. The processes used are mainly “jointless” in character (that is, lost wax or frozen mercury). This gives, among other things, a general improvement in dimensional accuracy by the elimination of the parting line. The introduction of newer alloys such as Elektron 21 with reduced reactivity has resulted in renewed interest in this form of casting.











whether metal cores (permold) or destructible sand cores (semipermanent mold) are used. It is not currently possible to match the size or complexity obtainable by the sand-casting process. The fact that the mold or die can be repeatedly reused is a very big advantage over sand casting; however, the total number of castings required from a given die must be capable of amortizing the high initial capital cost of the die, and so forth. In general, the surface finish and dimensional tolerances are superior to those obtainable from a comparable sand casting. A distinct disadvantage to the process is the high cost of modifying the casting design or the running system, once the die is constructed. Nearly all of the alloys that can be sand cast are suitable for use in this process. One exception is the high-zinc alloy with no rare earths present (that is, ZK61A), which is extremely hot short and is deemed uncastable.

If steps are taken to accommodate the general tendency for hot shortness, a whole range of castings can be economically produced. The precautions that should be taken to minimize the effects of hot-shortness cracking include provisions for adequate draft. Extreme care is needed when retracting the metal cores not to put undue stress on the hot casting; when two or more cores are being used, they should be extracted simultaneously.

There are also practical limitations on the intricacy of the shapes that can be cast. In particular, deep ribs and complex coring are not practical. In general, because of faster solidification, mechanical properties tend to be superior to those of sand castings. Low-Pressure Die-Casting Process. In this process the metal dies that are used are basically similar to those employed for permanent mold casting. The main difference between the two processes is that in the low-pressure process the molten metal enters the mold from below, under a low pressure. The solidification pattern is inverted. The feed to the solidifying casting comes from below. This method has become popular for the manufacture of highquality automotive wheels (Ref 15).

Die Casting The die-casting process is ideally suited to high-volume production of the type of cast components shown in Fig. 7 and Fig. 8. The high cost of the die can be amortized by the large production volume. The process is a fast production method capable of a high degree of automation, one for which certain magnesium alloys are ideally suited. The number of alloys that can successfully be cast by this process is much more restricted than is the case with sand castings. The alloys typically cast include AZ91A, B,

(a)

(b)

(c)

(d)

Permanent Mold Casting The salient features of the permanent mold casting method are:
This process should be considered when the

design of the part is within the capability of the process. There are two main types of permanent mold casting, depending on

Fig. 7

AZ91 die-cast magnesium alloy used in automotive applications. (a) Door frame for hidden headlight assembly weighing 0.370 kg (0.816 lb). (b) Air intake grille weighing 3.240 kg (7.143 lb). (c) Air cleaner cover (shown mounted on a vehicle engine) weighing 2.307 kg (5.086 lb). (d) Brake and clutch pedal bracket weighing 0.637 kg (1.40 lb). Courtesy of International Magnesium Association

1110 / Nonferrous Alloy Castings From 0.45 kg (1 Ib) of zinc

1 part is produced.

0.45 kg From 0.45 kg (1 Ib) of aluminum

2½ parts are produced.

0.45 kg

From 0.45 kg (1 Ib) of magnesium

32/3 parts are produced.

Fig. 9 (a)

(b)

Comparison of the density of magnesium with the densities of zinc and aluminum to illustrate the lightweight properties of magnesium alloy die castings

Fig. 8

Components illustrating wide diversity of applications using AZ91 magnesium die castings in both commercial and consumer products. (a) Carriage frame for computer daisy wheel. (b) Wheel frame assembly for a wheelchair. Courtesy of International Magnesium Association

and E; AM60; AS41; and AZ61. Four hundred shots per hour from a hot-chamber machine have been reported. Two important advantages shown by magnesium die castings are that molten magnesium does not solder onto the die in the same way as does aluminum; therefore, the expensive dies last much longer. Dies last two to three times longer than with aluminum. Also, magnesium can be machined four times faster than aluminum. The relative machinability ratings of magnesium compared to other metals are: Metal

Magnesium alloys Aluminum alloys Brass Cast iron Mild steel Nickel alloys

Relative power required

1.0 1.3 2.3 3.5 6.3 10.0

These factors, combined with the fact that magnesium is only two-thirds the weight of aluminum (Fig. 9), account for the growing popularity of magnesium die casting, particularly in the automotive field. The excellent machining characteristics of magnesium offer the following advantages:
Reduction in machining time
Tool life as much as ten times longer than





that of other commercial metals Excellent surface finishes with one cut Minimum tool buildup Well-broken chips, minimizing handling Lower machining costs

The same methods of supplying metal to the die, such as the hand-ladling and autopour methods, can be applied.

One other process that can also be advantageously applied to magnesium is the hot-chamber process, in which the gooseneck is immersed in the bath of molten alloy, thereby keeping it replenished and at the correct pouring temperature. This latter process is made possible only by the fact that iron is not attacked by the molten magnesium alloy at the bath and pouring temperatures involved in die casting magnesium. Diecasting Machines. A large number of machines of both the hot- and cold-chamber type are currently in use. The cold-chamber method is often preferred because of the higher injection pressure that can be employed, allowing larger parts to be fabricated. The machines must be equipped with a means for ensuring that no water comes into contact with the magnesium. Dies. The materials used for the manufacture of the dies are similar to those used for aluminum, with dies and cores typically of hot-work tool steels (for example, H11 and H13) heat treated to 44 to 48 HRC. The die surfaces are often nitrided. The die life can be appreciably lengthened if the die surface is finished to a 0.38 mm (15 min.) finish. Repolishing the surface of the die after approximately 25,000 shots is sometimes carried out to retain this high finish. Under modern conditions, any process variables (such as too rapid an injection speed, giving rise to larger dimensions across the parting line) can be fairly easily controlled. Lubrication of the die surface by using a fine mist of one of the water-free proprietary lubricants is the standard method and is best done using automated lubricating systems. Die temperature should be in the optimum range of 260  C þ 14  C (500  F þ 25  F). After   the dies are preheated into this temperature range, the heat supplied by repeated casting is normally

sufficient to maintain the required temperature. In fact, there is usually a need to cool the die to stay within this range. Thus, dies are provided with a means for water cooling. Such control over die temperature is necessary to maintain casting quality. If die temperature is too low, misruns are likely to occur; if it is too high, die soldering, hot tears, and shrinkage can result. Control of Metal Composition. The production of sound magnesium alloy die castings that are free from any contamination from oxide or flux and show superior corrosion resistance depends largely on the degree of control exercised over metal cleanliness and on the method of melting and handling the metal. Very rarely is it possible or desirable to recycle the typical die-casting process scrap directly into the casting furnace. This is because scrap is normally highly contaminated with lubricant, and the slug or biscuit from the runner system normally requires a refining process, before it can be reused. Some die casters carry out this procedure themselves, while others have the material recycled on a contract basis. Much closer attention must presently be paid to the whole aspect of metal handling because of the growing importance of the high-purity AZ91D alloy. The very good corrosion resistance, which is a feature of this alloy, depends on retaining a very low nickel, iron, and copper content. Recycling magnesium die-casting alloys consists of heating the melt charge to approximately 705  C (1300  F), adding flux, and stirring the melt thoroughly to allow the flux to absorb the oxide present. The melt is then allowed to stand for 10 to 15 min to allow the oxide-laden flux to settle as sludge to the bottom of the crucible. The flux generally used is a mixture of magnesium chloride, potassium, and other chlorides, plus a smaller amount of calcium fluoride. The

Magnesium and Magnesium Alloys / 1111

The iron to manganese ratio must be controlled. Beryllium is sometimes added in quantities up to 0.001% to reduce the burning tendency of the molten alloy. Design Parameters. Considerable effort has been applied to the formulation of design criteria (Ref 17). As a general rule of thumb for the designer, wall thicknesses can be the distance to the gate: t¼

D 100

(Eq 1)

where t is the uniform wall thickness and D is the distance from the gate. These thicknesses can be as small as 0.9 mm (0.035 in.). Typical relationships between minimum wall thickness and surface area are: Minimum wall thickness mm

1.78–2.51 2.51

Surface area of die cast magnesium part

in.

m2

in.2

0.070–0.099 0.100

0.052 >0.052

80 >80









h

t h=t

h t

h>t Cored out to avoid undesirable heavy section 2/3t

r

h

t Rounded-off corner is desirable, r = 0.5 to 1 mm min

Ample fillets Rounded corners Blended sections Avoidance of remote heavy sections Strengthening thin sections by the use of ribs Radii that are as large as possible (that is, larger than for zinc alloy die castings)

Casting Finish. The best surface finish on magnesium alloy die castings is obtained by close control over the process variables. These include die temperature, cavity filling, die lubrication, metal temperature, holding time in the die, and the smoothness of the die surfaces. Tolerances. Typical tolerances are shown in the following tables. Typical tolerances for noncritical dimensions are:

Uniform wall thickness rib reinforcement r1 = t r2 = r1 + t

r1 t

r2

r1 = t r2 = 0

r1 Tolerance Dimensions

mm

Length up to 25 mm (1 in.) þ0.25 þ0.010   Additional tolerance for each 25 mm (1 in.) in length 25–305 mm (1–12 in.) þ0.038  þ0.0015  >305 mm (>12 in.) þ0.025 þ0.001  

Typical tolerances for critical dimensions are: Tolerance Dimensions

mm

t

in.

in.

Length up to 25 mm (1 in.) þ0.10 þ0.004   Additional tolerance for each 25 mm (1 in.) in length 25–305 mm (1–12 in.) þ0.038 þ0.0015   >305 mm (>12 in.) þ0.025  þ0.001 

These tolerances represent ordinary production practices and are based on dimensions that are not affected by die parting or moving parts. If those factors cannot be ignored, these tolerances must be increased to compensate for their effect on the die-casting dimensions. Casting Defects. In die castings, misruns and cold shuts probably represent the most frequent form of defect. There can be several causes of these two types of defects, mostly process related. They may, for example, be caused by

r2 t1

t1 2(t1 + t2)


0.005% Fe (Fe = 0.032 Mn maximum)
0.0010% Ni (10 ppm)
0.0400% Cu (400 ppm)

The various relationships between draft and wall depth for interior walls and between draft and cored hole depths are all calculable. Most surfaces vertical to the parting plane of the die should be tapered so that the casting can be readily ejected. Shrinkage forces are considerable, and it is important to have sufficient ejectors to ensure that the casting is not damaged. Draftless magnesium castings can also be made to the competitive advantage of magnesium. Also, since many of the die-casting alloys show a tendency toward stress corrosion, it is important for designers to be able to calculate the induced stress levels to ensure that these fall below limiting values for the alloy in question. Some of these important design features are shown in Fig. 10. Today’s design data replace much of the earlier rule-of-thumb practices. Some basic requirements still apply, such as:

2(t1 + t2)

amount of flux varies from 1 to 3% of the melt, depending on the cleanliness of the charge. Care must be taken to make sure that any recycled turnings are dry before attempts are made to melt them. Also, in the casting process, contamination of the melt with flux may result in corrosion of the final casting. Charred lubricant products or oxides impair the castability and mechanical properties of the casting. During the metal-handling procedure, any excessive tendency to oxidize and any noticeable difference in viscosity are indications that the metal has not been refined properly. Fluxless Melting Parameters. This method of melting, which has already been described, is the preferred method of melting for die casting. It prevents many of the problems associated with flux, including the risk of corrosion caused by flux inclusions. There is also an economic advantage to the use of the fluxless process in that melting losses, primarily caused by the entrapment of metal globules in the sludge formed with flux melting, are greatly reduced. Problems with sludge removal from the flux-melting process are either eliminated entirely or very much reduced. Periodic chemical analysis of the melt is necessary because with repeated melting, losses of beryllium, manganese, and aluminum can occur. Aluminum is present in the alloy to provide strength, hardness, and castability, but too much aluminum results in brittleness. Zinc improves the castability of the alloy and, to a degree, its corrosion resistance. Zinc, like aluminum, improves the mechanical properties. Hot shortness, however, increases with increasing zinc content. Manganese, up to its limit of solubility, improves the corrosion resistance. With the high-purity AZ91D alloy the limits for the contaminating elements are (Ref 16):

r t2

Angular transition increases strength.

Fig. 10

r t1

t2 + t1 2

2(t1 + t2)

Design parameters to aid stress reduction in magnesium alloy castings

extremely slow filling of the die, excessive application of lubrication, incorrect metal or die temperature, dirty metal (that is, excessively oxide laden), slow casting speed, or incorrect shot weight in the cold-chamber process. There is also a greater tendency for porosity in die castings, most often resulting from gas trapped within the casting. If this effect is serious, it can be minimized by conducting an evaluation of the running, gating, venting, and lubricating system. An increase in total vent area to at least half the gate area often lessens the

1112 / Nonferrous Alloy Castings number of cold shuts. An increased injection pressure may eliminate cold shuts altogether. Because of the possibility of gas-related porosity, die castings are not normally thermally treated other than with the occasional use of a low-temperature stabilizing treatment.

Applications of DC magnesium products include forged wheels for aircraft or racing cars, forged helicopter gearbox components, extruded bar and profiles, and rolled plate and sheet.

High-Temperature Alloys Direct Chill Casting Direct chill or (DC) casting involves pouring an alloy into a water-cooled permanent mold. The bottom of the mold is capable of being lowered, once the metal has begun to solidify, to produce a billet or ingot of considerable depth. The product is a solid block of metal that is generally free of porosity and has a good microstructure due to the rapid cooling effect. The billets may be used as a solid block to machine parts from, but are more usually further processed by forging, extrusion, or rolling to produce wrought products. The DC molds are usually made of copper of diameters from 7.5 to 50 cm (3 to 20 in.). It is also possible to cast rectangular shapes for rolling stock. The traditional mold is relatively deep with a water spray ring applied to the outside of the mold to allow solidification of the magnesium. The movable base of the mold is loosely sealed to prevent metal leakage. The alloy is prepared by conventional magnesium processes, that is, fluxed or fluxless melting in a steel crucible. The metal is then transferred to the DC mold either by pouring through a launder system, by pumping, or by siphoning. Once solidification of the metal has commenced, the base is slowly lowered to allow more liquid metal into the top of the mold. Several techniques are employed to maintain a steady level of metal in the mold. These include manual adjustment of metal flow, electrical contactors, and laser level systems. One or more water spray rings are placed below the mold through which the billet is lowered. This enables the direct chill effect onto the hot alloy. The following parameters need careful control to produce satisfactory magnesium alloy DC cast billet:







Liquid metal temperature Mold design Metal height in the mold Mold base drop speed Position of water spray rings Cooling water flow rates at each of the spray rings
Water temperature Once cast, the billets are usually machined to remove the cast surface and then inspected for any defects. The level of inspection will depend on the application, but usually includes visual inspection for surface anomalies, ultrasonic testing for internal inclusions, and eddy current tests for surface cracks. The billets may also require a heat treatment.

Magnesium alloy components often are required to operate at elevated temperatures; examples include:
Helicopter gearbox cases
Jet engine components
High-performance cars

At the higher temperatures there is a need for good tensile properties and creep resistance. These properties are important if magnesium alloys are to remain competitive (Ref 18–23). Some of the older magnesium alloys did have these properties, an example being the thoriumcontaining sand-casting alloys ZH62 and HZ32. These alloys had good tensile properties at the higher temperatures and excellent creep resistance. Use of radioactive thorium is no longer acceptable, and these alloys are no longer specified for new applications. A silver-containing alloy, QE22 (MSR), also has good high-temperature properties, but does not possess the excellent corrosion resistance of newer alloys. Considerable alloy development in recent years has led to several new alloys specifically designed to meet the needs of modern applications. Sand-casting alloys WE43, WE54, and more recently Elektron 21 are now specified for aerospace and motor sport components where high-temperature mechanical properties and excellent corrosion performance are essential (Ref 24, 25). Other alloys are also in the course of development. Die-casting alloys, including AE44 and AJ62 have been developed to target the automotive market, in particular gearbox cases. Again the operating temperatures of the cases have risen over the years, and creep resistance is extremely important (Ref 26–30). Good corrosion resistance is essential since most of these components have no paint or other protective treatments. Alloy development continues in this area, and more alloys are emerging to fill this market area.

Welding Magnesium alloys can be readily welded by gas metal arc welding and by resistance spot welding. Rods of approximately the same composition as the base metal are generally satisfactory. Arc welds in some magnesium alloys, specifically the Mg-Al-Zn series and alloys containing more than 1% Al, are subject to stress-corrosion cracking and thermal treatment must be used to remove the residual stresses that cause this condition. The other types of magnesium alloys, including those containing manganese, rare earths, zinc, or zirconium, are not sensitive to stress corrosion and

normally do not require stress relief after welding (Ref 31). Repaired castings are generally heat treated again after welding. The postweld heat treatment of magnesium casting alloys depends on the desired final temper of the castings. Alloys such as QE22A, EQ21A, WE43B, WE54A, and Elektron 21 require solution heat treatment followed by artificial aging following weld repair. A reduced solution treatment time and temperature will assist in prevention of grain growth in the weldment area. However, if complete solution heat treatment is not desired, welded castings should always be stress relieved. ACKNOWLEDGMENT This article has been updated and revised from H. Proffitt, Magnesium and Magnesium Alloys, Casting, Vol 15, Metals Handbook, ASM International, 1988, p 798–810. REFERENCES 1. W. Unsworth, “Application Guide Lines for Various Magnesium Alloys,” Paper presented at Magnesium Symposium, Westinghouse Electric Corporation, Lima, Sept 1984 2. W. Unsworth, Meeting the High Temperature Aerospace Challenge, Proc. 43rd Annual World Magnesium Conference, International Magnesium Association, June 1986 3. R. Carnahan, R. Decker, E. Nyberg, R. Jones, and S. Pitman, Development of Semi-Solid Molded Magnesium Components from Alloys with Improved High Temperature Creep Properties, Magnesium Technology 2000, as held at the 2000 TMS Annual Meeting (Nashville, TN), March 12–16, 2000, p 403–409 4. E.F. Emley and P.A. Fisher, The Control of Quality of Magnesium-Base Alloy Castings, J. Inst. Met., Vol 85, Part 6, 1956–1957 5. J.W. Frueling, “Protective Atmospheres for Molten Magnesium,” Ph.D. thesis, University of Michigan, 1970 6. J.W. Frueling and J.D. Hannawalt, Protective Atmosphere for Melting Magnesium Alloys, Trans. AFS, 1969, p 159 7. J.D. Hannawalt, Practical Protective Atmospheres for Molten Magnesium, Met. Eng., Nov 1972, p 6 8. G. Schemm, Sulphur Hexafluoride as a Protection Against Oxidation, Giesserei, Vol 58, 1971, p 558 9. J.D. Hannawalt, “SF6—Protective Atmospheres for Molten Magnesium,” Paper G-775-111, Society of Die Casting Engineers, 1975 10. R.S. Busk and R.B. Jackson, Use of SF6 in the Magnesium Industry, Proc. 37th Annual World Magnesium Conference, International Magnesium Association, June 1980 11. H.J. Proffitt, Magnesium Sand Casting Technology, Proc. Special Conference on Recent Advances in Magnesium Technology, AFS/ CMI, June 1985

Magnesium and Magnesium Alloys / 1113 12. E.F. Emley, Principles of Magnesium Technology, Pergamon Press, 1966 13. L. Petro, “Premium Quality Magnesium Sand Castings—Production and Application,” Paper presented at Magnesium Symposium, Westinghouse Electric Corporation, Lima, Sept 1984 14. G. Betz and N. Zeumer, Production of Light Metal Castings With Precast Lubrication Passageways, Proc. 43rd Annual World Magnesium Conference, International Magnesium Association, June 1986 15. J. Bolstad, What Improved Metal Handling Practices Can Do to Automotive Castings, Proc. 44th Annual World Magnesium Conference, International Magnesium Association, May 1987 16. K.N. Riechek, K.J. Clark, and J.E. Hillis, Controlling the Salt Water Corrosion Performance of Magnesium ZA91 Alloy in High and Low Pressure Cast Forms, Proc. Special Conference on Recent Advances in Magnesium Technology, AFS/CMI, June 1985 17. R.S. Busk, Magnesium Products Design, Marcel Dekker, 1987 18. P. Lyon, I. Syed, and S. Heaney, Elektron 21—An Aerospace Magnesium Alloy for Sand Cast and Investment Cast Applications, Proc. Seventh International Conference on Magnesium Alloys and Their Applications, D.G.M., 2006

19. G. Sigaudo, Development of Aerospace Casting in WE43 Poured by the DPSC Process, Proc. Third International Magnesium Conference, 1996 20. J-M. Arlhac and J-C. Chaize, New Magnesium Alloys and Protections in New Helicopters, Proc. Third International Magnesium Conference, 1996 21. B. Geary, Advances in the Application of Magnesium in Helicopter Gearboxes, Proc. Third International Magnesium Conference, 1996 22. L. Duffy, Magnesium Alloys: the Light Choice for Aerospace, Mater. World, Vol 4, 1996, p 127–130 23. J. King, P. Lyon, and K. Savage, Influence of Rare Earth Elements and Minor Additions on Properties and Performance of Magnesium-Yttrium Alloys in Critical Aerospace Applications, Proc. 59th World Magnesium Conference, Montreal 2002 24. H. Karimzadeh, “The Microstructure and Mechanical Properties of Some Magnesium Alloys Containing Yttrium and Heavy Rare Earths,” Ph.D. Thesis, University of Manchester, 1985 25. W. Unsworth, The Role of Rare Earth Elements in the Development of Magnesium Alloys, Int. J. Mater. Prod. Technol., 1989

26. P. Bakke, A.C. Bowles, and H. Westengen, Elevated Temperature Alloys—Paths for Further Performance Gains in AE44, Proc. Seventh International Conference on Magnesium Alloys and Their Applications, D.G.M., 2006 27. H.G. Gjestland and H. Westengen, Advancement in High Pressure Die Casting of Magnesium, Proc. Seventh International Conference on Magnesium Alloys and Their Applications, D.G.M., 2006 28. M. Pekguleryuz and A. Kaya, Creep Resistant Magnesium Alloys for Powertrain Applications, Proc. Sixth International Conference, Magnesium Alloys and Their Applications, 2003 29. J. Abthoff, W. Gelse, and J. Long, Magnesium Alloys and Their Applications as Housing Materials for Sports Prototype— and Passenger Cars, Proc. Third International Magnesium Conference, 1996 30. J.M.A. Willekens, Magnesium Seat-Frames: History and Evaluation, Proc. Third International Magnesium Conference, 1996 31. S. Houch, B. Mikucki, and A. Stevenson, “Selection and Application of Magnesium and Magnesium Alloys, Properties and Selection: Nonferrous Alloys and SpecialPurpose Materials, Vol 2, ASM Handbook, ASM International, 1990

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1114-1118 DOI: 10.1361/asmhba0005335

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Cobalt and Cobalt Alloy Castings Steve Sikkenga, Cannon-Muskegon

COBALT-BASE ALLOYS have been used in demanding applications for over 80 years. From the introduction of Stellite (Deloro Stellite, Inc.) in the late 1920s, the dental alloy Vitallium (Howmedica Inc.) in the 1930s, the turbocharger alloys such as HS-31 and X-40 in the 1940s, through to today’s (2008) technology, cobalt alloys have contributed significantly to the performance of advanced-technology products and processes. Cast cobalt alloys are used for three primary purposes in industrial, aerospace, and medical applications:  To resist wear  To resist chemical corrosion  To resist high-temperature stress and surface

degradation Today (2008), cast cobalt alloys play an important role in aero- and land-based gas turbines. While vacuum cast nickel alloys predominate in the hot sections of modern aero turbine engines, cobalt alloys are routinely specified for particularly demanding applications, such as fuel nozzles and vanes for industrial gas turbines. Cast Co-Cr-Mo alloys compete directly with titanium and cobalt forgings for medical prosthetic implant devices. Cast cobalt-base “hard alloys” (e.g., Co-WC cemented carbide) are often found in applications involving severe wear conditions. Cobalt alloys are relatively expensive. Cost considerations dictate that these alloys are used in applications where their particular blend of properties meets unique material demands.

Physical Metallurgy Cast cobalt alloys are strengthened primarily by the presence of carbides in the microstructure, by substitutional solid-solution strengthening, and by grain-boundary strengthening. Stacking faults play a role in the development of properties in these alloys.

Crystallography The crystal structure of pure cobalt is hexagonal close-packed (hcp) at room temperature. Above 420  C (785  F), the face-centered cubic

(fcc) crystal structure is stable. However, the transformation from fcc to hcp upon cooling is sluggish. Alloying elements tend to stabilize one structure or the other and serve to increase or decrease the transformation temperature. The fcc crystal structure is stabilized by the addition of carbon, niobium, iron, manganese, nickel, tantalum, titanium, and zirconium. The hcp structure is stabilized by the presence of chromium, molybdenum, silicon, and tungsten. The fcc structure is necessary for high-temperature strength and ductility at all temperatures. The composition of the fcc matrix phase is a function of overall alloy composition and the phases that precipitate during solidification, upon heat treatment, or in service. Stacking faults are regions of the hcp configuration contained in the fcc matrix and are an important crystallographic feature of cobalt alloys. The tendency to form stacking faults depends on stacking fault energy (SFE), which is a function of composition. High SFE reduces the tendency for stacking faults to form. Elements that increase SFE tend to stabilize the fcc structure. Because of its complete solubility in the matrix, nickel is often employed in cobalt-base alloys to increase SFE and stabilize the fcc structure. When encountered, regions of the hcp matrix phase are also known as the E (epsilon) phase. Stacking faults interact with dislocations directly and also are sites for precipitation of carbide and topologically close-packed (tcp) phases such as s (sigma) and Laves phases. High stacking fault density tends to increase strength and wear resistance but can reduce tensile and stress-rupture ductility. In order of effectiveness, the solid-solution hardening elements are tantalum, tungsten, niobium, and molybdenum, followed by zirconium, titanium, and carbon. The effectiveness of the solution strengthening is proportional to the difference between atomic radii and the lattice parameter of the matrix and is also a function of solubility in the matrix.

Compositions The nominal compositions of representative cast cobalt alloys are shown in Table 1. Many of the elements present in cobalt alloys serve multiple roles by promoting solid-solution

hardening, facilitating carbide formation, and exerting an influence on SFE. Chromium. All of the commonly used cobalt alloys contain chromium. This element provides protection against oxidation and hot corrosion. Chromium also participates in carbide precipitation. Refractory elements such as tungsten, molybdenum, tantalum, and niobium are added to provide strengthening through solid-solution hardening and carbide formation. Carbon is a critical element in all the cobalt alloys. Nickel is included in many of the compositions for austenite (fcc) stabilization. The element is completely soluble in cobalt and does not form carbides. Aluminum and titanium are less-often employed for strengthening in cobalt alloys than they are in nickel alloys. The coherent gamma-prime strengthening phase (Ni3Al, Ti) found in nickel-base superalloys does not have a counterpart in cobalt-base alloys. Residual gases such as oxygen, nitrogen, and hydrogen are typically controlled to low levels, except for those instances where nitrogen is intentionally added for solid- solution strengthening. Sulfur is detrimental in cast grades because it can contribute to hot tearing tendencies and also promotes undesirable wetting between the alloys and refractories during subsequent melting and casting. Boron is often included in cobalt alloys, where the element improves high-temperature strength, rupture life, and ductility by improving grain-boundary structure. Many of the tramp and trace elements, which are controlled to low levels in nickel-base superalloys, are also of concern in cobalt alloys. These elements include lead, tin, antimony, thallium, tellurium, zinc, bismuth, and so on. These are typically held to low parts per million (ppm) concentrations in cobalt high-temperature alloys and critical medical components. These elements tend to segregate to grain boundaries and can reduce high-temperature rupture life and ductility.

Phases and Microstructure The microstructure of most cast cobalt alloys is composed of the fcc matrix phase, carbides,

Cobalt and Cobalt Alloy Castings / 1115 Table 1

Nominal compositions and designations of representative cobalt alloys Element

Purpose

Wear-resistant alloys

Corrosion-resistant alloys Temperature-resistant alloys

Effect of elements in cast cobalt alloys Carbide formation Solid-solution strengthening Surface stability Grain-boundary strengthening Increase face-centered cubic stability

Alloy trade name

AMS No.

UNS No.

C

Cr

Ni

Mo

Ta

W

Nb

Zr

Ti

B

Others

Stellite 3 Stellite 6 Stellite 12 Stellite 19 Tribaloy T-400(a) 98M2 Star J ASTM F-75 Haynes Ultimet(b) MAR-M-509 MAR-M-302 ECY 768 Stellite 25 Stellite 31 X-40 X-45 FSX-414 Stellite 21 WI-52

... 5373, 5387 ... ... ... ... ... ... ... ... ... ... ... 5382 ... ... ... 5385 ...

... R30006 R30012 ... R30400 ... ... ... R31233 ... ... ... ... R30031 ... ... ... R30021 ...

2.5 1.0 1.5 1.8 0.05 2.0 2.5 0.25 0.06 0.6 0.85 0.6 0.1 0.5 0.5 0.25 0.25 0.25 0.45

30 28 30 31 8 30 33 28 26 23 21.5 23 20 25 25 25 29 27 21

... ... ... ... ... 4 ... ... 9 10 ... 10 10 10 10 10 10 2.5 ...

... ... ... ... 28 ... ... 6 5 ... ... ... ... ... ... ... ... 5.5 ...

... ... ... ... ... ... ... ... ... 3.5 9 3.5 ... ... ... ... ... ... ...

12 4 8 10 ... 18 18 ... 2 7 10 7 15 7.5 7.5 7.5 7.5 ... 11

... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... 2

... ... ... ... ... ... ... ... ... 0.5 0.2 ... ... ... ... ... ... ... ...

... ... ... ... ... ... ... ... ... 0.2 ... 0.2 ... ... ... ... ... ... ...

... ... ... ... ... 1 ... ... ... ... 0.005 ... ... ... 0.01 0.01 0.01 ... ...

... ... ... ... 2.5 Si 4V ... ... 2 Fe, 0.08 N ... ... ... ... ... ... ... ... ... ...

 

  

... ... ... ...

 

 

 

 

... ... ...

... ... ... ...

... ... ...

... ... ... ... Fe, Mn

...

 

... ...



... ... ...

... ...



... ... ...

... ...



 





...

(a) DuPont de Nemours and Co. Inc. (b) Haynes International

and occasionally tcp phases such as s (sigma) and Laves phase. The role of carbon and carbides is paramount in these alloys. During solidification and subsequent cooling, carbides of various compositions are formed interdendritically and at grain boundaries. As carbon content and refractory element concentration are increased, the volume fraction of carbide increases proportionally. The composition, structure, and morphology of carbides depend on the overall alloy composition and thermal history of the casting. Representative microstructures are shown in Fig. 1 to Fig. 4. The M23C6-type carbides are found in nearly all cobalt alloys containing chromium. As such, they are the most commonly observed carbides. Carbides of the form MC are found in alloys containing niobium, tantalum, titanium, and zirconium, as are M7C3 and M3C2 when carbon content is relatively high. The M6C carbide is found in alloys containing molybdenum and tungsten. For cobalt alloy castings used at high temperatures (i.e., turbine engine applications), the volume fraction, composition, distribution, size, and morphology of carbides are not completely stable and often vary during the service life of the casting. Intermetallic or tcp phases, including Laves, sigma, and mu, can also form in cast cobalt alloys. These phases are generally undesirable due to their negative effects on ductility.

Properties Because cobalt-base alloys are costly, they are used in situations where the application

Fig. 1

As-cast Stellite 12, unetched. Primary MC carbides and face-centered cubic matrix. Original magnification: 100

Fig. 2

As-cast ASTM F-75, electropolished and unetched. Present are eutectic carbides, grain boundary, and face-centered cubic matrix. Original magnification: 100

ASTM F-75, solution annealed at 1220  C (2230  F), electropolished and unetched. Same sample as Fig. 2, showing carbides nearly completely dissolved. Also shown is distributed microporosity, carbide-free grain boundary, and triple point in the facecentered cubic matrix. Original magnification: 100

Fig. 3

Fig. 4

As-cast FSX-414, unetched. Face-centered cubic matrix and interdendritic M23C6 carbides. Original magnification: 400

1116 / Nonferrous Alloy Castings Charge Materials. Cobalt master alloys are often produced in controlled closed-loop reverting arrangements, although nearly any combination of virgin, revert, and select material sources can be accommodated. A typical charge may include 60% customer-owned revert combined with 40% virgin cobalt, chromium, tungsten or molybdenum, and nickel. Virgin/revert ratios are specified by individual foundries to manage revert pools effectively. While the optimum ratio would appear to be equal to the overall foundry casting yield, ratios are occasionally used that maximize economic gain in light of cobalt price volatility. Casting revert generated from AOD-refined master alloy can be recycled in-house within reason and where permitted by specification. Silicon and manganese levels should be monitored and corrected if needed. In many cases, iron tends to be the single element that limits the use of cobalt-base foundry revert or select materials and comes from several sources. First, raw cobalt may contain some iron. Second, machining chips routinely contain ironbase contaminants, usually stainless steel, that cannot be removed effectively. For this reason, machining chips are often premelted to determine their exact composition before being incorporated into production heats. A third source of iron contamination (as well as other elements) is mixed casting revert. Some iron also comes into cobalt alloys from shot blast media and from iron-containing refractories. Continuous Casting. Following AOD, 4500 kg (5 ton) cobalt heats are transferred to the continuous casting machine. Cobalt alloys are produced in nominal diameters of 64, 75, and 100 mm (2½, 3, and 4 in.). Recently, 50 mm (2 in.) diameter capabilities have been added. A typical remelt ingot measures 75 by 460 mm (3 by 18 in.) and weighs approximately 16 kg (36 lb). The total processing time, including induction melting and AOD refining, is approximately 6 h. Foundry Performance. Argon melt protection is recommended and often employed when making critical cobalt castings. The use of argon is almost always beneficial, regardless of the alloy being cast. In the case of cobalt

requires properties not available in other, more economical alloys. Physical and mechanical properties for select cobalt casting alloys are listed in Tables 2 and 3, showing typical property values for various applications. Other sources of property information include Properties and Selection: Nonferrous Alloys and SpecialPurpose Materials, Vol 2, ASM Handbook; Nickel, Cobalt, and Their Alloys, ASM Specialty Handbook; and property data sheets from individual producers.

Foundry Cobalt alloys are cast by several different foundry methods. Highly engineered components are usually investment cast in air or under a protective atmosphere, or by vacuum investment casting. Some wear-resistant parts are cast using green sand, resin-bonded sand, shell mold, or permanent mold techniques. (For information on these foundry techniques, see the Sections “Melting and Remelting” and “ExpendableMold Casting Processes with Permanent Patterns” in this Volume.) Argon-Oxygen Decarburization. The argon-oxygen decarburization (AOD) process originally developed for stainless steels is also effective for reducing levels of carbon, silicon, sulfur, and dissolved gases in nickel and cobalt alloys. When moderate-to-high carbon contents (relative to stainless steels) are desired, as is typical in cobalt alloys, AOD processing consists of modest oxidation and reduction steps combined with a pure argon stir or purge. Artificial slags rich in lime (CaO) or other basic oxides are used to control sulfur to extremely low levels. Sulfur levels of 10 ppm or less are common, and 2 ppm or less is achievable. Trace element control in cobalt alloys is accomplished through the refining action of AOD as well as through careful raw material selection. In AOD, elements with high vapor pressures (at steelmaking temperatures) are volatilized and effectively removed as a gas. The elements tin and antimony have relatively low vapor pressures and are kept low through careful control of revert, select, and raw materials.

Table 2 Physical properties of select cast cobalt alloys Density

Melting range

Purpose

Example alloy

g/cm3

lb/in.3

Wear-resistant alloys Corrosion-resistant Alloys Heat-resistant alloys

Stellite 6 ASTM F-75 FSX-414

8.3–10.0 8.0–8.9 8.3–9.4

0.30–0.36 0.29–0.32 0.30–0.34



C

1260–1360 1315–1370 1340–1400

Modulus of elasticity F

GPa

106 psi

2300–2475 2400–2500 2450–2550

200–220 200–214 209–248

29–32 29–31 30–36



alloys, the higher cost of the alloys and the critical nature of the component usually justify the use of argon. Through effective use of argon melt protection combined with good foundry practice, the oxygen increase from ingot to casting can be 20 ppm or less. With appropriate foundry practices, cobalt alloys can be cast in air to be very clean. Oxygen contents of 100 ppm or less in AOD-refined meltstock contribute to this cleanliness. In any airmelt operation, mechanical removal of the small amount of dross present on the melt surface before casting is crucial. The use of argon melt protection will reduce the amount of dross formation but will not eliminate the need to remove this small amount of dross. Inoculants. Many cobalt alloys respond well to grain size control through shell inoculation. Incorporation of 2 to 6% cobalt aluminate, cobalt metasilicate, or cobalt oxide yield substantial grain size reductions near casting surfaces with careful control of casting temperatures. The amount of grain size reduction is proportional to the amount of inoculant used. Inoculants also shorten slurry life, so the amount used in practice is a compromise. Filtration. While the best practice will always be to start with clean alloy and keep it clean, filters can be useful. Filters can reduce fill rates, acting as chokes to reduce turbulence and entrapped air if placed in the in-gates. However, by acting as a restriction, filters may require additional superheat to maintain the ability to feed thin sections or long distances. In any case, filters should be viewed as the “last line of defense” against nonmetallic inclusions rather than the primary means by which clean castings are produced. Castability. Cobalt alloys exhibit good-toexcellent castability, especially compared to nickel alloys and low-carbon stainless steels. The alloys are characterized by relatively high carbon contents, and some have appreciable silicon and/or boron, all of which provide alloy fluidity. With melting ranges somewhat lower than the stainless steels, pouring temperatures rarely exceed 1650  C (3000  F). Using a fixed superheat-over- liquidus approach (+140  C, or 250  F, for example) suggests that pouring temperatures of 1670  C (3040  F) or less are reasonable. The alloys generally possess good hot strength (which is often the basis for their use in the first place) and are quite resistant to hot tearing. However, some cracking may be experienced in the high-carbon, wear-resistant alloys when mechanical impact or severely restrictive gating arrangements are encountered. The

Table 3 Mechanical properties of select cast cobalt alloys Ultimate tensile strength

Yield strength

Purpose

Example alloy

MPa

ksi

MPa

Wear-resistant alloys Corrosion-resistant alloys Heat-resistant alloys

Stellite 6 ASTM F-75 FSX-414

830 690–860 690–970

120 100–125 100–140

Near ultimate tensile strength 450–620 450–690

ksi

Ductility, % elongation

Hardness, HRC

65–90 65–100

nil–4 8–20 4–10

37–50 30–36 30–42

Cobalt and Cobalt Alloy Castings / 1117

Postcasting Treatment

hardness while decreasing tensile ductility. As in many alloy systems, the aging response of cobalt alloys can be quite sensitive to time and temperature, and specific aging conditions vary by alloy. The precipitation of carbide necessarily depletes the matrix of chromium and other elements. Negative effects on ductility and corrosion resistance should be evaluated before an aging heat treatment is specified. Medical implant castings may be subjected to a sintering heat treatment to bond mechanically applied metal Co-Cr-Mo beads, providing an open yet strongly adherent structure for maximum bone/implant fixation. Sintering temperatures approaching 1275  C (2325  F) are used to bond this porous coating to ASTM F-75 castings. To avoid incipient melting, slow ramp-up or multistep heating schemes are used.

Heat Treatment

Hot Isostatic Pressing (HIP)

The heat treatment of cobalt alloys is primarily intended to manipulate carbides for the optimization of properties. The fcc matrix can dissolve a substantial amount of carbides at elevated temperatures. This relationship of increasing carbide solubility with temperature provides for classic precipitation-hardening processes. Upon cooling, carbides are likely to precipitate in morphologies and distributions differing from those found in the as-cast condition. In high-temperature applications, carbides are also likely to precipitate in use over time, with a pronounced effect on stress-rupture life, rupture ductility, and impact strength. Many cobalt alloys are used in the as-cast condition, especially those intended for wear-resistant applications. Occasionally, wear-resistant grades are stress relieved at approximately 900  C (1650  F). Alloys intended for corrosion service are often solution annealed, a process that dissolves carbides, freeing chromium and other elements to protect the matrix. Annealing temperatures (and, when used, hot isostatic pressing temperatures) tend to be high, in the 1150 to 1270  C (2100 to 2320  F) range. To avoid carbide precipitation and the accompanying matrix depletion of chromium, rapid cooling rates from solution heat treat temperatures are required. Techniques include forced-air cooling, water or oil quenching, or gas fan quenching of vacuum heat treated castings. Alloys for high-temperature applications are often solution annealed or annealed and aged in order to break down primary carbides, allowing for prolific, fine carbide precipitation. Similar to the corrosion-resistant alloys, annealing temperatures for high-temperature alloys are usually high, in the 1090 to 1220  C (2000 to 2225  F) range. Aging at an intermediate temperature to intentionally precipitate carbides may, in some cases, follow solution annealing. Most carboncontaining cobalt alloys respond to aging at temperatures of approximately 760  C (1400  F). Aging generally increases strength and

High-performance castings are often subjected to HIP to close solidification shrinkage and microporosity. Castings are heated in a pressure vessel to temperatures similar to or higher than those used for solution annealing. Argon gas pressures of 100 to 179 MPa (15,000 to 25,000 psi) exert isostatic stress on the casting and effectively move internal porosity to external surfaces. The high-cycle fatigue resistance of castings is improved by HIP.

nitrogen-containing alloys, such as F-75 and Haynes Ultimet (Haynes International), are stable in terms of nitrogen pickup or loss. Consistent properties and casting performance are facilitated by uniform heat-to-heat alloy composition and process controls. Specific casting considerations and foundry practices are covered in other sections of this Volume. Quality. Casting defects to be avoided are similar to those found in other cast metals. Solidification shrinkage, gas porosity, entrapped air, nonmetallic inclusions, and contamination from other alloys in the foundry are potential pitfalls that must be avoided by careful process engineering and process control.

Coatings Various techniques are used to apply performance-enhancing coatings on cobalt-base castings. Plasma spray, high-velocity oxyfuel thermal spray, physical vapor deposition, and chemical vapor deposition are used to deposit specialized surface layers. For gas turbine applications, aluminide corrosion- and oxidationprotection layers are employed. Ceramic thermal barrier coatings allow higher operating gas temperatures. Medical implant castings may be coated to provide improved bone ingrowth and adhesion. Sintered beads and plasma-sprayed overlays are employed to enhance bonding between bone and implant.

Welding Cobalt-base alloys may be specified for a given application, in part, because of generally good weldability compared to alternative (usually nickel-base) alloys. Welding of cobalt alloys can be accomplished with most welding techniques. For example, ASTM F-75 is readily welded using manual or automated gas tungsten arc welding techniques using direct current with electrode-negative, pure argon shielding gas and matching filler wire. Best results are obtained by minimizing heat input. The F-75 castings should be solution annealed after welding. The ductility of cobalt-base cast alloys varies widely, and weld cracking can be

problematic in some grades. Preheating may be used to minimize cracking.

Applications Wear-Resistant Alloys Example alloys include Stellite 6, Stellite 3, Stellite 12, Stellite 19 (Deloro Stellite, Inc.), Tribaloy T-400 (DuPont de Nemours and Co. Inc.), Star J, 98M2, and Haynes Ultimet (Haynes International). The history of the use of cobalt alloys for wear-resistant applications begins with the 1907 patents by Elwood Haynes for his Stellite alloys based on cobalt and chromium. Carbon (and occasionally boron) is used in conjunction with chromium and one or more refractory elements to produce a high-hardness, carbide-rich material. These alloys feature good hot hardness, reasonable toughness (compared to cemented carbide), and properties intermediate between high-speed steels and cemented carbides. Hardness values increase proportionally with MC, M6C, and M23C6 carbide volume fraction that accompanies additions of tungsten, molybdenum, and carbon to the composition. While strength and hardness increase with carbide volume fraction, ductility and toughness decrease. The melting ranges for these highcarbon alloys are typically lower and narrower than those of the high-temperature and medical alloys. As such, most of the wear-resistant cobalt alloys exhibit very good casting fluidity. Applications for wear-resistant cast cobalt alloys include glass-forming tools, wire dies, wear plates and slides, valve balls, valve seat inserts, scraper blades, and similar components. Stellite 6 is one of the most common cobalt wear-resistant alloys. Its ability to withstand sliding and abrasive wear comes from the high volume fraction of carbides formed from the 1.2% C, 5% W, and 30% Cr in the alloy. The alloy has a Rockwell hardness of approximately 43 HRC but limited tensile ductility. An alternative approach to wear resistance is employed by Haynes Ultimet. This alloy contains rather low carbon content (and thus a low volume fraction of carbide). By careful control of the alloy balance to achieve a low stacking fault energy fcc matrix and subsequently high work-hardening rate, Ultimet achieves excellent wear resistance uniquely combined with a high degree of corrosion resistance. The alloy also shows good tensile ductility and weldability. The alloy features an intentional nitrogen content and good foundry performance. Molybdenum is used in the Tribaloy grades, with wear-resistant properties owing to a silicon-stabilized Laves phase rather than carbides. The microstructure consists of approximately 50% Laves phase with 1100 HV hardness in a soft fcc matrix. The cast Tribaloy alloys have very low ductility, which limits the designs of parts that can be made without cracking.

1118 / Nonferrous Alloy Castings Corrosion-Resistant Alloys Example alloys include ASTM F-75 and Haynes Ultimet. Corrosion-resistant cast cobalt alloys are characterized by relatively high chromium contents and lower carbon contents (compared to wearand temperature-resistant grades). Chromium provides corrosion resistance in oxidizing acids, while molybdenum provides resistance to pitting corrosion and chloride attack. The corrosionresistant grades are typically lower in carbon content than high-temperature or wear-resistant alloys, with properties developed by substitutional and interstitial solid-solution strengthening and a modest contribution from precipitated carbides. Corrosion-resistant alloys are typically used in the solution-annealed condition. ASTM F-75 offers remarkable corrosion resistance in the human body, with an order of magnitude improvement in corrosion rate compared to stainless steels. It is therefore the most commonly specified alloy used for cast orthopedic implants in humans. Haynes Ultimet, in addition to offering exceptional wear resistance, is also specified for its excellent corrosion properties. The alloy is uniquely designed to resist the combined degrading effects of corrosion and wear. Similar in some respects to F-75, Ultimet contains significant levels of chromium and molybdenum, which impart resistance to corrosion. Wear properties result from a high work-hardening rate rather than from a high volume fraction presence of carbides.

Alloys for Orthopedic Implants An example alloy is ASTM F-75. The use of artificial orthopedic implants has been well established for many years. Advances in medical technology have led to implants improving the quality of life for an everincreasing range of patients. Two metallic alloy families have dominated in terms of prosthetic implants: titanium and cobalt alloys. While

titanium offers an advantage in weight, Co-CrMo is unparalleled in its clinical history, market acceptance, and widespread use. In the late 1920s, Austenal Laboratories searched for an alternative to gold for fabricating denture supports. That alternative was found in a derivative of Haynes Stellite, which was subsequently named Vitallium, and has been used since the 1930s. Today (2008), this alloy is used for orthopedic implants, most notably artificial hips and knees. The alloy is generically referred to by its ASTM International designation, F-75 (also by ISO 5832 part 4), and contains approximately 29% Cr and 6% Mo. While the ASTM International specification limits carbon to 0.35%, implant manufacturers have opted for lower levels of carbon and an intentional addition of nitrogen. The addition of nitrogen as an intentional alloying element has allowed Co-Cr-Mo to achieve high levels of strength with good ductility without sacrificing corrosion resistance. The alloy F-75 typically contains approximately 0.15% intentionally added nitrogen. This level of nitrogen appears to provide interstitial solid-solution strengthening of the matrix fcc phase and may partially substitute for carbon in carbides. Highly polished castings are used for femoral heads for total joint replacement hips and femoral knee components. Other cobalt medical castings include acetabular cups and tibial trays as well as smaller components for shoulder, ankle, and finger joint replacement. In all cases, but especially in hip components, casting quality is imperative because parts are subjected to heavy loads in service and subject to fatigue. Articulating metal-on-metal implant designs eliminate the ultrahigh-molecular-weight polyethylene used in some contemporary replacement joints but further raise the requirement for casting quality. Implant castings are often given multistep heat treatments, including solution annealing, HIP, and sintering of boneingrowth structures and materials.

Alloys for Aero and Land Gas Turbines Example alloys include MAR-M-509, FSX414, HS-31 (X-40/X-45), HS-25, and ECY 768. Cobalt alloys are well suited to high-temperature creep and fatigue-resistant, primarily nonrotating applications where stress levels are lower than for rotating components. For this reason, turbine vanes and other static components are frequently designed in cobalt alloys. Cast cobalt alloys are used in aerospace and industrial gas turbine applications because they possess good stress-rupture properties and creep resistance at very high temperatures. Cobalt alloys possess slightly lower coefficients of thermal expansion and slightly better (higher) thermal conductivities than nickel-base superalloys. This combination of properties, combined with fairly flat stress-rupture curves, makes cobalt alloys appropriate for larger components with good thermal fatigue resistance. Cobalt alloys are not as oxidation resistant as most nickel-base superalloys because of the high aluminum/titanium contents in the nickel alloys. However, hot corrosion properties of cobalt alloys can be superior to those of the nickel alloys because of differences in the melting temperature of Co-Co4S3 (in cobalt alloys) compared to Ni-Ni3S2 (in nickel alloys) and in the diffusivities of sulfur in nickel and cobalt. SELECTED REFERENCES  J.D. Donaldson and S.J. Clark, Cobalt in

Superalloys, Cobalt Development Institute, London, 1985  C. Sims and W. Hagel, The Superalloys, J. Wiley & Sons, New York, 1972  H.J. Wagner and A. Hall, The Physical Metallurgy of Cobalt-Base Superalloys, Defense Metals Information Center, Columbus, OH, 1962

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1119-1127 DOI: 10.1361/asmhba0005336

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Nickel and Nickel Alloy Castings* Selc¸uk Kuyucak, CANMET—MTL, Natural Resources Canada

NICKEL-BASE ALLOY CASTINGS are widely used in corrosive-media and high-temperature applications. The principal alloys are identified by designations of the Alloy Casting Institute (now called the High Alloy Product Group) of the Steel Founders’ Society of America and are included in ASTM International specifications A 494 and A 297 and U.S. Navy specifications (QQ). There are also many proprietary grades for severe-corrosion applications, as well as heat-resistant alloys. In addition to these conventionally processed alloys, directionally solidified (DS) and single-crystal (SC) alloys are also being processed. The various types of cast alloys can be classified as:
Nickel
Nickel-copper

Table 1








Nickel-chromium-iron Nickel-chromium-molybdenum Nickel-molybdenum Nickel-base proprietary Directional solidification/single crystal

The cast nickel-base alloys, with the exception of some high-silicon and proprietary grades, have equivalent wrought grades and are frequently specified as the cast components of a wrought-cast system. Compositions of cast and equivalent wrought grades differ in minor elements because workability is the dominant factor in wrought alloys, while castability and soundness are the dominant factors in cast alloys. The differences in minor elements do not result in significant differences in serviceability.

Tables 1 and 2 provide designations and compositions for corrosion-resistant, heat-resistant, and DS/SC alloys. It should also be noted that extensive data on mechanical properties, microstructural characteristics, and corrosion properties of nickel-base castings can be found in Volumes 1, 9, BA and 13B, respectively, of the ASM Handbook.

Compositions Cast Nickel. The ASTM A 494 CZ-100 grade covers the requirement for a cast nickel alloy. A higher-carbon or a higher-silicon grade is occasionally specified for greater resistance to wear and galling. The minor elements within

Compositions of corrosion-resistant nickel-base casting alloys Composition, %

Alloy

Cast nickel CZ-100 Nickel-copper M-35-1 M-35-2 M-30H M-25S M-30C QQ-N-288-A QQ-N-288-B QQ-N-288-C QQ-N-288-D QQ-N-288-E QQ-N-288-F Nickel-chromium CU5MCuC

C

Si

Mn

Cu

Fe

Cr

P

S

Mo

Others

1.0

2.0

1.5

1.25

3.0

...

0.03

0.03

...

...

0.35 0.35 0.30 0.25 0.30 0.35 0.30 0.20 0.25 0.30 0.40–0.70

1.25 2.0 2.7–3.7 3.5–4.5 1.0–2.0 2.0 2.7–3.7 3.3–4.3 3.5–4.5 1.0–2.0 2.3–3.0

1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5

26.0–33.0 26.4–33.0 27.0–33.0 27.0–33.0 26.0–33.0 26.0–33.0 27.0–33.0 27.0–31.0 27.0–31.0 26.0–33.0 29.0–34.0

3.5 3.5 3.5 3.5 3.5 2.5 2.5 2.5 2.5 3.5 2.5

... ... ... ... ... ... ... ... ... ... ...

0.03 0.03 0.03 0.03 0.03 ... ... ... ... ... ...

0.03 0.03 0.03 0.03 0.03 ... ... ... ... ... ...

... ... ... ... ... ... ... ... ... ... ...

... ... ... ... 1.0–3.0Nb

... ... ... ... 1.0–3.0 (Nb + Ta) ...

0.05

1.0

1.0

1.5–3.5

...

19.5–23.5

...

...

2.5–3.5

Nickel-chromium-iron CY-40 0.40

3.0

1.5

...

11.0

14.0–17.0

0.03

0.03

...

...

Nickel-chromium-molybdenum CW-12MW 0.12 CW-7M 0.07 CW-2M 0.02 CW-6MC 0.06 CW-6M 0.07 CX-2MW 0.02

1.0 1.0 0.8 1.0 1 1.0

1.0 1.0 1.0 1.0 1 0.8

... ... ... ... ... ...

4.5–7.5 3.0 2.0 5.0 3 2–6

15.5–17.5 17.0–20.0 15.0–17.5 20–23.0 17.0–20.0 20.0–22.5

0.04 0.04 0.03 0.015 ... ...

0.03 0.03 0.03 0.015 ... ...

16.0–18.0 17.0–20.0 15.0–17.5 8.0–10.0 18.5 12.5–14.5

0.20–0.40V, 3.75–5.25W ... 0.20–0.60V 3.15–4.50Nb ... 2.5–3.5W, 0.35V

Nickel-molybdenum N-12MV N-7M

1.0 1.0

1.0 1.0

... ...

4.0–6.0 3.0 max

1.0 1.0

0.04 0.04

0.03 0.03

26.0–30.0 30.0–33.0

0.20–0.60V ...

0.12 0.07

Note: Aim compositions are given in a range; single values are maximums.

*Updated and revised from John M. Svoboda, “Nickel and Nickel Alloys,” Casting, ASM Handbook, Vol 15, ASM International, 1988

0.6–1.2Nb

1120 / Nonferrous Alloy Castings specified limits (Table 1) provide for excellent castability and the production of sound, pressure-tight castings. The usual practice is to produce the alloy with 0.75% C and 1.0% Si. When treated with magnesium, similar to ductile iron production, the carbon is present in the matrix as a finely distributed spheroidal graphite. A reduction in mechanical properties results if flake graphite is present. A maximum carbon content of 0.10% or less is occasionally specified when castings are welded into wrought nickel systems. Low-carbon CZ-100, however, is a difficult material to cast, with no significant advantage over the higher-carbon option under any known service conditions. Cast nickel-copper alloys are designated in ASTM A 494 as M-35-1, M-35-2, M-30H, M-25S, and M-30C and in U.S. Navy specification QQ-N-288 as A, B, C, D, E, and F (Table 1). The lower-silicon grades (2.0% Si) are commonly used in corrosion applications in conjunction with wrought nickel-copper and coppernickel alloys as pump, valve, and fitting castings. The higher-silicon grades (>2.0% Si) combine corrosion resistance with high strength and wear resistance. Grades containing the highest silicon contents (>4% Si) are used when exceptional resistance to galling is desired. The high-carbon composition (QQ-N-288-F) is used where improved machinability is more important than ductility and/or weldability. Nickel-Chromium-Iron Alloys. The cast nickel-chromium-iron CY-40 is included in ASTM A 494. Its composition is given in Table 1. The cast alloy differs from the parallel

wrought grade in carbon, manganese, and silicon contents for improved castability and pressure tightness. Heat-resistant nickel-chromium-iron grades HW and HX (Table 2) are covered in ASTM A 297. HW alloy (60Ni-12Cr-23Fe) is especially well suited for applications requiring thermal fatigue resistance. In addition, HW exhibits excellent resistance to carburization and high-temperature oxidation. This alloy performs satisfactorily at temperatures of up to 1120  C (2050  F) in strongly oxidizing atmospheres and up to 1040  C (1900  F) in reducing atmospheres, provided that sulfur is not present in the combustion gas. HW alloy is widely used for intricate heat treating fixtures that are quenched with the load and for many other applications (such as furnace retorts and muffles) that involve thermal shock, steep temperature gradients, and high stresses. Its structure is austenitic and contains carbides in amounts that vary with carbon content and thermal history. In the as-cast condition, the microstructure consists of a continuous interdendritic network of elongated eutectic carbides. Upon prolonged exposure at service temperatures, the austenitic matrix becomes uniformly peppered with small carbide particles, except in the immediate vicinity of eutectic carbides. This change in structure is accompanied by an increase in room-temperature strength but no change in ductility. HX alloy (66Ni-17Cr-12Fe) is similar to HW but contains more nickel and chromium. Its higher chromium content gives it substantially

better resistance to corrosion by hot gases (even sulfur-bearing gases), which permits it to be used in severe-service applications at temperatures up to 1150  C (2100  F). Cast nickel-chromium-molybdenum alloys, designated CW-12MW, CW-7M, CW-2M, and CW-6MC in ASTM A 494 (Table 1), are used in severe-service conditions that usually involve combinations of acids at elevated temperatures. The molybdenum in these compositions improves both resistance to nonoxidizing acids and high-temperature strength. Cast nickel-molybdenum alloys, designated N-12MV and N-7M in ASTM A 494 (Table 1), are most frequently applied in handling hydrochloric acid at all concentrations and temperatures, including the boiling point. Nickel-molybdenum alloys are also produced under proprietary names. Nickel-Base Proprietary Alloys. In addition to the more common nickel-base alloys, there are a number of trademarked, proprietary alloys and other nickel-base alloys that are widely used in corrosive service. Many of these alloys have excellent general corrosion resistance and are most commonly used in applications for which stainless steels are inadequate. Others are used in specialized applications and should not be considered substitutes for stainless steel. Producers should be consulted when applying these alloys, particularly in applications for which there is no history of use. Heat-Resistant Alloys. Castings are classified as heat resistant if they are capable of sustained operation while exposed, either

Table 2 Compositions of heat-resistant nickel-base casting alloys Nominal composition, % Alloy designation

B-1900 CMSX-2 (SC) CMSX-3 (SC) CM-247-LC (SC) HW(b) HX(b) Alloy X Alloy 100 Alloy 738X Alloy 792 Alloy 713C Alloy 713LC Alloy 718 Alloy X-750 M-252 MAR-M 200 (DS) MAR-M 246 MAR-M 247 (DS) NX 188 (DS) Rene´ 77 Rene´ 80 Rene´ 100 TRW-NASA VIA Udimet 500 Udimet 700 Udimet 710 Waspaloy WAZ-20 (DS) CX-2MW CU5MCuC

C

Ni

Cr

0.1 <30 ppm <30 ppm 0.07 0.55 0.55 0.1 0.18 0.17 0.2 0.12 0.05 0.04 0.04 0.15 0.15 0.15 0.15 0.04 0.07 0.17 0.18 0.13 0.1 0.1 0.13 0.07 0.20 0.02 max 0.05

64 66 66 62 60 66 50 60.5 61.5 60 74 75 53 73 56 59 60 bal 74 58 60 61 61 53 53.5 55 57.5 72 bal 38–44

8 8 8 8.1 12 17 21 10 16 13 12.5 12 19 15 20 9 9 9.4 ... 15 14 9.5 6 18 15 18 19.5 ... 20.0–22.5 19.5–23.5

Co

10 4.6 4.6 9.2 ... ... 1 15 8.5 9 ... ... ... ... 10 10 10 10 ... 15 9.5 15 7.5 17 18.5 15 13.5 ... ... ...

Mo

Fe

6 0.6 0.6 0.5 0.5 0.5 9 3 1.75 2.0 4.2 4.5 3 ... 10 ... 2.5 0.7 18 4.2 4 3 2 4 5.25 3 4.2 ... 12.5–14.5 2.5–3.5

... ... ... ... 23 12 18 ... ... ... ... ... 18 7 ... 1 ... ... ... ... ... ... ... 2 ... ... 1 ... 2–6 ...

Al

6 5.6 5.6 5.6 ... ... ... 5.5 3.4 3.2 6 6 0.5 0.7 1 5 5.5 5.5 8 4.3 3 5.5 5.5 3 4.25 2.5 1.2 6.5 ... ...

B

Ti

W

Zr

Others

0.015 ... ... 0.015 ... ... ... 0.01 0.01 0.02 0.012 0.01 ... ... 0.005 0.015 0.015 0.015 ... 0.015 0.015 0.015 0.02 ... 0.03 ... 0.005 ... ... ...

1 1.0 1.0 0.7 ... ... ... 5 3.4 4.2 0.8 0.6 0.9 2.5 2.6 2 1.5 1 ... 3.3 5 4.2 1 3 3.5 5 3 ... ... ...

... 8 8 9.5 ... ... 1 ... 2.6 4 ... ... ... ... ... 12.5 10 10 ... ... 4 ... 6 ... ... 1.5 ... 20 2.5–3.5 ...

0.10 ... ... 0.015 ... ... ... 0.06 0.1 0.1 0.1 0.1 ... ... ... 0.05 0.05 0.05 ... 0.04 0.03 0.06 0.13 ... ... 0.08 0.09 1.5 ... ...

4Ta(a) 6.0Ta 6.0Ta, 0.10Hf 3.2Ta 2.0Mn, 2.5Si 2.0Mn, 2.5Si ... 1V 1.75Ta, 0.9Nb 4Ta 2Nb 2Nb 0.1Cu, 5Nb 0.25Cu, 0.9Nb ... 1Nb(c) 1.5Ta 1.5Hf, 3Ta ... ... ... 1V 0.4Hf, 0.5Nb, 0.5Re, 9Ta ... ... ... ... ... 0.80Mn max, 0.35V max 1.00Mn max 0.6–1.2Nb

DS, directionally solidified; SC, single crystal. (a) B-1900 + Hf also contains 1.5% Hf. (b) Alloy Casting Institute designation. (c) MAR-M 200 + Hf also contains 1.5% Hf.

Nickel and Nickel Alloy Castings / 1121 continuously or intermittently, to operating temperatures that result in metal temperatures in excess of 650  C (1200  F). Many alloys of the same general types are also used for their resistance to corrosive media at temperatures below 650  C (1200  F), and castings intended for such service are classified as corrosion resistant. Although there is usually a distinction between heat-resistant alloys and corrosionresistant alloys, based on carbon content, the line of demarcation is vague—particularly for alloys used in the range from 480 to 650  C (900 to 1200  F). Table 2 lists a number of nickel-base casting alloys used for high-temperature applications. In addition to the HW and HX grades mentioned previously in the discussion on nickelchromium-iron alloys, a number of proprietary alloys are listed. These materials, often referred to as superalloys, contain appreciable amounts of chromium and cobalt, with aluminum and titanium added for strengthening. The effects of aluminum and titanium on the structure and the resulting properties of nickel-base alloys are discussed in the section “Structure and Property Correlations” in this article. Directionally solidified and single-crystal alloys have the highest creep resistance of any of the nickel-base alloys. Directional solidification is accomplished by removing all of the heat from one end and supplying heat at the other end of the casting, to establish a strong thermal gradient at the solidification front throughout the solidification process. In this way, large, elongated columnar grains are produced, eliminating transverse grain boundaries. This structure

Fig. 1

prolongs creep life in service against shear forces normal to the columnar grains. Alloys developed for single-crystal casting do not need the grain-boundary strengthening elements such as carbon, boron, zirconium, and hafnium. The absence of these alloying elements results in materials with high incipient melting temperatures. Figure 1 compares the macro grain structures of equiaxed (conventional), directionally solidified, and single-crystal nickel-base alloy turbine blades. Table 2 lists several compositions of DS/SC alloys.

Structure and Property Correlations Cast Nickel. The mechanical property requirements for CZ-100 are listed in Table 3. Cast nickel has excellent toughness, thermal resistance, and heat-transfer characteristics. Nickel-Copper Alloys. Tensile properties of the nickel-copper castings are controlled by the solution-hardening effect of silicon or silicon plus niobium. The tensile properties of M-351, M-35-2, and composition A are controlled by a carbon-plus-silicon relationship; composition E and M-30C tensile properties are determined by a silicon-plus-niobium relationship. At approximately 3.5% Si, an age-hardening effect appears; at approximately 3.8% Si, the solubility of silicon in the nickel-copper matrix is exceeded, and hard, brittle silicides begin to appear. The combination of an aging effect plus silicides in composition D results in an alloy with exceptional resistance to galling. As the silicon content is increased above 3.8%, the amounts

of hard, brittle silicides in the tough nickel-copper matrix increase; ductility decreases sharply; and tensile and yield strengths increase. As a result, hardness is the only mechanical property specified for composition D. The toughness of nickel-copper alloys decreases with increasing silicon content, but all grades retain their roomtemperature toughness down to at least 195  C (320  F). Tensile requirements for nickel-copper alloys are listed in Table 3. Nickel-Chromium-Iron Alloys. Minimum mechanical properties for both corrosion-resistant and heat-resistant alloys are given in Table 3. Alloy CY-40 is frequently used for elevated-temperature fittings in conjunction with the wrought alloy of similar base composition. Typical elevated-temperature properties are listed in Table 4. Applications for HW and HX alloys were discussed previously. Nickel-Chromium-Molybdenum Alloys. The CW-12MW and CW-7M grades have a relatively high yield strength (Table 3) due to the solution-hardening effects of chromium, molybdenum, and silicon in CW-7M and of tungsten and vanadium in CW-12MW. Ductility is excellent up to the limit of solid solubility. Inadequate heat treatment or improper composition balance, however, may result in the formation of a hard, brittle phase and in a significant loss of ductility. Careful control within the specified composition range is therefore necessary to meet the specified ductility. Carbon and sulfur contents should be kept as low as practicable. Nickel-Molybdenum Alloys. The N-12MV and N-7M grades have good yield strengths

Comparison of equiaxed (left), directionally solidified (center), and single-crystal (right) nickel-base alloy turbine blades for an aircraft engine. Courtesy of Howmet Corporation, Whitehall Casting Division

1122 / Nonferrous Alloy Castings Table 3

Tensile requirements for cast nickel-base alloys Tensile strength

Yield strength

Alloy

MPa

ksi

MPa

ksi

Elongation in 50 mm (2 in.), %

Hardness, HB

Corrosion-resistant CZ-100 M-35-1 M-35-2 M-30H M-25S M-30C N-12MV N-7M CY-40 CW-12MW CW-7M

345 450 450 690 ... 450 525 525 485 495 495

50 65 65 100 ... 65 76 76 70 72 72

125 170 205 415 ... 225 275 275 195 275 275

18 25 30 60 ... 32.5 40 40 28 40 40

10 25 25 10 ... 25 6 20 30 4 25

... ... ... 243–294 300 (min) 125–150 ... ... ... ... ...

Heat-resistant HW HX

415 415

60 60

415 415

60 60

... ...

... ...

Table 4 Elevated-temperature mechanical properties of alloy CY-40 Temperature

Tensile strength

Yield strength

F

MPa

ksi

MPa

ksi

Elongation in 25 mm (1 in.), %

MPa

ksi

Room 480 650 730 815 925

900 1200 1350 1500 1700

486 427 372 314 187 ...

70.5 62 54.5 45.5 27.1 ...

293 ... ... ... ... ...

42 ... ... ... ... ...

16 20 21 25 34 ...

... ... 165 103 62 38

... ... 24 15 9 5.5

Fig. 2

Stress-rupture for 1000-h life of selected cast nickel-base heat-resistant alloys



C



Stress to rupture in 100 h

because of the solution-hardening effect of molybdenum (Table 3). Ductility is controlled by the carbon and molybdenum contents. For optimum ductility, carbon content should be as low as practicable, and molybdenum content should be adjusted to avoid the formation of intermetallic phases. Heat-Resistant Alloys. Nickel-base heatresistant casting alloys, often referred to as superalloys, generally contain substantial levels of aluminum and titanium (Table 2). These elements strengthen the austenitic matrix through precipitation of Ni3(Al,Ti), an ordered face-centered cubic (fcc) compound referred to as gamma prime (g0 ). Various ratios of aluminum and titanium are used in the different nickel-base heatresistant alloys; generally, titanium atoms can replace aluminum atoms up to a ratio of 3 Ti to 1 Al without altering the ordered fcc crystallographic structure of g0 . When excess titanium is present, Ni3Ti, an ordered close-packed hexagonal compound known as eta phase (Z), precipitates. Because g0 is coherent with the matrix, precipitation of this phase has a greater strengthening effect than precipitation of Z. In addition to the strengthening imparted by g0 precipitation, solid-solution strengthening is conferred by the addition of refractory elements, and grain-boundary strengthening by additions of boron, zirconium, carbon, and hafnium. Hafnium also enhances grain-boundary ductility. Stress-rupture curves for various nickel-base alloys are shown in Fig. 2. The strength of these alloys is complemented by superior corrosion resistance, which is conferred by chromium and aluminum (titanium may be more favorable than aluminum under hot-corrosion conditions). Coatings are used on most nickel alloys for temperatures exceeding approximately 815  C (1500  F) to provide adequate protection from oxidation and corrosion at these temperatures. Nickel-base heat-resistant alloy castings are produced by investment casting under vacuum, and improvements in properties have been made not only through control of composition but also through more precise control of microstructure. A significant advance in microstructure control was the development of a columnar grain structure produced by directional solidification and single-crystal technology (see subsequent discussion). Extensive use of nickel alloy castings essentially began with alloy 713, and alloys are available that can be used at temperatures up to approximately 1040  C (1900  F). In addition to creep strength and corrosion resistance, two other properties—stability and resistance to thermal fatigue—are important considerations in the selection of nickel-base heat-resistant casting alloys. Thermal fatigue resistance is partially controlled by composition, but it is also significantly affected by grainboundary area and alignment relative to applied stresses. The crystallographic orientation of grains also influences thermal stresses because the modulus of elasticity, which directly influences thermal stresses, varies with grain

Nickel and Nickel Alloy Castings / 1123

Fig. 3

Fig. 4

Various radial and axial turbine wheels made from MAR-m-247 alloy. Countesy of Howmet Corporation, Whitehall Casting Division

Directionally solidified turbine vane made from CM-247-LC alloy. Courtesy of Thyssen Guss AG

orientation. The stability of property values is directly influenced by metallurgical stability; any microstructural changes that take place during long-term exposure at high temperatures under stress cause attendant changes in properties. For example, if the g0 phase coarsens, strength decreases. Also, potentially deleterious

topologically close-packed (tcp) secondary phases, such as s, Laves, and m, may form. Coarsening of g0 can be controlled to some degree by adjusting alloy additions. Formation of tcp phases is controlled by adjusting the composition of the matrix to minimize the electron vacancy number, Nv. A high Nv indicates a tendency toward the formation of tcp phases. In general, an Nv value below 2.4 indicates minimal formation of deleterious phases; however, this relationship varies with base-alloy composition. The metallurgical structures of both cast and wrought heat-resistant alloys are discussed in greater detail in Metallography and Microstructures, Volume 9 of ASM Handbook, formerly Metals Handbook, 9th ed. Alloys 713C and 713LC are closely related investment casting alloys used principally for low-pressure turbine airfoils in gas turbines. Intended for operation at intermediate temperatures from 790 to 870  C (1450 to 1600  F), these alloys are generally used in uncooled airfoil designs. Alloy 738X is an investment casting alloy similar in strength to alloy 713C and Udimet 700 but with outstanding sulfidation resistance. It is used principally for latter-stage turbine airfoils and for hot-corrosion-prone applications such as industrial and marine engines.

Udimet 700, although primarily a wrought alloy, is also used in investment cast high-pressure turbine blades. In cast form, it is similar in strength to alloy 713C but offers better hotcorrosion resistance. It is designed for operation at intermediate temperatures from 730 to 900  C (1350 to 1650  F). A stability-controlled version of U-700 is known as Rene´ 77. Alloy 100 is designed for use at metal temperatures up to approximately 980  C (1800  F) in cooled and uncooled airfoils. A stability-controlled version of alloy 100 is known as Rene´ 100. B-1900, to which 1% Hf is usually added to improve ductility and thermal-fatigue resistance, is designed for use at metal temperatures up to approximately 980  C (1800  F) in cooled and uncooled airfoils. Rene´ 80 offers excellent corrosion resistance in sulfur-bearing environments. It is designed for use at metal temperatures up to approximately 950  C (1750  F). Alloy 792 is designed for use in applications similar to those of Rene´ 80. It is one of the most sulfidation-resistant nickel alloys available. MAR-M 246 and MAR-M 247 are designed for use at metal temperatures of approximately 980 to 1010  C (1800 to 1850  F) in cooled and uncooled airfoils and radial and axial wheels (Fig. 3). DS MAR-M 200 + Hf is produced by directional solidification (see subsequent discussion) and is designed for metal temperatures of approximately 1010 to 1040  C (1850 to 1900  F). It is used in cooled airfoils. Other alloys (such as Udimet 500) are occasionally used in turbine airfoil applications, and alloy 718 has been cast into large static structures for gas turbines. Alloys for directional and single-crystal solidification possess high creep resistance. Directionally solidified turbine blades have columnar grain boundaries aligned parallel to the solidification direction. This minimizes the grain-boundary sliding in this direction, the main cause for creep strain at high temperatures. Single-crystal alloys have no grain boundaries and therefore have no need for grain-boundary strengthening elements. For this reason, they can be solution heat treated at higher temperatures than conventional alloys, precipitating a greater amount of the g0 strengthening phase. The lack of grain boundaries also enhances the corrosion resistance of these materials. Table 2 lists several DS/SC alloy compositions. A turbine vane made from CM-247-LC DS alloy is shown in Fig. 4. Properties and performance of DS/SC alloys are detailed in Ref 1 to 4.

Melting Practice and Metal Treatment Coreless electric induction furnaces have become the mainstay of the foundry industry for small heat sizes, especially when a number of different alloys are produced. They are also

1124 / Nonferrous Alloy Castings the least expensive of the major furnace types to install. The foundry industry uses these furnaces in sizes ranging from 9 kg to 18 Mg (20 lb to 20 tons); however, most electric induction furnaces are in the 25 to 1350 kg (50 to 3000 lb) range. More detailed information on induction furnaces can be found in the article “Induction Furnaces” in this Volume. Vacuum Melting. Nickel-base alloys containing more than approximately 0.2% of the reactive elements aluminum, titanium, and zirconium are not suitable for melting and casting in oxidizing environments such as air. At the higher alloying levels, these elements readily oxidize, resulting in gross inclusions, oxide laps, and poor composition control. Consequently, such alloys generally require inert gas injection or vacuum melting and casting methods. Extra low gas contents, which can be obtained by vacuum melting, are also required for certain nickel-base alloys. Vacuum melting processes, which are described in the Section “Melting and Remelting” in this Volume, are always used for directional solidification and single-crystal casting alloys. Argon-oxygen decarburization (AOD) has been used to achieve some of the results that vacuum melting can produce. The AOD unit is similar to a Bessemer converter with tuyeres in the lower sidewalls for the injection of argon and oxygen. These units are charged with molten metal from an arc or induction furnace and approximately 20%, but usually less, solid charge. The continuous injection of gases causes a violent stirring action and intimate mixing of slag and metal, which can lower sulfur values to below 0.005%. The gas contents (hydrogen, nitrogen, and oxygen) may be even lower than those of vacuum induction melted alloys. Ladle furnaces are used for melt refining and trim additions to finalize the melt composition. Steadily increasing requirements for alloy cleanliness have led producers to adopt several novel refining technologies and process routes, many involving increased use of the ladle as a refining vessel. Such procedures require longer holding times in the ladle, which necessitate either increased superheats in the furnace or external heating in the ladle to avoid early solidification. Higher superheat, in addition to requiring excessive energy expenditure, can contribute to the problem of melt contamination. The preferred solution is to supply heat to the ladle, maintaining the alloy at minimal superheat during refining. This can be accomplished by induction heating, or the transferred arc plasma torch, with the added benefit of enhanced refining reactions that aid in the production of clean metal with low levels of residual elements. See the article “Degassing” in this Volume.

Properties” and “Selection and Evaluation of Steel Castings” in this Volume). Specific aspects of foundry practice discussed here include pouring, gating and risering, cleaning, welding, and heat treatment of conventional corrosion-resistant nickel-base alloy castings.

Foundry Practice

Gating Systems

Foundry practice for nickel-base alloys is, for the most part, similar to that used for cast stainless steels (see the articles “Steel Castings

As a rule of thumb in methoding castings, the pouring time in seconds is designed to be the square root of poured weight in pounds. Thus,

Pouring Practice Three types of ladles are used for pouring nickel-base castings: bottom-pour, teapot, and lip-pour. Ladle capacity normally ranges from 45 kg to 36 Mg (100 lb to 40 tons). The bottom-pour ladle has an opening in the bottom that is fitted with a refractory nozzle. The nozzle is plugged by a stopper head connected to a stopper rod when it is desired to contain the molten metal in the ladle. When flow is desired, the stopper rod is manually pulled up via a slide-and-rack mechanism. It is recommended to have a completely on or off flow, since “throttling” would split the stream and cause turbulence. Typically, these ladles fill several castings before they start to leak. Special nozzles have been developed to prevent this leakage if it is desired to interrupt the flow to fill numerous castings. Bottom pouring is preferred for pouring large castings from large ladles, because it is difficult to tip a large ladle and still control the stream of molten steel. Inclusions, pieces of ladle lining, and slag float to the top of the ladle; thus, bottom pouring greatly reduces the risk of passing nonmetallic particles into the mold cavity. However, because of its high head-height, the bottom-pouring stream has greater turbulence and tendency for air entrainment into the mold cavity. The teapot ladle incorporates a ceramic wall, or baffle, that separates the bowl of the ladle from the spout. The baffle extends almost four-fifths of the distance to the bottom of the ladle (Fig. 5). As the ladle is tipped, hot metal flows from the bottom of the ladle up the spout and over the lip. Because the metal is taken from near the bottom of the ladle, it is free of slag and pieces of eroded refractory. The teapot design is feasible in various sizes, generally covering the entire range of casting sizes that are below the minimum size for which the bottom-pour ladle is used. Lip-pour ladles resemble the teapot type but have no baffles to hold back the slag. Because the hot metal is not taken from the bottom of the ladle, this type of ladle pours a more contaminated metal and is seldom used to pour high-alloy castings. Nevertheless, it is widely used as a tapping ladle (at the melting furnace) and as a transfer ladle to feed smaller ladles of the teapot type.

the average pouring rate in lb/s also is numerically equal to the pouring time. The pouring rate is determined by the sprue head height, the smallest cross-sectional area in the gating system, which is also termed the choke, and frictional factors that depend on the complexity of the gating system and the casting. The choke is typically placed at the sprue base. The sprue itself is tapered or widened in cross section above the choke to slow down the metal and prevent air aspiration into the gating system at the sprue-to-pouring-cup connection. It is desirable to fill the castings quickly to minimize the required superheat but also with adequate yield of casting-to-poured-weight ratio. Thus, the rule of thumb presented previously serves as a good compromise for methoding engineers to design a gating system for a given casting. In bottom-pouring, the ladle nozzle controls the flow rate, and the gating must be sized to accommodate the maximum flow rate from the ladle. The bottom-pouring system can also benefit from a pouring basin as an intermediary vessel that would allow for a choked design in the gating and eliminate entrained air from the turbulent pouring stream (Ref 5). In lip-pouring, the pourer has control over the pouring rate and, for optimum filling, ensures that the pouring cup is full throughout the pouring process. Tile gating is extensively used in bottompouring operations. They are installed like a core during molding and simplify the molding operation. The ceramic material is more durable than sand against erosion from the high-velocity metal, and often a “gooseneck” design, with a single horizontal runner and a vertical ingate connected to the bottom of the casting, is preferred for its simplicity. Gating for lip-poured systems for smaller castings are rammed in sand with a semicircular or rectangular cross section for the gates and runners. To ensure the metal rises up in the gates, it is common practice to place the runners in the drag and the ingates in the cope. Uniform filling of the gates is also important. In the gating system shown in Fig. 6, the runner area decreases progressively by an amount equal to the area of each gate it passes. This practice ensures that all gates fill at the same time. A large difference in the flow between gates creates swirls of

Fig. 5

Typical teapot ladle used to pour small- to medium-sized castings

Nickel and Nickel Alloy Castings / 1125 metal in the mold about vertical axes, in addition to those occurring about horizontal axes, increasing turbulence. A good gating design minimizes turbulence and mold erosion. When metal impinges on a mold surface at 90 bends, such as that which occurs at sprue-to-runner and runner-to-ingate locations in Fig. 6, extra dead-metal is allowed in the direction of the incoming flow to cushion the flow impact. Sprue well and runner extensions serve this purpose. Typically, the runner extensions are longer than necessary to also serve as a trap for the first, and usually the most contaminated, metal to enter the system. Similarly, the runners and ingates are curved wherever a change in direction occurs. This streamlining reduces turbulence in the metal and minimizes mold erosion. In pouring nickel-base alloys, as well as other high-alloy steels, a high gating ratio (the ratio of the cross-sectional area of the choke to that of all of the runners emanating from the sprue basin and to all of the gates, as illustrated in Fig. 6) is desirable at a slight cost in yield to slow down the metal past the choke, in the runner and ingates, to minimize turbulence in the mold cavity. It is typical to design with gating ratios of 1:2:2 to 1:4:4 or higher. More detailed information on gating practice can be found in the article “Gating Design” in this Volume.

Risers Nickel-base alloys volumetrically contract 3% on solidification. This contraction creates shrinkage cavities and microporosity in locations that solidify last. Risers (feeders) with greater thermal modulus than that of the cast part are used to supply molten metal to the solidifying casting to compensate for its shrinkage and draw the shrinkage porosity to the riser body. Risers also compensate for contraction from superheat, where nickel-base alloys undergo a volumetric contraction of 1.5% per 100  C (0.8% per 100  F) as they cool from the pouring temperature to the solidification temperature.

It is advantageous to use the metal head in the riser to pressurize the casting for better feeding. Therefore, risers are located at the top of a casting, and usually their cavity is open to the top of the mold, except in difficultto-reach areas, where blind risers may be used. During molding, exothermic or insulating riser sleeves are placed on the pattern and molded into the cope half of the mold. The risers are sized to have 20% greater thermal modulus (ratio of section volume to its cooling surface area) than that of the section it is feeding. A neck-down design is employed in riser-to-casting connections to avoid porosity at this location, where otherwise the riser itself would create a T-section. This design is advantageous also for easier removal of risers from casting. In feeding a casting, the thicker sections act as reservoirs for feeding the thinner sections, which solidify first. Thus, risers are placed over thick sections that cannot be fed by other areas of the casting. Demonstrating this principle, the gear blank casting shown in Fig. 7 is provided with a large riser over the central hub and six smaller risers spaced equally around the rim of the gear to ensure adequate feeding. Metal enters the mold at the two gates, which are 180 apart. The feeding distances determine the frequency of riser placement over the length of a section. For plates of 25 to 50 mm (1 to 2 in.) thick section, up to 5 thicknesses in length can be fed from a riser with an end-effect (section ending at sand mold) and 4 thicknesses between two risers. The number of thicknesses that can be fed for an end–effect decreases with decreasing plate width, for example, for 25 to 50 mm (1 to 2 in.) thick bars, the end-effect feeding distance decreases to 3.5 thicknesses. The feeding distance also decreases with increasing section size. For 100 to 150 mm (4 to 6 in.) thick sections, the end feeding distance is 4 thicknesses, and the feeding distance between risers is 3 thicknesses. An end-chill from a single riser or a midchill between two risers adds another thickness to the feeding distance. The feeding distance of a given section can be extended by adding a taper, thereby establishing a favorable temperature gradient for unidirectional solidification. A tapered pad of exothermic material placed in the mold along the length of the casting may also be used to

extend the feeding distance. More information is found in the article “Riser Design” in this Volume. Patternmaker’s Shrinkage. The solidified nickel-base alloys undergo 2.2% linear contraction on cooling from solidification to room temperature. This contraction must be built into any patternwork to obtain accurate final dimensions of the part.

Welding Cast Nickel. Alloy CZ-100 can be readily repair welded or joined to other castings or to wrought forms by using any of the usual welding processes with suitable nickel rod and wire. Joints or cavities must be carefully prepared for welding, because small amounts of sulfur or lead cause weld embrittlement. Nickel-Copper Alloys. The weldability of the nickel-copper alloys decreases with increasing silicon content but is adequate up to at least 1.5% Si. Niobium can enhance weldability, particularly when small amounts of low-melting residuals are present. Careful raw material selection and proper foundry practice, however, have largely eliminated any difference in weldability between niobium-containing and niobium-free grades. The higher-silicon compositions (3.5% Si) are not considered weldable. They can be brazed or soldered, as can the lower-silicon grades. Nickel-Chromium-Iron Alloys. The CY-40 castings can be repair welded or fabrication welded to matching wrought alloys by any of the usual welding processes using rod and wire of matching nickel-chromium contents. Postweld heat treatment is not required after repair welding or fabrication, because the heat-affected zone is not sensitized by the weld heat. Nickel-Chromium-Molybdenum Alloys. Alloys CW-12MW and CW-7M can be welded by any of the usual welding processes, using wire or rod of matching composition. For optimum weldability, carbon content should be as low as practicable. The usual practice is to solution treat and quench after repair welding to remove sensitization and intermetallic precipitates. Nickel-Molybdenum Alloys. Alloys N-12MV and N-7M can be welded by using any of the usual welding processes with wire or rod of matching composition. Postweld heat treatment is usually performed because these alloys are subject to the precipitation of intermetallic compounds in the heat-affected zone.

Heat Treatment

Fig. 6

Gating system for good metal flow

Fig. 7

Gating and feeding system used to cast gear blanks

Cast nickel (alloy CZ-100) is used in the ascast condition. Some other alloys are also used as-cast, but most require some type of thermal treatment to develop optimum properties.

1126 / Nonferrous Alloy Castings Nickel-copper alloys are used in the as-cast condition. Homogenization at 815 to 925  C (1500 to 1700  F) may, under some conditions, improve corrosion resistance slightly, but in most corrosive conditions, alloy performance is not affected by the minor segregation present in the as-cast alloy. At approximately 3.5% Si, silicon begins to have an age-hardening effect. The resultant combination of aging and the formation of hard silicides when the silicon content exceeds approximately 3.8% can cause considerable difficulty in machining. Softening is accomplished by a solution heat treatment, which consists of heating to 900  C (1650  F), holding at temperature for 1 h per 25 mm (1 in.) of section thickness, and oil quenching. Maximum softening is obtained by oil quenching from 900  C (1650  F), but such treatment is likely to result in quench cracks in castings with complex shapes or varying section thickness. In the solution heat treatment of complicated or varying-section castings, it is advisable to charge them into a furnace below 315  C (600  F) and heat to 900  C (1650  F) at a rate that limits the maximum temperature difference within the casting to approximately 56  C (100  F). After being soaked, castings should be transferred to a furnace held at 730  C (1350  F), allowed to equalize in temperature, and then oil quenched. Alternatively, the furnace can be rapidly cooled to 730  C (1350  F), the casting temperature can be equalized, and the castings can be quenched in oil. Solution heat treated castings are age hardened by placing them in a furnace held at 315  C (600  F) or below, heating uniformly to 595  C (1100  F), and holding at 595  C (1100  F) for 4 to 6 h. At the end of age hardening, castings may be furnace or air cooled. Nickel-Chromium-Iron Alloys. Alloy CY40 is used in the as-cast condition because it is insensitive to the intergranular attack encountered in as-cast or sensitized stainless steels. A modified cast nickel-chromium-iron alloy for nuclear applications with 0.12% C (max) is usually solution treated and quenched as an additional precaution. Sensitization in the heat-affected zone is not a problem with CY-40. Unless residual stresses pose a problem, postweld heat treatment is therefore not required. Nickel-Chromium-Molybdenum Alloys. The high chromium and molybdenum contents of CW-12MW and CW-7M result in the precipitation of carbides and intermetallic compounds in the as- cast condition, which can be detrimental to corrosion resistance, ductility, and weldability. These alloys should therefore be solution treated at a temperature of 1175 to 1230  C (2150 to 2250  F) and water or spray quenched. Nickel-Molybdenum Alloys. Slow cooling in the mold is detrimental to the corrosion resistance, ductility, and weldability of N-12MV and N-7M. These alloys should therefore be

solution treated at a minimum temperature of 1175  C (2150  F) and water quenched.

temperatures

feedwater in nuclear plants because of a greater margin of safety over stainless steels. Nickel-chromium-molybdenum alloys are probably the most common materials for upgrading a system in which service conditions are too demanding for either standard or special stainless steels because of severe combinations of acids and elevated temperatures. These cast alloys can be used in conjunction with similar wrought materials, or they can serve to upgrade pump and valve components in a wrought stainless steel system. Nickel-molybdenum alloys have specialized application areas, primarily in the handling of hydrochloric acid at all temperatures and concentrations. Applications should not be based on upgrading in areas where stainless steels are inadequate, because the nickel-molybdenum alloys are unsuitable for handling most oxidizing solutions for which stainless steels are used. Alloys for directional and single-crystal solidification are used as blades for aircraft and some land-based turbines (Fig. 1 and 4). Directionally solidified blades have excellent creep resistance against applied shear stresses normal to their surface, whereas single-crystal blades have creep resistance in all directions.

thermal fatigue)

REFERENCES

Specific Applications Corrosion-resistant nickel-base castings are primarily used in fluid-handling systems with matching wrought alloys; they are also commonly used for pump and valve components or for applications with crevices and velocity effects requiring a superior material in a wrought stainless system. Because of their relatively high cost, nickel-base alloys are usually selected only for severe-service conditions in which maintenance of product purity is of great importance and for which less costly stainless steels or other alternative materials are inadequate. Detailed information on the corrosion resistance of nickel-base alloys in aqueous media is available in the article “Corrosion of Nickel and Nickel-Base Alloys” in Corrosion: Materials, Volume 13B of ASM Handbook. In the application of heat-resistant alloys, considerations include:
Resistance to corrosion (oxidation) at elevated
Stability (resistance to warping, cracking, or
Creep strength (resistance to plastic flow)

Numerous applications of cast heat-resistant nickel-base alloys were discussed earlier in this article. Cast Nickel. Nickel castings are most commonly used in the manufacture of caustic soda and in processing with caustic. As the temperature and caustic soda concentration increase, austenitic stainless steels are useful only up to a point. The nickel-copper and nickel-chromium-iron alloys take over as useful alloys under these conditions, while cast nickel is selected for the higher caustic concentrations, including fused anhydrous soda. Minor amounts of such elements as oxygen and sulfur can have profound effects on the corrosion rate of nickel in caustic. Detailed corrosion data should therefore be consulted before making a final alloy selection. Nickel-Copper Alloys. The principal advantages of the Ni-30Cu alloys are high strength and toughness, coupled with excellent resistance to mineral acids, organic acids, salt solution, food acids, strong alkalis, and marine environments. The most common applications for nickel-copper castings are in the manufacture of, and processing with, hydrofluoric acid and the handling of saltwater, neutral and alkaline salt solutions, and reducing acids. Nickel-chromium-iron alloys are commonly used under oxidizing conditions to handle hightemperature corrosives or corrosive vapors where stainless steels may be subject to intergranular attack or stress-corrosion cracking. In recent years, the CY-40-type alloy has found large-scale application in handling hot boiler

1. K. Harris, G.L. Erickson, and R.E. Schwer, “Development of the Single-Crystal Alloys CM SX-2 and CM SX-3 for Advanced Technology Turbine Engines,” Technical Paper 83-GT-244, American Society of Mechanical Engineers 2. K. Harris, G.L. Erickson, and R.E. Schwer, “Directionally Solidified DS CM 247 LC— Optimized Mechanical Properties Resulting from Extensive g0 Solutioning,” paper presented at the Gas Turbine Conference and Exhibit (Houston, TX), March 1985 3. K. Harris, G.L. Erickson, R.E. Schwer, J. Wortmann, and D. Froschhammer, “Development of Low-Density Single- Crystal Superalloy CMSX-6,” technical paper, Cannon- Muskegon Corporation 4. K. Harris, G.L. Erickson, and R.E. Schwer, “CMSX Single Crystal, CM DS and Integral Wheel Alloys Properties and Performance,” paper presented at the Cost 50/501 Conference, High Temperature Alloys for Gas Turbines and Other Applications (Lie`ge,) Oct 1986 5. S. Kuyucak, “Water Modeling Studies of Bottom- Pouring Ladle Operations,” Paper 06-095, AFS. Trans., 2006 SELECTED REFERENCES
W.J. Jackson, Ed., Steel Castings Design

Properties and Applications, Steel Castings Research and Trade Association, 1983
J.D. Redmond, Selecting Second-Generation Duplex Stainless Steels, Chem. Eng., Oct 1986 and Nov 1986

Nickel and Nickel Alloy Castings / 1127
Steel Castings Handbook, Supplement 8,


Steel Castings Handbook, Supplement 9,


P.F. Wieser, Ed., Steel Castings Handbook,

High Alloy Data Sheets, Corrosion Series, Steel Founders’ Society of America, 1981

High Alloy Data Sheets, Heat Series, Steel Founders’ Society of America, 1981

5th ed., Steel Founders’ Society of America, 1980

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1128-1142 DOI: 10.1361/asmhba0005337

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Titanium and Titanium Alloy Castings Mustafa Guclu, Stanley Associates Inc.

SINCE THE INTRODUCTION OF TITANIUM and titanium alloys in the early 1950s, these materials have become one of the backbone materials for the aerospace, energy, and chemical industries (Ref 1). The combination of high strength-to-weight ratio, excellent mechanical properties, and corrosion resistance makes titanium the best material for many critical applications. Titanium alloys are used for static and rotating gas turbine engine components. Some of the most critical and highly stressed civilian and military airframe parts are made of these alloys. Net-Shape Technology Development. The use of titanium expanded from applications in nuclear power plants to food processing plants, from oil refinery heat exchangers to marine components and medical prostheses (Ref 2). However, the high cost of titanium alloy components may limit their use to applications for which lower-cost alloys, such as aluminum and stainless steels, cannot be used. The relatively high cost is often the result of the

intrinsic raw material cost of the metal, fabricating costs, and, usually most important, the metal-removal costs incurred in obtaining the desired end shape. As a result, a substantial effort has been focused on the development of net-shape or near-net-shape technologies to make titanium alloy components more competitive (Ref 3). These titanium net-shape technologies include powder metallurgy (P/M), superplastic forming (SPF), precision forging, and precision casting. Precision casting is by far the most fully developed and the most widely used net-shape technology. Casting Industry Growth. The United States annual shipments of titanium castings (Ref 4) grew rapidly between 1978 and 1997 (Fig. 1). The titanium castings shipments dropped precipitously as golf club production moved to low-cost countries after 1997. Shipments have been increasing since 2003 due to increased production demand for industrial castings. The titanium industry is relatively young compared to other metal casting industries.

1100 2220

1000 900

Shipments, Mg

700 1420 600 500 1020 400 300

620

200 220

Fig. 1

19 94 19 96 19 98 20 00 20 02 20 04

19 92

19 90

19 88

19 86

19 84

19 82

19 80

19 78

100

United States titanium casting shipments from 1978 to 2003. Source: Ref 4

Shipments, Ib ⫻ 103

1820

800

Even at current levels (approaching 500 Mg, or 1
106 lb, annually), castings still represent less than 2% of total titanium mill product shipments. This is in sharp contrast to the ferrous and aluminum industries, where foundry output is 9% (Ref 5) and 14% (Ref 6) of total output, respectively. This suggests that the growth trend of titanium castings will continue as users become more aware of industry capability, suitability of cast components in a wide variety of applications, and the net-shape cost advantages. Foundries and Capacities. Table 1 summarizes the use and capacities of the various titanium casting practices by a number of foundries in several countries. Properties Comparable to Wrought. The term castings often connotes products with properties generally inferior to wrought products. This is not true with titanium cast parts. Mechanical properties of titanium castings are generally comparable to their wrought counterparts. Toughness, crack growth resistance, and creep resistance are generally superior and strength is almost the same, while high-cycle fatigue is normally a little lower. As a result, titanium castings can be reliably substituted for forged and machined parts in many demanding applications (Ref 7, 8). This is due to several unique properties of titanium alloys. One is the a + b-to-b allotropic phase transformation at a range of 705 to 1040  C (1300 to 1900  F), which is well below the solidification temperature of the alloys. As a result, the cast dendritic b structure is wiped out during the solid-state cooling stage, leading to an a + b platelet structure (Fig. 2a), which is also typical of b-processed wrought alloy. Further, the convenient allotropic transformation temperature range of most titanium alloys enables the ascast microstructure to be improved by means of postcast cooling rate changes and subsequent heat treatment. Reactivity. Another unique property is the high reactivity of titanium at elevated temperatures, leading to an ease of diffusion bonding. As a result, hot isostatic pressing (HIP) of titanium castings yields components with no subsurface porosity. At the HIP temperature range (900 to 950  C, or 1650 to 1740  F) titanium dissolves any microconstituents deposited on

Titanium and Titanium Alloy Castings / 1129 Table 1

Status and capacity of titanium foundries in the United States, Russia, and Europe in 2007 Approximate maximum envelope size

Maximum pour weight Titanium foundry

kg

Rammed graphite

lb

mm

Investment casting in.

Alba Industrial (Gorky Zavod, Zelenodolsk, Russia) (Norway) 4500 9911 2450 diam
1950 Alcoa Howmet Corporation (MI and VA) 730 1600 . . . ATI Wah Chang (Allvac-Oremet) (OR) 600 1300 1525 diam
1830 Balashikha Casting (Moscow, Russia) 200 450 . . . Castings Technology International, Rotherham (United Kingdom) 120 264 . . . Doncasters Settas (Belgium) 820 1800 1525 diam
220 Pacific Cast Technologies (OR) 908 2000 . . . PCC-Structurals (OR) 772 1700 . . . PCC France (France) 270 600 990 diam
990 PCC-Schlosser (OR) 200 450 . . . Selmet (OR) 227 500 . . . Ti Squared Technologies (Ramcast) (OR) 27 60 . . . Tital (Germany) 180 400 1145 diam
760

96 diam ... 60 diam ... ... 60 diam ... ... 39 diam ... ... ... 45 diam

mm


77 1000 diam
1950 1575 diam
1829
72 . . . 600 diam
800 500 cube
48 610 diam
610 1372 diam
889 2006 diam
2500
39 1194 diam
1270 1065 diam
915 1143 diam
610 406 cube
30 1015 diam
635

B

40 diam 62 diam ... 24 diam 20 cube 24 diam 54 diam 79 diam 47 diam 42 diam 45 diam 16 cube 40 diam

Melt stock


77
72

Billet Billet Billet
31 Billet Billet
24 Billet
35 Billet
100 Billet
50 Billet
36 Billet
24 Billet Billet
25 Billet

and revert and and and and and and and and and and

revert revert revert revert revert revert revert revert revert revert

Use of postcast HIP

Often Always Seldom Often Always Often Always Always Always Often Often Seldom Always

also by the U.S. Bureau of Mines (Ref 16), led to the production of complex shapes. This process, and its derivations, are used to produce parts for marine and chemical-plant components such as the pump and valve components shown in Fig. 3(a). Some aerospace components such as the aircraft brake torque tubes, landing arrestor hook, and optic housing shown in Fig. 3(b) have also been produced by this method. Since this early work, other molding systems have been used on relatively simpleshaped parts that allow metal volumetric shrinkage to occur without restrictions.

A

E

In.

C

B

D

C

Physical Metallurgy (a)

170 µm

(b)

5 µm

Fig. 2 Comparison of the microstructures of (a) as-cast versus (b) cast + HIP Ti-6Al-4V alloys illustrating lack of porosity in (b). Grain-boundary a (B) and a plate colonies (C) are common to both alloys; b grains (A), gas (D), and shrinkage voids (E) are present only in the as-cast alloy.

internal pore surfaces, leading to complete healing of casting porosity as the pores are collapsed during the pressure and heat cycle. Both the elimination of casting porosity and the promotion of a favorable microstructure improve mechanical properties. However, the high reactivity of titanium, especially in the molten state, presents a special challenge to the foundry. Special, and sometimes relatively expensive, methods of melting (Ref 9), moldmaking, and surface cleaning (Ref 7, 8) may be required to maintain metal integrity.

Historical Perspective of Casting Technology Although titanium is the fourth most abundant structural metal in the earth’s crust (0.4 to 0.6 wt%) (Ref 9), it has developed more recently as a technical metal. This is the result of the high reactivity of titanium, which requires complex methods and high energy input to win the metal from the oxide ores.

The required energy per ton is 1.7 times that of aluminum and 16 times that of steel (Ref 10). From 1930 to 1947, metallic titanium extracted from the ore as a powder or sponge form was processed into useful shapes by P/M methods to circumvent the high reactivity in the molten form (Ref 11). Melting Methods. The melting of small quantities of titanium was first experimented with in 1948 using methods such as resistance heating, induction heating, and tungsten arc melting (Ref 12, 13). However, these methods never developed into industrial processes. The development during the early 1950s of the cold crucible, consumable-electrode vacuum arc melting process, “skull melting,” by the U.S. Bureau of Mines (Ref 13, 14) made it possible to melt large quantities of contamination-free titanium into ingots or net shapes. First Castings. Shape casting of titanium was first demonstrated in the United States in 1954 at the U.S. Bureau of Mines using machined high-density graphite molds (Ref 13, 15). The rammed graphite process developed later,

Pure titanium exists in two allotropic forms as alpha (a) and beta (b) at ambient pressure and temperatures. The alpha (a) phase characterized by a hexagonal close-packed (hcp) crystalline structure is stable from room temperature to approximately 882  C (1620  F). The beta (b) phase in pure titanium has a body-centered cubic (bcc) structure and is stable from approximately 882  C (1620  F) to the melting point of about 1670  C (3038  F). The transformation temperature from a or from a plus b to all b is known as the b transus temperature. This transformation temperature can be raised or lowered depending on the type and amount of impurities or alloying additions. The transformation of the bcc b phase to hcp a phase in titanium alloys can occur either by a diffusion-controlled nucleation and growth process or martensitically depending on cooling rate and alloy composition. Heat treatment temperatures are defined in reference to the b transus temperature. The most important titanium phase diagram is the Ti-Al system (see ASM Handbook, Vol 3, Alloy Phase Diagrams), which is the basis for the structure-property optimization during heat treating of the titanium alloys. Apart from the a and b phases, several intermetallic phases can form depending on temperature and composition. The Ti3Al intermetallic technically relevant since it is brittle. The aluminum content in

1130 / Nonferrous Alloy Castings

(a)

Fig. 3

(b) Typical titanium parts produced by the rammed graphite process. (a) Pump and valve components for marine and chemical-processing applications. (b) Brake torque tubes, landing arrestor hook, and optic housing components used in aerospace applications

titanium alloys is limited to about 6 wt% to minimize the amount of Ti3Al intermetallic precipitates in the a phase. The reaction is promoted by the increased oxygen and/or aluminum. The titanium alloy classes denote the general type of microstructure after processing. There are four basic classes of titanium-base materials, each favoring specific characteristics. The alloy classes are commercially pure titanium, alpha and near-alpha, alpha-beta, and beta and near-beta alloys. Most commercially pure titanium, alpha, and near-alpha alloys will have a minimal amount of b phase due to tramp elements or minor additions of alloying elements. An alpha-beta alloy consists of a plus transformed b phases. Beta and near-beta alloys tend to retain b phase on cooling to room temperature, but precipitate secondary phases during heat treatment depending on chemical composition. With the wide selection of titanium alloys available, optimum choice for a given environment is possible.

Alloy Systems Commercially pure (CP) titanium is less expensive, generally more corrosion resistant, lower in strength than its alloys, and not heat treatable. It is highly weldable and formable and used primarily in applications requiring corrosion resistance and high ductility where strength is not a prime consideration. Grade 1 is the purest, used in many chemical and marine applications; grade 2 is considered an industrial “workhorse;” grade 3 has higher strength than grade 2 and is often used in the same applications; grade 4 is often used in marine applications. There are also grades with a small palladium content that increases resistance to crevice corrosion at high temperatures and low pH. Alpha and Near-Alpha Alloys. The alpha alloys are non-heat-treatable and generally are very weldable. They have low to medium strength, good notch toughness, and reasonably good ductility, and they possess excellent mechanical properties at cryogenic temperatures. The more highly alloyed alpha and near-alpha alloys offer optimum high-

temperature creep strength and oxidation resistance as well. Like all the hcp phase materials, they do not exhibit ductile-brittle transformation. Strengthening effects in alpha alloys are achieved by solid-solution additions of the alloying elements. Alpha-Beta Alloys. Most of the alloys in use fall into the alpha-beta class, including Ti-6Al4V, which comes as close as possible to being a general-purpose titanium alloy. These alloys are heat treatable, and most are weldable. Their strength levels are medium to high. Their hot forming qualities are good, but the high-temperature creep strength is not as good as most alpha alloys. Alpha-beta alloys are capable of excellent combinations of strength, toughness, and corrosion resistance. Beta and Near-Beta Alloys. Most beta titanium alloys contain small amounts of alpha stabilizers that permit second-phase (alpha) strengthening to high levels at room to intermediate temperatures. Beta alloys are prone to ductile-brittle transformation and thus are not used for cryogenic applications. The beta alloys are readily heat treatable to high strengths, generally weldable, and have good creep resistance to intermediate temperatures. Excellent formability can be expected in the solution-treated condition. Beta-type alloys also have good combinations of properties in thin and heavy sections.

Effects of Alloying Elements Alloying elements in titanium are classified into a or b stabilizing additions depending on whether they increase or decrease the a/b transformation temperature of 882  C (1620  F) of pure titanium. The selective addition of alloying elements to titanium enables a wide range of physical and mechanical properties to be obtained. Basic effects of a number of alloying elements are:  Alpha stabilizers. Certain alloying addi-

tions, notably aluminum and interstitials (oxygen, nitrogen, carbon) tend to stabilize the alpha phase, that is, raise the b transus temperature at which the alloy will be

transformed completely to the beta phase. The maximum oxygen content of titanium alloys is generally limited to less than 0.25 wt% because of reduced ductility and increased susceptibility to stress-corrosion cracking.  Beta stabilizers. Most alloying additions such as chromium, niobium, copper, iron, manganese, molybdenum, tantalum, and vanadium stabilize the beta phase by lowering the b transus temperature. It should be mentioned that hydrogen is a beta eutectoid forming element. The maximum hydrogen content in titanium alloys is limited to about 125 to 150 ppm because of hydrogen embrittlement.  Neutral additives. Some elements, notably tin and zirconium, behave as neutral solutes in titanium and have little effect on the transformation temperature, acting as solidsolution strengtheners of the alpha phase. Alloying elements to improve corrosion resistance include nickel, molybdenum, palladium, and ruthenium. Additionally, silicon is added to titanium alloys to improve creep resistance in amounts ranging from 0.08 to 0.45 wt%. More detailed information of the influence of the various alloying elements on the stability of the alpha and beta phases, and mechanical properties are provided in ASM Handbook, Vol 2, Properties and Selection: Nonferrous Alloys and Special-Purpose Materials in the article “Wrought Titanium and Titanium Alloys.”

Microstructures of Titanium Castings The microstructure of titanium alloys are described by a and b phases. The microstructures can be a, b, or a mixture of a + b based on the alloy composition. The microstructure can be a fine or coarse arrangement of a, b, or a mixture of a + b phases depending on cooling rate from the b phase field. The b-to-a transformation can occur by nucleation and growth or martensitically, depending on the alloy composition and/or cooling rate. The martensitic product is usually hcp and is designated a0 ,

Titanium and Titanium Alloy Castings / 1131 or orthorhombic martensite, designated a00 , which forms in alloys that contain higher concentrations of refractory elements such as molybdenum, tantalum, or niobium. More detailed information of the phase transformations in titanium alloys is given in the “Selected References” listed at the end of this article. Typical cast microstructures consist of acicular a, a platelets, and a colonies separated by thin b lathe (intergranular b), which are dependent on alloy composition and cooling rate. In near-alpha and alpha-beta alloys, the a platelet morphology is a transformation product of the b phase when cooled below the b transus. During slow cooling, the a platelets may form colonies of similarly aligned platelets, thus sharing a common crystallographic orientation. Larger colonies are developed by slow cooling rates through the b transus. In addition, during slow cooling through the a plus b phase field, grain-boundary a forms on prior b grain boundaries, which thickens becomes more continuous a at slower cooling rates. The acicular or platelet a produced by transformation from b phase has special aspects as far as casting properties are concerned. These kinds of microstructures have good fracture toughness and creep properties. As would be expected, fine microstructures produced by a beta solution treatment and aging provides higher tensile strength with reduced ductility. The grain size, grain shape, grain-boundary alpha size, and size of alpha colonies and platelets in titanium have a very significant influence on mechanical properties. It is the ability to manipulate the phases/grains present as a result of alloy composition, heat treating temperature, holding times, and cooling rate that is responsible for the variety of properties that can be produced in titanium and its alloy castings. Quantitative metallography is used to determine areal fraction and size of various phases and their effect on mechanical properties. Transformed b-phase products in alloys can affect tensile strength, ductility, toughness, and cyclic properties. Typically, the finer grain sizes and microstructures associated with faster cooling in the thin sections exhibit higher strengths and ductility, while the coarser grain sizes and microstructures associated with the slower cooling in the heavier sections exhibit lower strengths and ductility. However, it is also important to recognize that cooling effects can vary Table 2

Generally, castings of titanium alloys are hot isostatically pressed to close casting porosity. HIP conditions may affect the resultant properties as HIP is another heat treatment as far as microstructure is concerned. It also should be noted that test results are often from small separately cast test coupons and will not necessarily reflect the property level achievable with similar processing on a full-scale cast part. Property levels of actual cast parts, especially larger components, will probably be somewhat lower, the result of coarser grain structure due to slower cooling rates achieved. The property variations based on the test coupon origins are given in the section “Final Evaluation and Certification” in this article. Specifications. Industry-wide specifications, listed in Table 4 for reference, give more detail on mechanical property guarantees and process control features. In addition, most major aerospace companies have comparable specifications. Metallic Materials Properties Development and Standardization (MMPDS-03) (formerly, MIL Handbook V ) presently includes titanium alloy investment castings (Ref 17). The A-basis design tensile properties of HIP and mill-annealed Ti-6Al-4V alloy investment castings are shown in Table 5. These properties are covered by the AMS 4992 specification. As with wrought products, commercially pure titanium casting pumps and valves are the principal components made using titanium casting technology for corrosion-resistant service and are used extensively in chemical and petrochemical plants. Newer Alloys. As aircraft engine manufacturers seek to use cast titanium at higher operating temperatures, Ti-6Al-2Sn-4Zr-2Mo and Ti-6Al-2Sn-4Zr-6Mo are being specified more frequently. Other titanium alloys for service to 595  C (1100  F) are being developed as castings. Extra-low interstitial grade Ti-6Al-4V has been used for critical cryogenic space shuttle service where fracture toughness is an important design criteria. The most recent alloys to receive attention in the casting industry are the metastable b alloys Ti-3Al-8V-6Cr4Zr-4Mo (Beta C) and Ti-15V-3Al-3Cr-3Sn (Ti-15–3). Originally developed as a highly cold-formable and subsequently age-hardened sheet material, these alloys are highly castable and are readily heat treated to an 1170 MPa

substantially from one casting to another in identical section size based on the pour weight. Compared to wrought materials, castings typically have coarser grain microstructures that favor improved crack propagation resistance, better creep, and higher fracture toughness. Alternatively, the finer grain structures associated with wrought materials favor higher tensile and HCF strengths. Various types of microstructure examples for the titanium alloys are provided in ASM Handbook, Vol 9, Metallography and Microstructure in the article “Titanium and Titanium Alloys.”

Cast Alloys All production titanium castings are based on traditional wrought product compositions. The extra-low interstitial (ELI) grades specified for some titanium alloys implicitly recognize the effect of reduced interstitials such as oxygen, carbon, nitrogen, and hydrogen on toughness at both room and subzero temperatures. ELI grades are used for critical applications where enhanced ductility and toughness are produced by keeping interstitials at a very low level. As in the wrought products, the Ti-6Al-4V alloy dominates casting applications and is the benchmark alloy against which others are compared. However, other wrought alloys have been developed, for special applications, with better room-temperature or elevated-temperature strength, creep, or fracture toughness characteristics than those of Ti-6Al4V. Cast titanium alloys have properties generally comparable to those of their wrought counterparts. Chemistry and Properties. The chemical composition of cast titanium alloys are shown in Table 2. The table also indicates the approximate share of each alloy. Typical room-temperature tensile properties of various cast titanium alloys are shown in Table 3. Virtually all existing data have been generated from alloy Ti-6Al-4V; consequently, the basis for most cast alloy property data is Ti-6Al-4V. Because the microstructure of cast titanium alloy parts is comparable to that of wrought material, many properties of cast plus HIP parts are at levels similar to those for wrought alloys. These properties include tensile strength, creep strength, fracture toughness, and fatigue crack propagation.

Chemical composition of cast titanium alloys with estimates of relative casting usage

Alloy

Ti-6Al-4V Ti-6Al-4V ELI Commercially pure titanium grade 2 Ti-6Al-2Sn-4Zr-2Mo Ti-6Al-2Sn-4Zr-6Mo Ti-5Al-2.5Sn Ti-3Al-8V-6Cr-4Zr-4Mo Ti-15V-3Al-3Cr-3Sn Total (a) Superior, relative to Ti-6Al-4V

Estimated relative usage of castings

90% 2% 5% 2% <1% <1% <1% <1% 100%

Nominal composition, wt% O

N

H

Al

Fe

0.18 0.11 0.25 0.10 0.10 0.16 0.10 0.11

0.015 0.010 0.015 0.010 0.010 0.015 0.015 0.015

0.006 0.006 0.006 0.006 0.006 0.006 0.006 0.006

6 6 ... 6 6 5 3.5 3

0.13 0.10 0.15 0.15 0.15 0.2 0.2 0.2

V

Cr

Sn

Mo

Zr

Special properties(a)

4 4 ... ... ... ... 8.5 15

... ... ... ... ... ... 6 3

... ... ... 2 2 2.5 ... 3

... ... ... 2 6 ... 4 ...

... ... ... 4 4 ... 4 ...

General purpose Cryogenic toughness Corrosion resistance Elevated-temperature creep Elevated-temperature strength Cryogenic toughness Strength Strength

1132 / Nonferrous Alloy Castings Table 3 Typical room-temperature tensile properties of titanium alloy castings (bars machined from castings) Specification minimums are less than these typical properties. Yield strength Alloy(a)

Commercially pure (grade 2) Ti-6Al-2Sn-4Zr-2Mo, annealed Ti-6Al-4V, annealed Ti-6Al-4V-ELI(b) Ti-6Al-2Sn-4Zr-6Mo, b-STA(c) Ti-3Al-8V-6Cr-4Zr-4Mo, b-STA(c) Ti-15V-3Al-3Cr-3Sn, b-STA(c)

Alloy class

MPa

ksi

Alpha Near alpha Alpha+beta Alpha+beta Alpha+beta Beta Beta

448 910 855 758 1269 1241 1200

65 132 124 110 184 180 174

Ultimate strength MPa

ksi

552 1006 930 827 1345 1330 1275

80 146 135 120 195 193 185

Elongation,%

18 10 12 13 1 7 6

Reduction of area, %

33 22 21 23 1 12 12

(a) Solution treated and aged (STA) heat treatments may be varied to produce alternate properties. (b) ELI, extra low interstitial. (c) b-STA, solution treatment in b-phase field followed by aging.

Table 4 Standard industry specifications applicable to titanium castings Standards

AMS AMS AMS AMS AMS AMS AMS AMS AMS AMS AMS

Title

2175 2249 2444 2488 2643 2694 2801 2804 4954 4956 4985

AMS 4991 AMS 4992 AMS-H-81200 AMS-STD-2154 AMS-T-81915 ARP 1331 AS 1814 ASTM B 367 ASTM B 600 ASTM E 1320 ASTM E 1409 ASTM ASTM ASTM ASTM ASTM ASTM ASTM ASTM ASTM ASTM

E E E E E E E E E E

1417 1742 1820 23 2371 3 340 399 407 466

ASTM E 606 ASTM E 647 ASTM E 8 NAVSEA T9074-ASGIB-010/271

Castings, Classification and Inspection of Chemical-Check Analysis Limits for Titanium and Titanium Alloys Coatings Titanium Nitride PVD Anodic Treatment Ti and Ti Alloys Standard Examination of Ti Alloys, Chemical Etch Inspection Procedure Repair Welding of Aerospace Castings Heat Treatment of Titanium Alloy Parts Identification, Castings Titanium Alloy Welding Wire 6Al-4V Titanium Alloy Welding Wire 6Al-4V, Extra Low Interstitial Titanium Alloy Investment Castings 6Al-4V 130 UTS, 120 YS, 6% Hot Isostatically Pressed Anneal Optional or When Specified Titanium Alloy Castings, Investment 6Al-4V Hot Isostatically Pressed Anneal Optional Castings, Structural, Titanium Alloy 6Al-4V Hot Isostatically Pressed Heat Treatment of Titanium and Titanium Alloys Inspection, Ultrasonic, Wrought Metals, Process For Titanium and Titanium Alloy Castings, Investment The Chemical Milling Process Terminology for Ti Microstructures Standard Specification for Titanium and Titanium Alloy Castings Descaling and Cleaning Ti and Ti Alloy Surfaces Standard Reference Radiographs for Titanium Castings Standard Test Method for Determination of Oxygen and Nitrogen in Titanium and Titanium Alloys by the Inert Gas Fusion Technique Standard Practice for Liquid Penetrant Testing Standard Practice for Radiographic Examination Standard Test Measurement of Fracture Toughness Standard Test Methods for Notched Bar Impact Testing of Metallic Materials Standard Test Method for Analysis of Titanium Alloys by AEPS Preparation of Metallographic Specimens Standard Practice for Macroetching Metals and Alloys Standard Test Method for Linear-Elastic Plane-Strain Fracture Toughness KIc of Metallic Materials Standard Practice for Microetching Metals and Alloys Standard Practice for Conducting Force Controlled Constant Amplitude Axial Fatigue Tests of Metallic Materials Standard Practice for Strain-Controlled Fatigue Testing Standard Test Method for Measurement of Fatigue Crack Growth Rates Standard Test Methods of Tension Testing of Metallic Materials Requirements for Nondestructive Testing Methods

(170 ksi) strength level, making them serious candidates for the replacement of high-strength precipitation-hardened stainless steels such as 17–4PH. The full density advantage of titanium of about 40% is preferred because strength levels are comparable in both materials. Titanium-aluminide castings are being developed for application in the compressor sections of aircraft gas turbine engines subjected to the highest temperatures. Compositions based on both the a2 (Ti3Al) and g (TiAl) ordered phases have been cast experimentally. The low ductility of these alloys at room temperature has been the major producibility challenge. The service potential for titanium aluminides in the 595 to 925  C (1100 to 1700  F) temperature range makes them attractive. The difficulty in machining shapes in these brittle alloys is the advantage of net-shape methods such as castings. Because Ti-6Al-4V dominates the industry, much more metallurgical work has been accomplished with this alloy, and it is discussed next.

Alloy Ti-6Al-4V Microstructures of Ti-6Al-4V Cast Microstructure. To understand the relatively high mechanical property levels of titanium alloy castings and the many improvements made in recent years, it is necessary to understand the microstructures of castings and their influence on the mechanical behavior of titanium. The phase transformation from b to a + b leads to the elimination of the dendritic cast structure due to the narrow freezing range. The existence of such dendrites is evident in the surface morphology of shrinkage pores (Fig. 4). The phase transformation, which in the alloy Ti-6Al-4V is typically initiated at 999  C (1830  F), results in the microstructural features shown in Fig. 2(a). This microstructure is very similar to a b-processed wrought microstructure

Table 5 A-basis design tensile properties of HIP and mil-annealed Ti-6Al-4V titanium alloy investment casting (integrally cast or specimens machined from a casting) Cross-section thickness, mm (in.) Property

Tensile strength, MPa (ksi) Yield strength at 0.2% offset, MPa (ksi) Elongation in 4D, % (S-basis) Source: Ref 17

Less than 12.70 (0.500)

862 (125) 765 (111) 5

12.70–38.10 (0.500–1.500)

848 (123) 772 (112) 4

38.10–101.60 (1.500–4.000)

827 (120) 758 (110) 3

20 µm

Fig. 4

Dendritic structures present in the shrink cavity of an as-cast titanium component

Titanium and Titanium Alloy Castings / 1133 23, 24). Many postcast thermal treatments attempt to reduce the size of this phase to improve fatigue life. Alpha Plate Colonies. Alpha platelets (C, in Fig. 2a) are the transformation products of the b phase when cooled below the b transus temperature. The hcp orientation of these plates is related to the parent body-centered cubic (bcc) b phase orientation through one of the 12 possible variants of the Burgers relationship (Ref 25, 26):

and has similar properties. Thus, in the study and development of titanium alloy castings, it is possible to draw much information from conventional ingot metallurgy. Hot isostatic pressing is a standard practice for all titanium cast parts produced for the aerospace industry (Table 1). As a result, cast + HIP microstructure also needs to be considered (Fig. 2b). Because the HIP temperature is typically well below the b transus temperature, the as-cast (Fig. 2a) and the cast + HIP microstructures look alike, except for the lack of porosity in the latter. As-Cast and Cast + HIP Microstructures. Because most castings for demanding applications are produced with Ti-6Al-4V alloy (Table 2), only microstructures of this a + b alloy are reviewed here. Beta grains (A, in Fig. 2a) develop during the solid-state cooling stage between the solidus/liquidus temperature and the b transus temperature. As a result, large sections, which cool slower, show larger b grains. The size range of the b grains is 0.5 to 5 mm (0.02 to 0.2 in.). As is further discussed, large b grains may lead to large a plate colonies. This is beneficial for fracture toughness, creep resistance, and fatigue crack propagation resistance (Ref 18, 19) and detrimental for low- and high-cycle fatigue strength (Ref 20, 21). Acicular or Platelet Alpha. Alpha begins to form below the b transus temperature when this alloy is slowly cooled from the beta region. The a forms in platelikes (light areas in Fig. 2a) with a crystallographic relationship to the b. The length of these plates can equal the b grain radius. Grain-Boundary Alpha. This a phase (B, in Fig. 2a) is formed on prior b grain boundaries when cast material is cooled through the a + b phase field (in Ti-6Al-4V this is typically from 999  C, or 1830  F, down to room temperature). This has been found to be detrimental to fatigue crack initiation at room temperature (Ref 22, 23) and at elevated temperatures (Ref

Table 6

f110gb k ð0001Þa

When cooling rates are relatively slow, such as in thick-section castings , many adjacent a platelets transform into the same Burgers variant and form a colony of similarly aligned and crystallographically oriented platelets. The large colonies (like those marked C in Fig. 2a) may be associated with early fatigue crack initiation (Ref 21), the result of heterogeneous basal slip across the plates (Ref 27). At the same time, the large colony structure is beneficial for fatigue crack propagation resistance (Ref 28, 29). Because a platelet colonies cannot grow larger than the b grains, titanium castings with large b grains typically have large colonies. The a platelets are typically 1 to 3 mm (40 to 120 min.) in thickness and 20 to 100 mm (800 to 4000 min.) in length (Ref 30, 31), and the typical colony size in castings is 50 to 500 mm (0.002 to 0.02 in.) (Ref 22, 30, 31). As a general rule, slower cooling rates, such as in thick cast sections, result in microstructures with larger b grains, a longer and thicker grain-boundary a phase, thicker a platelets, and larger a platelet colonies. Porosity. Gas (D, in Fig. 2a) and shrinkage (E) voids are typical phenomena in as-cast titanium products. Hot isostatic pressing, however, closes and heals these pores. This is demonstrated by comparing the as-cast microstructure in Fig. 2(a) with the cast + HIP structure in Fig. 2(b). Hot isostatic pressing also causes a degree of a plate coarsening.

Methods for modifying the microstructure of a + b titanium alloy net-shape products Hydrogenation temperature

Method(a)

(Eq 1)

h111ib k h1120ia

Modification of Microstructure. The cast titanium alloy microstructures and mechanical property combinations are accomplished principally through appropriate heat treatment performed either above or below the beta transus. However, there is no known heat treatment that can produce grain refinement in titanium alloys such as is possible in steel. Virtually all titanium castings produced commercially today are supplied in the annealed condition. The following section reviews several of the previous studies and their results. Modification of microstructure is one of the most versatile tools available in metallurgy for improving mechanical properties of alloys. This is commonly achieved through a combination of cold or hot working followed by heat treatment. Net shapes such as castings or P/M products cannot be worked, which limits the options for controlling microstructures. A substantial amount of work has been done in recent years to improve the microstructures of titanium alloy net-shape products, with an emphasis on Ti6Al-4V material. Most treatment schemes can be successfully applied to both cast parts (Ref 8) and P/M compacts (Ref 32, 33). In the case of titanium alloy castings, the main goal has been to eliminate the grain-boundary a phase, the large a plate colonies, and the individual a plates. This is accomplished either by solution treatments or by temporary alloying with hydrogen. In some cases, the hydrogen and solution treatments are combined. The details of some of the methods tried and their areas of applications are listed in Table 6 (Ref 34–45). The typical resulting microstructures of the a-b solution treatment (ABST), b solution treatment (BST), broken-up structure (BUS), and high-temperature hydrogenation (HTH) methods are shown in Fig. 5(a), 5(b), 5(c), and 5(d), respectively. As can be seen from the micrographs, these treatments are successful in eliminating the large a plate colonies and the grain-boundary a phase. As discussed

Typical solution treatment



C



F

Intermediate treatment 

C

Dehydrogenation temperature 

F



C



F

Typical annealing or aging treatment

Applied to product forms

Ref

Cast, P/M, I/M

34–37

Cast

38

Cast, I/M

39

Cast, I/M

39

P/M, I/M

40, 41

BUS

1040  C (1900  F) for ½ h

...

...

...

...

...

...

GTEC

1050  C (1925  F) for ½ h

...

...

...

...

...

...

BST

1040  C (1900  F) for ½ h and GFC(b) 955  C (1750  F) for 1 h and GFC ...

...

...

...

...

...

...

...

...

...

...

...

...

650

1200

870 (glass encapsulated)

1600

760

1400

845  C (1550  F) for 24 h 845  C (1550  F) for ½ h and 705  C (1300  F) for 2 h 540  C (1000  F) for 8 h 540  C (1000  F) for 8 h ...

1040  C (1900  F) for ½ h ...

595 870

1100 1600

760 815

1400 1500

... ...

Cast, P/M, I/M Cast

41–43 44

...

900

1650

Cool to RT(c) No intermediate step (continuous process) Cool to RT

705

1300

...

Cast, P/M, I/M

45

ABST HVC (Hydrovac process) TCT CST HTH

(a) Most data apply to Ti-6Al-4V. bt temperature approximately 999  C (1830  F). (b) GFC, gas fan cooled. (c) RT, room temperature

1134 / Nonferrous Alloy Castings below, a substantial improvement of both tensile and fatigue properties is achieved with these processes.

Oxygen Influence. Figure 6 is a frequency distribution of tensile properties from separately cast test bars representing hot isostatically pressed and annealed Ti-6Al-4V castings (Ref 46). Note that oxygen, a carefully controlled alloy addition, is in the 0.16 to 0.20% range, which is common for many aerospace specifications. Some specifications allow a 0.25% maximum oxygen content. The resultant tensile and yield properties with oxygen in the 0.20 to 0.25% range are typically about 20 to 50 MPa (3 to 7 ksi) higher than those shown in Fig. 6 with slightly lower ductility and fracture toughness

levels (Ref 47). In this case, it is possible to guarantee 827 MPa (120 ksi) yield strength and 896 MPa (130 ksi) ultimate tensile strength levels with 6% minimum elongation. This strength level is the same minimum guarantee for wrought-annealed Ti-6Al-4V. Microstructure Influence. Because the microstructure of titanium alloy cast parts is very similar to b-processed wrought or ingot metallurgy (I/M) material, many properties of hot isostatically pressed castings, such as tensile strength, fracture toughness, fatigue crack propagation, and creep, are at nearly the same levels as forged and machined parts. As mentioned, special heat treatments to modify the microstructure of HIP processed Ti-6Al-4V alloy castings have been studied (Ref 37–39, 41, 44, 45, 48–50). Table 7 provides mean tensile strength and fracture toughness properties

(a)

(b)

Mechanical Properties of Ti-6Al-4V

of Ti-6Al-4V alloy heat treated as shown in Table 6. The fracture toughness data are available for only a few of the conditions. As can be seen, some of the treated conditions present properties in excess of I/M b-annealed material. However, it should be noted that tests were done on relatively small cast coupons. Properties of actual cast parts, especially large components, could be somewhat lower, the result of coarser grain structure or slower cooling rates. Of special interest are the hydrogen-treated conditions (such as thermochemical treatment, or TCT; constitutional solution treatment, or CST; and HTH, in Table 6) that result in very high tensile strength (as high as 1124 MPa, or 163 ksi) with tensile elongation as high as 8% (Table 7). Fatigue and Fatigue Crack Growth Rate. The fatigue crack growth rate (FCGR) behavior of cast Ti-6Al-4V is also, as expected, very similar to that of b-processed wrought Ti-6Al4V (Ref 51–53). This is demonstrated in Fig. 7 in which the scatterband of FCGR of cast and cast plus HIP alloy is compared to b-processed I/M (Ref 18, 54). The scatterbands of smooth axial fatigue results of cast, cast HIP (Ref 15, 48, 49, 55– 58), and wrought Ti-6Al-4V are shown in Fig. 8. This figure clearly indicates that the HIP process can result in a substantially improved fatigue life well into the wroughtannealed region. The fatigue properties of aerospace quality castings have always been an important issue, because in most other cast alloy systems this is the property that is most degraded when compared to wrought products. However, because of the complete closure and healing of gas (D, in Fig. 2a) and shrinkage (E, in Fig. 2a) pores by HIP and the b-annealed microstructure, it is possible to get fatigue life comparable to wrought material in premium investment cast and hot isostatically pressed parts. As indicated previously (Table 6), substantial work has been done to modify the microstructure of cast parts to produce fatigue properties either equivalent or superior to the best wrought-annealed products. Figure 9 compares the smooth fatigue life of Ti-6Al-4V test bars treated by ABST, BST, BUS, CST, Garrett treatment (GTEC), and HTH (Table 6) to wrought material scatterband. As can be seen, all of these treatments were successful in improving fatigue life above wrought levels. The hydrogen treatments (CST and HTH) resulted in the highest improvement in fatigue strength. However, it should be noted that wrought products subjected to the same treatments result in comparable improvements in fatigue strength. Additionally, the use of above treatments for the cast parts may be limited due to dimensional distortion and/or added cost.

Casting (c)

Fig. 5 (d) HTH

(d) Micrographs of microstructures resulting from a variety of hydrogen and solution heat treatments used to eliminate large a plate colonies and grain-boundary a phase in titanium alloys. (a) ABST (b) BST (c) BUS

Casting Design The best casting design is usually achieved by means of a thorough review by the

Titanium and Titanium Alloy Castings / 1135 1075

1075

125

26

26

22

22

875 125

Elongation, %

135

925

Reduction of area, %

975

Ultimate tensile strength, ksi

875

Ultimate tensile strength, MPa

135

925

0.2% offset yield strength, ksi

975

18

14

18

14

10

10

6

6

825

825

115

115 0

10 20 30 Frequency, %

(a)

Fig. 6

30

145

145

775

30 1025

1025 0.2% offset yield strength, MPa

155

155

775

40

2

2 0

(b)

10 20 30 Frequency, %

40

0

10

(c)

20 30 40 Frequency, %

0

50

10

(d)

20 30 40 Frequency, %

Frequency distribution of tensile properties of hot isostatically pressed and annealed Ti-6Al-4V casting test bars. Percent frequency is plotted versus (a) 0.2% offset yield strength, (b) ultimate tensile strength, (c) percent reduction of area, and (d) percent elongation. Alloy composition is 0.16 to 0.20% O; sample size is 500 heats. Source:

Ref 46

Stress intensity (DK), ksi

As-cast Cast HIP BUS(b) GTEC(b) BST(b) ABST(b) TCT(b) CST(b) HTH(b) Typical wrought b-annealed

Yield strength MPa

ksi

MPa

ksi

896 869 938 938 931 931 1055 986 1055 860

130 126 136 136 135 135 153 143 153 125

1000 958 1041 1027 1055 1020 1124 1055 1103 955

145 139 151 149 153 148 163 153 160 139

Elongation, %

Reduction of area, %

pffiffiffiffiffiffi ksi in:

16 18 12 11 15 12 9 15 15 21

97 99 ... ... ... ... ... ... ... 83

8 10 8 8 9 8 6 8 8 9

107 109 ... ... ... ... ... ... ... 91

Ref

37, 37, 37, 38 39 39 41, 44 45 18,

48 39, 49 39

50

19

(a) All conditions (except as-cast) are cast plus HIP. (b) See Table 6 for process details.

manufacturer and user when the component is still in the preliminary design stage. Additional features may be incorporated to reduce machining cost, and components may be integrated to eliminate later fabrication. Specifications and tolerances may be reviewed vis-a`-vis foundry capabilities, producibility, and pattern tool concepts to achieve the most practical and costeffective design. When minimum cast part weight is critical, such as in aerospace components, the capability of the foundry to produce varying wall thicknesses, for example, may be beneficial. Often, cast features that cannot be economically duplicated by any other method may be readily produced. Titanium castings present the designer with few differences in design criteria, compared with other metals. Ideal designs do not contain isolated heavy sections or uniform heavy walls of large area so that centerline shrinkage cavities

and regions with a coarse microstructure may be avoided. From a practical sense, however, ideal tapered walls to promote directional solidification are not usually a reality. The advent of hot isostatic pressing to heal internal as-cast shrinkage cavities has offered the designer much more freedom; however, there still is a practical limit to the size of internal cavity that can be healed through hot isostatic pressing without contributing significant surface or structural deformation due to the collapse of internal pores. The lost wax investment process provides more design freedom for the foundry to properly feed a casting than does the traditional sand or rammed graphite approach. It is normal practice to add a gate and riser to hot isostatically pressed investment castings to achieve reasonably good internal x-ray quality so that hot isostatic pressing will not cause extensive surface or structural deformation.

10

20

in. 50

-3 10

Wrought b annealed

0.01

KIc

pffiffiffiffiffi MPa m

5

(0.02)

da/dN crack growth rate, mm/cycle

Material condition(a)

Ultimate tensile strength

2

-4

Cast and cast plus HIP

10

-3

10

-5

10 -4

10

-6

10

da/dN crack growth rate, in./cycle

Table 7 Tensile properties and fracture toughness of Ti-6Al-4V cast coupons compared to typical wrought b-annealed material

-5

10

-6 (2 ⫻ 10 )

-7

10 1

2

5

10

20

50

100

Stress intensity (DK ), MPa m

Fig. 7

Scatterband comparison of FCGR behavior of wrought b annealed Ti-6Al-4V to cast and cast plus HIP Ti-6Al-4V

The usual required minimum practical wall thickness for investment castings is 2.03 mm (0.080 in.); however, sections as small as 1.14 mm (0.045 in.) are routinely made. Even thinner walls may be achieved by chemical milling beyond that required for a case removal; however, as-cast wall variation is not improved and

1136 / Nonferrous Alloy Castings 1200 160

120

800 Cast parts

600

Wrought anneal

80

Cast plus HIP 400

Maximum stress, ksi

Maximum stress, MPa

1000

40 200

0 100

Fig. 8

103

104

105 Cycles to failure, Nf

106

107

0 108

Comparison of smooth axial fatigue rate in cast and wrought Ti-6Al-4V at room temperature with R = +0.1

1200

1000

BUS

140

CST

HTH BST

800

120 GTEC 100

600

ABST

Maximum stress, ksi

Maximum stress, MPa

160

80

Wrought anneal 60

400 103

104

105

106

107

108

Cycles to failure, Nf

Fig. 9

Comparison of Ti-6Al-4V investment cast test bars subjected to various thermal and hydrogen treatments. Smooth axial fatigue measured at room temperature. R = +0.1; frequency = 5 Hz using triangular wave form.

Table 8

General linear and diametrical tolerance guideline for titanium castings Size

Total tolerance band(a)

mm

in.

Investment cast

Rammed graphite process

25 to <102 102 to <305

1 to <4 4 to <12

0.76 mm (0.030 in.) or 1.0%, whichever is greater 1.02 mm (0.040 in.) or 0.7%, whichever is greater

305 to <610 610

12 to <24 24

1.52 mm (0.060 in.) or 0.6%, whichever is greater 0.5%

1.52 mm (0.060 in.) 1.78 mm (0.070 in.) or 1.0%, whichever is greater 1.0% 1.0%

Examples 254

10

508

20

1.78 mm (0.070 in.) total tolerance band or 0.89 mm (0.035 in.) 3.05 mm (0.120 in.) total tolerance band or 1.52 mm (0.060 in.)

2.54 mm (0.100 in.) total tolerance band or 1.27 mm (0.050 in.) 5.08 mm (0.200 in.) total tolerance band or 2.54 mm (0.100 in.)

(a) Improved tolerances may be possible depending on the specific foundry capabilities and overall part-specific requirements.

becomes a larger percentage of the resultant wall thickness. Sand or rammed graphite molded castings have a usual minimum wall thickness of 4.75 mm (0.187 in.), although 3.05 mm (0.12 in.) is not unreasonable for short sections. Fillet radii should be as generous as possible to minimize the occurrence of hot tears and surface porosity. While 0.76 mm (0.030 in.) radii are produced, the preferred minimum is 3.05 mm (0.12 in.). A rule of thumb is that a fillet radius should be 0.5 times the sum of the thicknesses of the two adjoining walls. With proper tool design, zero draft walls are possible. To promote directional solidification, a 3 included draft angle may be preferred. Hot isostatic pressing will close any centerline shrinkage cavities in zero draft walls, making it unnecessary to provide draft. Draft requirements are also dependent on foundry practice, with rammed graphite tooling usually requiring draft, and investment casting typically not requiring draft. Tolerances. Typically, the major area of concern is true position of a thin-section surface with respect to a datum. Surface areas of approximately 129 cm2 (20 in.2) or greater in sections of less than approximately 3.00 mm (0.120 in.) thickness are susceptible to distortion, depending on adjoining sections. The high strength of titanium compared with aluminum and low elastic modulus compared with steel present challenges in straightening and in maintaining extremely tight, true positions. General tolerance band capabilities for linear dimensions are shown in Table 8. Hot sizing fixtures have been increasingly used to help control critical casting dimensions. This technique typically involves the use of steel fixtures to “creep” the casting into final tolerances in an anneal or stress-relief heat treatment by the weight of the steel or the use of differential thermal expansion of the steel relative to the titanium. Standard casting industry thickness tolerances of 0.76 mm (0.030 in.) for rammed graphite and 0.25 mm (0.010 in.) for investment cast walls are more difficult to maintain with titanium primarily because of the influence of chemical milling. As mentioned previously, for critical applications it is necessary to mill all surfaces chemically to remove the residue a case. This operation is subject to variation because of part geometry and bath variables, and because it is usually manually controlled. Standard industry surface finishes are shown in Table 9.

Melting and Pouring Vacuum Arc Remelting (VAR). The dominant method of melting titanium is with a consumable titanium electrode lowered into water-cooled copper crucibles while confined in a vacuum chamber. This skull-melting technique prevents the highly reactive liquid titanium from dissolving the crucible because it is contained in

Titanium and Titanium Alloy Castings / 1137 a solid “skull” frozen against the water-cooled crucible wall. When an adequate melt quantity has been obtained, the residual electrode is quickly retracted, and the crucible is tilted for pouring into the molds. A “skull” of solid titanium remains in the crucible for reuse in a subsequent pour or for later removal. The other melting method is to use cold wall induction crucible melting (induction skull melting). Superheating. The consumable electrode practice affords little opportunity for superheating the molten pool because of the cooling effect of the water-cooled crucible. Because of limited superheating, it is common either to pour castings centrifugally, forcing the metal into the mold cavity, or to pour statically into preheated molds to obtain adequate fluidity. Postcast cooling takes place in a vacuum or in an inert gas atmosphere until the molds can be safely removed to air without oxidation of the titanium. The cold wall induction crucible melting provides somewhat greater superheat and better stirring action to promote uniformity of temperature and composition. Electrode Composition. Consumable titanium electrodes are either ingot metallurgy forged billet, consolidated revert wrought material, selected foundry returns, or a combination of all of these. Casting specifications or user requirements can dictate the composition of revert materials used in electrode construction. Figure 10 shows a typical centrifugal casting furnace arrangement.

Molding Methods Rammed Graphite Molding. The traditional rammed graphite molding process uses powdered graphite mixed with organic binders. Patterns typically are made of wood. The mold material is pneumatically rammed around the pattern and cured at high temperature in a reducing atmosphere to convert the organic binders to pure carbon. The molding process and the tooling are essentially the same as for cope and drag sand molding in ferrous and nonferrous foundries. In the 1970s, derivations of rammed graphite mold materials were developed using components of more traditional sand foundries along with inorganic binders. This resulted in more dimensionally stable and less costly molds that were capable of containing molten titanium without undue metal/mold reaction. Lost Wax Investment Molding. The principal technology that allowed the proliferation of titanium alloy castings in the aerospace industry was the investment casting method, introduced in the mid-1960s. This method, used at the dawn of the metallurgical age for casting copper and bronze tools and ornaments, was later adapted to enable production of high-quality steel and nickel-base cast parts. The adaptation of this method to titanium casting technology required the development of ceramic slurry materials with minimum reaction with the extremely reactive molten titanium.

Proprietary lost wax ceramic shell systems have been developed by the several foundries engaged in titanium casting manufacture. Of necessity, these shell systems must be relatively inert to molten titanium and cannot be made with the conventional foundry ceramics used in the ferrous and nonferrous industries. Usually, the face coats are made with special refractory oxides and appropriate binders. After the initial face coat ceramic is applied to the wax pattern, more traditional refractory systems are used to add shell strength from repeated backup ceramic coatings. Regardless of face coat composition, some metal/mold reaction inevitably occurs from titanium reduction of the ceramic oxides. The oxygen-rich surface of the casting stabilizes the a phase, usually forming a metallographically distinct a case layer on the cast surface, which is removed later by chemical milling. Foundry practice focuses on methods to control both the extent of the metal/mold reaction and the subsequent diffusion of reaction products inward from the cast surface. Diffusion of reaction products into the cast surface is time-at-temperature dependent. Depth of surface contamination can vary from nil on very thin sections to more than 1.52 mm (0.06 in.) on heavy sections. On critical aerospace structures, the brittle a case is removed by chemical milling. The depth of surface contamination must be taken into consideration in the initial wax pattern tool design. Hence, the wax pattern and casting are made slightly oversize, and final dimensions are achieved through careful chemical milling. Metal superheat, mold temperature and thermal conductivity, g force (if centrifugally cast), and rapid postcast heat removal are other key factors in producing a satisfactory product. These parameters are interrelated; that is, a high g force centrifugal pour into cold molds may achieve the same relative fluidity as a static pour into heated molds.

Postcasting Practice Chemical Milling The oxygen-rich surface of the casting stabilizes the a phase, usually forming a metallographically distinct a case layer on the cast surface. This low-ductility layer may be

removed through chemical milling. This eliminates the possibility of diffusion of these contaminants into the part during subsequent HIP or heat treatment. Castings are made slightly oversize, and final dimensions are achieved by carefully controlled chemical milling. For reasons mentioned previously, chemical milling uses solutions to dissolve metal layers from castings and other titanium product forms. Solutions (several types are available) are based on hydrofluoric and nitric acid mixtures plus additives. The additives are designed to enhance surface finish and control hydrogen pickup. Hydrogen solubility, and hence pickup is greater in the b phase and is also affected by etch rate and bath temperature. Vacuum annealing may be used to remove hydrogen that is picked up during chemical milling. The general objective is to remove the entire as-cast surface uniformly to the extent of maximum a case depth and to retain the dimensional integrity of the part.

Hot Isostatic Pressing Titanium is extremely reactive at elevated temperatures, even in the solid state. Gases (mainly, oxygen, hydrogen, and nitrogen) readily dissolve, leading to porosity. For highly critical applications, hot isostatic pressing is used to eliminate subsurface, internal gas, and shrinkage porosities, which may be introduced during casting, by diffusion bonding. Additionally, the fatigue strength is improved significantly by the HIP treatment. For example, hot isostatic pressing may be used to ensure complete elimination of internal gas (D, in Fig. 2a) and shrinkage (E, in Fig. 2a) porosity. The cast part is chemically cleaned and typically subjected to argon pressure of 103 MPa (15 ksi) at 900 to 950  C (1650 to 1760  F) for a 2 h hold time and cooled under argon to below 427  C (800  F) then air cooled to room temperature (Ti-6Al-4V alloy) for void closure and diffusion bonding. This practice has been shown to reduce the scatterband of fatigue property test results and significantly improve fatigue life (Fig. 8). The HIP temperature may coarsen the microstructure, causing a slight debit in tensile strength, but the benefits of HIP normally exceed this slight decrease in strength, and the practice is widely used for aerospace titanium alloy cast parts. Vacuum chamber

Table 9 Surface finish of titanium castings Process

NAS 823 surface comparator

mm

min.

Investment As-cast Occasional areas of

C-12 C-25

3.2 6.3

125 250

Rammed graphite As-cast Occasional areas of Hand finished

C-30–40 C-50 C-12–25

7.5–10 12.5 3.2–6.3

300–400 500 125–250

Note: RMS, root mean square

Consumable titanium electrode

RMS equivalent

Water-cooled copper crucible Stacks of molds

Centrifuge

Fig. 10

A centrifugal vacuum casting furnace

1138 / Nonferrous Alloy Castings Welding Weld repair of titanium castings is an integral part of the manufacturing process and is used to eliminate or repair surface related defects, such as HIP-induced surface depressions, laps, voids, and inclusions (Ref 59). In fact, weld repair is a significant cost and long delivery element in the production of a complex casting shapes. Gas tungsten arc welding (GTAW) practice in argon-filled glove boxes typically is used with weld filler metal of the same composition as the parent metal. Generally, all weld-repaired castings are stress relieved or annealed. Excellent quality weld deposits are routinely obtained with proper practice, and the subsequent weld deposits may have higher strength and lower ductility than the parent metal because of microstructural differences caused by the faster cooling rates associated with the welding process. If needed, those differences may be reduced by a postweld solution heat treatment, but the standard practice is for stress relief or anneal only. Higher-ductility ELI-grade wire can also be used. Because titanium is very reactive to atmospheric gases above about 400  C (750  F), additional precautions must be taken to protect the weld metal and adjacent base-plate heataffected zone until sufficient cooling has occurred. Shielding gas comprising argon or a mixture of helium and argon is required to protect the weld area from the atmosphere during and immediately after welding wherever the temperature exceeds about 400  C (750  F). More detailed information on welding of titanium and its alloys is provided in ASM Handbook, Vol 6, Welding, Brazing, and Soldering in the article “Welding of Titanium Alloys.”

Ti-6Al-4V alloy castings is shown in Table 10. As described previously, the various types of heat treatments are not applicable to all titanium alloys. While solution treatment above the beta transus then water quenching or gas fan cooling, followed by aging may be beneficial for obtaining microstructural refinement, its use can lead to dimensional distortion due to creep at high temperature and rapid quenching. These issues normally are controlled through the use of fixtures, but at an added cost. Because HIP is a heat treatment, the noncritical titanium castings can be used after HIP. Unlike most aluminum alloys and many heat treatable ferrous alloys, hardness is not a good measure of the adequacy of the heat treating process accomplished during heat treatment of titanium alloys. Instead, mechanical property tests such as tensile, Charpy, and fracture toughness, along with metallographic evaluation are used to verify the heat treated properties of titanium alloy castings. More detailed information of the influence of the various heat treatments and mechanical properties are provided in ASM Handbook, Vol 4, Heat Treating in the article “Heat Treating of Titanium and Titanium Alloys.”

Final Evaluation and Certification Titanium castings are produced to numerous quality specifications. Typically, these require some type of x-ray and dye penetrant inspection in addition to dimensional checks using layout equipment, dimensional inspection fixtures, and coordinate measuring machines. Metallurgical Table 10

Heat Treatment of Titanium Castings

Product Applications The titanium castings industry is relatively young by most foundry standards. The earliest commercial applications, in the 1960s, were for use in corrosion-resistant service in pump and

Summary of cast Ti-6Al-4V alloy dominant heat treatments

Heat treatment designation

Hot isostatic pressing (HIP)

The modification of the cast microstructure of titanium castings can only be made by heat treatment. While there are a number of heat treatments used for titanium castings, the three dominant treatments are the mill anneal (MA), anneal (AN), and stress relief (SR). It should be noted that most of the titanium literature does not distinguish between mill anneal and anneal even though the heat treatments can result in very different mechanical properties because of the higher heat treating temperature employed in annealing. Heat treatment of titanium castings typically is conducted in furnaces having air, inert, or slightly oxidizing atmospheres to prevent any hydrogen pickup. As with HIP, castings must be chemically clean prior to heat treatment if diffusion of surface contaminants is to be avoided. Typical heat treating process variables are the chemical composition, the furnace atmosphere, the phases present, the time held, the heat treating temperature, the quench delay time, and the cooling rate. A summary of dominant heat treatments and purposes for the

certifications may include HIP and heat treatment run certifications, and chemistry, tensile properties, and microstructural examination of representative coupons for absence of surface contamination. Standard industry specifications applicable to titanium casting are listed in Table 4, as indicated previously. The American Society for Testing and Materials (ASTM) Standard Practice E 1320 reference radiographs are used for the titanium castings (Table 4). Because internal discontinuities in titanium do not necessarily appear the same as they do in other metals, it is necessary to have a certified technician evaluate radiographs for proper interpretation. As pointed out in the section “Cast Alloys,” the origin of the test bars is to be reported (or specified) when certifying the mechanical properties of cast titanium alloys because of cooling rate effects that cause property variations.Typically, data are reported from separately cast bars for reasons of cost and convenience. Table 11 compares cast Ti-6Al-4V alloy room-temperature mean tensile properties from three sources of test bars such as separately cast, cast on parts, and machined from castings (Ref 60). As expected, there are slight differences in tensile properties of the test bars, with those excised from the cast part being lowest.

Mill-anneal (MA) Anneal (AN)

Stress relieve (SR)

Heat treatment cycle

Comments

Apply argon pressure of 103 MPa (15 ksi) at 900–960  C (1650–1760  F) for 2 h, under argon vacuum cool to below 427  C (800  F) then air cool to room temperature Mill anneal at 650–790  C (1200–1450  F) for 30 min to 2 h, air cool to room temperature Anneal at 790–900  C (1450–1650  F) for 30 min to 2 h, furnace cool to 540  C (1000  F) then air cool to room temperature Stress relieve at 480–650  C (900–1200  F) for 2–4 h, air cool to room temperature

Elimination of subsurface voids by diffusion bonding and improved fatigue properties

Good combination of room temperature mechanical properties and machinability Improved fracture toughness high fatigue crack growth resistance and resistance to SCC, but reduced tensile strength, machinability, straightening, full stress relieving, and dimensional stability Reduced residual stresses

Note: Approximate beta transus temperatures: Ti-6Al-4V, STD, R56400 Tb = 999  C (1830  F). Ti-6Al-4V, ELI, R56401 Tb = 977  C (1790  F)

Table 11 Comparison of cast Ti-6Al-4V room-temperature tensile properties from three sources of test material Mechanical property(a)

Tensile strength, MPa (ksi) Yield strength, MPa (ksi) Elongation, % Reduction in area, %

Separately cast

Cast on part

Machined from casting

907 (131.6) 821 (119.1) 10.8 20.8

886 (128.5) 810 (117.5) 10.1 22.2

851 (123.4) 780 (113.2) 8.8 16.6

Heat treatment: HIP at 900  C (1650  F), 103 MPa (15 ksi), 2 h plus mill annealing at 732  C (1350  F), 2 h. (a) Each value average of minimum 292 tests. Source: Ref 60

Titanium and Titanium Alloy Castings / 1139 valve components. These applications continue to dominate the rammed graphite production method; however, in more recent years, some users have justified the expense of lost wax investment tooling for some commercial corrosion-resistant casting applications (see Fig. 11). Aerospace use of rammed graphite type castings became a production reality in the early 1970s for aircraft brake torque tubes, missile wings, and hot gas nozzles. As the more precise investment casting technology developed and the commercial use of HIP became a reality in the mid-1970s, titanium casting applications quickly expanded into critical airframe (Fig. 12) and gas turbine engine (Fig. 13) components. The first applications were primarily in Ti6Al-4V, the workhorse alloy for wrought aerospace products, and castings were often substituted for forgings, with the addition of some net-shape features; this trend continues. With continuing experience in manufacturing and specifying titanium castings, applications expanded from relatively simple less-critical components for military engines and airframes to large, complex structural shapes for both military and commercial engine and airframe programs. Titanium cast parts are routinely produced for critical structural applications such as space shuttle attachment fittings, complex airframe structures, engine mounts, compressor cases and frames of many types, missile bodies and wings, and hydraulic housings (Fig. 14). Quality and dimensional capabilities continue to be advanced. Titanium castings are used for framework for very sensitive optical equipment due to their stiffness and the compatibility of the coefficient of thermal expansion of titanium with that of glass (Fig. 15). Applications evolving for engine airfoil shapes (Fig. 16) include individual vanes and integral vane rings for stators, as well as a few rotating parts that would otherwise be made from wrought product. Growth will continue as users seek to take advantage of the flexibility of design inherent in the investment casting process and the improvement in economics of net and near-net shapes. In spite of the wide acceptance of titanium castings for airframe applications, growth has been somewhat hindered because of the lack of an industry-wide database to establish casting factors accurately, relative to wrought material. Such standards were being established in the 1980s with probable application of lower design casting factors (Ref 61). Foundry size capabilities are expanding to allow the manufacture of larger airframe and static gas turbine engine structures. Widespread routine use of aerospace titanium castings is anticipated as the titanium foundry industry matches with well-established quality and product standards, and user understanding and confidence continue to be gained from satisfactory product performance. Concurrent with the above trend, investment cast titanium is increasingly being specified for medical prostheses because of its inertness to body fluids, elastic modulus approaching that

Fig. 11

Investment cast titanium components for use in corrosive environments

Fig. 12

Investment cast titanium airframe parts

Fig. 13

Typical investment cast titanium components used for gas turbine applications

1140 / Nonferrous Alloy Castings

Fig. 14

Recent applications of titanium castings are the Lightweight 155mm Howitzer (LW155). The U.S. Marines’ and U.S. Army’s next-generation of the LW155 is a titanium-intensive weapon system (Fig. 18–20). Near-net-shape titanium castings were selected for reduced weight, increased mobility, and improved survivability over its aging, M198 steel-based predecessor (Ref 62–64). The specific titanium components for the LW155 were described in a recent paper (Ref 64, 65) including their manufacturing techniques, microstructures, mechanical properties, and implementation progress. Titanium hydraulic housings produced by the investment casting process

Fig.

18

Investment cast Ti-6Al-4V stabilizer arm approximately 39.9 cm (16 in.) wide, 194.6 cm (77 in.) long, and 45 kg (100 lb.). Source: Ref 64

Fig. 15

Titanium housings for military optical applications produced by the investment casting process

Fig. 17

Fig. 16

Investment cast titanium engine airfoil components

Fig. 19

Investment cast Ti-6Al-4V spade. Source: Ref 64

Fig. 20

Investment cast Ti-6Al-4V saddle. Source: Ref 65

Titanium knee and hip implant prostheses manufactured by the investment casting process

of bone, and the net-shape design flexibility of the casting process. Custom-designed knee and hip implant components (Fig. 17) are routinely produced in volume. Some of these are

subsequently coated with a diffusion-bonded porous surface to facilitate bone ingrowth or an eventual fixation of the metal implant with the patient’s bone structure.

Titanium and Titanium Alloy Castings / 1141

ACKNOWLEDGMENT This article has been updated and revised from J.R. Newman, D. Eylon, and J.K. Thorne, Titanium and Titanium Alloys, Casting, Vol 15, Metals Handbook, ASM International, 1988, p 824–835. REFERENCES 1. H.B. Bomberger, F.H. Froes, and P.H. Morton, Titanium—A Historical Perspective, Titanium Technology: Present Status and Future Trends, F.H. Froes, D. Eylon, and H.B. Bomberger, Ed., Titanium Development Association, 1985, p 3–17 2. Titanium for Energy and Industrial Applications, D. Eylon, Ed., The Metallurgical Society, 1981, p 1–403 3. Titanium Net Shape Technologies, F.H. Froes and D. Eylon, Ed., The Metallurgical Society, 1984, p 1–299 4. “Titanium Shipment Data 1978–2003,” USGS Minerals Information, http://minerals.usgs.gov/minerals/pubs/commodity/ titanium/, accessed on 12 Nov 2007 5. American Foundrymen’s Society, private conversation, 1987 6. Aluminum Association, private conversation, 1987 7. D. Eylon, F.H. Froes, and R.W. Gardiner, Developments in Titanium Alloy Casting Technology, J. Met., Vol 35 (No. 2), Feb 1983, p 35–47; also, in Titanium Technology: Present Status and Future Trends, F.H. Froes, D. Eylon, and H.B. Bomberger, Ed., Titanium Development Association, 1985, p 35–47 8. D. Eylon and F.H. Froes, Titanium Casting— A Review, Titanium Net Shape Technologies, F.H. Froes and D. Eylon, Ed., The Metallurgical Society, 1984, p 155–178 9. H.B. Bomberger and F.H. Froes, The Melting of Titanium, J. Met., Vol 36 (No. 12), Dec 1984, p 39–47; also, in Titanium Technology: Present Status and Future Trends, F.H. Froes, D. Eylon, and H.B. Bomberger, Ed., Titanium Development Association, 1985, p 25–33 10. E.W. Collings, Physical Metallurgy of Titanium Alloys, American Society for Metals, 1984 11. “Titanium: Past, Present and Future,” NMAR-392, National Materials Advisory Board, National Academy Press, 1983; also, PB83–171132, National Technical Information Service 12. W.J. Kroll, C.T. Anderson, and H.L. Gilbert, A New Graphite Resistor Vacuum Furnace and Its Application in Melting Zirconium, Trans. AIME, Vol 175, 1948, p 766–773 13. R.A. Beahl, F.W. Wood, J.O. Borg, and H.L. Gilbert, “Production of Titanium Castings,” Report 5265, U.S. Bureau of Mines, Aug 1956, p 42

14. A.R. Beall, J.O. Borg, and F.W. Wood, “A Study of Consumable Electrode Arc Melting,” Report 5144, U.S. Bureau of Mines, 1955 15. R.A. Beahl, F.W. Wood, and A.H. Robertson, Large Titanium Castings Produced Successfully, J. Met., Vol 7 (No. 7), July 1955, p 801–804 16. S.L. Ausmus and R.A. Beahl, “Expendable Casting Molds for Reactive Metals,” Report 6509, U.S. Bureau of Mines, 1964, p 44 17. Metallic Materials Properties Development and Standardization (MMPDS-03), Battelle Memorial Institute, Columbus, Ohio, 1 Oct 2006, p 5–61 18. G.R. Yoder, L.A. Cooley, and T.W. Crooker, Fatigue Crack Propagation Resistance of Beta-Annealed Ti-6Al-4V Alloys of Differing Interstitial Oxygen Content, Metall. Trans. A, Vol 9A, 1978, p 1413– 1420 19. R.R. Boyer and R. Bajoraitis, “Standardization of Ti-6Al-4V Processing Conditions,” AFML-TR-78–131, Air Force Materials Laboratory, Boeing Commercial Airplane Company, Sept 1978 20. D. Eylon, T.L. Bartel, and M.E. Rosenblum, High Temperature Low Cycle Fatigue of Beta-Annealed Titanium Alloy, Metall. Trans. A, Vol 11A, 1980, p 1361–1367 21. D. Eylon and J.A. Hall, Fatigue Behavior of Beta-Processed Titanium Alloy IMI685, Metall. Trans. A, Vol 8A, 1977, p 981–990 22. D. Eylon, Fatigue Crack Initiation in Hot Isostatically Pressed Ti6Al-4V Castings, J. Mater. Sci., Vol 14, 1979, p 1914–1920 23. D. Eylon and W.R. Kerr, The Fractographic and Metallographic Morphology of Fatigue Initiation Sites, Fractography in Failure Analysis, STP 645, American Society for Testing and Materials, 1978, p 235–248 24. D. Eylon and M.E. Rosenblum, Effects of Dwell on High Temperature Low Cycle Fatigue of a Titanium Alloy, Metall. Trans. A, Vol 13A, 1982, p 322–324 25. W.G. Burgers, Physics, Vol 1, 1934, p 561–586 26. J.C. Williams, Kinetics and Phase Transformation, in Titanium Science and Technology, Vol 3, R.I. Jaffee and H.M. Burte, Ed., Plenum Press, 1973, p 1433–1494 27. D. Schechtman and D. Eylon, On the Unstable Shear in Fatigued Beta-Annealed Ti-11 and IMI-685 Alloys, Metall. Trans. A, Vol 9A, 1978, p 1273–1279 28. G.R. Yoder and D. Eylon, On the Effect of Colony Size on Fatigue Crack Growth in Widmansta¨tten Structure Alpha + Beta Alloys, Metall. Trans. A, Vol 10A, 1979, p 1808–1810 29. D. Eylon and P.J. Bania, Fatigue Cracking Characteristics of Beta-Annealed Large Colony Ti-11 Alloy, Metall. Trans. A, Vol 9A, 1978, p 1273–1279

30. R.J. Smickley and L.P. Bednarz, Processing and Mechanical Properties of Investment Cast Ti6Al-4V ELI Alloy for Surgical Implants: A Progress Report, in Titanium Alloys in Surgical Implants, STP 796, H.A. Luckey and F. Kubli, Ed., American Society for Testing and Materials, 1983, p 16–32 31. R.J. Smickley, Heat Treatment Response of HIP’d Cast Ti6Al-4V, Proc. WesTech Conference, ASM International and Society of Manufacturing Engineers, 1981 32. F.H. Froes, D. Eylon, G.E. Eichelman, and H.M. Burte, Developments in Titanium Powder Metallurgy, J. Met., Vol 32 (No. 2), 1980, p 47–54 33. F.H. Froes and D. Eylon, Powder Metallurgy of Titanium Alloys—A Review, Titanium, Science and Technology, Vol 1, G. Lutjering, U. Zwicker, and W. Bunk, Ed., DGM, 1985, p 267–286; also, in Powder Metall. Int., Vol 17 (No. 4), 1985, p 163– 167 and continued in Vol 17 (No. 5), 1985, p 235–238; also, in Titanium Technology: Present Status and Future Trends, F.H. Froes, D. Eylon, and H.B. Bomberger, Ed., Titanium Development Association, 1985, p 49–59 34. D. Eylon and F.H. Froes, Method for Refining Microstructures of Cast Titanium Articles, U.S. Patent 4,482,398, Nov 1984 35. D. Eylon and F.H. Froes, Method for Refining Microstructures of Prealloyed Powder Metallurgy Titanium Articles, U.S. Patent 4,534,808, Aug 1985 36. D. Eylon and F.H. Froes, Method for Refining Microstructure of Blended Elemental Powder Metallurgy Titanium Articles, U.S. Patent 4,536,234, Aug 1985 37. D. Eylon, F.H. Froes, and L. Levin, Effect of Hot Isostatic Pressing and Heat Treatment on Fatigue Properties of Ti-6Al-4V Castings, Titanium, Science and Technology, Vol 1, G. Lutjering, U. Zwicker, and W. Bunk, Ed., 1985, p 179–186 38. D.L. Ruckle and P.P. Millan, Method for Heat Treating Cast Titanium Articles to Improve Their Mechanical Properties, U.S. Patent 4,631,092, Dec 1986 39. D. Eylon, W.J. Barice, and F.H. Froes, Microstructure Modification of Ti6Al-4V Castings, Overcoming Material Boundaries, Vol 17, Society for the Advancement of Material and Process Engineering, 1985, p 585–595 40. W.R. Kerr, P.R. Smith, M.E. Rosenblum, F.J. Gurney, Y.R. Mahajan, and L.R. Bidwell, Hydrogen as an Alloying Element in Titanium (Hydrofac), Titanium’80, Science and Technology, H. Kimura and O. Izumi, Ed., The Metallurgical Society, 1980, p 2477–2486 41. R.G. Vogt, F.H. Froes, D. Eylon, and L. Levin, Thermo-Chemical Treatment (TCT) of Titanium Alloy Net Shapes, Titanium Net Shape Technologies, F.H. Froes and D. Eylon, Ed., The Metallurgical Society, 1984, p 145–154

1142 / Nonferrous Alloy Castings 42. L. Levin, R.G. Vogt, D. Eylon, and F.H. Froes, Method for Refining Microstructures of Titanium Alloy Castings, U. S. Patent 4,612,066, Sept 1986 43. L. Levin, R.G. Vogt, D. Eylon, and F.H. Froes, Method for Refining Microstructures of Prealloyed Powder Compacted Articles, U.S. Patent 4,655,855, April 1987 44. R.J. Smickley and L.E. Dardi, Microstructure Refinement of Cast Titanium, U.S. Patent 4,505,764, March 1985 45. C.F. Yolton, D. Eylon, and F.H. Froes, High Temperature Thermo-Chemical Treatment (TCT) of Titanium With Hydrogen, Proc. Fall Meeting, The Metallurgical Society, 1986, p 42 46. W. J. Barice, Investment Casting of Large Components, Titanium Net Shape Technologies, The Metallurgical Society of AIME, 1984, p 179–191 47. M. Guclu, I. Ucok, and J.R. Pickens, Effect of Oxygen Content on Properties of Cast Alloy Ti-6Al-4V, Cost-Affordable Titanium, F.H. Froes, M.A. Imam, and D. Fray, Ed., The Minerals, Metals and Materials Society, 2004, p 135–143 48. F.C. Teifke, N.H. Marshall, D. Eylon, and F.H. Froes, Effect of Processing on Fatigue Life of Ti6Al-4V Castings, Advanced Processing Methods for Titanium, D. Hasson, Ed., The Metallurgical Society, 1982, p 147–159 49. R.R. Wright, J.K. Thorne, and R.J. Smickley, Howmet Turbine Components Corporation, Ti-Cast Division, private communication, 1982; also, Technical Bulletin TB 1660, Howmet Corporation 50. L. Levin, R.G. Vogt, D. Eylon, and F.H. Froes, Fatigue Resistance Improvement of Ti-6Al-4V by Thermo-Chemical Treatment, Titanium, Science and Technology, Vol 4, G. Lutjering, U. Zwicker, and W. Bunk, Ed., 1985, p 2107–2114 51. L.J. Maidment and H. Paweltz, An Evaluation of Vacuum Centrifuged Titanium Castings for Helicopter Components,

52.

53.

54.

55. 56. 57. 58. 59.

60.

61.

Titanium’80, Science and Technology, H. Kimura and O. Izumi, Ed., The Metallurgical Society, 1980, p 467–475 J.-P. Herteman, “Properties d’Emploi de l’Alliage de Titane T.A6V Moule Densifie ou non par Compaction Isostatic a` Chaud,” Technical Report 30/M/79, CEAT, July 1979 W.H. Ficht, “Centrifugal Cast Titanium Compressor Case,” Paper presented at the Manufacturing Technology Advisory Group Meeting, General Electric Company, Aircraft Engine Group, Lynn, MA, 1979 D. Eylon, P.R. Smith, S.W. Schwenker, and F.H. Froes, Status of Titanium Powder Metallurgy, Industrial Applications of Titanium and Zirconium: Third Conference, STP 830, R.T. Webster and C.S. Young, Ed., American Society for Testing and Materials, 1984, p 48–65 J.K. Kura, “Titanium Casting Today,” MCIC-73–16, Metals and Ceramics Information Center, Dec 1973 J.R. Humphrey, Report IR-162, REM Metals Corporation, Nov 1973 M.J. Wynne, Report TN-4301, British Aircraft Corporation, Nov 1972 H.D. Hanes, D.A. Seifert, and C.R. Watts, Hot Isostatic Processing, Battelle Press, 1979, p 55 J.D. Cotton, L.P. Clark, and H.R. Phelps, Titanium Investment Casting Defects: A Metallographic Overview, J. Met., Vol 58 (No. 6), 2006, p 13–16 Personal communication to Margaret Tuttle by McLellan from R.R. Boyer, Boeing Commercial Airplane Company, October, 1998, cited by M.T. McLellan, Ti-6Al-4V, Casting, Aerospace Structural Metals Handbook, Vol 4, Code 3801, Supplement 1999, p 53 R.J. Tisler, Fatigue and Fracture Characteristics of Ti-6Al-4V HIP’ed Investment Castings, Proc. International Conference on Titanium, Titanium Development Association, Oct 1986, p 23–41

62. C. Hatch and R. Nestor, Military Makeover the Investment Casting Way, Mod. Cast., Dec 2003, p 30–32 63. F. Hoerster and J. Boulet, Investment Castings Pioneer New Applications, Foundry Manage. Technol., June 2004, p 14–17 64. K. Klug, I. Ucok, M.N. Gungor, M. Guclu, L.S. Kramer, W.T. Tack, L. Nastac, N.R. Martin, and H. Dong Near-Net-Shape Manufacturing of Affordable Titanium Components for M777 Lightweight Howitzer: Research Summary, J. Met., Vol 56 (No. 11), 2004, p 35–41 65. L. Nastac, M.N. Gungor, I. Ucok, K.L. Klug, and W.T. Tack, Advances in Investment Casting of Ti-6Al-4V Alloy: A Review, Int. J. Cast Metal Res., Vol 19 (No. 2), 2006, p 73–93

SELECTED REFERENCES  Alloy Phase Diagrams, Vol 3, ASM Hand-

book, ASM International, 1992

 R. Boyer, G. Welsch, and E.W. Collings,       

Materials Properties Handbook: Titanium Alloys, ASM International, 1994 M.J. Donachie, Titanium—A Technical Guide, 2nd ed., ASM International, 2000 Heat Treating, Vol 4, ASM Handbook, ASM International, 1991 C. Leyens and M. Peters, Editors, Titanium Alloys, Wiley-vch Verlag GmbH and Co KGaA, Germany, 2003 G. Lutjering and J.C. Williams, Titanium, Springer-Verlag Berlin Heidelberg, Germany, 2003 Metallography and Microstructures, Vol 9, ASM Handbook, ASM International, 2004 Properties and Selection: Nonferrous Alloys and Special-Purpose Materials, Vol 2, ASM Handbook, ASM International, 1990 Welding, Brazing, and Soldering, Vol 6, ASM Handbook, ASM International, 1993

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1143-1148 DOI: 10.1361/asmhba0005338

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Zirconium and Zirconium Alloy Castings Richard C. Sutherlin, ATI Wah Chang, An Allegheny Technologies Company

ZIRCONIUM casting technology was originally developed at the U.S. Bureau of Mines Albany Research Center (currently known as the National Energy Technology Laboratory, NETL—Albany) in 1957 (Ref 1). The initial program to manufacture zirconium castings was an effort to produce various shapes for nuclear reactor components that were less costly than the forged and machined parts. This program was sponsored by the Atomic Energy Commission and was done in conjunction with a development program for titanium castings. Both titanium and zirconium are reactive metals and require similar processing to produce acceptable castings. Even today, similar processes can and are used to produce titanium and zirconium castings. Although the initial process development was for use in nuclear applications, zirconium castings are primarily produced for the chemical processing industry (CPI). Many complex shapes are produced for the CPI (Fig. 1). This article describes typical foundry practices used to commercially produce zirconium castings. The processes discussed include rammed graphite casting as well as investment casting. The article covers primarily the rammed graphite process because it is the most widely used process for zirconium cast parts. Processes discussed include mold preparation, melting techniques, static and centrifugal casting, final evaluation, and testing processes, in addition to information on the mechanical and chemical properties of zirconium castings.

Molten zirconium reacts with and is embrittled by most materials, including organic and inorganic compounds, iron, and most other metals. Exposure to air by molten zirconium results in an instantaneous increase in oxygen and nitrogen throughout the molten metal. Because of its high reactivity, zirconium must be melted and cast under vacuum or inert gas atmospheres. It is also important that zirconium be cast into nonreactive molds to prevent contamination of the surfaces. In the past, many types of mold materials have been studied and

used, including expendable graphite mold materials with various binders (Ref 2). Other high-stability refractories such as zirconia, thoria, and yttria have also been used to help prevent contamination of the casting by the mold (Ref 3, 4). To minimize the contamination of the zirconium, investment casting includes controlling the ash content of the pattern wax. Because of its high melting point, the fluidity of the molten zirconium is relatively poor. Poor fluidity can cause dispersed shrinkage voids and no-fill defects. All of these factors make zirconium

Zirconium Reactivity Considerations Zirconium has a melting point of 1852  C (3365  F). It reacts vigorously with interstitial elements (e.g., hydrogen, nitrogen, oxygen, and carbon) at elevated temperatures. This reaction begins to occur at temperatures much lower than the melting point of zirconium. Even small levels (e.g., <100 ppm of hydrogen or nitrogen) of interstitial elements can be detrimental to reactive metals such as zirconium. Contamination of zirconium can lead to increased hardness, reduction of ductility, and reduced impact properties.

Fig. 1

Zirconium pump and valve components. Courtesy of ATI Wah Chang

1144 / Nonferrous Alloy Castings metal a much more difficult material to cast than other more common materials of construction.

Melting Processes Zirconium casting uses two melting methods: vacuum arc skull melting and induction melting (either vacuum or skull). Both methods are commonly used for melting reactive alloys. Vacuum arc skull melting furnaces use consumable electrodes melting into water-cooled copper crucibles. When sufficient metal has been melted, the crucible is tipped and the metal is poured into the mold positioned below. Skull melting furnaces are capable of using all types of mold systems and are limited only by the weight of the melted metal and the size constraint of the furnace. Castings can be produced with the receiving molds in a static, that is, stationary, mode as well as by centrifugal casting. Induction skull melting (ISM) uses a watercooled copper crucible to prevent contamination of the reactive metal. The ISM crucible is segmented to allow the use of an induction power source to apply an oscillating magnetic field to the metal inside the crucible. Without the metal segments, the induction coil would

Fig. 2

serve only to melt the copper crucible. With the segments, the magnetic field supplied by the induction coil passes through the crucible segments and couples with the metal inside the crucible. When the charge is melted, a thin shell freezes along the crucible base and walls. The shell that contains the molten metal is known as a “skull” because the metal is melted within a solid shell of the same material. This type of melting configuration is commonly referred to as “skull melting.” Vacuum induction melting (VIM) is used to produce small casting pours. One current furnace design uses a split-copper crucible with a melting capacity of approximately 27 kg (60 lb). The advantage of a VIM system is the ability to pour small single-casting pours with minimal skull losses. As with skull melting, VIM furnaces have the capacity to use most types of molding systems (Ref 5).

Casting Processes The two types of casting processes used for the production of zirconium castings are rammed graphite and investment (i.e., lost-wax) molding. The rammed graphite process is by far the most

Process flow outline of the rammed graphite process. Courtesy of ATI Wah Chang

common method used for zirconium castings. The rammed graphite process can be used for all sizes of castings up to 1040 kg (2300 lb). (See Fig. 2 for a process flow outline of the rammed graphite process.) The investment casting process has been used for smaller castings and may be practical if smooth surfaces, tight dimensional tolerances, or complex shapes are required. Casting can be done in a static mode or centrifugally.

Rammed Graphite Casting Molds for zirconium can be produced using wood, plastic, composite, or metal patterns. These molds are produced from patterns similar to those used for other metals. Conventional rammed graphite molding uses the standard cope and drag patterns, with or without cores. Cope and drag patterns are split patterns that are mounted or cast on separate plates. Cope and drag patterns using wood, fiberglass, or metal patterns can be used repeatedly for hundreds of castings. This is the preferred method for high-volume production of castings. Most often, standard patterns can and are used repeatedly to produce duplicate molds. The

Zirconium and Zirconium Alloy Castings / 1145 types of pattern materials generally are chosen depending on the number of castings to be produced, the final casting design, and the dimensional requirements required. Patterns range from simple shapes with loose dimensional requirements to very complex shapes with extremely tight tolerances. Zirconium, zirconium alloys, and other reactive metals are extremely reactive in the molten state as mentioned previously. Because of this, it is very important that the process used will minimize the contamination by interstitial elements such as oxygen, nitrogen, and hydrogen and other gases. Zirconium will also react with all conventional molding materials. The extent of the zirconium metal reactions when molten will depend on the casting size and mass to be poured, casting configuration, molding material, and mold surface roughness. Because of zirconium reactivity, graphite generally is used as a mold medium. Graphite powder with a controlled particle size and size distribution, is mixed with binders to enhance green strength for handling and fired strength for pouring and is rammed against a wooden or fiberglass pattern to form a mold. The pitch, corn syrup, and starch binders give the best combination of graphite particle coating and optimum mold properties and will burn off during the firing process (Ref 6). This process of mechanically ramming the graphite into molds is best performed manually to ensure that a proper compaction pressure and complete fill is achieved. The corn syrup and starch binders give the mold sufficient green strength to withstand the drying process, which could take up to 24 h at temperatures up to 200  C (393  F). Molds are arrayed on drying racks in a drying furnace to shorten the curing time. The drying time also depends on the size (thickness) and shape of the mold being dried. Drying (baking) the molds will remove moisture and prevent steam generation during the higher-temperature mold firing process. After drying, the molds are fired in a reducing-atmosphere furnace to prevent oxidation of the graphite. The firing temperature should be in excess of 816  C (1500  F) to drive off any remaining volatile materials in the mold binders. These elevated temperatures will yield molds with high strength and hardness. The molds are then cleaned and assembled, and the gates and risers are cut into the molds. Because of the poor fluidity of molten zirconium, it is extremely important to properly design and position the gates and risers in the mold. Gating provides proper flow of the liquid metal into the mold cavity during the casting process. The design must account for the proper metal fill in the mold cavities to prevent excessive shrinkage in the final casting. Shrinkage of the casting will coincide with the particular alloy being cast and the physical properties of that alloy. Typical casting shrinkage is 16 to 26 mm/m (3/16 to 5/16 in./ft) (Ref 5). Care must be taken to provide a smooth surface for the mold/metal interface since rough

surfaces on the molds will allow greater reaction between the mold and the metal. Graphite, however, has a pronounced chilling effect on molten zirconium that helps to minimize the mold/metal reaction. Internal core areas, however, may be susceptible to excessive metal/ mold reaction. The typical reaction layer is 0 to 0.13 mm (0 to 0.005 in.) thick, but large masses of metal will be prone to greater metal/mold reaction (Ref 5). Sand systems have been identified as having potential use as mold media for zirconium. The first is zircon sand/waterglass using zirconium silicate sand and sodium silicate as the binder. The advantage of this system is reduced shrinkage of the mold to provide a more dimensionally precise casting. Another advantage is that sodium silicate is a no-fire binder requiring only a 200  C (390  F) cure, which would eliminate the multiple-stage firing process needed for graphite. This mold system requires a zirconia mold wash to protect the molten metal from reaction with the sand. Castings of a commercial size 23 kg (50 lb) were produced using this system in 1979 with moderate success (Ref 3, 4). The casting surface was rougher than with graphite molds, and the cores showed higher mold/metal reaction. The second system, an olivine sand/bentonite binder, also requires the use of a zirconia mold wash to minimize the mold-mold/metal reaction. No commercial-size castings, however, are known to have been produced with this system (Ref 4). Depending on the particular size and configuration of the casting, one or more molds are assembled for the casting operation. The maximum size of a zirconium pour is about 1040 kg (2300 lb), but larger (Ref 7) castings have been produced in multiple pours using smaller cast parts that are assembled by welding. In terms of dimensional size, a cast part of 1270 mm (50 in.) in diameter by 1830 mm (72 in.) is generally the maximum for a single pour. Wall thicknesses as thin as 4.75 mm (0.1875 in.) have been produced by the rammed graphite method. One of the largest zirconium castings ever poured was produced in 1997 for use in a CPI application. The final 1829 mm (72 in.) casting weighed (2745 lb) (Ref 8, 9).

Investment Casting Investment casting of zirconium can be accomplished using the ISM process because it offers advantages for reactive alloy investment casting (Ref 10). This process uses a water-cooled copper crucible, which helps to minimize any surface contamination caused by the use of ceramic crucibles. This process allows for the use of revert and scrap for production of lower-cost castings. Investment casting may offer advantages of a near-net shape, tight dimensional tolerance, or complex geometry for reactive metals such as zirconium and zirconium alloys.

Static Casting Static casting is the most common and simplest type of casting. This type of casting process is performed for final castings that are the easiest to fill. If the casting is complex and has many types of channels, it should be centrifugally cast.

Centrifugal Casting Centrifugal casting is accomplished by mounting the molds on a turntable. This process uses centrifugal force to feed molten metal into the mold cavities. The setup utilizes a center sprue with a runner system to feed from the center sprue to the molds located radially about the sprue. An advantage of this process is that it allows the production of more complex cast shapes as opposed to static casting. The mold is filled against the centrifugal forces, allowing a slower fill rate and reducing the potential for entrapped gases in the casting. The centrifugal casting process will eliminate or reduce the metal lost to gates and risers. A limitation of centrifugal casting in a vacuum arc skull melt furnace is that the pour must have a balanced setup; that is, molds equally spaced around the center sprue. The mold media and mold design must be strong enough to support the forces exerted on the setup by the molten metal.

Processing After casting the molten metal into the mold, the graphite mold is physically removed from the casting. Care must be used when removing the graphite mold and core because damage may occur to the casting if struck sharply. Castings with thicknesses as great as 127 mm (5 in.) have been produced commercially with the rammed graphite mold system. After “knockout” or metal removal, gates and risers are removed by abrasive cutoff or by oxyacetylene torch. If torch cutting is used, the slag area and subslag-contaminated material must be removed. All metal surfaces in contact with the mold will have a contaminated layer. For graphite mold systems, it is a carbon-stabilized layer. For refractory mold systems, an oxide layer known as “alpha case” forms. Depending on the application, this hard, brittle layer should be removed from the surface because it may be a source of contaminants during subsequent weld repair and may adversely affect the mechanical properties of the casting. In some applications, corrosion resistance may also be affected. Abrasive grit, shot blasting, or grinding can be used to remove most of this contaminated layer.

Final Surface Finishes Grinding. Final surface finishes on some portions of the castings may be very rough and might need to be ground when the cast

1146 / Nonferrous Alloy Castings parts are conditioned or when impellers are balanced. “Dedicated” (i.e., only used on one material) grinding wheels must be used for zirconium. Grinding wheels used for iron- or nickel-base materials will cause embedded particles on the final surface of the zirconium that may cause localized corrosion of the impeller or the acid-wetted surfaces if exposed to environments such as hydrochloric acid (Ref 11). Chemical Milling. For internal surfaces or blind (i.e., hard-to-reach) areas, chemical milling is more efficient than abrasive grit or shot blasting and is the preferred cleaning method. Chemical milling solutions are of the following general chemical composition: 25 to 50% nitric acid, 3 to 10% hydrofluoric acid, and the remainder water. Temperature for the solution should be kept below 55  C (130  F) to minimize hydrogen absorption by the zirconium casting and generation of noxious NOx fumes.

Casting Evaluation Processes Castings are evaluated using a number of destructive and nondestructive techniques. Castings can be evaluated using radiographic examination, dye penetration, mechanical testing, and chemical analysis. Radiographic Testing of Castings. Radiographic inspection can be used to detect internal shrinkage, porosity, and inclusions in the final casting. Liquid dye penetrant testing is generally performed to identify surface flaws or defects in the metal casting. Any cracks, laps, or surface pores can be located by dye penetrant testing and repaired. Chemical Analysis of Castings. All final castings must meet chemical standards to meet the ASTM B 752 specification requirements. A chemical analysis of each pour must be made from a sample or test bar that is representative of the pour. Table 1 lists the chemical property requirements for both wrought and cast zirconium and zirconium alloys (Ref 12, 13). Mechanical Testing. To certify to ASTM standards for mechanical strength, castings must be tested using mechanical methods. These test

Table 1 ASTM chemical requirements for wrought and cast Zr702 and Zr705 Composition, % Chemical element

Zr + Hf, min Hafnium, max Hydrogen, max Nitrogen, max Oxygen, max Carbon, max Iron + chromium, max Phosphorous Niobium N/A, not applicable

Zr702 wrought

Zr702 cast

Zr705 wrought

Zr705 cast

99.2 4.5 0.005 0.025 0.16 0.05 0.2

98.8 4.5 0.005 0.03 0.25 0.10 0.3

95.5 4.5 0.005 0.025 0.18 0.05 0.2

95.1 4.5 0.005 0.03 0.30 0.10 0.3

N/A N/A

0.01 N/A

N/A 2.0–3.0

0.01 2.0–3.0

specimens are generally removed from either special test blocks or bars from the casting mold, or cut from representative areas of the casting. Table 2 lists the mechanical property requirements for both wrought and cast zirconium and zirconium alloys (Ref 12, 13).

Weld Repair of Casting Defects As-cast zirconium may contain casting defects such as gas hole porosity, cold shuts (or surface laps), and shrinkage cavities. Gas hole porosity is generated from the metal-mold reaction and the evolution of gases from the mold. This type of defect can be minimized by proper firing of the mold, but certain portions of a casting may still contain these small spherical voids. Cold Shuts. Because of the high melting point of zirconium, 1852  C (3365  F) and the chilling effect the mold media has on the metal, zirconium typically exhibits more cold shuts than titanium or other castings. However, these defects are surface phenomena only and depending on the application, may be left as-is with no detrimental effects. Shrinkage Cavities. Zirconium has less fluidity than titanium, which is the closest material to compare zirconium to for melting parameters and casting methodology. This low fluidity may result in shrinkage cavities forming in areas of a casting where flow conditions are upset. Therefore, zirconium requires larger or greater numbers of metal distribution gates and additional risers to minimize shrinkage voids. Weld Repair. All castings have volumetric shrinkage voids. Unless a given casting is designed for use with these voids present, it will be necessary to remove them. Suitable excavation and weld repair then become necessary steps in the production of zirconium castings. Weld repair is performed using gas tungsten arc welding (GTAW). Welding can be accomplished on a bench setup with sufficient inert-gas protection, but for productionscale processes it is best to perform the weld repair in a weld chamber. Zirconium is extremely reactive to atmospheric gases, and if contaminated, the weld will be embrittled, possess adverse mechanical properties, and perhaps exhibit reduced corrosion resistance. Table 2 Mechanical properties of selected zirconium-base alloys Mechanical property

Zirconium alloy Zr702

Zr702C

Ultimate tensile 380 (55) 380 (55) strength, MPa (ksi) Yield strength, 205 (30) 276 (40) MPa (ksi) Elongation, 16 12 %, min Hardness, N/A 210 (B96) HB, max N/A, not applicable

Zr705

Zr705C

550 (80)

483 (70)

380 (55)

345 (50)

18

12

N/A

235 (B99)

Preparation of the weld area is necessary for sound welds. The area must be cleaned of any residual grease or oil. The filler rod should also be degreased before use. For repair of internal defects, the defect is exposed by drilling or machining and without lubricants or coolants in order to minimize contamination of the weld area. The surrounding area must be cleaned after the machining operation. The cavity itself does not necessarily need to be cleaned. After placing the casting in the weld chamber, either a vacuum pumped down followed by backfilling with inert gas or a flow-through purge of inert gas is required to protect the metal during welding. To ensure the weld chamber is adequate to produce a quality weld, it is advisable to initially strike an arc on the test plate and observe the color of the weld spot or nugget. If discoloration of the weld spot is noted then there might be sufficient air left in the chamber to cause contamination to the zirconium weld. The chamber should either be evacuated again or the purge gas continued until no coloration is noted on subsequent spot tests. The filler metal should also be cleaned using a solvent and be of the same nominal composition as the casting.

Postweld Heat Treatment of Zirconium Castings Postweld heat treatment (stress relief) of zirconium castings, if necessary, is carried out at 538 to 620  C (1000 to 1150  F) for 30 min per 25.4 mm (1 in.) of thickness. It is critical that any welds on zirconium grade 705 alloy castings are stress relief annealed at the stress relief temperature of 538 to 620  C (1000 to 1150  F) for 30 min per 25.4 mm (1 in.) of thickness to prevent delayed hydride cracking of the weld (Ref 14). For some specific applications, such as high sulfuric acid concentrations, welds may require a 770  C (1420  F) for 1 h per 25.4 mm (1 in.) of thickness heat treatment for optimum corrosion resistance properties (Ref 14).

Hot Isostatic Pressing Hot isostatic pressing (HIP) is a process that subjects the cast parts to high pressures at elevated temperatures. The atmosphere used in the furnace is argon gas. Hot isostatic pressure chamber conditions use argon pressure applied at about 110 MPa (15 ksi) at temperatures of about 815  C (1550  F). Hot isostatic pressing eliminates internal voids and microporosity through a combination of plastic deformation, creep, and diffusion. This process increases the density of the zirconium castings and improves the mechanical properties. Zirconium castings typically can be processed in the same hot isostatic chamber with titanium with acceptable results. Hot isostatic pressing will close or “heal” the internal pores and voids but will create surface shrinkage cavities. These cavities may require weld repair to fill them in.

Zirconium and Zirconium Alloy Castings / 1147 Machining Cast zirconium has machining characteristics similar to those of the wrought product. The final casting can be machined and/or drilled using either high-speed steel or tungsten carbide tooling. As with all zirconium machining, care should be taken in the handling and storage of turnings and other finely divided particles because of their flammable nature.

Physical Properties of Zirconium Castings (a) (a)

50 µm

(b) (b)

Fig. 3

50 µm

Micrographs of zirconium grade 702C cast structure at (a) 100 and (b) 200 (original magnification). Courtesy of ATI Wah Chang

50 µm

50 µm

Zirconium castings are similar in composition to wrought zirconium product. Molten zirconium first solidifies to a body-centered cubic structure that transforms into a hexagonal close-packed Widmansta¨tten a phase upon cooling (see Fig. 3 and 4). Cast zirconium alloys generally have higher strength than the wrought alloys because of the higher interstitial element content of the alloy. Cast ductility is generally less than that of the similar wrought product alloys. Table 1 lists the chemical composition requirements of the zirconium cast alloys compared to the wrought products (Ref 12, 13). Table 2 lists the minimum roomtemperature tensile properties of cast zirconium alloys compared to those of the wrought alloys (Ref 12, 13). Figures 5 and 6 show typical mechanical properties for zirconium castings from room temperature through 316  C (600  F). As the figures show, the Zr705C alloy has greater strength.

Fig. 4

Micrographs of zirconium grade 705C cast structure at (a) 100 and (b) 200 (original magnification). Courtesy of ATI Wah Chang

Impact Properties of Zirconium Castings Impact strength is not a property that is normally tested for zirconium castings. Zirconium Charpy impact properties have been shown to be somewhat less than that of wrought products in the same alloy grade (Ref 11). Work has been done to study and improve the impact properties of zirconium castings (Ref 15).

ACKNOWLEDGMENT This article has been updated and revised from J.P. Laughlin, Zirconium and Zirconium Alloys, Casting, Vol 15, Metals Handbook, ASM International, 1988, p 836–839. REFERENCES

Fig. 5

Typical mechanical properties of Zirconium Grade 702C at temperatures of 25 to 371  C (70 to 700  F). Source: Courtesy of ATI Wah Chang

1. S.L. Ausmus, F.W. Wood, and R.A. Beall, “Casting Technology for Titanium, Zirconium and Hafnium,” Report of Investigation 5686, Bureau of Mines, 1960 2. S.L. Ausmus and R.A. Beall, “Expendable Casting Molds for Reactive Metals,”

1148 / Nonferrous Alloy Castings

Fig. 6

Typical mechanical properties of Zirconium Grade 705C at temperatures of 25 to 371  C (70 to 700  F). Source: Courtesy of ATI Wah Chang

Report of Investigation 6509, Bureau of Mines, 1964 3. R.K. Koch, J.L. Hoffman, M.L. Transue, and R.A. Beall, “Casting Titanium and Zirconium in Zircon Sand Molds,” Report of Investigation 8208, Bureau of Mines, 1977 4. R.K. Koch and J.M. Burrus, “Shape Casting Titanium in Olivine, Garnet, Chromite and Zircon Rammed and Shell Molds,” Report of Investigation 8443, Bureau of Mines, 1980

5. J.P. Laughlin, Zirconium and Zirconium Alloys, Casting, Vol 15, Metals Handbook, ASM International, 1988, p 836–839 6. Dan Kihlstadius, Rammed Graphite Molds, Casting, Vol 15, Metals Handbook, ASM International, 1988, p 273 7. W.A. Budd, Titanium and Zirconium Rammed Graphite Castings for Industrial Applications, Reactive Metals in Corrosive Applications, Conference Proceedings (Sunriver, OR), Oremet-Wah Chang, 1999, p 150

8. Cast Zirconium Pump the Answer to Rohm and Haas Corrosion Problem, Outlook, Vol 19 (No. 2), Oremet-Wah Chang, Albany, OR, 2nd Quarter, 1998, p 1 9. Oregon Metallurgical Corporation Casts the World’s Largest One-Piece Zirconium Pump from Zircadyne, Outlook, Vol 12 (No. 1), Teledyne Wah Chang, Albany, OR, Summer 1991, p 1 10. S. Reed, G. Felzien, and T. Spence, Investment Casting of Zr and Ti, Corrosion Solutions Conference (Sunriver, OR), Wah Chang, 2003, p 146 11. T. Spence and S. Reed, Localized Corrosion of Cast Zr702, Corrosion Solutions Conference (Sunriver, OR), Wah Chang, 2001 12. “Standard Specification for Casting, Zirconium Base, Corrosion Resistant, for General Application,” B 752-02, Annual Book of ASTM Standards, ASTM International 13. “Standard Specification for Zirconium and Zirconium Alloy Strip, Sheet and Plate,” B 551/B-551M-04, Annual Book of ASTM Standards, ASTM International 14. R.C. Sutherlin, Procedures for Heat Treatment of Fabricated Reactive Metal Equipment, Corrosion Solutions Conference Proceedings (Sunriver, OR), ATI Wah Chang, 2005, p 65–74 15. S. Reed, G. Felzian, and T. Spence, Investment Casting of Zr and Ti, Reactive Metals in Corrosive Applications, Conference Proceedings, (Sunriver, OR), Oremet-Wah Chang, 1999, p 145–148

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1149-1164 DOI: 10.1361/asmhba0005339

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Synthesis and Processing of Cast Metal-Matrix Composites and Their Applications Pradeep K. Rohatgi, University of Wisconsin-Milwaukee Nikhil Gupta, Polytechnic Institute of New York University Atef Daoud, Central Metallurgical Research and Development Institute (CMRDI)

METAL-MATRIX COMPOSITES (MMCs) are an important class of engineered materials that are increasingly replacing a number of conventional materials in the automotive, aerospace, and sports industries, driven by the demand for lightweight and higher-strength components (Ref 1–6). MMCs are engineered combinations of two or more materials (one of which is a metal or alloy) in which tailored properties, not achievable in monolithic metals or alloys, are achieved by systematic combinations of different constituents. MMCs, in general, consist of continuous or discontinuous fibers, whiskers, or particles dispersed in a metallic alloy matrix. MMCs can be synthesized by vapor phase, liquid phase, or solid phase processes. In this article, the discussion emphasizes the liquid-phase processing where solid reinforcements are incorporated in the molten metal or alloy melt, which are then allowed to solidify to form a composite. Cast MMCs are not new to the industry. Although traditionally not called a composite, the aluminum-silicon eutectic alloy (Fig. 1a) is composed of silicon needles embedded in an aluminum matrix. Such a microstructure may be called an in situ composite. Another example of an in situ composite commonly produced by foundries is ductile cast iron (Fig. 1b), in which graphite nodules are dispersed in a ferrite matrix. A limitation of composites such as the aluminum-silicon eutectic alloy and ductile cast iron is that the volume percentages of the two phases are restricted to narrow ranges predicted by their phase diagrams. Additionally, the morphology and spatial arrangement of reinforcements cannot be varied as freely as in synthetically produced composites, which are synthesized by physically mixing a reinforcement phase in a metallic melt. Therefore, in situ composites are excluded from the discussion in this article in order to focus on synthetic composites.

In synthetic composites, one can change, almost at will, the chemistry, shape, volume percentage, and distribution of the second phase (reinforcement). The restrictions, if any, are rheology and flow of metal between the reinforcements. Given the large number of metals and alloys used in synthesizing composite materials, it is not possible to cover all types of composites in this short discussion. Therefore, major issues pertaining to processing of cast MMCs are discussed by means of examples related to aluminum-matrix composites.

Classification of MMCs MMCs are classified into three broad categories depending on the aspect ratio of the reinforcing phase. These categories are defined as:

Fig. 1

 Continuous fiber-reinforced composites  Discontinuous or short fiber-reinforced

composites

 Particle-reinforced composites

The structures of the three types of MMCs are schematically shown in Fig. 2. A unidirectionally aligned continuous fiber-reinforced composite is illustrated in Fig. 2(a). The fiber material can be carbon or a ceramic such as silicon carbide (SiC), and the matrix can be a metal such as titanium, copper, or aluminum. Extensive literature is available on mechanical and thermal properties, such as modulus and thermal conductivity, of fiber-reinforced MMCs. Generally, mechanical properties, including strength and stiffness, are higher in the longitudinal direction than in the transverse direction to the fibers. Figure 2(b) illustrates a short fiber-reinforced composite. The short

Phase-diagram-restricted metal-matrix composites. (a) Aluminum-silicon alloy. (b) Ductile cast iron

1150 / Nonferrous Alloy Castings

Fig. 2

Classification of metal-matrix composites. (a) Aligned continuous fiber-reinforced composite. (b) Short fiber-reinforced composite. (c) Particulate composite

Fig. 4

Injection of nickel-coated graphite particles in molten aluminum alloys in an initial experiment on casting aluminum-graphite particle composites at Merica Laboratory, International Nickel Company, 1965

Fig. 3

Magnitude of stiffening and strengthening of composites relative to the matrix as a function of reinforcement aspect ratio and volume fraction, f

fibers can be either aligned or randomly oriented in the matrix material. Figure 2(c) demonstrates a particle-reinforced composite in which irregularly shaped particles of a second phase are dispersed in a metallic matrix. The particles or whiskers incorporated in these composites can be made of graphite or ceramics, such as SiC, alumina silica, and the matrix can be a metal, such as aluminum, copper, titanium, or magnesium. Fiber-reinforced composites are used in applications where higher specific strength and specific modulus are required. Particulate composites normally provide significant improvement in wear properties of composites but only a marginal improvement in mechanical properties compared to the matrix alloy. Unlike fiber-reinforced MMCs, the particle-reinforced MMCs are generally isotropic and more amenable to secondary processing practices, such as rolling, extrusion, and forging after casting. Figure 3 schematically illustrates the magnitude

and range of stiffening and strengthening of composites in relation to the matrix as a function of reinforcement aspect ratio (ratio of length to diameter) and volume fraction ( f ). Particle-reinforced MMCs differ from dispersion-hardened materials, in which the particle size varies between 0.001 and 1 mm (0.039 and 39 min.) and the volume fraction of particles is below 0.10. Dispersion-hardened alloys are excluded from this discussion on synthetic composites.

Solidification Processing of MMCs and Possible Effects of Reinforcement on Solidification The initial discovery of the synthesis of cast MMCs was made in 1965 at the Merica Laboratory of the International Nickel Company (Fig. 4) (Ref 7). The initial work confirmed that

nickel-coated graphite particles could be either introduced to molten aluminum through an inert gas stream (Fig. 4) or stir mixed in the melt using an impeller rotating in the melt to form cast aluminum-graphite composites. Tests showed that cast aluminum-graphite composites could be run against aluminum alloys without seizing or galling, even under conditions of boundary lubrication. The initial work also included incorporation of SiC and alumina in the matrix of cast aluminum alloys. The initial work on small-sized engines at the Associated Engineering Company demonstrated that aluminum-graphite liners were able to run against aluminum pistons without seizing in the Alpha Romeo and the Ferrari automobiles used in Formula One races. Figure 5 shows a centrifugally cast aluminum-graphite composite cylinder block of a motorcycle engine, where the graphite particles were concentrated near the inner periphery of the cylinder due to the centrifugal force during casting. The aluminum-graphite cylinder liner shown in Fig. 5 was fitted to demonstrate that one could run aluminum cylinders against aluminum pistons without seizing, as a result of solid lubrication provided by the graphite particles incorporated in the castings. Since this initial work, considerable progress has been made in the field of cast MMCs. Table 1 shows the wide variety of combinations of metal and alloy matrices and the reinforcements that have been used to synthesize composite materials. This table is by no means comprehensive but illustrates the MMC development activities. In solidification processing of composites, the reinforcement phase is incorporated in the

Synthesis and Processing of Cast Metal-Matrix Composites and Their Applications / 1151 Table 1 Selected matrix-reinforcement combinations used to make cast metal-matrix composites Matrix

Aluminum base

Copper base

Steel

Tin-base Babbit metal Tin base Magnesium base Zinc base

Fig. 5

Dispersoid

Graphite flake Graphite granular Carbon microballoons Shell char Al2O3 particles Al2O3 discontinuous SiC particles SiC whiskers Mica SiO2 Zircon Glass particles Glass beads (spherical) MgO (spherical) Sand TiC particles Boron nitride particle Si3N4 particle Chilled iron ZrO2 TiO2 Lead Graphite Al2O3 ZrO2 TiO2 CeO2 Illite clay Graphite microballoons Graphite particle Al2O3 Zinc Graphite fibers Nickel-coated graphite Lead

(a) Centrifugally cast aluminum-graphite composite. (b) Cast iron cylinder block of a motorcycle engine fitted with an aluminum-silicon graphite liner

(a) In wt% unless otherwise indicated

liquid metal, which is then solidified in a mold. The possible effects of reinforcements on solidification microstructure, the property motivation for using MMCs, and present applications are discussed next. The solidification processing of composites can be divided into two main classes: stir casting and melt infiltration.

facilitating the development of processes to synthesize different cast MMCs (Ref 12). In several instances, coatings of wettable metals are deposited on reinforcements to facilitate their transfer into melts. For example, nickel or copper coatings are used on graphite particles that are incorporated in aluminum melts. In other cases, wetting agents, including magnesium and lithium, are added to the melts to promote wetting between the melts and reinforcements. Stir mixing and casting are now used for large-scale production of particulate MMCs, and companies such as ALCAN market ingots of select composites including Al-SiC, which can be remelted and cast in a variety of shapes (Ref 8). Various metals, such as aluminum, magnesium, nickel, and copper, have been used as the matrix, and a wide variety of materials, such as graphite, SiC, SiO2, Al2O3, Si3N4, and ZrSiO4, have been used as reinforcements. The study of cast composites has significantly contributed to the understanding of the solidification of conventional monolithic castings, especially to the issues related to the effects of inclusions on the fluidity, viscosity, nucleation, growth, particle settling, and particle pushing. Sand and Permanent Mold Casting. Metallic melts containing suspended ceramic particles can be cast in sand as well as permanent molds. The slow solidification rates obtained

Stir Casting In stir casting, molten metal is stirred with the help of either a mechanical stirrer (Ref 8) or highintensity ultrasonic treatment (Ref 9–11) while the particles are added to the melt. This action disperses the reinforcing phase to create a composite melt, containing a suspension of reinforcements. Figure 6 shows a schematic view of the stir-mixing process using an impeller submerged in the melt (Ref 1). Processing of MMCs by stir mixing and casting requires special precautions, including temperature control and design of pouring and gating systems. The stir-mixed suspension of reinforcement in melts can be cast into desired shapes using a variety of conventional casting processes. The thermodynamics and kinetics of transfer of reinforcements from the gas phase to the liquid phase through the oxide film and finally from the liquid to the solid phase is known,

Size, mm

20–60 15–500 40, thickness 1–2 125 3–200 3–6 mm long, 15–25 mm diam 9–12 5–10 40–180 5–53 40 100–150 100 40 75–120 46 46 40 75–120 5–80 5–80 ... ... 11 5 8 10 753 ... ... ... ... ... ... ...

Amount(a)

0.9–0.815 1–8 ... 15 3–30 0–23 vol% 3–60 10, 0–0.5 vol% 3–10 5 0–30 8 30 10 36 vol% 15 8 10 36 vol% 4 4 10 ... 74 vol% 2.12 vol% ... ... 3 ... ... ... ... 40–60 vol% ... 7

in insulating sand molds permit buoyancydriven segregation of particles. Depending on the intrinsic hardness and density of the dispersed particles, high volume fractions of reinforcements can be concentrated at the top or bottom of a component cast by this method, to serve as selectively reinforced surfaces. Tailor-made lubricating or abrasion-resistant contact surfaces for various tribological applications have been made by this technique. Use of smaller-sized particles, thin sections of sand cast composites, and the codispersion of two types of ceramic particles, as in the case of hybrid composites, can reduce the segregation of reinforcements. Centrifugal casting of composite melts containing particle dispersions results in the formation of two distinct zones in the solidified material: a particle-rich zone and a particle-free zone. If the particles are less dense than the melt (for example, graphite, mica, or fly ash hollow spheres in aluminum), then the particle-rich zone forms at the inner circumference. The outer zone is particle rich if the particles are denser than the melt (for example, zircon or SiC in aluminum). Centrifugal casting has been used to produce cylindrical bearings of aluminum- and copperbase composites. In these composites, the solid lubricants such as graphite are segregated at the inner periphery where they are needed for

1152 / Nonferrous Alloy Castings alloy slurry, which is held between its liquidus and solidus temperatures. Under mechanical agitation, such an alloy slurry exhibits thixotropy in that the viscosity decreases with increasing shear rate. This effect appears to be time-dependent and reversible. Increasing the particle residence time in the slurry promotes wetting and bond formation due to chemical reactions at the interface. This semifusion process reduces deformation resistance and allows near-net shape fabrication by extrusion or forging.

Infiltration Processes In the infiltration technique, liquid metal is infiltrated through the narrow crevices between the fibers or particulate reinforcements, which are arranged in a preform to fix them in space (Ref 13, 14). Unlike the stir-mixing process where the reinforcements are free to float or settle in the melt, the reinforcements do not have freedom to move in the infiltration processes. As the liquid metal enters between the fibers or particles during infiltration, it cools and then solidifies, producing a composite. In general, the infiltration technique is divided into three distinct operations: the preform preparation (the reinforcement elements are assembled together into a porous body), the infiltration process (the liquid metal infiltrates the preform), and the solidification of liquid metal. In this method, a transient layer of solidified metal can form as soon as the liquid metal comes in contact with the cold fiber. Melt infiltration can be achieved with the help of mechanical pressure (Ref 15), inert gas pressure, or vacuum (Ref 16, 17). Techniques of pressureless infiltration have also been developed (Ref 18, 19), along with electromagnetic and centrifugal infiltration. The fundamental phenomena involved in infiltration processes are surface thermodynamics, surface chemistry, fluid dynamics, and heat and solute transport. The processing variables governing the evolution of microstructures are:     

Fig. 6

(a) Schematic view of stir-casting process. Courtesy of N. Chawla. Microstructures of typical particulate cast metal-matrix composites made by stir casting: (b) Al-Si/20 vol% Grp, (c) Al-Si/20 vol% spherical Al2O3p, made by Comalco Australia, and (d) Al-SiCp, made by Duralcan Corporation

lubrication. This technique is being used to produce selectively reinforced brake rotors, where the hard SiC reinforcement particles segregate on the outer surface of the component, improving the wear resistance. Compocasting. Particles and discontinuous fibers of silicon carbide, titanium carbide,

alumina, silicon nitride, graphite, mica, glass, slag, magnesium oxide, and boron carbide have been incorporated into vigorously agitated, partially solidified aluminum alloy slurries by the compocasting technique. The discontinuous ceramic phase is mechanically entrapped between the proeutectic phase present in the

Fiber preheat temperature Metal superheat temperature Interfiber spacing Infiltration pressure Infiltration speed

If the melt or fiber temperature is too low, poorly infiltrated or porous castings will be produced. If the temperatures are too high, excessive fiber/metal reaction will degrade casting properties. Finally, if the plunger speed is too fast, the preform can be deformed on infiltration. A threshold pressure is required to initiate melt flow through a fibrous preform or powder bed. This pressure begins in the thermodynamic surface energy barrier because of nonwetting. The threshold pressure increases with increasing infiltration rates because wetting is a function of time. Pressure Infiltration Casting. Various pressure infiltration techniques have been used to manufacture cast composites. In pressure

Synthesis and Processing of Cast Metal-Matrix Composites and Their Applications / 1153 infiltration casting, the mold and the preforms are kept at a temperature close to the melting point of the metal; this gives the metal time to flow and fill the mold without choking (Fig. 7). Once the metal has filled major voids in the mold and preform, the pressure can be increased rapidly. After this, a uniform pressure can be maintained through the mold, and the remaining voids can be infiltrated isostatically. This makes it possible to use relatively thin-

walled molds and to infiltrate very fragile preforms with little damage. Even though the mold and preform start out at a high temperature, use of thin mold walls allows the mold to be cooled rapidly to minimize matrix-reinforcement interfacial reaction. Squeeze Casting. In the pressure infiltration process, the liquid metal is initially under hydrodynamic pressure, which changes to hydrostatic pressure at the conclusion of preform infiltration but prior to solidification. On the other hand, during squeeze casting a premixed suspension of chopped fibers, whiskers, or particles in the matrix melt is solidified under a large hydrostatic pressure with virtually no movement of the liquid (Fig. 8) (Ref 1). Recently, squeeze casting was developed to involve unidirectional pressure infiltration of fiber preforms or particle beds in order to produce void-free, near-net shape castings of composites. Aluminosilicate fiber-reinforced aluminum alloy pistons for use in heavy diesel engines have been produced using squeeze casting. A considerable amount of work has been done on the squeeze casting of ceramic particle

Fig. 7

Formation of composite during nonisothermal infiltration of a fiber preform by liquid metal

Fig. 8

Schematic representation of squeeze casting process. Courtesy of N. Chawla

and fiber-reinforced MMCs for industrial applications. The fibers or particles exert a considerable influence on grain size, coarsening kinetics, and microsegregation in the matrix alloy. Squeeze casting promotes a fine, equiaxed grain structure because of large undercoolings and rapid heat extraction. Vacuum infiltration is effected by creating a negative pressure differential between the preform and the surroundings, which drives the liquid through preform interstices against the forces of surface tension, gravity, and viscous drag. Vacuum infiltration is usually done in conjunction with chemical methods of wettability enhancement, for example, fiber surface modification or addition of wetting-enhancing elements to the matrix alloy. Vacuum infiltration has been used by AVCO Specialty Materials (Textron) to infiltrate silicon-coated SiC monofilaments with aluminum, and by DuPont to infiltrate alumina fiber preforms with an aluminum-lithium alloy. Electromagnetic Infiltration. The use of electromagnetic forces to infiltrate liquid metal into nonwetting preforms can dispense with the need for application of inert gas pressure and mechanically driven rams (Ref 20). Alumina fiber/aluminum composites have been fabricated using the electromagnetic infiltration technique. In this technique, the metal is subjected to a high-frequency electromagnetic field that interacts with the eddy currents produced within the metal. The interaction results in infiltrating the metal into the preform when the preform is suitably oriented with respect to the direction of the force. In a typical experimental design, a fiber preform is immersed in liquid metal contained in a ceramic crucible. A battery of capacitors is used to produce intense bursts of high-frequency axial magnetic field. A flux concentrator intensifies the magnetic field around the crucible. The entire assembly is configured such that body forces produced within the metal are oriented radially inward to enable melt ingress into the preform. Porosity, preform deformation, and fiber breakage are absent in the castings at the completion of infiltration and solidification. Aluminum-matrix composites reinforced with up to 25 vol% alumina fiber (Saffil, Saffil Ltd.) have been successfully produced using this technique. Centrifugal Infiltration. Infiltration can be achieved using centrifugal force to fabricate MMCs. High rotational speeds are able to generate enough centrifugal force to overcome the capillary forces for melt penetration and viscous forces for metal flow in the preform. Aluminum-matrix composites containing fly ash and carbon microballoons were successfully fabricated by using centrifugal infiltration (Ref 21). The microballoons were packed in an alumina tube, which acts as a crucible for the molten metal poured onto the powder bed. The infiltration and dispersion of powders were carried out at rotational speeds of up to 4000 rpm, and the rotation was continued until the solidification was completed. During rotation,

1154 / Nonferrous Alloy Castings the centrifugal force pressurizes the molten metal into the interstices of the particle bed. As the rotation is continued through the solidification process, the particles are “frozen-in” at their final position in the casting. Pressureless Spontaneous Infiltration. Spontaneous infiltration of preforms or permeable bodies of reinforcements by metals can be effected if the compositions of the melt and preform, temperatures, and gas atmosphere are controlled such that good wetting conditions are achieved. A cost-effective pressureless infiltration process to fabricate void-free, net-shape MMCs having high structural integrity has been developed by Lanxide Corporation, USA. The technique permits fabrication of composite components containing a wide range of reinforcements. Near-net shape SiC-reinforced aluminum-matrix composite components for electronic applications such as heat sinks have been synthesized by using pressureless infiltration of preforms. In the directed-metal oxidation process (a form of the Lanxide process), an ingot of aluminum alloy containing magnesium is placed on top of a permeable body of reinforcement or preform contained within a refractory vessel. The entire assembly is heated to a suitable temperature in an atmosphere of a free-flowing, nitrogen-bearing gas mixture. Spontaneous infiltration of the permeable reinforcement is assisted by the presence of magnesium in the alloy and an oxygen-free atmosphere. In a modified pressureless metal infiltration process (called PRIMEX, M Cubed Technologies, Inc.), magnesium in the matrix alloy volatilizes and reacts with nitrogen in the atmosphere to form a thin Mg3N2 coating on the reinforcement surface that makes the reinforcement system wettable. During infiltration, Mg3N2 reacts with aluminum to form aluminum nitride (AlN) and magnesium, which is released in the alloy. The AlN could be formed through direct reaction of nitrogen in the atmosphere with liquid aluminum. The A1N phase is present mainly as small precipitates in the composite, although thin films of this compound are also found as coating on the reinforcement surface. The spontaneous infiltration step can be followed by a mixing step prior to casting. Advanced Pressure Infiltration Casting. MMCs containing hard reinforcements such as SiC are difficult to machine. Therefore, development of near-net shape components of MMCs is desirable. Generally, pressure infiltration technology is directed to producing nearnet shape components. Special preforms can be prepared for these MMCs. These preforms have fine interstices through which the liquid metal is infiltrated. The preforms are encapsulated in a graphite mold and placed in a furnace. Liquid metal is poured and pressure is applied, causing the melt to pass through the mold and infiltrate the preform. Figure 9 shows alumina preforms of piston connecting rods developed by MMCC, Inc. and Fig. 10 schematically shows the advanced pressure

Fig. 9

Fig. 10

Alumina preforms printed at MMCC, Inc.

Schematic of the advanced pressure infiltration casting (APIC, MMCC, Inc.) process

infiltration casting (APIC, MMCC, Inc.) process that can use such preforms (Ref 5). The main feature of this process is that the melting of the matrix metal and the preheating of the preform take place outside of the autoclave. This technique is now available to produce near-net shape pressure-infiltrated parts of MMCs.

 Influence the viscosity and fluidity of the

Effects of Reinforcement Present in the Liquid Alloy on Solidification

  

The solidification microstructure of the matrix alloy is affected by the presence of reinforcement phase. The reinforcement can:  Act as a solute and diffusion barrier within

the melt

 Catalyze heterogeneous nucleation in the

melt

 Influence the conductivity of the liquid

melt and impart thixotropic properties

 Restrict fluid convection due to restriction of   

liquid in narrow interstices between fibers or particles Settle or float due to density differences Influence morphological instabilities, including planer-to-dendritic solidification Be pushed or entrapped ahead of the solidifying interface Influence dendrite arm coarsening Influence grain size Reduce the amount of latent heat to be extracted out of the casting

The aforementioned effects have been discussed in several studies. Particles have been shown to settle or float in the melt and are frequently pushed in the last freezing interdendritic liquid by the growing solid-liquid interface (Ref 22–24). The effect of reinforcements on solidification processes, more specifically, on

Synthesis and Processing of Cast Metal-Matrix Composites and Their Applications / 1155

Fig. 12

Solidification microstructure of discontinuously reinforced Al-SiC alloy composites showing the influence of spacing between SiC platelets on the microsegregation pattern in aluminum-copper alloys

Fig. 11

(a) Graph of dendrite arm spacing (DAS) versus solidification time. (b), (c), to (d) Microstructural formation in the interfiber regions of an Al-4.5Cu-alumina fiber composite. The solidification times will vary with material.

dendrite formation and microsegregation, is discussed in some detail. Figure 11 shows the effect of alumina fiber spacing on the microstructure of Al-4.5% Cu alloy (Ref 25). The primary phase morphology and distribution of second phases depend on the relative magnitudes of dendrite arm spacing (DAS) and the interfiber spacing. Figure 11 shows that when the DAS is less than the interfiber spacing, a normal solidification microstructure is observed, and the fiber does not alter the microstructure. When the DAS is equal to the interfiber spacing, the second phase, that is, Al-CuAl2 eutectic, segregates onto the fiber surface. Finally, when the DAS is more than the interfiber spacing, the dendritic morphology is ill-defined, and there is less precipitation of secondary phase. In this case, the fibers act as barriers to solute precipitation, and, as a result, the concentration of solute in aluminum will be more than the equilibrium value. This is important because it is possible to create microstructures with no dendrites and reduce microsegregation, resulting in decreasing homogenization time during heat treatment. Figure 12 shows the microstructure of SiC platelets incorporated in the matrix of an

aluminum-copper alloy by pressure infiltration. It shows frequent enrichment of Al-CuAl2 eutectic at the SiC platelet-matrix interface as a result of nucleation of the a-phase away from the platelets (Ref 26). Figure 12 also shows significant effects of platelet spacing on microsegregration. It is clear from Fig. 11 and 12 that reinforcements alter the solidification microstructure of alloys to a great extent, due to the restriction of spaces between which the liquid solidifies. This can be exploited to create improved microstructures in composites.

Property Motivation for Using MMCs In MMCs, the second phases (fibers, whiskers, particles) are incorporated in metals or alloys to achieve properties that are not obtainable in conventional monolithic materials. Desired improvements in mechanical, tribological, electrical, and thermal properties can be achieved by intelligently selecting the reinforcement materials and their size, shape, and

volume fraction. Table 1 shows the selected reinforcements and metallic matrices commonly used for synthesis of MMCs. Figure 13 shows a plot of specific modulus versus specific strength for a number of materials (Ref 27). Ideally, one would like a material to be near the upper right-hand corner of the graph, where one can obtain a combination of high specific strength and high specific modulus to minimize the weight of a component. It is clear that MMCs provide a better combination of specific strength and modulus compared to monolithic alloys of aluminum, magnesium, copper, nickel, and iron. Only a few materials are better than reinforced-aluminum composites. These include graphite-epoxy composites (along the fiber direction), pure ceramics, and diamond, although they are more expensive and some of them are brittle. Figure 14 shows that the specific cost of organic composites is much higher than MMCs for similar specific stiffness (Ref 27). In many applications, the resistance to distortion from thermal or mechanical loads is important, especially in precision devices. The higher the stiffness and thermal conductivity of the composite and the lower the coefficient of thermal expansion (CTE), the higher will be its resistance to mechanical and thermal distortion.

1156 / Nonferrous Alloy Castings

Fig. 14

Fig. 13

Specific modulus and specific strength of various structural materials

Specific cost of various structural materials in primary product forms

ceramic phase in metallic matrices improves the friction and wear behavior. Figure 17 shows the wear resistance of A356 alloy and A356SiC composite against 4140 steel. The wear loss of the composite is approximately 10 times less than that of the matrix alloy. The wear resistance of aluminum-matrix composites is better than that of cast irons, which are much heavier. Further improvement in the wear resistance of alloys can be achieved by developing hybrid composites reinforced with hard phases and soft-phase lubrication (Ref 29). For example, the transition from mild to severe wear occurred in the range of 225 to 230  C (435 to 445  F) for A356 alloys (Ref 30, 31). With the addition of 20 vol% SiC to the A356 alloy, the transition temperature increased to 400 to 450  C (750 to 840  F). A hybrid A356-matrix composite containing 20 vol% SiC and 10 vol% graphite remained in a mild wear regime even at the test temperature of 460  C (860  F).

Components of MMCs Currently in Use

Fig. 15

Materials selection chart for resistance to mechanical and thermal distortions

Figure 15 shows the materials selection chart for resistance to mechanical and thermal distortions (Ref 27). It is observed that the closer a material is to the top right-hand corner of the chart, the higher will be its resistance to thermal and mechanical distortion. This figure shows that MMCs have one of the highest resistances to mechanical and thermal distortion, with the exceptions of beryllium and diamond. MMCs are also becoming very popular for thermal management applications (Ref 28). For such applications, the CTE, thermal conductivity, and density are the important properties. Figure 16 shows a plot of CTE versus specific thermal conductivity. It is desired that for a given

CTE value, the thermal conductivity should be as high as possible. This figure shows that for the allowable range of thermal expansion for thermal management in electronic packaging, MMCs have one of the highest specific thermal conductivities, making them a desirable choice. In fact, only beryllium-containing materials and diamond are better than MMCs. Therefore, an increasing role of MMCs in thermal management applications is expected. The hardness of MMCs is significantly higher than that of the unreinforced-matrix alloys. Hardness increases with increasing volume fraction of reinforcement and is affected by processing variables. Also, the addition of

MMCs, in view of their excellent combination of properties such as high levels of strength and stiffness, improved thermal stability, low CTE, and improved wear and seizure resistance, have already found several applications. It is estimated that in the year 1999, the worldwide composite use was approximately 2.5 million kg (2750 ton) and increased to approximately 3.5 million kg (3850 ton) in the year 2004 (Ref 32, 33). The annual growth rate of MMCs has been approximately 6% in recent years (Ref 34). One of the reasons for limited use of MMCs is their cost. The higher cost of MMCs compared to the corresponding base alloy is due to the higher cost of fiber, more rigorous processing methods, and the difficulty in machining. However, in recent times, decreasing cost of fibers, modification of conventional metal-processing methods for processing composites, and the development of net or nearnet shape components are contributing to the decrease in the cost of MMC components.

Synthesis and Processing of Cast Metal-Matrix Composites and Their Applications / 1157

Fig. 16

Fig. 17

Volume loss of aluminum alloy, aluminum composite, and cast iron during sliding wear

Fig. 18

An Al-SiCp composite brake rotor developed for the Chrysler Prowler

Materials selection chart for thermal management applications

Table 2 Year

Use of metal-matrix composites (MMCs) in automotive applications Company

1980

Toyota

1990

Honda

1992 1996 1997 1997 1997 1997 1997 1997 1998 1998

Toyota Toyota Toyota Toyota Porsche GM GM GM Honda GM

1998 1999 1999 1999 2000 2000

Chrysler Mazda Lotus Volkswagen Honda Toyota

Composite and component

First commercially produced MMC diesel piston by squeeze infiltration in ceramic preform Four-cylinder 2.2 L MMC engine block selectively reinforced with mixture of Saffil, alumina, and graphite fibers Selectively reinforced low Coefficient of thermal expansion fiber engine pulley Alumina-boria whisker-reinforced piston Al/SiCp front brake rotors for an electric vehicle (RAV4-EV) Heat-spreader plates of an electronic cooling device (Prius hybrid vehicle) Porsche Boxter selectively reinforced cylinder bores Al-SiCp MMC engine cradle Extruded aluminum MMC propshaft on the Corvette Al-SiCp driveshaft for extended S/T trucks Produced selectively reinforced bore V6 engine for Acura NSX Used several MMC components (rear brake drum, electronic cooling package) for GM electric vehicle (GM-EVI) Used Al-SiCp brake rotors for its Prowler model Produced fiber/particle MMC piston Released Al-SiCp MMC brake rotors in Elise Low-fuel 3 L version of the Lupo model with rear-wheel Al-SiCp brake drums Bore-reinforced in-line four-cylinder engine for its S2000 roadster In-line four-cylinder engine for high-performance Celica with MMC bores

Source: Ref 35

There has been a dramatic decrease in the cost of particle-reinforced MMCs primarily due to the low cost of stir-mixing and casting processes and the use of inexpensive particles.

MMCs in Automotive Industries A number of automotive components, including pistons, cylinder liners, brake rotors, and connecting rods, have been made of aluminum-matrix composites. Table 2 shows some significant early developments of composite automotive components (Ref 35). Properties of interest to the automotive engineer include increased specific stiffness, wear resistance, and improved high-cycle fatigue resistance. While weight-savings is also

important in automotive applications, the need to achieve performance improvements with much lower cost premiums than tolerated by aerospace applications drives attention toward low-cost materials and processes. The raw material cost of cast iron or steel is less than $0.82/kg ($0.37/lb), and discontinuously reinforced cast aluminum ingot currently costs approximately $4.0 to 5.0/kg ($1.8 to 2.3/lb) for large-scale production, so the raw material cost of MMCs is higher than that of the material typically replaced. However, the MMC component would typically weigh 60% less than steel, significantly improving the cost comparison. Examples of MMCs in successful automotive applications, in which the combination of properties and cost satisfied particular needs, are discussed as follows.

The first commercially made automotive MMC component was a diesel engine piston produced by Toyota, Japan, in the early 1980s (Ref 6). The piston was produced by squeeze casting technology, in which liquid aluminum was infiltrated in alumina-silica short fibers. The ceramic preform was selectively placed in the ring portion of the piston. Later on, this technology was refined by blending nickel powder into an alumina fiber preform. In this method, nickel powder reacted with aluminum and formed NiAl3, which improved the adhesive wear resistance of the ring (Ref 6). Selectively reinforced aluminum composite diesel engine pistons have now become very popular throughout the world. In fact, this has triggered a flurry of activities in making other engine components from cast composites. Further, Toyota developed another MMC piston in which the preform was made of alumina-boria whisker and was selectively reinforced in the lip of the combustion bowl of the piston. In 1999, Mazda began producing pistons of fiber/particle-reinforced MMCs. Another major use of Al-SiCp composite is for brake rotor applications. Testing has shown that composite brake rotors are as effective as cast iron rotors in braking applications, while having a higher capacity for heat dissipation and 50 to 60% lower weight. Figure 18 shows an Al-SiCp composite brake rotor used in the Chrysler Prowler. Toyota has also used A359/ 20 vol% SiCp composite brake rotors in their electric vehicle (RAV4-EV), as shown in Fig. 19 (Ref 36). The Lotus Elise was also released with

1158 / Nonferrous Alloy Castings

Fig. 19

An A359/20 vol% SiCp composite brake rotor for a Toyota electric vehicle

Fig. 20

Aluminum-SiC composites as heat-spreader plates of an electronic cooling device for the world’s first hybrid vehicle, the Prius

Fig. 22

Aluminum-SiC composite components. (a) Automobile brake drum. (b) Apex insert for

cyclone

Fig. 21

Metal-matrix composite (MMC) crankshaft pulley made by infiltration of SIALON preform with aluminum

Al-SiCp composite brake rotors. The Volkswagen low-fuel version of the Lupo model also had AlSiCp composite brake rotors in the rear wheels.

Aluminum-SiC composites were found to be very effective as heat-spreader plates of an electronic cooling device for the world’s first

hybrid vehicle, the Prius (Fig. 20) (Ref 36). Toyota has also developed a cylinder block with MMC bores for a high-performance spark ignition engine. The ceramic preform was made from alumina-silica short fibers filled with mullite particles. The MMC bore was made by a laminar flow die casting process. This cylinder block, with MMC cylinder bore, was introduced in the 2000 model Celica sports car. Toyota developed an MMC crankshaft pulley that has a low CTE and does not expand away from the shaft during thermal cycling (Fig. 21). In this application, a ceramic preform was made of SIALON (Si-Al-O-N) and was infiltrated with aluminum. Figure 22 shows prototypes of an automobile brake drum and a refrax apex insert made of Al-SiCp composites produced at the laboratories of Council of Scientific and Industrial Research in Bhopal, India. There is a large market for cast composites in Asia, where the energy costs are very high, and replacing energy-intensive materials such as aluminum

Synthesis and Processing of Cast Metal-Matrix Composites and Their Applications / 1159

Fig. 23

Boron-aluminum composite tubular struts used as the frame and ribs in space shuttle Orbiter. Struts are not cast but represent a category of metal-matrix composite applications.

Fig. 25

Discontinuously reinforced aluminum metalmatrix composites. (a) Multi-inlet SiCp-Al truss node. (b) Grp-Al composite components for electronic packaging applications

primary motivation for the insertion of MMCs into aerospace applications is the excellent balance of specific strength and stiffness offered by MMCs relative to competing structural materials. The SiC particle-reinforced MMCs are used in the Lockheed F-16 fighter aircraft. For instance, the use of MMCs reduces the maximum deflection of the leading-edge tip of the ventral fin in an F-l6 aircraft by over 50%, reducing the aerodynamic loads and increasing the service life from 400 to 8000 h. Aluminum alloy skins used in the honeycomb structure of the fin were replaced by 6092 Al-17.5 vol% SiCp composite sheet material. The service life of the skins was significantly increased because of the increased specific stiffness of the aluminum-matrix composite material (Ref 37). Although not produced by casting, one of the major space-based applications of continuous fiber-reinforced MMCs is structural tubes for space shuttles. Figure 23 shows space shuttle Orbiter’s midfuselage main frame MMC tubes (Ref 38). These tubes were made of 50 vol% B fiber-reinforced aluminum-6061 alloy composites, produced by the diffusion-bonding technique. There are 243 MMC tubes of 25 to 92 mm (1 to 3.6 in.) diameter and 0.6 to 2.4 m (2 to 8 ft) length. The MMC tubes saved approximately 145 kg (320 lb) over the initial design of aluminum tubes. Another major development is the application of MMCs in an antenna wave-guide mast on the Hubble Space Telescope (Fig. 24) (Ref 38). The mast is made by infiltration technology using a preform containing 40% P-100 carbon fibers. Aluminum alloy 6061 was used as the matrix material. The mast is rectangular and has dimensions of 43 by 86 by 2000 mm (1.7 by 3.4 by 79 in.) (Ref 38). Cast Al-SiCp and Al-SiCw composites have also found applications as multi-inlet truss nodes and electronic packaging (Fig. 25) (Ref 38). General Electric Company designed an advanced-composite optical system gimbal, using 6061–40 vol% SiCp composite in selected areas, requiring low expansion and high wear resistance.

MMCs for Electronic Packaging Applications Fig. 24

The P-100/6061 aluminum high-gain antenna wave guides/boom for the Hubble Space Telescope (HST). (a) Before integration in the HST. (The ruler is 12 cm, or 5 in., long.) (b) On the HST as it is deployed in low-Earth orbit from the space shuttle Orbiter

with particulate reinforcements can result in considerable energy-savings (Ref 4).

MMCs for Space Applications The extreme environment in space presents both a challenge and an opportunity for material scientists. In the near-Earth orbit, typical

Fig. 26

Aluminum-SiC microwave radio frequency packaging for communication satellites

spacecrafts encounter naturally occurring phenomena such as vacuum, thermal radiation, atomic oxygen, ionizing radiation, and plasma, along with factors such as micrometeoroids and human-made debris. Therefore, aerospace applications were the original driving force for the development of MMCs. This development was due to the quest for dimensionally stable structures and weight reduction for improved performance and payload capability. The

MMCs are also finding applications in spacecraft thermal management systems. Cast MMCs are very suitable for thermal management applications due to their high conductivity and low CTE. Figure 26 shows a typical radio frequency packaging for microwave transmitters used in commercial low-Earth orbit communication satellites (Ref 27). This Al-SiC composite part is cast to near-net shape. Aluminum-SiC is not only lighter than the existing materials, but it is much more affordable. The PRIMEX pressureless melt infiltration process is used to produce high-volume-fraction SiC-reinforced aluminum composites for such applications.

1160 / Nonferrous Alloy Castings MMCs consisting of carbon fibers having high conductivity are also being employed in packaging and thermal management applications. There are relatively few applications to date of MMCs reinforced with continuous fibers, because they are relatively expensive. However, recent developments in fabrication methods have resulted in significant cost reductions for discontinuous-fiber MMCs, and these materials are now entering production lines.

Development of Select Cast MMCs Aluminum-Graphite Composite Cast aluminum-graphite particulate composites have been used in several applications, such as pistons, engine cylinder liners, and connecting rods, as shown in Fig. 27. These are the forerunners of aluminum-graphite engines being developed for antiseizing applications (Ref 12).

Lead-Free Copper Alloy/Graphite Composites Since lead is banned in a number of copper alloys for bearing and plumbing applications, lead-free copper-graphite composites have been developed as a substitute. Graphite particles have been shown to impart machinability and lubrication characteristics similar to lead at much lower costs and without the associated toxicity. Graphite is also much cheaper and abundantly available compared to bismuth and selenium, which are being proposed as alternatives to lead. Figure 28 shows a collection of plumbing fixtures and bearings cast in coppergraphite composites at the University of Wisconsin-Milwaukee. By centrifugally casting copper alloy/graphite composite, the graphite particles can be concentrated on the inner periphery, where they selectively reinforce and provide solid lubrication for bearing applications. Figure 29 shows the microstructure from the cylinder graphite-rich zone near the inner periphery, which shows dendrites of copper and graphite particles present in the interdendritic regions (Ref 39). A comparison of wear losses of copper-graphite and lead-copper against 1040 steel reveals that the coppergraphite composite shows much less wear than the lead-copper alloy, demonstrating its suitability for bearing applications (Ref 39).

Hybrid Composites

Fig. 27

Components made of aluminum-graphite composites. (a) Cylinder. (b) Cylinder liner. (c) Piston. (d) Connecting rods

Figure 30(a) shows hybrid composites containing both small SiC particles, which are hard, and large graphite particles, which are soft, in the matrix of a cast aluminum alloy. The production process is simplified by the simultaneous presence of two types of particles (the lighter graphite and heavier SiC). Together, they result in a buoyancy-neutral mixture where SiC is unable to settle down (due to the graphite, which is lighter and is trying to float up), and the graphite particles are unable to float up (due to the hindering effect of the SiC, which is trying to settle down). Tests at International Nickel Company and other

Fig. 29

Fig. 28

Montage of lead-free copper-graphite composite castings

(a) Schematic view of centrifugally cast copper-graphite alloy. (b) Microstructure of the graphite-rich zone

Synthesis and Processing of Cast Metal-Matrix Composites and Their Applications / 1161

Fig. 32

(a) Microstructure of A356–10 vol% fly ash composite. (b) Intake manifold made of such

a composite

Fig. 30

(a) Microstructure of Al-SiC-graphite hybrid composite. (b) Cylinder liners made out of hybrid composites. (c) Hybrid composite disc brake. (d) Hybrid composite brake rotor

organizations have shown that the wear resistance of these hybrid Al-SiC-graphite composites is greater than that of cast iron, and they are much easier to machine compared to Al-SiC composites. Figure 30(b–d) show cylinder liners, a brake rotor, and a disc brake, all made from aluminum-hybrid composites that are being tested as a replacement material for much heavier cast iron components (Ref 40). Several other hybrid composites, including Al-aluminum-graphite, have been developed, and their applications are being explored.

Fly-Ash-Filled Syntactic Foams

Fig. 31

(a) Fly ash cenospheres. (b) Fly ash cenospheres in a-aluminum matrix

In recent years, techniques have been developed to introduce and disperse fly ash into molten metal, resulting in castings with reduced weight, cost, and CTE and higher abrasion resistance than the matrix alloy. Such composites containing hollow particles in their structure are called syntactic foams. Fly ash is a readily available waste by-product of coal combustion, and its use as a filler can result in cost reduction of MMCs because it replaces the more expensive matrix alloy in the composite. Fly ash is already used as an additive to cement, but most of it is still disposed of in landfills. Syntactic foams can be made by pressure infiltration (Ref 41). Figure 31(a) shows cenospheres (hollow particles of fly ash), which are packed loosely in a tube and infiltrated by aluminum alloy to create aluminum/fly ash syntactic foam. Figure 31(b) shows the microstructure

of cenospheric fly ash dispersed in a cast aluminum matrix. The density of this cast aluminumbase foam is 1400 kg/m3 (87 lb/ft3) (approximately half the density of the matrix alloy), which demonstrates the potential of reducing the weight of aluminum by the incorporation of cenospheres. This opens up the possibility of also producing syntactic foams with other types of hollow particles, not just fly ash, in foundries. These foam materials are likely to have very high damping capacity and energyabsorption characteristics of interest to automotive industries. Figure 32(a) shows fly ash particles dispersed in a matrix of aluminum-silicon alloy produced by stir mixing and casting from a 180 kg (400 lb) melt. From such melts, it has been possible to make a variety of prototype castings, including the sand cast intake manifold shown in Fig. 32(b). This intake manifold is made of aluminum-10% fly ash and is lighter and cheaper than the conventional aluminum alloy manifolds. Figure 33 shows other sand castings, pressure die castings, and squeeze castings made from aluminum/fly ash melts, demonstrating the castability of such melts using a variety of casting processes (Ref 4). Fly ash has been filled in a variety of other matrices, including lead and magnesium, to synthesize lightweight composites.

Cast Metallic Foams The problem of stabilization of gas bubbles by the reinforcement phase in composite melts can be converted into an opportunity to create new materials (Ref 42). Figure 34 shows a gas

1162 / Nonferrous Alloy Castings

Fig. 35

Fig. 33

Sand casting, pressure die casting, and squeeze casting of aluminum/fly ash composite. (a) Sump pump cover. (b) Greenlee hydraulic tool handle cast at Eck Industries. (c) Mounting bracket die cast at Eck Industries. (d) Motor mounts cast at Thompson Aluminum

bubble stabilized by copper-coated carbon fibers in an aluminum melt (Ref 43). The copper-coated carbon fibers have stabilized the gas bubble in the melt, creating porosity, which results in making a stable foam structure. A similar phenomenon is observed in Al-SiCp melts, which helped in synthesizing composite foams (Ref 44, 45). The structure of an Al-SiC composite foam is shown in Fig. 35 (Ref 45). These types of foams from melts of MMCs have potential uses in energy-absorbing applications in automotive and aerospace structures.

Research Imperatives

Fig. 34 fibers

Gas bubble stabilized in aluminum alloy melt by the presence of copper-coated carbon

There are still many research challenges in the area of cast MMCs. To date, researchers have incorporated fibers or particles mainly in conventional monolithic alloy matrices, but have not been able to obtain the best advantage of cast MMCs. It is necessary to develop special alloys to serve as matrix materials in cast MMCs. These alloys can be specially tailored to bond to the reinforcement by having favorable thermodynamics and kinetics for desired interfacial reactions. There is a need to develop special particulate reinforcements, including surface-treated particles, for cast MMCs. There has been a significant amount of work on developing special fibers as reinforcements, but comparatively very little work has been done to

Foam material created by introducing gas in an Al-SiC melt

develop particles specially suited for cast MMCs. For the most part, SiC that was developed for grinding applications is being incorporated in aluminum matrices to make cast MMCs. There is a need to develop techniques for rapid infiltration of preforms, including techniques for rapid pressureless infiltration. Current pressureless infiltration processes take a long time, while costly tooling is required for pressure infiltration. There is further need to decrease the cost of synthesis of cast MMCs to the point where their life-cycle cost, as well as their initial cost, is lower than the existing materials. Attempts should be made to develop the use of low-cost reinforcements such as fly ash in cast MMCs to bring the cost of cast MMCs below monolithic alloys. Applications of such lightweight composites in the transportation industry can result in the conservation of significant amounts of energy. There is also a need to enhance the ductility and fracture toughness of cast MMCs through the use of special matrices, reinforcements, and processing techniques. Basic research is needed to promote nucleation of the primary phase on reinforcements so that the particles or fibers will be encapsulated within the grain or dendrite and can act as grain refiners. Research is needed to promote engulfment of particles by growing dendrites to obtain a more uniform distribution of particles in the microstructure of cast MMCs. This is likely to increase ductility and fracture toughness of cast MMCs. Techniques to cast thin-sectioned MMC components need to be developed for electronics packaging. Low-cost techniques for machining cast MMCs, especially for drilling and tapping, also need to be developed. Techniques for near-net

Synthesis and Processing of Cast Metal-Matrix Composites and Their Applications / 1163 shape production of components are needed. Nondestructive techniques must be developed to characterize the percentage of reinforcements and porosity in cast MMC components for facilitating quality control and ensuring uniformity in successive components. Techniques for degassing and grain refining composite melts as well as technologies for recycling of cast MMCs are needed to either recover the matrix alloy or reuse the composite itself. Techniques to improve the distribution and bonding of reinforcements and the matrix are required. In addition, it is necessary to ensure that similar distribution and bonding is achieved in successive components during bulk manufacture. Porous and cellular materials are receiving a great deal of attention, both in automotive and space applications, and can be produced using MMCs. Polymer-matrix cellular composites have been developed as functionally graded materials, nanocomposites, smart composites, and biomedical implants. Similar microstructures can also be developed in MMCs, which can have additional advantages related to the matrix metals. ACKNOWLEDGMENTS The authors would like to express gratitude to the National Science Foundation for funding under the grant CMMI-0726723. The help of Benjamin F. Schultz in preparing the manuscript is gratefully acknowledged. The authors thank David Weiss, Daniel Miracle, Suraj Rawal, James Cornie, Warren Hunt, Nikhilesh Chawla, and others for providing figures. REFERENCES 1. N. Chawla and K.K. Chawla, Metal Matrix Composites, Springer-Verlag, New York, 2006 2. J.M. Kunze and C.C. Bampton, Challenges to Developing and Producing MMCs for Space Applications, JOM, Vol 53, 2001, p 22–25 3. P. Rohatgi, Cast Aluminum-Matrix Composites for Automotive Applications, JOM, Vol 43, 1991, p 10–15 4. P.K. Rohatgi, D. Weiss, and N. Gupta, Applications of Fly Ash in Synthesizing Low-Cost MMCs for Automotive and Other Applications, JOM, Vol 58, 2006, p 71–76 5. J.S. Shelley, R. LeClaire, and J. Nichols, Metal-Matrix Composites for Liquid Rocket Engines, JOM, Vol 53, 2001, p 18–21 6. T. Suganuma, and A. Tanaka, Application of Metal Matrix Composites to Diesel Engine Pistons, J. Iron Steel Inst. Jpn., Vol 75, 1989, p 1790–1797 7. F.A. Badia and P.K. Rohatgi, Dispersion of Graphite Particles in Aluminum Castings Through Injection of the Melt, Am. Foundrymen Soc. Trans., Vol 77, 1969, p 402

8. M.D. Skibo, and D.M. Schuster, Process for Preparation of Composite Materials Containing Nonmetallic Particles in a Metallic Matrix, and Composite Materials Made Thereby, U.S. Patent 4,786,467, 1988 9. J.U. Ejiofor and R.G. Reddy, Developments in the Processing and Properties of Particulate Al-Si Composites, JOM, Vol 49, 1997, p 31–37 10. L. Ma, F. Chen, and G. Shu, Preparation of Fine Particulate Reinforced Metal Matrix Composites by High Intensity Ultrasonic Treatment, J. Mater. Sci. Lett., Vol 14, 1995, p 649–650 11. Y. Yang, J. Lan, and X. Li, Study on Bulk Aluminum Matrix Nano-Composite Fabricated by Ultrasonic Dispersion of NanoSized SiC Particles in Molten Aluminum Alloy, Mater. Sci. Eng. A, Vol 380, 2004, p 378–383 12. P.K. Rohatgi, R. Asthana, and S. Das, Solidification, Structures and Properties of Cast Metal-Ceramic Particle Composites, Int. Met. Rev., Vol 31, 1986, p 115–139 13. D.H. Baldwin, and R.L. Sierakowski, Fracture Characteristics of a Metal Matrix Composite, Composites, Vol 6, 1975, p 30–34 14. A. Mortensen, J. Cornie, and M. Flemings, Columnar Dendritic Solidification in a Metal-Matrix Composite, Metall. Mater. Trans. A, Vol 19, 1988, p 709–721 15. A.J. Cook and P.S. Werner, Pressure Infiltration Casting of Metal Matrix Composites, Mater. Sci. Eng. A, Vol 144, 1991, 189–206 16. Y. Kun, V. Dollhopf, and R. Kochendorfer, CVD SiC/Al Composites Produced by a Vacuum Suction Casting Process, Compos. Sci. Technol., Vol 46, 1993, p 1–6 17. J. Yang, and D. Chung, Casting Particulate and Fibrous Metal-Matrix Composites by Vacuum Infiltration of a Liquid Metal under an Inert Gas Pressure, J. Mater. Sci., Vol 24, 1989, p 3605–3612 18. M.K. Aghajanian, J.T. Burke, D.R. White, and A.S. Nagelberg, New Infiltration Process for the Fabrication of Metal Matrix Composites, SAMPE Q., Vol 20, 1989, p 43–46 19. E. Breval, M.K. Aghajanian, and S.J. Luszcz, Microstructure and Composition of Alumina/Aluminum Composites Made by Directed Oxidation of Aluminum, J. Am. Ceram. Soc., Vol 73, 1990, p 2610–2614 20. A. Evans, C.S. Marchi, and A. Mortensen, Metal Matrix Composites in Industry: An Introduction and a Survey, Kluwer Academic Publishers, 2003 21. J. Sugishita, Fabrication and Some Properties of Centrifugally Cast Surface Composites Based on Aluminum-11% Silicon Alloy, J. Jpn. Foundrymen’s Soc., Vol 57, 1985, p 102–107 22. J.W. Garvin, and H.S. Udaykumar, Particle-Solidification Front Dynamics Using a Fully Coupled Approach, Part I:

23.

24.

25.

26.

27. 28.

29.

30.

31.

32.

33.

34.

35.

36.

Methodology, J. Cryst. Growth, Vol 252, 2003, p 451–466 J.W. Garvin and H.S. Udaykumar, ParticleSolidification Front Dynamics Using a Fully Coupled Approach, Part II: Comparison of Drag Expressions, J. Cryst. Growth, Vol 252, 2003, p 467–479 J.W. Garvin, and H.S. Udaykumar, Drag on a Particle Being Pushed by a Solidification Front and Its Dependence on Thermal Conductivities, J. Cryst. Growth, Vol 267, 2004, p 724–737 A. Mortensen, M.N. Gungor, J.A. Cornie, and M.C. Flemings, Microstructures of Cast Metal Matrix Composites, JOM, Vol 38, 1986, p 30–35 P.K. Rohatgi and R. Asthana, Melt Infiltration of SiC Composites: Evaluation of Solidification Microstructure, Z. Metallkd., Vol 84, 1993, p 44–47 D. Miracle, Metallic Composites in Space: A Status Report, JOM, Vol 53, 2001, p 12–13 D.B. Miracle and B. Maruyama, “Metal Matrix Composites for Space Systems: Current Uses and Future Opportunities,” National Space and Missile Materials Symposium (Dayton, OH), 2000 D. Jun, L. Yao-Hui, Y. Si-Rong, and L. Wen-Fang, Dry Sliding Friction and Wear Properties of Al2O3 and Carbon Short Fibre Reinforced Al-12Si Alloy Hybrid Composites, Wear, Vol 257, 2004, p 930–940 S. Wilson, and A.T. Alpas, Effect of Temperature on the Sliding Wear Performance of Al Alloys and Al Matrix Composites, Wear, Vol 196, 1996, p 270–278 L. Yao-Hui, D. Jun, Y. Si-rong, and W. Wei, High Temperature Friction and Wear Behaviour of Al2O3 and/or Carbon Short Fibre Reinforced Al-12Si Alloy Composites, Wear, Vol 256, 2004, p 275–285 D. Weiss, You Want Me to Buy What? Justifying Composites in an Era of Cost Reduction, 2000 TMS Fall Meeting (St. Louis, MO), 2000, p 245–250 D. Weiss, Design of Aluminum Metal Matrix Components for the Casting Process, 2000 TMS Fall Meeting (St. Louis, MO), 2000, p 259–264 N. Gupta and W.H. Hunt, Jr., “Solidification Processing of Metal Matrix Composites: Rohatgi Honorary Symposium,” TMS Annual Meeting 2006 (San Antonio, TX), 2006 D.A. Gerard, T. Suganuma, P.H. Mikkola, and A. Mortensen, Solidification-Processed Metal Matrix Composites for the Transportation Industries, Proceedings of the Merton C. Flemings Symposium on Solidification and Materials Processing, MIT, Cambridge, MA, 2000, p 475 Y. Nii, Y. Baba, and S. Abo, Intelligent IGBT Module for the Hybrid Vehicle, Proceedings of the Conference of JSAE, 1998, p 61–64

1164 / Nonferrous Alloy Castings 37. B. Maruyama, Promise and Progress in Aluminum Composites, Adv. Mater. Process., June 1999, p 47–50 38. S. Rawal, Metal-Matrix Composites for Space Applications, JOM, Vol 53, 2001, p 14–17 39. M. Kestursayta, J.K. Kim, and P.K. Rohatgi, Friction and Wear Behavior of Centrifugally Cast Lead Free Copper Alloy Containing Graphite Particles Against Steel, Metall. Mater. Trans. A, Vol 32, 2001, p 2115–2125 40. T.F. Stephenson, A.E.M. Warner, S. Wilson, A.T. Alpas, and P.K. Rohatgi, Processing Properties and Application of Cast Metal Matrix Composites, Proceedings of the ASM Materials Week, 1996, p 337

41. P.K. Rohatgi, J.K. Kim, N. Gupta, S. Alaraj, and A. Daoud, Compressive Characteristics of A356/Fly Ash Cenosphere Composites Synthesized by Pressure Infiltration Technique, Compos. Part A: Appl. Sci. Manuf., Vol 37, 2006, p 430–437 42. T.W. Clyne, in Comprehensive Composite Material, A. Kelly, Ed., Elsevier Science, 2001 43. Z. Cao, B. Li, G. Yao, and Y. Wang, Fabrication of Aluminum Foam Stabilized by Copper Coated Carbon Fibers, Mater. Sci. Eng. A, Vol 486, No. 12, 2008, p 350–354 44. W. Jiejun, L. Chenggong, W. Dianbin, and G. Manchang, Damping and Sound Absorption Properties of Particle Reinforced Al

Matrix Composite Foams, Compos. Sci. Technol., Vol 63, 2003, 569–574 45. O. Prakash, H. Sang, and J.D. Embury, Structure and Properties of Al-SiC Foam, Mater. Sci. Eng. A, Vol 199, 1995, p 195–203

SELECTED REFERENCES  T.W. Clyne and F. Simancik, Ed., Metal

Matrix Composites and Metallic Foams, Wiley-VCH,Weinheim, Germany, 2000  S. Suresh, A. Mortensen, and A. Needleman, Ed., Metal Matrix Composites, ButterworthHeinemann, Boston, 1993

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1167-1173 DOI: 10.1361/asmhba0005340

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Approaches to Measurement of Metal Quality* Rathindra DasGupta, National Science Foundation

THE FUNCTIONALITY of aluminum alloy castings is strongly dependent on casting soundness. This implies that such castings be virtually free of hydrogen porosity and entrained nonmetallic inclusions. Negligence in reducing the dissolved hydrogen content of the melt, mitigating the formation of inclusions or removal of impurities, can result in premature failure of castings, broken machine tools, or poor mechanical properties. Figures 1 and 2 show the types of defects resulting from the use of “contaminated” metal before pouring/casting.

Metal Cleanliness Assessment Techniques Several qualitative, semiquantitative, and quantitative tests are available to estimate and control metal cleanliness, particularly inclusion concentration of aluminum alloys. A brief description of a few of the select techniques is provided below.

Chemical Analysis

(a)

(b)

Fig. 1

Inclusion (dross from melt) on machined surface led to broken machine tool. (a) Inclusion on machined surface. (b) Fracture through inclusion

Fig. 2

Oxide inclusions in tensile bars machined from actual castings contributing to poor tensile properties. Magnification: 8 

Elemental composition of an impurity is determined either wet or via spectrographic analysis in unfiltered and filtered samples (Ref 1). An efficiency parameter (Z) is then used to describe the filter performance (Ref 1): Z ¼ 100% ðCi  Co Þ=Ci

(Eq 1)

where Ci and Co are the concentrations of the specific element at inlet of the filter and after filtration, respectively. Detection methods (sample weight ranging from 0.5 to 30 g) based on the above principle and types of impurities (all sizes) analyzed include (Ref 1):  Emission spectroscopy for Li, B, Na, and    

heavier elements Hot extraction for hydrogen Combustion analysis for carbon Neutron activation for oxygen Gas chromatography for aluminum carbides and calcium carbides

*Adapted with permission from NADCA.

1168 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis Although the above methods have been used by many over the years, their limitations include (Ref 1):  Variability in data due to nonuniform distri-

bution of inclusions

 Samples analyzed may not be representative

of the bulk melt direct correlation between oxygen concentration (obtained via neutron activation) and the concentration of inclusions > 20 mm  Not practical for assessing inclusion levels on the shop floor  No

Pressure Filter Tests

A balance located below the filtrate crucible allows the operator to obtain the right amount of metal.

Fig. 3

Porous Disc Filtration Apparatus (PoDFA). Courtesy of ABB Group

1.4 0. 0

2 /kg

1.2

−8

2

mm

2 0.23 mm /kg

0.29

2 /kg

mm

mm 2/k

g

2 0.83 mm /kg

2 /kg

5.72

0.54

1 Mass filtered, kg

In this technique, a certain volume of metal is passed through a filter, thus forcing the inclusions in the melt to settle and concentrate on the filter. The trapped inclusions are subsequently analyzed metallographically. Methods of analysis based on the above principle include the Porous Disc Filtration Analysis (PoDFA) (ABB Bomem), Liquid Aluminum Inclusion Sampler (LAIS), and the Prefil Footprinter (ABB Bomem). In the PoDFA technique (Ref 1) (see Fig. 3), approximately 2 kg of molten metal are made to pass through a filter. Following sectioning of the filter with the residue, metallographic analysis is conducted to reveal the inclusion concentration, i.e. area of inclusions in the sectioned part (mm2/kg). Size, distribution, morphology, and type of inclusions can be further studied using optical microscopy and scanning electron microscopy (SEM). Measurement of the filtration rate may also be used as a measure of metal cleanliness assessment in the PoDFA technique. Figure 4 shows the variation in mass of metal passing though the filter as a function of time for various inclusion concentrations. It is evident that the PoDFA technique helps distinguish between very “dirty” metal (>1 mm2/kg) and relatively clean metal (<1 mm2/kg) (Ref 1). The LAIS technique (Ref 2, 3) (also an offline test, like the PoDFA) requires sampling cylinders into which the metal is pulled through a filter (see Fig. 5). Once the cylinders are prepared, the vacuum assembly is taken near the sampling area. A cylinder is fitted to a crossover that has another cavity where the sampling cup is placed. The upper end of the cylinder is connected to the vacuum pump having an appropriate vacuum gage. The vacuum is turned off, and the cylinder and sample assembly are immersed into the melt for preheating. The assembly is held using appropriate fixtures. After about 10 min of preheating, the vacuum is turned on and the melt begins to get sucked into the cylinder and the sampling cup through the crossover. The test is ended when the cylinder is filled completely (emission of fumes occurs). At this point, the vacuum is turned off and the cylinder assembly is placed on

mm

2

−6

12.9 mm

/kg

−4 Filtration rate test results

−2

0

1

2

3

4

5

6

7

8

9

10

Time, min

Fig. 4

Variation in mass of metal (through a filter) as a function of time for various inclusion concentrations. Source: Ref 1

a graphite chill block for unidirectional cooling. Thereafter, the cylinder is disassembled and the sample is removed for metallographic analysis (to determine level and/or type of inclusions). A sample obtained from the LAIS is typically a cylinder with the dimensions of 4.3  1.8 cm (1.7  0.7 in.). The sampling cup is a graphite cup having a frit (filter) running along its cross section about an inch from one end. The metal pulled resides in the hollow of the cylinder, and the oxides collect at the frit. Hence the frit has to be exposed and metallographically prepared to reveal the oxide count (mm2/kg) of the sample.

In a study (Ref 2) conducted by DasGupta and Apelian, cleanliness of a 319 aluminum alloy melt was determined using the LAIS technique. Four different melt conditions were evaluated, and for each melt condition several test bars were cast using a permanent mold test barmold. Figure 6 shows the variation in average percent elongation of a 319 aluminum alloy as a function of oxide count. As expected, higher elongation is achieved with cleaner metal (lower oxide count of the alloy) that results from either degassing or degassing + filtration being employed prior to pouring/ casting.

Approaches to Measurement of Metal Quality / 1169

To vacuum pump

C

Steel hollow cylinder

D

Sampling cup B

Molten aluminum

A

Filter/frit Porous graphite crossover Movement of aluminum during vacuum suction

Fig. 5

Schematic of Liquid Aluminum Inclusion Samples (LAIS) apparatus. Source: Ref 3

2.5

Fig. 7

Prefil apparatus. A, receptacle for collecting metal; B, heating element for preheating crucible; C, operating screen showing graph or footprint; D, insulated pressure chamber. The crucible is inside the insulated pressure chamber, D. Courtesy of ABB Group

Filtered and degassed metal

2

Elongation, %

Degassed metal but no filtration

1.5

No filtration, no degassing and metal allowed to regas for 10 h

1 Degassed metal but no filtration

0.5

0 0.2

Fig. 6

0.4

0.6

0.8 LAIS, mm2/kg

1

1.2

1.4

Variation in average percent elongation as a function of oxide count (LAIS). Source: Ref 2

Although both the LAIS and PoDFA tests have been used successfully in casting facilities, inclusion particles less than 10 mm in size have been known to influence the results (Ref 1). In the PREFIL Footprinter test (considered a reliable tool for on-line detection), a fixed volume of metal is allowed to flow at 83 kPa (12 psi) pressure through a fine filter at the base of the top crucible housed in the insulated pressure chamber (see Fig. 7). The metal flowing through the filter collects in the receptacle crucible at the bottom. The increase in weight is recorded in real time as metal flows through the filter in the top crucible. The faster the metal flows through the filter, the “cleaner” the metal (fewer inclusions such as oxide films left on the filter). If needed, the metal residue above the filter can be subsequently evaluated using optical microscopy and SEM.

Figure 8 shows the accumulated weight versus elapsed time for a 390 aluminum alloy. The data were generated on separate days from metal in a dip-well (following degassing and filtration). These data provide the extremes of overall metal cleanliness observed (at a given location and for a given alloy) and, thus, can later be used as a point of reference (Ref 4). A comparison of the six curves in Fig. 8 also reveals the effect of decreasing metal cleanliness (increasing inclusion concentration) on the slope and overall shape of the curves. For example, the initial slope of the curve (20 to 30 s) representing the lowest flow rate (curve L) is lower than that of curve H representing the highest flow rate (“cleanest” metal). Also to be noted is the change in the slope of curve L as the buildup on the Prefil filter

increases with time. On the other hand, the slope of curve H remains unaffected. In a study using 356 aluminum alloy, the effect of furnace refractory and furnace filter condition on the overall metal cleanliness was evaluated. The results from this study are shown in Fig. 9. As expected, damaged furnace refractory along with a damaged furnace filter yielded the lowest mass flow rate. The highest mass flow rate (thus, the cleanest metal) was achieved after repairing the damaged refractory and employing a new furnace filter. A comparison of the curves in Fig. 9 shows that the initial slopes of the curves (20 s) are fairly comparable; however, changes in inclusion level (attributed to damaged furnace refractory and/or damaged furnace filter) eventually contribute to changes in slope/ overall shape of the curves. Other techniques used for identifying the presence of inclusions in a melt include theQualiflash and the K-mold. The Qualiflash unit2 (Ref 2) has a crucible into which the melt to be analyzed is poured. The metal is made to pass through a filter located at the bottom of the crucible and flows into a “stepped” mold containing eight steps. A “dirtier” melt would fill fewer steps than a melt relatively free of inclusions. The number of steps filled is noted and used later to obtain a Qualiflash cleanliness rating. In the K-mold technique (a step bar permitting fracture of the knife point) (Ref 5) the number of inclusions (>400 mm) observed on fracture surfaces (four samples of five fractures each need to be examined, (Ref 5) divided by 20 provides the inclusion rating for the melt.

1170 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis 1600

1400 Highest flow rate (H)

1200 Date

Mass, g

1000

4/5/01 4/11/01 8/9/01 9/24/01 11/20/01 1/2/02

800

600 Lowest flow rate (L)

400

200

0 0

20

40

60

80

100

120

140

Time, s

Fig. 8

Prefil Footprinter data for 390 aluminum alloy recorded on various days

1400

1200 Refractory repaired; filter removed 1000 Refractory repaired; new filter added Mass, g

800

Fig. 10 600 Refractory repaired; filter still damaged 400 Poor furnace refractory; damaged filter

200

0 0

20

40

60

80

100

120

140

160

Time,s

Fig. 9

Prefil Footprinter data for 356 aluminum alloy

Electric Resistivity Tests Methods of analysis based on the above principle include the Liquid Metal Cleanliness Analyzer (LiMCA) (ABB Bomem); see Fig. 10. LiMCA is considered a reliable tool (Ref 3) for online detection of inclusions >15 mm. In this technique (Ref 1, 3) a small volume of metal is made to pass through an orifice of 0.1 to 1 mm while there is a large electric current passing through the aperture. The presence of

an inclusion particle passing through the orifice creates an increase in the total resistance of the current path, thereby generating a momentary voltage drop (displayed real-time on a computer screen) proportional to the volume of the particle. During the LiMCA test, the momentary voltage drops are noted and classified according to their magnitudes (Ref 3). Consequently, a measure of the size distribution of the inclusion particles (>15 mm) in the melt can be obtained.

Liquid Metal Cleanliness Analyzer. Courtesy of ABB Group

In a study by Shirandasht et al. (Ref 6) the behavior of various inclusions typically observed in aluminum-silicon alloys was examined by intentionally introducing Al2O3, Al4C3, MgO, CaO, TiB2, and TiAl3 inclusions in Al6%Si alloy melts through the addition of master alloys, powder, or other sources. In addition to the LiMCA data obtained (in the form of plots of total concentration number and volume concentration of inclusions as a function of time and particle size), the solidified samples from the LiMCA probe were subsequently examined using optical microscopy and electron probe microanalysis. The study revealed agglomeration of TiB2 inclusions as captured by the LiMCA probe tube, thus showing that the LiMCA was capable of measuring such clusters. Although the use of LiMCA in the foundry industry is limited by its high cost, its ability to be unaffected by the presence of TiB2 inclusions in the melt should offset its cost disadvantage in the long run. The study (Ref 6) states that LiMCA is the only technique capable of capturing the TiB2 agglomerates, compared to other techniques such as PoDFA and Prefil.

Approaches to Measurement of Metal Quality / 1171 N2

N2 + H2

Metal surface Molten aluminum

1

Note: • Ceramic probe • Gas bubbles (N2 + H2) collected • Depth of immersion is critical

2

Fig. 12

Telegas probe. Source: Ref 7

N2

N2 + H2

Fig. 11

AISCAN apparatus for hydrogen measurement. 1, probe and thermocouple immersed in melt; 2, hydrogen content measuring unit. Courtesy of ABB Group

Metal surface Molten aluminum Note:

Closed-Loop Recirculation In closed-loop recirculation (CLR) instruments, a purge gas (argon or nitrogen) is continuously circulated (via an immersed probe) through the metal being tested (Ref 7). This process continues till the hydrogen diffusing into the carrier gas is in equilibrium with the hydrogen in the molten metal (Ref 7). The diffusion of hydrogen into the purge gas results in a drop in thermal conductivity (of the gas mixture) that is then used to represent the partial pressure of hydrogen in the purge gas. The hydrogen concentration in molten aluminum alloys can then be determined by (Ref 7). Hydrogen in solution (H) ¼ So

sffiffiffiffiffiffiffiffiffiffiffiffiffiffi  ffi P1 760

(Eq 2)

where So is the solubility of hydrogen for a given temperature at a pressure of 760 mm Hg pressure and P1 is the partial pressure of hydrogen in the purge gas. The two techniques are based on the CLR principle include Telegas and AlSCAN (see Fig. 11). Both are used for measurement of dissolved hydrogen (reported in cm3 or mL per 100 g aluminum) in aluminum melts and are considered by many to be accurate and reproducible. Figures 12 and 13 illustrate the difference in probes used for the two methods. In a study conducted by DasGupta and Apelian (Ref 2), the hydrogen content of an A356 aluminum alloy melt (7% Si, 0.2% Fe, 0.2% Cu, 0.10% Mn, 0.46% Mg, 0.1% Zn, and 0.28% Ti) was determined using the AlSCAN apparatus. Three different conditions of the melt were evaluated, and for each melt condition several test bars were cast using a test bar mold. Figure 14 shows the effect of increasing hydrogen content on the range and average

• Probe consists of a square ceramic disc with two stainless steel capillary tubes • Gas exchanges occur at the probemetal interface • No gas bubbles collected (as with Telegas)

Fig. 13

AISCAN probe. Source: Ref 7

percent elongation of the A356 aluminum alloy. As expected, higher hydrogen content (resulting from ineffective or no melt treatment) is observed to decrease percent elongation. In another study conducted by DasGupta and Apelian (Ref 2), the hydrogen content of a 319.1 aluminum alloy melt (6% Si, 0.5% Fe, 3% Cu, 0.5% Mn, 0.1% Mg, 0.35% Ni, 1.0% Zn and 0.24% Ti) was determined using the AlSCAN technique. In this investigation, four different melt conditions were evaluated, and for each melt condition several test bars were cast. The effect of increasing hydrogen content on the range and average percent elongation of the 319.1 aluminum alloy is shown in Fig. 15. Higher hydrogen content (resulting from ineffective or no melt treatment) is again observed to decrease percent elongation.

Reduced-Pressure Test The reduced-pressure test (RPT), often referred to as the Straube-Pfifer test, provides qualitative information of the overall cleanliness (combinational effects due to inclusion content and hydrogen) of the melt. The RPT

(see Fig. 16) involves solidifying a sample of molten metal (between 100 and 200 g) under reduced pressure (typically 26 mm Hg). By reducing the pressure over the sample during solidification, hydrogen bubble nucleation and growth is enhanced (Ref 8). The bubble formation is caused by the decrease in hydrogen solubility with decreasing pressure according to the law of mass action: S¼

  fH pffiffiffiffi P

(Eq 3)

where S is the solubility of hydrogen gas in the melt; f is the activity coefficient of hydrogen; H is the amount of hydrogen in the melt; and P is the pressure of hydrogen vapor over the melt. Appearance of a “mushroom” head (convex surface) during solidification indicates a high hydrogen content, while a concave surface occurs when hydrogen content of the melt is low. The solidified sample is subsequently sectioned and visually examined for porosity nucleated by inclusions during the test. When the cross section of the sample is compared with a visual progression chart (see Fig. 17), an

1172 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis 5 4.5 4 2

Elongation,%

3.5

Molten metal purged with argon + chlorine for 40 min

3

4

Molten metal purged with argon for 30 min

2.5

1

No melt treatment

2 3

1.5 1

Fig. 16

Reduced pressure test apparatus. 1, vacuum chamber; 2, vacuum gage; 3, pressure regulator; 4, vacuum pump

0.5 0 0.1

0.11

0.12

0.13

0.14

0.15

0.16

0.17

0.18

0.19

mL/100 g hydrogen

Fig. 14

Effect of increasing hydrogen content of melt on percent elongation of A356 aluminum alloy. Source: Ref 2

3

2.5

Elongation,%

2

Filtered and degassed metal

No degassing, no filtration; melt allowed to regas for 10 h

Degassed melt but no filtration

1.5

1 Degassed melt but no filtration 0.5

Fig. 17 0 0.14

0.16

0.18

0.2

0.22

0.24

0.26

0.28

0.3

mL/100 g hydrogen

Fig. 15

Effect of increasing hydrogen content of melt on percent elongation of a 319 aluminum alloy. Source: Ref 2

Hydrogen content, mL /100 g

0.28 0.24 0.2 0.16 0.12 0.08 R square - 0.678 0.04 2.5

2.54

2.58

2.62

2.66

RPT density; g/cm3

Fig. 18

Correlation between RPT density and hydrogen content of 356 alloy. Source: Ref 8

2.7

Visual progression chart for a given alloy; lower number/rating indicates “cleaner” melt

estimate of the overall melt cleanliness can then be obtained (a lower number/rating equates to “cleaner” melt). The density of the RPT sample can also be determined and used as a measure of porosity. The optimum procedural steps for performing the RPT and measuring the density of the RPT sample are documented in the study by DasGupta et al. (Ref 8). DasGupta et al. (Ref 8) examined the correlation of RPT densities to hydrogen content (AlSCAN measurements) of an unmodified and not grain-refined 356 aluminum alloy. The correlation curve, shown in Fig. 18, was obtained from 100 data points (AlSCAN and RPT density measurements from 25 different heats) (Ref 8) that were fitted using polynomial regression. The correlation curve can be useful in that one can estimate hydrogen content in the melt from the density of an RPT sample. The correlation curve corroborates earlier studies that showed the RPT density to be inversely proportional to the hydrogen content of the melt (Ref 8).

Approaches to Measurement of Metal Quality / 1173 Ultrasonic Technique The ultrasonic technique, developed for online assessment of the cleanliness of molten aluminum alloys, relies on guide steel rods to introduce high-frequency sound waves into the melt and to receive the reflected signals (Ref 9). The study (Ref 9) reveals that the ultrasonic sensor allows online evaluation and monitoring during molten aluminum alloy processing, that the sensor is sensitive to grain refinement and degassing practices, and that the probe response time is less than a minute. REFERENCES 1. D. Apelian, How Clean is the Metal You Cast? The Issue of Assessment: A Status Report, Proc. Third International Conference on Molten Metal Aluminum Processing , Nov 9–10, 1992, American Foundry Society, p 1–15

2. S. DasGupta and D. Apelian, Interaction of Initial Melt Cleanliness, Casting Process and Product Quality: Cleanliness Requirements Fit for a Specific Use, Proc. Fifth International Conference on Molten Metal Aluminum Processing , Nov 8–10, 1998, American Foundry Society, p 234–258 3. S. Makarov, D. Apelian, and R. Ludwig, Inclusion Removal and Detection in Molten Aluminum: Mechanical, Electromagnetic, and Acoustic Techniques, AFS Trans., Vol 107, 1999, p 727–736 4. P.G. Enright, J.R. Hughes, and J. Pickering, “Characterization of Molten Metal Quality Using the Pressure Filtration Technique,” Paper No. 03–086, American Foundry Society, Congress, 2003 5. “Metal Melting and Handling,” NADCA In-Plant Study Course, 302-IP, North American Die Casting Association 6. J. Shirandasht, A.M. Samuel, F.H. Samuel, H.W. Doty, and S. Valtierra,

Microstructural Characterization of Inclusions in Al-6%Si Cast Alloy Measured by the LiMCA II Technique, AFS Trans., Paper No. 06–006 (02), Vol 114, 2006, p 45–58 7. AlSCAN Engineering Bulletin No. 5, Feb 1991 8. S. DasGupta, L. Parmenter, and D. Apelian, Relationship between the Reduced Pressure Test and Hydrogen Content of the Melt, Proc. Fifth International Conference on Molten Metal Aluminum Processing, Nov 8–10, 1998, American Foundry Society, p 285–300 9. R. Gallo, H. Mountford, and I. Sommerville, “Ultrasound for On-Line Inclusion Detection in Molten Aluminum Alloys: Technology Assessment,” American Foundry Society, International Conference on Structural Aluminum Castings, Nov 2003

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1174-1177 DOI: 10.1361/asmhba0005341

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Nondestructive Testing of Components* Rathindra DasGupta, National Science Foundation

NONDESTRUCTIVE TESTING of cast products is necessary to monitor the quality of the various components being manufactured. The level of quality required of these components often dictates the frequency of testing and types of tests being needed. The commonly used techniques include:
Liquid penetrant inspection (LPI)
Radiographic inspection
Fluoroscopic inspection and automated defect






recognition Ultrasonic inspection Eddy current inspection Process-Controlled resonant testing Leak test Electrical conductivity measurements

This article summarizes the application of these nondestructive tests to castings. The general subjects of nondestructive testing and quality control and thorough details for these and other tests can be found in the ASM Handbook, Vol 17, Nondestructive Evaluation and Quality Control (1989).

Liquid Penetrant Inspection In this test (Ref 1, 2), the liquid penetrant flows over a surface and enters various types of minute surface openings (0.1 mm (0.4 min.) wide) by capillary action. Consequently, this technique helps detect surface cracks, laps, laminations, and similar flaws. It should be noted that surface condition (cleanliness, roughness) of the part being inspected is critical since false indications can arise from poor surface preparation or condition. The three basic penetrant systems commonly used are:
Water-soluble:

The penetrants can be removed directly with water.
Postemulsifiable: These penetrants have an oil base and, thus, cannot be removed directly with water. Consequently, an emulsifier is needed (following application of the penetrant) to make the penetrant soluble in water.
Solvent-removable: These also have an oil base and employ a solvent for both precleaning and removal of excess penetrant. *Adapted with permission from NADCA.

Each of the aforementioned systems uses either a visible or a fluorescent penetrant material. The visible-penetrant inspection uses a liquid that is typically red in color and, thus, reveals red indications under visible light. On the other hand, the fluorescent penetrant inspection employs penetrants that fluoresce under ultraviolet light. Components with complex designs (sharp corners or radii) and/or made with alloys that are prone to hot tearing are often subjected to liquid penetrant inspection.

Radiographic Inspection The term radiography refers to a method that produces a two-dimensional permanent image on film or paper that has been exposed to unabsorbed radiation passing through the test piece. This technique requires subsequent development of the exposed film/paper so that the image is visible. Radiography is used extensively for evaluation of large components (e.g., steel castings) that exhibit a difference in thickness or physical density (Ref 3). Unlike liquid penetrant and eddy current inspection, radiography can satisfactorily detect flaws that are completely internal and located well below the surface of the component. If the workpiece is properly oriented during inspection, radiography is capable of detecting defects such as cracks. Voids and inclusions having measurable thickness in all directions can be also detected as long as they are not too small in relation to section thickness.

Fluoroscopic Inspection and Automated Defect Recognition This system has replaced conventional radiography (film x-ray) and consists of an x-ray generator and a detector (an image intensifier or a digital detector). When the x-rays generated are made to pass through the component (an aluminum casting in this example), the aluminum absorbs and attenuates a portion of the

x-rays and transmits the rest. The degree to which a part attenuates x-rays is a function of several variables including x-ray wavelength, velocity, and part density. The transmitted x-rays impinge on the detector that is located behind the casting. These rays selectively affect the detector and create an image that can be viewed on a screen (cathode ray tube). If there is a defect, it will absorb more or less x-ray energy (influenced by the nature of the defect) than the surrounding aluminum matrix. Consequently, the transmitted rays will affect the detector differently. In this way, the image created by the detector provides a visual impression of the defects that may be present in the casting. The ease with which a detector can discern a defect depends on its density relative to the parent material. The image created by a detector can be converted to a digital image that may be subsequently processed by a computer program (ADR program). Based on the difference in gray scales of the image, the program can detect defects in the image with great success (see Fig. 1). The ADR is used extensively in high-volume detection environments to reduce dependency on a human operator to make decisions.

Ultrasonic Inspection An ultrasonic transducer emits ultrasonic waves that pass through a certain section of a component. Upon impinging a defect, a portion of the wave is reflected back to the source, and a portion is transmitted to the bottom of the section. Upon hitting the bottom, part of the wave is reflected back to the source, and part is transmitted to the environment. By sensing the magnitude of these reflections and the time in between them, one can get a fair idea of the presence of a defect in the cross section as well as its location relative to the thickness of the component being tested. As shown in Fig. 2, the two large peaks correspond to the reflection off of the front wall (top of section) and back wall (bottom of section). Only these peaks are expected if there is no defect in the part. How-

Nondestructive Testing of Components / 1175 ever, in the presence of a defect, one or more secondary peaks will appear (between these large peaks) corresponding to reflections from the defect. If the defect is large and/or oriented in a random fashion, it may cause very little of the wave to transmit to the back wall. In this case, there will be a substantial loss in magnitude of the back wall peak (Ref 2). Owing to complex surface configurations and somewhat random locations of internal defects, both straight beam and angle beam techniques are needed, as well as some special frequency transducers to properly locate the defects.

Eddy Current Inspection

(a)

Fig. 1

(b) (a) X-ray inspection equipment. (b) Digital image of two aluminum castings showing no defects

Process-Controlled Resonant Testing (Contributed by Quasar International, Inc.)

Transducer

Front wall

Back wall

Fig. 2

Ultrasonic inspection equipment showing reflection of waves through a workpiece

Process-controlled resonant testing (PCRT) accounts for acceptable part-to-part structural variations and is used to separate structurally acceptable from unacceptable parts. It is used primarily on the factory floor because of its speed and accuracy in properly sorting parts. PCRT is used on parts that can resonate, such as most metals, ceramics, and some composite materials. PCRT differentiates between an acceptable and an unacceptable part based on comparison of the resonance pattern of a part compared to a pattern determined from acceptable parts used in pattern-recognition training. Resonant frequencies are determined by the dimensions and material properties of the “whole part” according to Eq 1: fr ¼ k=m

Coil running AC

Eddy currents

Defect

Fig. 3

As shown in Fig. 3, a coil running alternating current (AC) (exciting current) close to a conducting part induces “eddy current” in the part as a result of electromagnetic induction. The presence of flaws in the part alters the magnitude of this current. By monitoring the voltage across the coil, surface and subsurface flaws can be detected.

Eddy current inspection. The presence of a flaw alters the magnitude of current.

(Eq 1)

where fr is the resonant frequency, k is the stiffness (material property, e.g., Young’s modulus), and m is the mass (dimensions, density). A major advantage of PCRT (Fig. 4) is that the output results relate directly to the structural integrity of a part, not just to an indication, and that normal, acceptable structural variations are taken into account. It is fast (typically a few seconds per part), inexpensive, and requires no human judgment when in operation. Table 1 shows typical automotive parts where PCRT has been successfully employed. PCRT is effective on materials that are stiff and resonant. This includes most metals, ceramics, and a few composites. Metals that are not suitable include very soft metals, such as lead, and metals that have been made to dampen vibrations such as a few grades of gray iron. PCRT is acceptable for almost all structural parts used in the automotive and related industries.

1176 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis Table 1 A partial list of automotive part types and materials amenable to PCRT Category

Parts type

Cast steel, engine Stamped/brazed steel, engine Cast aluminum, suspension Cast ductile iron, suspension Cast iron, engine Sintered powder metal, drive line Cast titanium, turbine Forged and machined steel, transmission Ceramic, engine sensor Silicon, electronic chips Silicon nitride, bearing elements Various materials, turbines

(450)

1779

(400)

1557

(350)

1334

(300)

1112

(250)

890

(200)

Resonance inspection equipment

PCRT compares to other nondestructive test (NDT) methods in that it measures the structural response of a whole part. Other NDT methods look for “indications” of a flaw. With rare exception, most other NDT methods do not correlate with structural integrity. To illustrate this, Fig. 5 compares a PCRT score to the breaking strength of flawed powder metal parts in a laboratory breaking test. Note the nearly perfect correlation. PCRT is used most effectively on production parts that have high intrinsic value or high safety value. It is generally reasonable to test these parts at 100%. PCRT works most efficiently as an on-the-floor factory test. Depending on the circumstances, it can be used with premachined parts or machined parts. The potential limitations to the use of PCRT include:

Yield force, N(lb)

Fig. 4

2002

Investment cast steel rocker arms Stamped and brazed steel rocker arms Cast aluminum knuckles and control arms Cast ductile iron control arms Cast cylinder heads and cylinder blocks Sintered powder metal gears and clutch plates Cast gas turbine blades Machined forged steel transmission gears and shafts Ceramic oxygen sensors Silicon wafers Silicon nitride bearing elements Gas turbine components (not yet in production)

Correlation = 99.8%

0

5

10

15

20

Fig. 5

not with others (e.g., vibration-absorbing coatings).

anomalies not tied to structural performance. It may be necessary to provide additional visual inspection of parts to identify visually unacceptable indications. Process variation of parts may limit the applicability of PCRT. For example, mass/ dimensional variations exceeding 5% may cause PCRT to be unusable. Specific anomaly identification is highly unlikely. PCRT is a whole body measurement, and differentiating between a crack and a void in the same location is generally not possible. It may be possible to differentiate some anomalies by using multiple patterns and training sets. PCRT will only work with very stiff objects that provide resonances whose frequency divided by their width at half of the maximum amplitude (Q) are greater than 400 to 500. Although steel parts may be very stiff and perfectly reasonable to use for PCRT, steel foil would generally not be. PCRT will not work with parts that are covered in a heavy coating of oil or other similar material. PCRT may work with some coatings (e.g., anodized aluminum), but

The essence of how resonant inspection works and why it correlates with structural integrity can be understood with a simple model reflecting the physical elements that actually determine the mechanical characteristics of a material. This simple model forms the basis for determining the elastic properties of a material, such as Young’s modulus, shear modulus, Poisson’s ratio, and so forth. The model provides the needed insight into how material resonances are determined by elastic properties and very specifically how material flaws that adversely impact mechanical properties cause resonances to change. The real power associated with PCRT is its ability to account for normal, acceptable process variation that causes resonant frequencies to shift. The resonant frequencies of real production parts are affected by both the presence of defects and by the normal process variations in stiffness, density, and dimensions. These variations can cause frequency shifts that are generally larger than the shift caused by all except the most severe defects. So the process variations tend to mask the defects, if only










30

35

40

Pattern recognition score versus yield force for powder metal clamp flanges


PCRT cannot separate parts based on visual




25

Pattern-recognition score

a simple, single frequency shift is used as a test to differentiate between acceptable and unacceptable parts. In practice, pattern-recognition software is used to compute the predicted frequency for one resonance (the diagnostic resonance) based on the measured frequency of several other resonances. The difference between the computed and measured frequency for the diagnostic resonance is the predictor error. The algorithm used minimizes the predictor error for good parts and maximizes the error for the bad parts. PCRT operates by driving a part at given frequencies (audio through ultrasound, depending on the part characteristics) and measuring its mechanical response. Figure 6 shows a schematic for the PCRT apparatus. The process proceeds in small frequency steps over a previously determined and predicted range of interest. During such a sweep, the drive frequency typically brackets the expected location of preselected resonant frequencies. When all of the frequencies selected to form the diagnostic pattern are measured, the software scores the resonant pattern and determines if a part is accepted or rejected. Acquiring resonance data requires precise part fixturing and also requires that variation

Nondestructive Testing of Components / 1177

Control computer measurement and real-time algorithm application

CW transmitter/ receiver for signal processing

Parts fixture with transmitting (excitation) and receiving transducers

Fig. 6

PCRT instrumentation and data flow

Table 2 Electrical conductivity measurements Sample No.

Average wt% strontium

1 2 3 4 5 6 7 8

0 0.005 0.010 0.017 0.026 0.035 0.054 0.078

Average %IACS

22.5 35 35 32 31.5 31 30 29

in a part provides a more direct path for electron flow and, thus, yields higher conductivity values. The readings obtained are expressed as a percentage of the International Annealed Copper Standard (%IACS). Electrical conductivity measurements can be used for determining the presence of inclusions or porosity. This technique can also be used for assessing the level of eutectic silicon modification caused by the addition of strontium (or some other agent) since higher conductivity values result from modified structures (see Table 2). REFERENCES 1. Metals Handbook Desk Edition, American Society for Metals, 1985 2. Quality Assurance Methods, Chapter 8, High Integrity Aluminum Die Castings (Alloys, Processes, and Melt Preparation), Item No. 307, North American Die Casting Association (NADCA), 2004, p 125–129 3. Nondestructive Evaluation and Quality Control, Vol 17, ASM Handbook, ASM International, 1989 SELECTED REFERENCE

(a)

(b)

Fig. 7

Cold flake (presolidified aluminum from shot sleeve) entrapment in a casting contributing to “leak.” (a) Aluminum cold flakes on fracture surface. Original magnification: 8.7. (b) Scanning electron micrograph of cold flakes, Arrow indicates regions of partial bonding with mating surface. Original magnification: 500

of the resonant frequencies of the parts be quantified. Some resonances are very sensitive to the exact placement on the transmitting and receiving (lead zirconate titanate, or PZT) transducers. Parts are generally mounted in a tripodal configuration of the transducers with very precise guides used for emplacement.

in components that need to be pressuretight. As shown in Fig. 7, the presence of aluminum cold flakes (presolidified aluminum from a shot sleeve) in a casting created a nonknitting condition, thereby allowing the fluid to pass through the resultant discontinuity (Ref 2).

Leak Test

Electrical Conductivity Measurement

Leak testing provides the rate at which a liquid or gas will pass through a discontinuity (void, crack) in a part. Gross shrink porosity and/or presence of inclusions can contribute to “leaks”

Electrical conductivity of metals is the reciprocal of the electric charge flux per unit voltage gradient (Ref 2). The absence of imperfections


Nondestructive Evaluation and Quality Con-

trol, Vol 17, ASM Handbook, ASM International, 1989

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1178-1182 DOI: 10.1361/asmhba0005342

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Casting Fracture Characteristics Craig C. Brown, Stork Technimet

THE CHARACTERISTICS OF CASTING FRACTURES are, in general, similar to those of wrought alloys, but in many cases can be more complex. When examining fractures in castings it is important to fully understand the fundamentals of fracture mechanisms and the basic fracture modes that are typical for the alloy involved. This article discusses the visual and microscopic characteristics of fractures of cast alloys, but the details on fracture mechanisms can be obtained from the pertinent articles in the ASM Handbook, Volume 11, Failure Analysis and Prevention (Ref 1–4).

General Fracture Examination Background Information When examining a casting that has fractured, it is important to know basic information about the part. Background data concerning the alloy type and grade, the expected mechanical properties, the overall heat treatment, and any special surface treatments or processes should be obtained. It is also essential to know the history of the part. Did the fracture occur during the manufacturing process or when the part was in service? It is important to understand the loads applied to the part as well as the operating environment. This background information can provide some insight into the expected fracture characteristics.

Examination Methods The evaluation of a failed casting must include visual inspection of the whole part and other related components as well as the actual fracture. The part should be examined thoroughly to evaluate both the overall features and any unusual features. Details such as surface damage, distortion, machining marks, fretting, corrosion deposits, and so forth all provide information about the environment and relationship to other components. The examination of the fracture should be done using good illumination to see the fine details of the surface. Oblique illumination should be used to evaluate the texture and

profile of the fracture. This type of illumination will help determine the location of the fracture origin. The visual inspection should be conducted both prior to and after cleaning or removal of surface contamination. The fracture should be examined with an optical or digital microscope at the critical regions of the origin, midfracture, and final fracture regions. This may allow detection of any casting flaws. In some cases the fracture origin can be at a subsurface anomaly, such as gas/shrinkage porosity, a nonmetallic inclusion, or anomalies from weld repair. The precise location of crack initiation may only be discernible with aid of an optical/digital microscope. In order to determine the fracture mode, which is how the casting broke, the critical regions of the fracture should be examined with a scanning electron microscope (SEM). The fine details that are characteristic of ductile rupture, transgranular brittle fracture, intergranular fracture or fatigue, and even some flaws, are not discernible with the unaided eye or with the optical/digital microscope. These features can only be properly observed when examined at magnifications above approximately 500. The SEM allows routine examinations at magnifications up to 10,000.

Regions of Interest The primary objective of the visual inspection is to determine the location of the fracture origin and evaluate how the crack propagated through the alloy. Knowing where it started and where it ended provides information about how the part was loaded. The origin region should be examined to determine where the crack initiated. Did the crack start at a single point or over a broad area? Cracks typically originate at a site of stress concentration such as a surface anomaly, grinding marks, cast or machined notch, or subsurface flaw. The midfracture region exhibits features that indicate how the crack propagated through the material. This gives insight as to whether the part was subjected to cyclic, tensile, or torsional loading. The edges of the fracture can exhibit slant fracture, or shear lips, that can help identify the origin region. This is especially true for cast steels.

Factors That Affect Fracture Appearance The fracture texture at the origin region can have features that indicate where the crack started. With overload fractures, steel castings will typically exhibit a chevron pattern that will point back to the origin region, as shown in Fig. 1, whereas aluminum castings can exhibit a relatively rough texture in the origin region and a smoother texture at the other side of the broken section. This is especially true when there is no distinct stress raiser in the origin region of the aluminum part. With fatigue fractures, most cast alloys will exhibit thumbnail-shaped beach marks or arrest lines that fan outward from the origin region. The fracture surface texture of cast components can exhibit a different texture compared to a similar wrought product. With some alloys, such as die-cast aluminum and improperly controlled gray or ductile iron, the near-surface region can have a chill zone that will exhibit a fine, smooth fracture texture. Away from the surface the fracture may expose various anomalies such as shrinkage or gas porosity and nonmetallic inclusions in the form of sand, dross, slag, or reoxidation products. Fatigue fractures can exhibit a rougher texture as the crack propagates through and around the anomalies. With heavy-section steel castings in

Fig. 1

Overload fracture through a low-alloy steel casting. Courtesy of Stork Technimet, Inc. New Berlin, WI

Casting Fracture Characteristics / 1179 the quench and tempered condition, alloy segregation can cause varying degrees of both ductile and brittle fracture due to differences in the proportion of martensite and bainite. Weld repairs can have a significant influence on the potential for crack initiation, at the surface or at subsurface regions. Residual stresses from weld repair and improper grinding as well as rough finishing and carbon arc gouging can cause crack initiation. Incomplete removal of casting defects and weld anomalies, such as slag, shrinkage cracks, or cracks from improper weld process control have been found to cause failures in steel castings.

Fracture Features The fracture mechanisms of cast alloys are similar to wrought alloys. The features are influenced by the crystal structure, microstructure, loading rate, and temperature. The microscopic features of the fracture are typically similar to wrought alloys, but can be more complex and in most cases it is essential for the analyst to have knowledge about the details of the microstructure as the phases, constituents, and anomalies in the alloy will influence the characteristics of the fracture.

silicon particles would provide sites for MVC, and the fracture would exhibit dimples, as shown in Fig. 5. The ductility, as measured by the percent elongation during tensile testing, would be substantially higher for a fully spheroidized alloy, and this correlates with the fracture appearance.

Transgranular Brittle Fracture Transgranular brittle fracture in cast components typically exhibits the usual characteristics of brittle overload, such as little or no distortion, flat fracture, bright crystalline appearance, and a chevron pattern that points to the origin. Examination at high magnification will show cleavage fracture through the grains. The transgranular fracture is related to the nilductility transition temperature (NDTT) of the alloy, the strain rate, or brittle phases in the microstructure. As with wrought products, the NDTT may be shifted to a temperature above the operating temperature of the component. This can occur with carbon and low-alloy steels as well as ductile iron. An example of cleavage fracture in a ductile iron is shown in Fig. 6. This contrasts with the ductile rupture features shown previously. In some cases, such

as ductile iron and low-alloy steel, the fracture may exhibit a mixture of both cleavage fracture and ductile rupture. A typical example of a mixed fracture mode is shown for a low-alloy steel casting in Fig. 7. In some cases, the cleavage fracture is associated with the pearlite and the ductile rupture with the ferrite in the microstructure. It is important to be aware of the microstructure of the material, as the chill zone of a ductile iron can exhibit brittle features when the fracture extends through the iron carbides in the iron. The example presented in Fig. 8 is from the same casting shown in Fig. 3, but from a region that was chilled during solidification and cooling. Another example of how the microstructure of an alloy can significantly influence the features of a fracture is gray iron. Gray iron is typically considered to be a brittle material as the fracture surface exhibits the characteristic macrofeatures of a brittle alloy. Examination at high magnifications shows that the fracture through a gray iron is primarily through and at the graphite flakes. Only scattered rupture through the matrix occurs between the large flakes. Therefore, the fracture will exhibit smooth curved surfaces that replicate the

Ductile Rupture Cast components typically exhibit the usual characteristics of ductile overload, such as necking and distortion, a dull fibrous fracture with little branching, and shear lips. On examination at high magnification, ductile rupture typically exhibits dimples, which are formed by microvoid coalescence (MVC). Nucleation typically occurs at nonmetallic inclusions or secondphase particles. Carbon, low-alloy, and stainless steel castings will typically exhibit classical dimples. A low-alloy steel in the quench and tempered condition is shown in Fig. 2. A similar fracture mode can occur in a ductile iron, but the appearance is altered due to the presence of the nodules, which provide sites for MVC, as shown in Fig. 3. Dimples formed around the nodules. The size and shape of the dimples is controlled by the nodule count and nodularity. Ductile rupture in cast silicon-aluminum alloys can have different characteristics, depending on the heat treatment of the alloy. In the ascast condition, the primary and eutectic silicon can have an angular and flakelike structure. Rupture through the alloy will occur at and through the silicon particles and also through the aluminum matrix. The fracture will exhibit brittle features at the silicon particles, but tear ridges and dimples will be present at the matrix. These features are illustrated in Fig. 4, which shows the fracture through a type 380.0 die-cast aluminum component. The larger the primary and eutectic silicon particles, the smaller the proportion of ductile features will be. Heat treatment of a similar alloy would cause spheroidization of the silicon particles. These

50 µm 10 µm

Fig. 2

Ductile rupture in a low-alloy steel casting. Original magnification: 3000. Courtesy of Stork Technimet, Inc. New Berlin, WI

Fig. 4

Overload fracture through a type 380.0 aluminum alloy in the as-cast condition. Original magnification: 1000. Courtesy of Stork Technimet, Inc. New Berlin, WI

100 µm

200 µm

Fig. 3

Ductile rupture in a ductile iron. Original magnification: 150. Courtesy of Stork Technimet, Inc. New Berlin, WI

Fig. 5

Overload fracture through a type 356.0 aluminum alloy in the T6 condition. Original magnification: 500. Courtesy of Stork Technimet, Inc. New Berlin, WI

1180 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis texture of the graphite, and only isolated locations with either cleavage or ductile rupture, as shown in Fig. 9. It is important to be aware that carbon and low-alloy steels can often exhibit quasi-cleavage. If cast steel is properly quenched and tempered to form a microstructure of tempered martensite, the fracture will typically exhibit dimples, which are indicative of ductile rupture. If the casting is not quenched at a sufficient cooling rate and has a microstructure of tempered martensite and some ferrite, the fracture will exhibit features that are a blend of cleavage fracture and ductile tear ridges, often called quasi-cleavage. A typical example for low-alloy steel is shown in Fig. 10.

Fatigue Fatigue fracture in cast components typically exhibits the usual features of a progressive fracture mode, such as relatively flat fracture surface with beach marks and arrest lines, ratchet marks between the origins, striations when examined at high magnification, and an overload fracture at the final fracture region.

Crack initiation will occur at a site of stress concentration. This may be at the surface or at a subsurface anomaly. With cast alloys the origin is often found at porosity or nonmetallic inclusions. An example of crack initiation at shrinkage porosity in a low-alloy steel casting is shown in Fig. 11. As with wrought alloys, a fatigue crack in carbon and low-alloy steels will propagate through the material via microscopic steps and produce fine striations, which cannot be seen with the unaided eye, but can be discerned with a SEM at magnifications above 500. Typical striations in a low-alloy steel suspension component are shown in Fig. 12. It is important to understand the difference between these striations and beach marks or arrest lines. Beach marks are visually evident on most fatigue fractures in cast components and are shown in Fig. 11. The visual, thumbnail, or curved lines that radiate from the origin region are caused by variations in the cyclic stresses and the crack propagation rate. These marks are not always indicative of fatigue fracture, but can also be caused by incremental overload steps, although the overload steps are typically less uniform and have a rougher texture. Confirmation of a fatigue

Cleavage fracture in a heavy-section ductile iron casting. Original magnification: 200. Courtesy of Stork Technimet, Inc. New Berlin, WI

Fig. 8

Brittle fracture through the near-surface chilled region of a ductile iron. Original magnification: 1000. Courtesy of Stork Technimet, Inc. New Berlin, WI

50 µm

Fig. 7

Mixture of cleavage fracture and ductile rupture (arrows) in a low-alloy steel casting. Original magnification: 1000. Courtesy of Stork Technimet, Inc. New Berlin, WI

50 µm

Fig. 10

Quasi-cleavage in a heavy-section low-alloy steel casting. Original magnification: 1000. Courtesy of Stork Technimet, Inc. New Berlin, WI

50 µm

200 µm

Fig. 6

fracture should be done by examination at high magnification with a SEM. It should be noted that fatigue striations are not always discernible on examination at high magnification. For example, the appearance of a fatigue fracture through a cast aluminum alloy will appear quite different from cast steel. The aluminum alloy will typically have a feathery

Fig. 11

Subsurface fatigue crack initiation in a heavy section low-alloy steel casting. Original magnification: 6. Courtesy of Stork Technimet, Inc. New Berlin, WI

200 µm

Fig. 9

Fracture through the matrix between the graphite flakes of a gray iron. Original magnification: 250. Courtesy of Stork Technimet, Inc. New Berlin, WI

Fig. 12

Fatigue striations in a low-alloy steel casting. Original magnification: 2000. Courtesy of Stork Technimet, Inc. New Berlin, WI

Casting Fracture Characteristics / 1181

100 µm

Fig. 13

Fatigue fracture through a type 356.0 aluminum alloy in the T6 condition. Original magnification: 500. Courtesy of Stork Technimet, Inc. New Berlin, WI

Fig. 16

Linear rupture through a low-alloy steel with rock candy fracture. Original magnification: 5000. Courtesy of Stork Technimet, Inc. New Berlin, WI

50 µm

10 µm

Fig. 14

Fatigue fracture through a type 356.0 aluminum alloy in the T6 condition. Original magnification: 5000. Courtesy of Stork Technimet, Inc. New Berlin, WI

Fig. 17

Quench cracking in a low-alloy steel lever casting. Original magnification: 1000. Courtesy of Stork Technimet, Inc. New Berlin, WI

50 µm

50 µm

Fig. 15

Hydrogen-assisted cracking in a heavy-section low-alloy steel casting. Original magnification: 1000. Courtesy of Stork Technimet, Inc. New Berlin, WI

appearance where the fatigue fracture propagates through the matrix and a rougher, more globular texture where the crack extends around the eutectic silicon particles in the microstructure, as shown in Fig. 13 for a laboratory fatigue test specimen. Even at higher magnification, the striations are difficult to discern, as shown in Fig. 14.

Fig. 18

Stress-corrosion cracking in a bronze alloy C83600 suction roll shell. Original magnification: 1000. Courtesy of Stork Technimet, Inc. New Berlin, WI

Environmentally Induced Fracture Hydrogen-Assisted Cracking (Intergranular Brittle Fracture). Hydrogen-assisted cracking, also called hydrogen embrittlement, can occur with carbon and low-alloy steels that have been

exposed to an environment where hydrogen is generated at the surface of the part. This can be from pickling or plating operations or may occur from severe surface corrosion. The fracture surface will typically exhibit separation at the grain boundaries and will also exhibit small regions with ductile fracture features called ductile hairlines, have some secondary grainboundary cracking, and show the presence of micropores. The features are similar to those found in wrought components, and an example of a heavy-section low-alloy steel with hydrogen-assisted cracking is shown in Fig. 15. Rock Candy Fracture (Intergranular Brittle Fracture). Heavy-section carbon and low-alloy steels are susceptible to “rock candy fracture” or concoidal fracture if the steel contains excessive amounts of nitrogen and aluminum. Aluminum-nitrogen precipitates form at the austenite grain boundaries if the casting is cooled slowly. The overload fracture surface exhibits separation at the grain boundaries, and the grains can typically be seen with an unaided eye. Examination at high magnification typically reveals fine quasi-cleavage or ductile fracture with linear ruptures where the precipitates are present. A typical linear rupture is shown in Fig. 16 for a heavy-section low-alloy steel. Quench Cracking (Intergranular Brittle Fracture). Quench cracking can occur when a steel casting is quenched from the austenitizing temperature at too fast a rate. Intergranular fracture occurs at the prior austenite grain boundaries. In most cases the original fracture surface is oxidized during the tempering cycle, as the cracks are usually not detected until after completion of the heat treatment process. The fracture surface typically exhibits relatively clean separation between the grains with minimal evidence of ductile fracture or secondary grain-boundary cracking. A typical example of a quench crack for a low-alloy steel lever is shown in Fig. 17. A coating of temper scale covers the fracture surface. A metallographic cross section through the cracked region is usually needed to confirm the presence of intergranular fracture. Stress-corrosion cracking. (SCC) can occur with cast alloys and is most often encountered with copper-base alloys, such as brass and some bronzes. Stress-corrosion cracking occurs when a susceptible alloy is subjected to applied or residual stresses in a corrosive environment. This is especially true for some copper alloys when ammonia-based compounds are present. The fracture characteristics can be somewhat complex and in some cases difficult to distinguish from corrosion fatigue. A typical example of SCC in a bronze alloy C83600 suction roll for the paper industry is shown in Fig. 18. The surface exhibits the characteristic jagged features with fine microcracks. It should be noted that fracture features are often only discernible at the crack-tip region, or at secondary cracks, as the fracture surface is typically coated with oxidation products caused by the corrosive

1182 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis will develop at the cell boundaries, and the fracture surface can often exhibit a dendritic appearance from the solidification structure of the alloy. A typical example of a heat-resistant stainless steel fitting is shown in Fig. 20. REFERENCES

200 µm

1 mm

Fig. 19

Stress-corrosion cracking at the tip of a crack in ductile iron. Original magnification: 200. Courtesy of Stork Technimet, Inc. New Berlin, WI

Fig. 20

environment. Other alloys, such as ductile iron and steel, can also incur this type of cracking. The crack-tip region of a ductile iron head is shown in Fig. 19. Elevated-Temperature Fracture (Creep). Cast alloys that have been exposed to elevated temperatures for extended periods can incur

creep failure if the operating temperature is above the range recommended for the specific alloy. The fine features of the fracture surface will typically be totally obliterated from oxidation, and the texture will correspond to the macrostructure of the alloy. Creep voids and cracks

Elevated-temperature rupture in a heatresistant stainless steel. Original magnification: 40. Courtesy of Stork Technimet, Inc. New Berlin, WI

1. B.A. Miller, Overload Failure, Failure Analysis and Prevention, Vol 11, ASM Handbook, ASM International, 2002, p 671–699 2. R.A. Lund, Fatigue Fracture Appearances, Failure Analysis and Prevention, Vol 11, ASM Handbook, ASM International, 2002, p 627–640 3. W.T. Becker, Mechanisms and Appearances of Ductile and Brittle Fracture in Metals, Failure Analysis and Prevention, Vol 11, ASM Handbook, ASM International, 2002, p 587–626 4. Failures Related to Casting, Failure Analysis and Prevention, Vol 11, ASM Handbook, ASM International, 2002, p 103–155

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1183-1191 DOI: 10.1361/asmhba0005343

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Casting Failure Analysis Techniques and Case Studies Craig Brown, Stork Technimet, Inc.

Introduction Failure analysis of cast components typically is conducted in accordance with the techniques described in several articles in Failure Analysis and Prevention, Vol 11 of the ASM Handbook (Ref 1–4). The objective of this article is to review the failure analysis process with specific reference to the considerations that should be addressed when a casting has failed. The failure analysis methodology is demonstrated for three failed cast components: an aluminum bracket, a bronze suction roll, and a steel automotive spindle. The techniques used for failure analysis of castings require a thorough understanding of how the casting was made and processed in order to properly identify how a cast part failed. The scope of the investigation depends on the definition of the problem, but in all cases the root cause for how and why a casting failed can only be determined if the proper background information is collected, representative castings are examined, thorough inspections are conducted, the appropriate material evaluations are completed, and the service conditions to which the castings are exposed are clearly understood. It is most important to clearly determine how a casting failed before it can be decided why it failed. The failure mechanism must be identified prior to conclusions being drawn as to what happened and before recommendations are made for fixing the problem. Therefore, it is important that an organized and properly structured investigation be conducted using the appropriate techniques for the issue involved. It is also imperative to have a cross-functional team consisting of metallurgists, manufacturing staff, product engineers, field service engineers, and application engineers working together to analyze the failure to understand the root causes. Depending on the type of casting failure, there are a variety of reasons for wanting to clearly understand how and why a part failed. In some cases the problem is related to an inprocess quality-control issue; in others it is related to problems at the customer’s assembly plant. Also, not all failures occur in the field,

and, in many cases, the failure investigation is in support of a product engineering department that is making changes to design or manufacturing processes. Often, the warranty department or legal representatives need to understand the root cause of a failure to resolve financial claims. Sometimes the failure is related to raw material and/or processing problems that have affected the quality of the casting. The techniques used for investigating a casting failure are influenced by the possible causes of the issue involved. The evaluation and selected tests should take into account the possible causes for casting failures, which may include the incorrect material used or specified, the presence of casting defects or anomalies, casting processing problems, design issues, improper service conditions or maintenance, and abuse of the casting with regard to the design intent. In order to properly conduct a failure analysis investigation, a structured approach must be used. This involves obtaining casting background information, planning the evaluation and selecting the appropriate casting for analysis, conducting a preliminary examination, conducting the proper material evaluations, and thoroughly evaluating the test data.

Background Information Before any testing is conducted, a significant amount of background information needs to be known about the casting under investigation. Although this stage of a failure analysis is not always thought of as an element of “failure analysis technique,” it is an essential step in the process. All of the subsequent test results are limited somewhat if the casting background information is not known. In many cases the available background information is limited, but the following data should be obtained: Casting information  Identification of component (name and part

number)

 Material composition (alloy grade)

 Applicable           

specifications and drawings (casting and machining) Casting manufacturing process (sand, permanent mold, die casting, investment, centrifugal) Number of cavities on pattern/die In-gate and riser locations Casting date and shift Heat treatment Mechanical properties Physical characteristics Manufacturing process steps Weld-repair practice Attachment method and location within the assembly Exposure to process fluids

Service Conditions  Date and shift of final product assembly and/

or shipment

 Service duration and storage  Operating environment: temperature, atmo-

sphere, corrosive conditions

 Frequency of occurrence of failures  Types and location of loading on the casting:

tension, compressive, torsion, cyclic, impact

 Anomalous service conditions  Repair and rework history

The manufacturing and service history of a casting is especially important with regard to failures related to corrosion problems, as knowledge about the residue on the surface and information concerning the exposure to protective or corrosive environments can aid in the interpretation of the data. The heat treatment and service history is important for stressrelated failures since the fracture mode can be influenced by the prior thermal treatments or anomalous operating conditions. It is important to know the process and service history of the part, as an assessment of the possible residual stresses within the casting and the applied stress from assembly and the service environment is needed to be able to interpret the results of the study. For example, weld repair or a quench and temper treatment can result in significant residual stress in a steel casting.

1184 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis

Planning and Sample Selection The success of a failure analysis investigation depends on the planning of the project and selection of appropriate material for evaluation. It is important that the scope of the project be defined at the initial stages of the investigation. The scope depends on a variety of factors that include the material available for evaluation, the timing for completion of the evaluation, and, of course, cost considerations. Under ideal conditions, there would be several samples to investigate, several weeks for the examination and testing, and no limit to the cost. In reality, most investigations have to be conducted with limited representative samples of the failure condition, a defined timetable, and at a minimum cost. Therefore, it is important that the material be carefully examined and that the samples selected are the most representative of the problem. Examples of failed parts must be thoroughly evaluated. In addition, it can be helpful to simultaneously evaluate comparable unfailed parts to differentiate any features from the failed parts. In many cases, old parts that have performed properly can be compared to new parts that are representative of the problem. Once the representative samples have been selected, the appropriate inspections and tests are defined for the issue at hand. It is also important to look at the entire system in which the failed casting was used. Some important hints for the failure may exist on a different part that may have been installed away from the casting in the system. Try to obtain the entire assembly in which the failure occurred, not just the failed casting.

Preliminary Examination Visual inspection of the failed component, and any unfailed parts if they are available, is a critical element of a failure analysis. Too often, this part of the study is not conducted properly out of eagerness to start the material testing. However, a significant amount of information about the casting and the history of its environmental exposure can be obtained at this stage of the investigation. Very often, small innocuous features are important clues in solving the puzzle of how the part failed. The failed part should be carefully examined and the features documented. The record can be written, but in most cases digital photographs should be taken of all pertinent features of the casting. Once subsequent testing has been conducted, the original features can be lost. A clear understanding and a complete record of the following items must be obtained:  Scope and extent of damage to the casting  Appearance of the fracture, if the casting is

cracked or broken  Determination of the fracture origin, the direction of crack propagation, and identification of the final fracture region

 Presence of nonmetallic inclusions, shrink     

age, or gas porosity or any other anomalies at or near the fracture origin Damage and wear to the surface of the part, both near to and away from the failure Dents, scuff marks, and collateral damage to the casting Presence of staining, fretting, or discoloration on the surface that appears to be related to the failure Extent of corrosion or pitting to the casting if the failure is corrosion related Deposits of debris, grease, or other material on the surface of the part Weld-repair evidence

Each of the various features provides information about the history of the casting. In many cases the features are appropriate for the casting, such as a shot-blasted surface or a drag mark from a die. In other cases, the features provide a record of the manufacturing process and service history of the part. It is important to understand what features should be present, based on the previously obtained background information, and what features are unusual or unique that would provide information about the failure. Before the part is destructively tested, it is important that there is a good understanding of all of the observations that have been obtained. Dimensional measurements should be taken during or after visual inspection, and the records identifying the locations should be studied. It is extremely beneficial to have somebody who is familiar with the product available for discussion during this stage, as some innocuous marks on an assembly away from the casting may hold a very important key for the identification of failure cause. For example, fretting marks in bolted joints away from the casting may be a sign of unexpected resonance vibration to the casting, which resulted from a loss of bolt torque in the joint. It is often very difficult to connect the casting failures to these remote wear marks without consulting with a product expert. It is very important to examine all fracture surfaces to determine the actual fracture origin. The relationship between the crack-initiation site and any anomalies in the casting must be understood. Castings may have shrinkage porosity, oxide inclusions, or other casting anomalies. In many instances, the fracture exposes these features. Just because they are present does not mean that they are related to the failure. Cracks will take the path of least resistance through the material, and this often exposes casting anomalies. It is important that there be clear evidence that the fracture started at the anomaly to conclude that it contributed to the failure. In some cases, anomalies are present near the fracture origin but are not the cause of the failure. It is important to determine if the part would have failed at the identified location even if the anomaly had not been present. This determination requires an understanding of how

the part is used and how the stresses were applied at the time of the failure. This issue emphasizes the importance of visual inspection as a critical component of the failure analysis process. Nondestructive evaluation (NDE) of the casting is often an important step in the failure analysis process. This typically involves the following processes:  Radiographic examination (RT) to evaluate

the casting for subsurface features such as cracks, shrinkage porosity, gas porosity, nonmetallic inclusions, and other casting anomalies  Ultrasonic inspection (UT) to evaluate the casting for subsurface features such as cracks, nonmetallic inclusions, or porosity  Magnetic particle inspection (MT) of ferrous castings to investigate the presence of surface and near-surface cracks and other linear indications  Liquid penetrant inspection (PT) of nonmagnetic castings to investigate the presence of surface cracks and porosity It is important to recognize that the nondestructive examination process can contaminate the casting surface. This is especially true for PT, MT, and UT. Therefore, if chemical analysis of the casting surface, fracture, or corrosion products is required, the chemical analysis should be conducted prior to any nondestructive testing. In some cases the NDE is intentionally not conducted as the value of the information obtained may be of less importance than the information that can be obtained from a subsequent chemical analysis. Careful visual examination often provides sufficient details concerning the failure of the part.

Material Evaluation After completion of the preliminary examination and the proper photographic documentation of the failed casting, an evaluation of the fracture and material properties should be conducted. This typically involves fractography and microscopic examination followed by evaluation of the chemical composition, mechanical properties, and microstructure. It should be remembered that one test is worth many expert opinions, and actual data are more meaningful than assumptions about the properties of the part. Test results that confirm that a property meets specification, or is “typical,” are just as important as data that show a property to be abnormal. Therefore, sufficient testing should be conducted to allow a proper evaluation of the casting.

Fractography Examination of the fracture surface should be conducted using both macroscopic and microscopic techniques. Macroscopic examination

Casting Failure Analysis Techniques and Case Studies / 1185 of the fracture surface, or other features of interest, would include examination of the casting with the unaided eye and with an optical stereomicroscope up to magnifications of about 50. It is most important to study the features at the fracture origin. The location of the origin in relation to the casting surface and any other features should be identified. The features of any anomalies in the material should be studied and the direction of crack propagation through the material should be determined. This is especially important for castings that show the presence of nonmetallic inclusions, oxide skins, porosity, and weld repairs. Macroscopic examination often reveals the direction of the fracture. For example, for the case of bending-induced fracture, the side of the fracture surface that lies normal to the casting surface fractured earlier than the side on which the fracture surfaces do not intersect normal to the casting surface. Cracks typically will initiate at a site where stress concentration has occurred, and this may not always be at the surface of the part. This is especially true for parts that have failed by fatigue. In addition to examining the fracture origin, other features of the failure, such as beach marks, arrest lines, and steps in the fracture should be studied. With most fractures, the direction of crack propagation or extension through the material can be determined by examination under oblique illumination. Ridgelike fractures features, such as chevron marks, will typically point back to the fracture origin. It is often helpful to examine the casting and draw ink arrows on the part adjacent to the fracture to indicate the direction of the crack propagation. It is important to study intersecting cracks to determine which crack occurred first. In this way, the origin can be clearly identified. The features at the origin will expose how the part failed. If the origin is not identified, then the mechanism of how the part failed may be only conjecture. Microscopic examination of the casting involves examination of the part at magnifications up to 10,000 or higher, typically using a scanning electron microscope (SEM). The critical features found during visual examination should be studied at high magnification. It is most important that the features at the origin(s) be examined. The shape, size, and characteristics of the fine details should be studied and representative micrographs taken. The fracture mechanism should be identified at the location where the fracture started. Too often, a study of the origin is not done properly and the root cause of a failure is missed. It is especially important to study the fracture origin regions of castings, as the parts can have a complex microstructure and may contain anomalies, such as nonmetallic inclusion, porosity, or weld-repair artifacts. The fracture mode should correlate with the microstructure of the material as well as the expected stresses on the part. Examination of the fracture surface away from the origin may reveal features that are different from those at the origin, and the crack could be

propagating via a different mechanism. For example, a casting may have crack initiation by stress-corrosion cracking, but crack propagation could be by fatigue. Corrective action for fatigue alone may not solve the problem. The crack-origin region should be examined at magnifications up to 5000. In many cases, the fine features that help characterize and identify the fracture mode cannot be discerned unless they are studied at magnifications above 1000. Other areas that should be studied are the regions slightly away from the origin and at the midfracture zone. This allows confirmation of the fracture mode at locations where the fracture extended through the material. It is also important that the fracture mode at the final fracture zone be identified to clarify whether the cast material was inherently ductile or brittle. As with all fracture studies, any other features found on visual inspection or with the SEM should be studied and documented. With castings, it is important to know the microstructure of the alloy since the crystallographic structure of the metal will influence the fracture features. The normal features need to be recognized, so that any abnormal characteristics can be identified. For example, fracture through a gray iron can be somewhat complex; it would have smooth, wavy surfaces where the separation occurred at the graphite flakes, ductile rupture at the ferrite-rich zones, and cleavage fracture at the pearlite-rich zones. Shrinkage porosity may exhibit globular cavities with a dendritic structure, which would be different from the fracture through the graphite flakes. Any deposits or nonmetallic materials should also be studied to characterize their features. The features help determine whether the deposits are anomalies in the material, corrosion products, postfracture contamination, or some other condition. The shape, size, and distribution of the deposits help tell the story of how the casting failed. There should always be an explanation for all of the features found on examination. It is often beneficial to take micrographs at low, medium, and high magnification at any selected location. This allows the investigator to correlate the features depicted at high magnification with the surrounding fracture surface.

Chemical Analysis The chemical composition of any failed casting should be determined to confirm that the part was made from the specified material. The analysis should be conducted using the appropriate analytical technique for the alloy, with the analysis verified against check standards of a similar alloy. Some of the more common analytical techniques include inductively coupled plasma spectroscopy (ICP) with combustion analysis for the carbon and sulfur, optical emission spectroscopy (OES), glow discharge spectroscopy (GDS), x-ray florescence spectroscopy (XRF), and other variations of these methods. X-ray

diffraction (XRD) may need to be used to characterize any deposits or unusual phases. It should be noted that some castings could exhibit significant variations in the composition caused by alloy segregation during solidification. This is especially true with heavy section castings and should be taken into account when the material is being sampled. For castings in which segregation is commonly observed, such as aluminum alloys, the chemical analysis samples for OES should be prepared by melting several specimens and casting them into sample disks for subsequent analysis. An analysis of a single piece cut from the casting can give an erroneous result. Variation can also occur from part to part when an additive has been added to the melt prior to pouring. For example, cast iron may exhibit a loss of magnesium when comparing a part from the beginning of the pour to a part from the end of the pour. At a minimum, the weight percent of the elements defined in the standard composition ranges should be determined. With some materials, such as carbon and low-alloy steels, consideration should be given to determining the concentrations of other elements, such as boron, titanium, niobium, vanadium, calcium, nitrogen, hydrogen, and oxygen as these elements influence the hardenability, precipitation of carbides, shape of the manganese sulfides, and susceptibility to brittle fracture. With siliconcontaining aluminum castings, determination of the strontium and sodium contents can help interpret the expected degree of modification of the eutectic silicon in the microstructure. Metals often contain alloying elements at low concentrations that drastically change the microstructures and properties. These elements may not always be specified in the standard ranges. For example, the manganese sulfide shape can be altered depending on small variations in the aluminum and titanium content of low-alloy steel. In-depth understanding of metallurgy is necessary to interpret the analysis of chemical composition against observed failures. Energy dispersive x-ray spectroscopy (EDS) is used with the scanning electron microscope to analyze the surfaces studied during a failure investigation. In most instances this technique does not provide quantitative results, but does provide the relative proportions of the elements detected. It is a very powerful tool, but the limitations of the technique should be appreciated. The compositions of very small areas and deposits may be determined, but it is important to interpret the results properly. EDS analysis should be conducted on the surfaces and deposits of interest, but also on clean areas of the base metal to compare the results with those expected for the alloy under the same test conditions. This is especially true for EDS analysis of fracture surfaces. Typically, fractures occur through and around phases and precipitates in the matrix, resulting in a disproportionately high concentration of those elements in the exposed phases. For example, the fracture through a silicon-containing aluminum alloy,

1186 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis such as 380.0, will expose a significant amount of the eutectic silicon particles. It is not unusual for EDS analysis of the fracture surface of a die casting to exhibit a silicon content of over 20%, while the base metal may only actually contain 10%. Therefore, the interpretation of the EDS test results must be handled in a similar manner as the fracture analysis, as the results are dependent on the fracture mode and the microstructure of the alloy. Another caution associated with the EDS analysis is the excitation volume of the x-ray. Under a typical analysis mode, the x-ray emission occurs within a layer that is up to 5 mm beneath the surface. Elements present in secondary phases below the surface that are not visible by the secondary electron images may show up in the x-ray spectrum. This can confuse an identification of contaminants in the analysis. The peak overlap is another source of confusion in this analysis. In some ferrous alloys, peaks of elements found in the slag can overlap with peaks of elements that are commonly observed in the base material.

Mechanical Testing Mechanical testing of a failed casting should be conducted to determine whether the properties of the part meet the specification requirements. Usually the hardness and tensile properties are determined, but with steel castings and other high-alloy materials, the impact properties as well as fracture toughness should also be considered for evaluation. In some cases, the shear or fatigue properties of the material may need to be determined. In most cases, if the specimens were extracted from the proper location, determination of these properties confirms that the castings met the expected values. The mechanical test results, along with the microstructure, often confirm that the casting was properly heat treated. Mechanical testing should be done for most failure investigations, as knowing that the mechanical properties are acceptable is just as important as knowing that they are not. Brinell hardness testing typically is used for failed castings since the indentation covers a greater surface area than the Rockwell hardness test resulting in less error from variations in the microstructure. Rockwell hardness testing of steel and aluminum castings provides acceptable data, but should not be used on iron castings because the graphite in the microstructure will result in values that are skewed to the soft side. It is important that the hardness impression be made on sound material and not on subsurface regions that have shrinkage porosity. With steel castings that have been heat treated, it is important that the test be conducted on material below any surface decarburization. Microhardness testing may be appropriate when localized heat treatments, such as carburizing or induction hardening, have been conducted on the failed casting. Knoop or Vickers microhardness tests can be conducted, but the choice of which test method to use is dictated

by the hardness conversion tables available for evaluation of the test results. The ASTM standard E 140 is used for converting the data to the customary Brinell and Rockwell hardness values. Hardness surveys through the selected area usually are conducted, but it is important to know the microstructure at the area tested because any anomalies, such as decarburization, will influence the test results. The case depth or thickness callout specified on a component drawing is often incomplete. The investigator should understand the correct meaning of the case-depth requirement. For example, is it the total case or the effective case depth to a 50 HRC equivalent? The standard or subsize tensile specimens usually provide acceptable test results if the specimens have been extracted from sound material. The objective of the mechanical testing is to determine the properties of the alloy as near to the fracture origin as possible from sound metal that is representative of the material. It is important that the tensile specimen not contain significant amounts of shrinkage porosity in the reduced section of the bar. If shrinkage porosity, oxide inclusions, or other anomalies are present, the specimen generally will exhibit ultimate tensile strength and percent elongation values less than expected for the sound material. In many cases, the machining of round tensile specimens shifts the reduced section of the tensile bar toward the centerline of the section, and this can result in low tensile and elongation values. In these cases, flat tensile specimens taken from the midradius region or near the surface of the casting will provide more representative values for the material. In some cases, the material specification provides allowance for extraction of the test specimens from castings as opposed to separately cast test coupons, but selection and agreement on the location from which the tensile specimens are extracted is an important part of a failure investigation. It should also be understood that the cast alloys can exhibit significant variation in properties at different locations within the part, even if sound specimens can be extracted. The cooling rate during solidification will influence the properties. For example, with cast aluminum alloys, there is a direct correlation between the dendrite arm spacing and the mechanical properties. This should be taken into account when evaluating the test data since the tensile specimen can seldom be extracted from the casting at the fracture origin location. Many commonly used cast metals specify their mechanical properties by separately cast tensile specimens. The investigators need to understand the differences between the separately cast tensile specimens and the failed castings. For ductile cast iron parts, the failed parts may be cast in different molding materials from the separately cast coupons. Secondary processes such as heat treatment may reduce the differences between the actual castings and separately cast tensile test specimens. For example,

high-pressure die-cast aluminum parts usually exhibit significantly lower properties than separately cast specimens. These castings are usually used in the as-cast condition; however, the difference in properties may be reduced after T6 heat treatment. The investigators need to understand the behaviors of each casting metal in detail to accurately interpret the results of mechanical testing during failure analysis.

Metallography Examination of the microstructure at the fracture origin is a critical component of a thorough failure analysis. It is important that the microstructure at the location where the failure started be inspected to determine if the part was properly heat treated and to determine if there are any casting anomalies that may have influenced the failure. Too often, failure investigators evaluate the casting at a location away from the origin and assume the microstructure is similar to that at the origin region. This is not always the case, due to the variations in the cooling rate during solidification or heat treatment, the presence of decarburization, and the presence of nonmetallic inclusions, porosity, or some other near-surface anomaly. The preparation of the cross section must be done in such a manner that the microstructure is not altered during cutting, grinding, mounting, and polishing. It is also important that proper specimen preparation be used to maintain good edge retention because the features at the surface of fracture origin are of primary interest. Action should also be taken to ensure a good bond to the mounting media is present or subsequent bleed-out of the etchant can obscure the microstructural features. The microstructure should be examined in both the unetched and etched conditions. With carbon and low-alloy steels it is important to evaluate the shape and distribution of the manganese sulfides in the unetched condition. In some cases it is difficult to see fine cracks in the material after the specimen has been etched. Therefore, the features in the unetched condition should be documented photographically prior to etching the cross section. The specimen should be etched with the appropriate etchant for the alloy involved. In some cases, more than one etchant may need to be used to identify phases such as ferrite, carbides, and unique constituents, such as sigma phase. The microstructure should be evaluated to determine if it is proper for the specified material at the fracture origin, the near-surface zone, and away from the fracture origin. Variations in the microstructure can provide information about the heat treatment and alloy segregation. Castings will often show a dendritic solidification structure with slight alloy depletion in the dendrites and alloy enrichment in the interdendritic regions. This variation can influence the relative proportions of the constituents of the microstructure and may have a bearing on the root cause of the failure.

Casting Failure Analysis Techniques and Case Studies / 1187 Special Testing

Case Studies

In some cases, special testing may be required as part of the failure analysis investigation. Stress analysis, strain gaging, and fatigue testing of components may need to be conducted to determine or confirm the properties of the component/material involved in the failure, or the component/material to be used as part of a corrective action. As with the original failure investigation, a substantial amount of information about the material and component properties is needed prior to testing to set up and conduct a valid test program. In other cases, environmental testing or accident reconstruction may need to be conducted.

The following three case studies demonstrate the application of the failure analysis techniques described. While the approach to determining the causes of the failures is the same, the case studies show how the methodology is adapted for differing materials, failure mechanisms, and failure circumstances.

Data Analysis and Report Preparation On completion of the material evaluation a thorough analysis of the background information, preliminary examination, and the test results must be conducted. The test results should show a correlation with the casting process and the manufacturing operations. It is important to keep an open mind and “let the data tell the story about the casting.” The “picture” of the failure should include all the pieces of the puzzle provided by the data. If any data or piece of information does not fit into the picture or description of the failure, then an improper conclusion can be drawn and further information or data are needed to make sense of the results. In most cases, the simplest explanation of the data is the best solution to the failure analysis problem. Often, very minor details are very important and must be explained in the conclusion. It should also be remembered that arriving at the wrong conclusion could be worse than having no conclusion. It is also important that there are sufficient data to come to a conclusion. If the information is inconclusive, or only partly explains the situation, then this should be stated. All failure analysis projects should conclude with a report that summarizes the background information, the results of the preliminary studies and the material tests, and the conclusions of the study. The scope of the report can range from a short summary statement to a very detailed presentation of the data. It is important to present the data in a logical and clearly written manner, so that the reader can follow the test plan. Typically, the data are presented with only a basic interpretation of the information in the body of the report. The detailed discussion and interpretation of all of the data should be in a separate section. It is important to separate facts from speculation. Different analysts may draw different conclusions from the same results. Therefore, the report writer should first present the data and then provide an interpretation of the findings.

Case Study No. 1: Cast Aluminum Bracket Background. A cast aluminum bracket for a suspension system failed during endurance testing. The casting was manufactured from A356.0 aluminum alloy in the T61 condition. Visual Inspection. Two fractures occurred through the part, and the primary fracture is shown in Fig. 1. Examination of the surfaces showed that the fracture at the top portion of the casting exhibited a small region at the corner of the part that had smoother features compared to the remainder of the surface, shown in Fig. 2. Some dark material, which was indicative of nonmetallic inclusions or oxides, was also present near the surface of the casting. The remainder of the primary fracture surface, and the surfaces of the second fracture, exhibit

features indicative of overload fractures through a cast aluminum alloy. Scanning Electron Microscopy (SEM). Examination of the apparent fracture origin with SEM confirmed the presence of nonmetallic material and porosity, as shown in Fig. 3. Inspection of the fracture surface at the origin region showed the presence of striations, indicating that fatigue fracture had initiated at and near the surface of the casting where the inclusions and porosity were present in the aluminum alloy, as shown in Fig. 4. Some postfracture damage and oxides were present on the surface from rubbing of the mating surfaces of the fracture. The fracture away from the origin exhibited ductile rupture through the aluminum matrix and brittle fracture through the eutectic silicon particles. The relative proportion of fatigue fracture to overload fracture was small. This would indicate that the part had likely been subjected to relatively high cyclic stresses during testing. Energy Dispersive X-ray Spectroscopy (EDS). EDS analysis of the nonmetallic particles showed that they contained primarily silicon and oxygen, with smaller amounts of aluminum, sodium, iron, potassium, magnesium, calcium, and titanium. A significant amount of carbon was also present. The rounded features and composition indicated that

1 mm

Fig. 1

The cast aluminum bracket fractured at the location indicated by the arrow.

Fig. 3

Scanning electron image of the origin region showing the presence of sand grains

100 mm

Fig. 2

The fracture initiated at the surface of the casting. The origin region exhibited both bright fracture features and the presence of oxides.

Fig. 4

Scanning electron image showing striations, which indicate that the fracture propagated through the aluminum alloy via fatigue

1188 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis the material was likely molding sand with residual binder that had been encapsulated by the molten metal during the mold-filling process. Chemical Composition. Chemical analysis of the casting confirmed that the part had been made from the proper alloy and that there were no anomalies or deficiencies in the composition. The results are shown in Table 1. Mechanical Properties. A tensile specimen was extracted from the bracket, and the results of tensile and hardness tests are shown in Table 2. The material exhibited a tensile strength below the minimum requirement for a separately cast specimen, but above the modified requirement for a specimen taken from a casting, based on ASTM B 26, section 26. The specimen broke at yield and had a low percent elongation. This was caused by a substantial amount of gas porosity in the material, which reduces the ductility, and in this case reduced the tensile strength such that the material broke when the alloy started to yield. The Brinell hardness of the part was typical for the specified material. Metallography. A cross section through the fracture origin region showed a substantial amount of gas porosity. This was consistent with the visual inspection, SEM studies, and the tensile specimen. The microstructure consisted of partially spheroidized eutectic silicon and Fe-Si-Al phases in a matrix of aluminum. Magnesium silicide precipitates were present in the matrix. The microstructure was consistent with a solution-treated and artificially aged material, but the degree of spheroidization was inadequate to optimize the mechanical properties. The cross section showed that away from the origin region, brittle fracture had occurred at the eutectic silicon particles, as shown in Fig. 5. This is consistent with the SEM studies. Conclusion. The study concluded that the casting failed by fatigue caused by cyclic stresses that exceeded the fatigue resistance of the material. Crack initiation occurred at a corner of the casting and extended a short distance into the material, with the remainder of the fracture occurring by both ductile and brittle fracture. Sand particles and gas porosity provided sites of stress concentration in the aluminum alloy. Although the casting had not been properly solution treated during heat treatment, this was likely not a major contributing factor because of the other anomalies in the material.

used to flush the fine holes that are drilled through the shell. The roll was taken out of service when a crack was found in the rubber cover. Further inspection of the roll showed extensive cracking at the interior surface of

Table 1 Chemical analysis results for failed aluminum bracket (wt%) Element

Bracket

A356.0 Limits

Silicon Iron Copper Manganese Magnesium Chromium Nickel Zinc Titanium Aluminum

7.36 0.14 0.03 0.01 0.28 <0.001 <0.001 0.024 0.07 bal

6.5–7.5 0.20 max 0.20 max 0.10 max 0.25–0.45 (1) (1) 0.10 max 0.20 max bal

Table 2

the bronze shell, which was under the rubber cover. Visual Inspection. A section of the cracked shell was cut from the roll at the apparent origin region and visually inspected. The section is shown in Fig. 6. Numerous large cracks were present at the interior surface that ran at a slight angle to the longitudinal axis of the shell. In addition, fine secondary cracks extended from the drilled holes of the shell in the areas adjacent to the primary cracks, as shown in Fig. 7. Examination of the crack surfaces revealed a relatively rough texture with a significant amount of oxidation and debris, as shown in Fig. 8. The crack tip regions showed only a slight amount of discoloration. Some corrosion had occurred on the surfaces of the drilled holes at the crack locations. A laboratory-induced fracture through the base metal exhibited features typical for a bronze alloy.

Tensile and hardness test results for failed aluminum bracket

Property

Bracket

Yield strength, 0.2% offset, MPa (ksi) Tensile strength, MPa (ksi) Elongation, % Brinell hardness, HBW 10/500/30

... 201.3 (29.2) 0.5 82.6

Sand cast A356.0-T6 limits

Sand cast A356.0-T6 modified limits

165.5 (24.0) min 234.4 (34.0) min 3.5 min 70–105 (typical)

124.1 (18.0) min 175.8 (25.5) min 0.9 min 70–105 (typical)

Fig. 5

Cross section through the casting at the origin region showing a microstructure typical for a solution treated and artificially aged A356.0 aluminum alloy, but with only limited spheroidization of the eutectic silicon

Fig. 7

Fine jagged cracks are present at the drilled holes in the bronze shell.

Fig. 8

Dark gray corrosion products extend to the tips of the cracks.

Case Study No. 2: Cast Bronze Suction Roll Shell Background. A cast bronze suction roll shell cracked after being in service for several years. The shell was vertically cast using a proprietary process from a C83600 copper alloy and was covered with a proprietary grade rubber material. The suction roll was used in the wet end of a paper mill and was exposed to white water from the paper pulp and to cleaning solutions

Fig. 6

Cracked suction roll shell section

Casting Failure Analysis Techniques and Case Studies / 1189 Scanning Electron Microscopy (SEM). Examination of a typical crack surface with a SEM revealed the presence of oxidation products and deposits on most of the fracture surface. Inspection of the crack tip region revealed features typical for stress corrosion cracking in a bronze alloy, as shown in Fig. 9. Although oxidized, the remainder of the crack surface exhibited a texture indicative of this fracture mode. Energy Dispersive X-ray Spectroscopy (EDS). EDS analysis of the oxidation products and debris showed the presence of elements that would be present in the base alloy (copper, zinc, tin, and lead) as well as elements found in the paper product such as carbon (from the cellulose) and aluminum and silicon (both from the clay additive). The deposits also showed the presence of chlorine and sulfur, which were likely in the form of chlorides and thiosulfates. These compounds can cause stress-corrosion cracking in bronze alloys. Chemical Composition. Chemical analysis via inductively coupled plasma atomic emission spectroscopy of the bronze casting confirmed that the part had been made from the proper alloy and that there were no anomalies or deficiencies in the composition. Results are shown in Table 3. Mechanical Properties. A tensile specimen extracted from the shell had mechanical properties, shown in Table 4, that were typical for the specified copper alloy C83600. Metallography. A cross section through the bronze shell near the interior surface at a drilled

hole showed a microstructure that was typical for the specified cast material. The section also showed fine cracks extending into the material, as shown in Fig. 10. The cracks exhibited jagged paths through the grains and showed the presence of corrosion products. The features were consistent with stress-corrosion cracking of a bronze alloy. Conclusion. The study concluded that the primary cause for the failure of the suction roll shell was stress-corrosion cracking. Crack initiation occurred at multiple locations at the interior surface of the shell at the corners of the drilled holes. The roll had been operating in a relatively severe environment. The combination of a susceptible bronze alloy, a corrosive environment, and applied stresses from operation of the roll resulted in the formation of numerous cracks. Although the crack surfaces that were examined exhibited features typical for stress-corrosion cracking, it is possible that some of the cracks may have propagated through the shell via fatigue, as the roll would have been subjected to cyclic stresses. Even though the shell had been made from the specified material, the environment was too severe for the bronze alloy.

quenched and tempered to a hardness of 241 to 311 HBW. A crack was found in the part during a maintenance inspection after it had been in service for several years. Visual Inspection. A circumferential crack over 20 cm (8 in.) long was present at the bottom surface of the spindle adjacent to a wheel bearing. Crack initiation occurred near the machined surface where the fracture exhibited a rough texture, as shown in Fig. 11. Beach marks extended outward from the origin area, indicating that a progressive fracture, such as fatigue, had likely occurred. Inspection of the fracture surface at the origin region revealed globular features indicative shrinkage porosity, as shown in Fig. 12. A narrow band of smooth fracture surface was present between the dendritic structure of the shrinkage porosity and the machined surface of the part. This smooth region also exhibited the beach marks or arrest lines that were present on the primary fracture surface. Scanning Electron Microscopy (SEM). Examination of the apparent fracture origin with SEM confirmed the presence of shrinkage porosity at the origin region. Crack initiation occurred at multiple locations at both the outboard and inboard sides of the porosity.

Case Study No. 3: Cast Steel Spindle Background. A cast steel spindle for an offroad vehicle failed in service. The part had been sand cast from Grade 6830 steel and had been

Table 4 Tensile and hardness test results for failed bronze suction roll Properties

Yield strength, 0.2% offset, MPa (ksi) Tensile strength, MPa (ksi) Elongation, % Brinell hardness, HBW

Fig. 9

Suction roll shell

Copper alloy C83600 (typical)

131.0 (19.0)

117.2 (17.0)

261.3 (37.9)

255.1 (37.0)

29 69.1

30 60

Fig. 11

The spindle fracture exhibits an origin region near the machined surface.

Fig. 12

The origin region exhibits a rough, globular region with features indicative of shrinkage

Scanning electron image of the crack tip area showing features indicative of stress-corrosion

cracking.

Table 3 Chemical analysis results for failed bronze suction roll (wt%) Element

Copper Tin Lead Zinc Iron Antimony Nickel Phosphorous Silicon

Suction roll shell

Copper alloy C83600 limits

85.3 4.1 4.3 4.5 0.1 0.04 0.7 0.03 0.005

84.0–86.0 4.0–6.0 4.0–6.0 4.0–6.0 0.30 max 0.25 max 1.0 max 0.05 max 0.005 max

Fig. 10

Cross section through the bronze alloy near a drilled hole shows jagged cracks filled with corrosion product, typical for stress-corrosion cracking.

porosity.

1190 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis On the outboard side, the cracks had joined together and propagated to the machined surface of the part, as shown in Fig. 13. On the inboard side, the cracks joined to form a wide crack that propagated through the full thickness of the spindle wall. Examination of the fracture surface at high magnification revealed the presence of striations, which confirmed that the casting had failed by fatigue. The striations are shown in Fig. 14. Chemical Composition. Chemical analysis of the casting, shown in Table 5, confirmed that the part had been made from a typical cast steel and that there were no anomalies or deficiencies in the composition. Mechanical Properties. A tensile specimen was extracted from the spindle near the fracture location, and the test results are presented in Table 6. The specimen cut from the casting had mechanical properties below the minimum requirement for a separately cast specimen. Three 10 by 10 mm (0.4 by 0.4 in.) Charpy V-notch impact specimens were also extracted from the spindle and tested at 20  C ( 4  F). The average impact toughness value of the specimens was well below the specified minimum, and the fracture surfaces exhibited brittle fracture features. The Brinell hardness was at the

low end of the specification. The low mechanical property values are not unusual for specimens cut from a heavy section casting with the measured chemical composition. The limited hardenability of the alloy and heavy mass of the spindle casting would make it difficult to obtain the proper microstructure needed to produce the optimum mechanical properties. Metallography. A cross section through the fracture origin region showed the presence of shrinkage porosity near the origin region and a microstructure of martensite and bainite, as shown in Fig. 15. The bainitic structure is consistent with the poor mechanical properties of the steel and indicates that the casting had not been quenched properly during heat treatment. The cross section also showed that the nearsurface region at the fracture origin had been weld repaired, as shown in Fig. 16. The repair had likely been conducted during the manufacturing process to remove anomalies in the steel casting, such as shrinkage porosity. Not all of the porosity was removed, and the remaining porosity was covered over with a layer of weld metal.

Microhardness Testing. Knoop microhardness tests were conducted on the cross section to determine the hardness of the repaired region. The results showed that the weld metal exhibited an equivalent hardness of 31 HRC, whereas the heat-affected zone of the base metal was considerably harder at 45 HRC. This indicates that the weld repair was conducted after the casting had been heat treated and that minimal or no preheat had been used. Conclusion. The study concluded that the casting failed by fatigue caused by cyclic bending stresses that exceeded the fatigue properties of the material. Crack initiation occurred below the surface of the casting at a cluster of shrinkage porosity. The small pores in the steel provided sites for stress concentration and crack initiation. The multiple fatigue cracks propagated toward the machined surface of the part and through the wall of the spindle. Although the mechanical properties of the steel did not meet the specification requirements, the values were typical for a heavy section steel casting made from an alloy with limited hardenability. The poor values are not considered a primary contributing factor to the failure. The evaluation showed that a primary cause for the failure was a poor weld repair that had not properly removed the near-surface shrinkage porosity and likely resulted in a complex residual stress pattern in the steel at the stressed region. ACKNOWLEDGMENT Case studies were provided by Stork Technimet, Inc., New Berlin, WI.

Table 5 Chemical analysis results for failed cast steel spindle (wt%)

Fig. 15 Fig. 13

Scanning electron micrograph of origin region showing multiple cracks initiating at the porosity and propagating toward the surface of the casting

Just below the casting surface the fracture shows features indicative of shrinkage porosity. The microstructure at the porosity consists of tempered martensite, whereas the remainder of the section is upper bainite and ferrite.

Element

Casting

Material specification

Carbon Manganese Silicon Sulfur Phosphorous Chromium Nickel Molybdenum Vanadium Aluminum

0.27 0.73 0.35 0.012 0.009 0.39 0.58 0.20 0.007 0.037

0.35 max Not specified Not specified 0.045 max 0.04 max Not specified Not specified Not specified 0.12 max 0.06 max

Table 6 Mechanical test results for failed cast steel spindle

Fig. 16 Fig. 14

Scanning electron image exhibiting striations indicative of fatigue fracture on the primary fracture in the steel casting

Cross section through the origin region showing the presence of a repair weld at the surface of the part. The repair was conducted after the casting had been heat treated.

Property

Spindle

Yield strength, 0.2% offset, MPa (ksi) Tensile strength, MPa (ksi) Elongation, % Reduction of area, % Charpy V-notch impact toughness, J (ft
lbf) Brinell hardness, HBW

582.6 (84.5) 765.3 (111.0) 11 22 16 (12)

Material specification

14 min 30 min 27 (20) min

241

241–311

658.4 (95.5) min 792.9 (115.0) min

Casting Failure Analysis Techniques and Case Studies / 1191 REFERENCES 1. Failures Related to Casting, Failure Analysis and Prevention, Vol 11, ASM Handbook, ASM International, 2002, p 103–155

2. D.P. Dennies, Organization of a Failure Analysis, Failure Analysis and Prevention, Vol 11, ASM Handbook, ASM International, 2002, p 324–350 3. I. Le May, Examination of Damage and Material Evaluation, Failure Analysis and

Prevention, Vol 11, ASM Handbook, ASM International, 2002, p 351–370 4. Practices in Failure Analysis, Failure Analysis and Prevention, Vol 11, ASM Handbook, ASM International, 2002, p 393–417

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1192-1202 DOI: 10.1361/asmhba0005344

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Common Defects in Various Casting Processes Rathindra DasGupta, National Science Foundation

THE FUNCTIONALITY OF CASTINGS is strongly dependent on casting soundness. Thus, castings need to be virtually free of defects that can result in premature failure of castings, broken machine tools, or poor mechanical properties. Traditionally, unique names have been used to describe various casting imperfections; however, the International Committee of Foundry

Technical Associations has identified seven basic categories of casting defects (Ref 1):

 Incorrect dimension  Inclusions or structural anomalies

    

Table 1 (Ref 1) presents some of the common defects in each of the seven categories. Also included in this article are select case studies relevant to inclusions, cavities (gas porosity, shrinkage), and discontinuities (hot tearing, cold shut).

Metallic projections Cavities Discontinuities Defective surfaces Incomplete casting

Table 1 International classification of common casting defects No.

Common name

Description

Sketch

Metallic Projections A 100:

Metallic projections in the form of fins or flash

A 110:

Metallic projections in the form of fins (or flash) without change in principal casting dimensions

A 111..........Thin fins (or flash) at the parting line or at core prints

A 112..........Projections in the form of veins on the casting surface

No.

Common name

Description

A 200:

Massive projectiions

A 210:

Swells

A 212(a)......Excess metal in the vicinity of the gate or beneath the sprue

Erosion, cut, or wash

A 213(a)......Metal projections in the form of elongated areas in the direction of mold assembly

Crush

Sketch

Joint flash or fins

Veining or finning A 220:

A 113..........Network of projections on Heat-checked the surface of die castings die

Projections with rough surfaces

A 221(a)......Projections with rough surfaces on the cope surface of the casting

Mold drop or sticker

(continued) (a) Defects that under some circumstances could contribute, either directly or indirectly, to casting failures. Adapted from International Atlas of Casting Defects, American Foundrymen’s Society, Des Plains, IL

Common Defects in Various Casting Processes / 1193 Table 1 No.

(continued) Common name

Description

A 114(a)......Thin projection parallel to a casting surface, in re-entrant angles

A 115..........Thin metallic projection located at a re-entrant angle and dividing the angle in two parts

A 120:

Sketch

No.

Description

Common name

Fillet scab

A 222(a)......Projections with rough surfaces on the drag surface of the casting (massive projections)

Raised core or mold element cutoff

Raised sand

Fillet vein

A 223(a)......Projections with rough surfaces on the drag surface of the casting (in dispersed areas)

A 224(a)......Projections with rough surfaces on other parts of the casting

Mold drop

Sketch

Metallic projections in the form of fins with changes in principal casting dimensions

A 123(a)......Formation of fins in planes related to directiion of mold assembly (precision casting with waste pattern); principal casting dimensions change

Cracked or broken mold

A 225(a)......Projections with rough surfaces over extensive areas of the casting

A 226(a)......Projections with rough surfaces in an area formed by a core

Corner scab

Broken or crushed core

Cavities B 100:

Cavities with generally rounded, smooth walls perceptible to the naked eye (blowholes, pinholes)

B 124(a) .... Small, narrow cavities in the form of cracks, appearing on the faces or along B 110: Class B 100 cavities internal to the casting, not extending to the surface, edges, generally only after discernible only by special methods, machining, or fracture of the machining casting B 111(a) .... Internal, rounded cavities, usually smooth-walled, of varied size, isolated or grouped irregularly in all areas of the casting

B 112(a) .... As above, but limited to the vicinity of metallic pieces placed in the mold (chills, inserts, chaplets, etc.)

Blowholes, pinholes

Dispersed shrinkage

B 200: Cavities with generally rough walls, shrinkage B 210: Open cavity of Class B 200, sometimes penetrating deeply into the casting

Blowholes, adjacent to inserts, chills, chaplets, etc.

B 211(a) .... Open, funnel-shaped cavity; wall usually covered with dendrites

Open or external shrinkage

B 212(a) .... Open, sharp-edged cavity in fillets of thick castings or at gate locations

Corner or fillet shrinkage

(continued) (a) Defects that under some circumstances could contribute, either directly or indirectly, to casting failures. Adapted from International Atlas of Casting Defects, American Foundrymen’s Society, Des Plaines. IL

1194 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis Table 1 No.

(continued) Common name

Description

B 113(a) .... Like B III, but accompanied by slag inclusions (G 122)

Sketch

No.

Description

Common name

Sketch

Slag blowholes B 213(a) .... Open cavity extending from a core

Core shrinkage

B 120: Class B 100 cavities located at or near the casting surface, largely exposed or at least connected with the exterior B 220: Class B 200 cavity located completely internal to the casting B 121(a) .... Exposed cavities of various sizes, isolated or grouped, usually at or near the surface, with shiny walls

Surface or subsurface blowholes

B 122(a) .... Exposed cavities, in re-entrant angles of the casting, often extending deeply within

Corner blowholes, draws

B 221(a) .... Internal, irregularly shaped cavity; wall often dendritic

Internal or blind shrinkage

B 222(a) .... Internal cavity or porous area along central axis

Centerline or axial shrinkage

B 300: Porous structures caused by numerous small cavities B 123 ....... Fine porosity (cavities) at the casting surface, appearing over more or less extended areas

Surface pinholes

B 311(a) .... Dispersed, spongy dendritic shrinkage within walls of casting; barely perceptible to the naked eye

Discontinuities

C 100:

Discontinuities, generally at intersections, caused by mechanical effects (rupture)

C 110:

Normal cracking

C 111(a) .... Normal fracture appearance, sometimes with adjacent indentation marks

B 310: Cavities according to B 300, scarcely perceptible ot the naked eye

Breakage (cold)

C 300:

Discontinuities caused by lack of fusioin (cold shuts); edges generally rounded, indicating poor contact between various metal streams during filling of the mold

C 310:

Lack of complete fusion in the last portion of the casting to fill

C 311(a) .... Complete or partial separation of casting wall, often in a vertical plane

C 320: C 120:

Macro- or microshrinkage, shrinkage porosity, leakers

Cold shut or cold lap

Lack of fusion between two parts of casting

Cracking with oxidation

C 121(a) .... Fracture surface oxidized completely around edges

Hot cracking

C 321(a) .... Separation of the casting in a horizontal plane

Interrupted pour

(continued) (a) Defects that under some circumstances could contribute, either directly or indirectly, to casting failures. Adapted from International Atlas of Casting Defects, American Foundrymen’s Society, Des Plaines. IL

Common Defects in Various Casting Processes / 1195 Table 1 No.

(continued) Description

Common name

Sketch

No. C 330:

C 200:

Discontinuities caused by internal tension and restraints to contraction (cracks and tears)

C 210:

Cold cracking or tearing

C 211(a) .... Discontinuities with squared edges in areas susceptible to tensile stresses during cooling; surface not oxidized

C 220:

C 222(a) .... Rupture after complete solidification, either during cooling or heat treatment

Lack of fusion around chaplets, internal chills, and inserts

C 331(a) .... Local discontinuity in vicinity of metallic insert

Chaplet or insert cold shut, unfused chaplet

C 400:

Discontinuities caused by metallurgical defects

C 410:

Separation along grain boundaries

C 411(a) .... Separation along grain boundaries of primary crystallization

Conchoidal or “rock candy” fracture

C 412(a) .... Network of cracks over entire cross section

Intergranular corrosion

Hot tearing

Quench cracking Defective Surface

D 112 ........ Surface shows a network of jagged folds or wrinkles (ductile iron)

Cope defect, elephant skin, laps

D 113 ........ Wavy fold markings without discontinuities; edges of folds at same level, casting surface is smooth

Seams or scars

D 114 ........ Casting surface markings showing direction of liquid metal flow (light alloys)

Flow marks

D 120: Surface roughness

Sketch

Cold tearing

Hot cracking and tearing

C 221(a) .... Irregularly shaped discontinuities in areas susceptible to tension; oxidized fracture surface showing dendritic pattern

Common name

Description

D 100:

Casting surface irregularities

D 110:

Fold markings on the skin of the casting

D 111 ....... Fold markings over rather large areas of the casting

Surface folds, gas runs

D 134 ........ Casting surface entirely pitted or pock-marked

Orange peel, metal mold reaction, alligator skin

D 135 ........ Grooves and roughness in the vicinity of re-entrant angles on die castings

Soldering, die erosion

D 140: Depressions in the casting surface

D 141 ........ Casting surface depressions in the vicinity of a hot spot

Sink marks, draw or suck-in

(continued) (a) Defects that under some circumstances could contribute, either directly or indirectly, to casting failures. Adapted from International Atlas of Casting Defects, American Foundrymen’s Society, Des Plaines, IL

1196 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis Table 1 No.

(continued) Description

Common name

D 121 ........ Depth of surface roughness is approximately that of the dimensions of the sand grains

Rough casting surface

D 122 ........ Depth of surface roughness is greater than that of the sand grain dimensions

Severe roughness, high pressure molding defect

Sketch

No.

Description

D 142 ........ Small, superficial cavities in the form of droplets of shallow spots, generally gray-green in color

D 131 ........ Grooves of various lengths, Buckle often branched, with smooth bottoms and edges

D 133 ........ Irregularly distributed depressions of various dimensions extending over the casting surface usually along the path of metal flow (cast steel)

Flow marks, crow’s feet

D 223 ........ Conglomeration of strongly adhering sand and metal at the hottest points of the casting (re-entrant angles and cores)

Metal penetration

Slag inclusions

D 210: Deep indentation of the casting surface

D 211 ........ Deep indentation, often over large area of drag half of casting

Rat tail

Sketch

D 200: Serious surface defects

D 130: Grooves on the casting surface

D 132 ........ Grooves up to 5.1 mm (0.2 in.) in depth, one edge forming a fold which more or less completely covers the groove

Common name

Push-up, clamp-off

D 220: Adherence of sand, more or less vitrified

D 221 ........ Sand layer strongly adhering to the casting surface

Burn on

D 222 ........ Very adherent layer of partially fused sand

Burn in

D 243 ........ Scaling after anneal

Scaling

Incomplete Casting

D 224 ........ Fragment of mold material embedded in casting surface

D 230:

Dip coat spall, scab

Plate-like metallic projections with rough surfaces, usually parallel to casting surface

D 231(a) ..... Plate-like metallic projections with rough surfaces parallel to casting surface; removable by burr or chisel

Scabs, expansion scabs

E 100:

Missing portion of casting (no fracture)

E 110:

Superficial variations from pattern shape

E 111 ........ Casting is essentially complete except for more or less rounded edges and corners

Misrur

E 112 ........ Deformed edges or contours due to poor mold repair or careless application of wash coatings

Defective coating (tear-dropping) or poor mold repair

(continued) (a) Defects that under some circumstances could contribute, either directly or indirectly, to casting failures. Adapted from International Atlas of Casting Defects, American Foundrymen’s Society, Des Plaines, IL

Common Defects in Various Casting Processes / 1197 Table 1 No.

(continued) Description

Common name

Sketch

No. E 120:

D 232(a) ..... As above, but impossible to eliminate except by machining or grinding

D 233(a) ..... Flat, metallic projections on the casting where mold or core washes or dressings are used

D 240:

Cope spall, boil scab, erosion scab

Adherent packing material

E 125 ........ Casting partially melted or seriously deformed during annealing

Fusion or melting during heat treatment

E 200:

Missing portion of casting (with fracture)

E 210:

Fractured casting

E 220:

Serious variations from pattern shape

E 121 ........ Casting incomplete due to premature solidification

Misrun

E 122 ........ Casting incomplete due to insufficient metal poured

Poured short

E 123 ........ Casting incomplete due to loss of metal from mold after pouring

Runout

E 124 ........ Significant lack of material due to excessive shot-blasting

Excessive cleaning

F 122 ........ Certain dimensions inexact

Irregular contraction

F 123 ........ Dimensions too great in the direction of rapping of pattern

Excess rapping of pattern

F 124 ........ Dimensions too great in direction perpendicular to parting line

Mold expansion during baking

F 125 ........ Excessive metal thickness at irregular locations on casting exterior

Soft or Insufficient ramming, mold-wall movement

Fractured casting

Piece broken from casting

E 221 ........ Fracture dimensions correspond to those of gates, vents, etc.

Sketch

Oxide scale

D 242 ......... Adherence of ore after malleablizing (white heart malleable)

E 211 ........ Casting broken, large piece missing; fractured surface not oxidized

Common name

Blacking scab, wash scab

Oxides adhering after heat treatment (annealing, tempering, malleabizing) by decarburization

D 241 ......... Adherence of oxide after annealing

Description

Broken casting (at gate, riser, or vent)

(continued)

(a) Defects that under some circumstances could contribute, either directly or indirectly, to casting failures. Adapted from International Atlas of Casting Defects, American Foundrymen’s Society. Des Plaines, IL

1198 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis Table 1 No. E 230:

(continued) Description

Common name

Sketch

No.

Description

Common name

Sketch

Fractured casting with oxidized fracture

E 231 ........ Fracture appearance indicates exposure to oxidation while hot

F 126 ........ Thin casting walls over Distorted general area, especially on casting horizontal surfaces

Early shakeout

F 200:

Casting shape incorrect overall or in certain locations

F 210:

Pattern incorrect

Incorrect Dimensions or Shape F 100:

Incorrect dimensions; correct shape

F 110:

All casting dimensions incorrect

F 111 ........ All casting dimensions incorrect in the same proportions

F 120:

F 212 ........ Casting shape is different from drawing in a particular area; pattern is correct

F 222 ........ Variation in shape of an internal casting cavity along the parting line of the core

F 223 ........ Irregular projections on vertical surfaces, generally on one side only in the vicinity of the parting line

Inclusions or Structural Anomalies

Shift

G 100:

Inclusions

G 110:

Metallic inclusions

Metallic inclusions

Shifted core

G 111(a) ..... Metallic inclusions whose appearance, chemical analysis or structural examination show to be caused by an element foreign to the alloy

Cold shot

Ramoff, ramaway

G 112(a) ..... Metallic inclusions of the same chemical composition as the base metal; generally spherical and often coated with oxide

G 113 ........ Spherical metallic inclusions inside blowholes or other cavities or in surface depressions (see A 311). Composition approximates that of the alloy cast but nearer to that of a eutectic

Internal sweating, phosphide sweat

Projection Joint

Projection Joint

Deformations from correct shape Pattern

F 231 ........ Deformation with respect to drawing proportional for casting, mold, and pattern

Pattern mounting error

Hindered contraction

Shift or Mismatch

F 221 ........ Casting appears to have been subjected to a shearling action in the plane of the parting line

F 230:

Improper shrinkage allowance

Certain casting dimensions incorrect

F 121 ........ Distance too great between extended projections

F 220:

F 211 ........ Casting does not conform to Pattern error the drawing shape in some or many respects; same is true of pattern

Deformed pattern

Mold Casting

G 120:

Nonmetallic inclusions; slag, dross, flux

(continued) (a) Defects that under some circumstances could contribute, either directly or indirectly, to casting failures. Adapted from International Atlas of Casting Defects, American Foundrymen’s Society, Des Plaines, IL

Common Defects in Various Casting Processes / 1199 Table 1

(continued)

No.

Description

F 232 ........ Deformation with respect to drawing proportional for casting and mold; pattern conforms to drawing

Common name

Sketch

No.

Pattern

Deformed mold, mold creep, springback

Description

Common name

G 121(a) ..... Nonmetallic inclusions whose appearance or analysis shows they arise from melting slags, products of metal treatment or fluxes

Slag, dross or flux inclusions, ceroxides

G 122(a) ..... Nonmetallic inclusions generally impregnated with gas and accompanied by blowholes (B 113)

Slag blowhole defect

Sketch

Mold Casting

Pattern F 233 ........ Casting deformed with respect to drawing; pattern and mold conform to drawing

Casting distortion

G 130:

Casting

F 234 ........ Casting deformed with respect to drawing after storage, annealing, machining

G 140:

Nonmetallic inclusions; mold or core materials

Mold

Warped casting

G 131(a) ..... Sand inclusions, generally very close to the surface of the casting

Sand inclusions

G 132(a) ..... Inclusions of mold blacking or dressing, generally very close to the casting surface

Blacking or refractory coating inclusions

G 143(a) .... Folded films of graphitic luster in the wall of the casting

Lustrous carbon films, or kish tracks

G 144 ....... Hard inclusions in permanent molded and die cast aluminum alloys

Hard spots

Nonmetallic inclusions; oxides and reaction products

G 141 ....... Clearly defined, irregular black spots on the fractured surface of ductile cast iron

Black spots

G 142(a) .... Inclusions in the form of oxide skins, most often causing a localized seam

Oxide inclusion or skins seams

(a) Defects that under some circumstances could contribute, either directly or indirectly, to casting failures. Adapted from International Atlas of Casting Defects, American Foundrymen’s Society, Des Plaines, IL

Case Studies Inclusions Case Study 1. Figure 1 shows an inclusion that caused tool breakage during machining of an aluminum die casting, and Fig. 2 shows the fracture (intentionally created) through the inclusion. A scanning electron micrograph of the defect is shown in Fig. 3. An x-ray spectrum

obtained from this defect reveals that the chemical composition is essentially magnesium = 14.2%, aluminum = 37.9%, and oxygen = 45.7%, indicating the oxide is spinel, that is, a magnesium aluminate compound with the general formula MgAl2O4. The most likely source of spinel is the melt. Inclusion occurrences (such as spinel) can be minimized by allowing settling time after furnace cleaning, using degassing and filtration

techniques, and avoiding excessive melt temperature. Case Study 2. Figures 4 and 5 are examples of wrinkled oxide skins (aluminum foil near dip well and/or from ladles) and refractory material entrapped in castings, respectively. Entrapment of these types of inclusions contributes to a “non-knitting” (incomplete fusion of metal) condition in the casting, thereby enabling premature failure of a casting during testing and/or use.

1200 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis

Fig. 1

Inclusion (arrow) on machined surface of a casting

(a)

Fig. 4

(b) Wrinkled oxide skins on fracture surface of a casting

100 µm

Fig. 2

Fracture through inclusion (arrow) shown in Fig. 1. Original magnification: 35

Fig. 5

Furnace refractory on the machined surface of a casting

Fig. 7

Scanning electron micrograph of cold flake designated as “1” in Fig. 6. Original magnification: 500. Ductile rupture (arrows) appearance indicates partial bonding between the flake and its mating surface.

1

1

2

iron oxide and Mg-Ce-Al-Si inoculant at the origin resulted in the failure of the casting. 2

Cavities: Gas Porosity and Shrinkage

200 µm

Fig. 3

Fig. 6

Case Study 3. A close-up view of a fracture surface of an aluminum casting containing presolidified aluminum (often referred to as “cold flake”) from a shot sleeve in a die-casting machine is shown in Fig. 6. Entrapment of these types of inclusions is also known to contribute to a “non-knitting” condition in the casting. The various factors contributing to such a defect include the percent fill in the sleeve,

dwell time in the sleeve, shot sleeve temperature, and metal temperature. A scanning electron micrograph of the chilled flake is shown in Fig. 7. The ductile rupture appearance in Fig. 7 indicates partial bonding between the flake and its mating surface. Case Study 4. Figure 8 reveals the origin of failure in a ductile iron casting. Entrapment of

Scanning electron micrograph of inclusion shown in Fig. 2. Original magnification: 150. Area identified as “1” is the inclusion while area “2” is the surrounding matrix.

Close-up view (stereomicroscope) of the presence of cold flakes designated as “1” and “2” on fracture surface. Original magnification: 8.7

Case Study 5. Figure 9 shows a typical micrograph of gas porosity. Trapped air or gas (generated from exposure of a carbonaceous lubricant to molten aluminum) in a casting may also lead to the formation of “blisters” on casting surfaces after heat treatment (see Fig. 10). Case Study 6. Figure 11(a) is a macrograph of the fracture surface (of a gravity permanent mold aluminum casting) revealing a dark gray stained area covering approximately 40% of the cross section. Scanning electron micrograph of the stained area (Fig. 11b) confirmed the presence of shrink porosity (dendrites) that is typically evident in the last regions to solidify (hot spots). Case Study 7. Figure 12 is a micrograph of a low-alloy steel shrinkage crack. The casting reportedly broke because of solidification shrinkage and weld shrinkage cracks.

Common Defects in Various Casting Processes / 1201

(a)

(b)

1 mm

Fig. 8

(a) Entrapment of iron oxide and Mg-Ce-Al-Si inoculant at the origin of a fractured ductile iron casting showing the origin. Original magnification: 7.5. (b) Higher magnification view of the origin. Original magnification: 50

(a)

Fig. 9

(b)

Typical micrograph of gas porosity. Original magnification: 100

200 µm

Fig. 10

Close-up view (stereomicroscope) of fracture through blisters (arrows) in a heat treated aluminum casting. Original magnification: 8. Blisters found are related to encapsulation of a carbonaceous material, such as a lubricant, during casting.

Fig. 11

(a) Close-up view (stereomicroscope) of dark gray stained area in a gravity permanent mold aluminum casting. (b) The occurrence of dendrites confirms the presence of shrink porosity in the stained area. Original magnification: 200

Shrinkage porosity

Fig. 12

Micrograph of low-alloy steel shrinkage crack. Original magnification: 7.5

Discontinuities: Hot Tearing and Cold Shut Case Study 8. “Hot tear” is often associated with shrinkage. In hot tears, rupture occurs

Fig. 13

Optical micrograph of a hot tear in a casting. Original magnification: 200

through the material that has not fully solidified. These hot spots remain fluid longer, thus serving as a focal point for the concentration of contraction strains. Factors affecting hot tears include improper alloying (with large

solidification range), poor product design (sharp corners), lack or absence of pressure in hot spots, and lack of cooling in hot spots. Figure 13 is a micrograph of a hot tear in an aluminum casting. Case Study 9. Figure 14(a) exhibits the presence of a leak path (often caused by the presence of shrink porosity, oxides, and/or cold flakes) at the parting line of an aluminum casting. Castings that leak gas or liquid are often referred to as leakers and often exist between a machined surface on one side and a hot spot on the other. The fracture surface (after intentionally fracturing through the defect) reveals a dark gray stained area (Fig. 14b) that was confirmed to be shrink porosity (Fig. 14c). Case Study 10. Finally, Fig. 15(a) shows a split at a cast hole for an insert and is an example of a cold shut (often caused by two metal streams not knitting together) in an aluminum casting. The various factors responsible for the occurrence of such a defect include: lower die (mold) temperature, increased fill time, lower

1202 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis

(a)

Fig. 14

(b)

(c)

50 µm

(a) Leaker at the parting line of a casting. (b) Exposed fracture surface reveals a dark gray stained area. (c) The occurrence of dendrites confirms the presence of shrink porosity in the stained area. Original magnification: 200

metal temperature, poor part design, improper gating design, excessive flashing, and oxides at the interface. The fracture surface (Fig. 15b) following intentional fracture through the defect reveals shiny oxide skins with some scattered fracture through the base metal. REFERENCE 1. F.A. Morrow et al., Testing and Inspection of Casting Defects (Processing of Castings), Casting, Vol 15, Metals Handbook, ASM International, 1988, p 544–560 SELECTED REFERENCE  L.C.F. Canale, R.A. Mesquita, and G.E. Tot-

ten, Ed., Failure Analysis of Heat Treated Steel Components, ASM International, 2008

(a)

Fig. 15

(b) (a) Split at cast hole for an insert. (b) Fracture surface shows shiny oxide skins.

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1203-1206 DOI: 10.1361/asmhba0005345

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Repair Welding of Castings Randall Counselman, Oldenburg Group Inc.

REPAIR WELDING, in general, is a necessary operation for most fabricators and can cost more than the price of the original component if performed improperly. It is always recommended that the source of the failure be determined, via full or limited failure analysis prior to repair as described in other articles in this Volume (see “Failure Analysis Techniques and Case Studies”), although this may not always be practical. It is also recommended that drawings and specifications for the original casting be obtained and reviewed to ensure that repairs are performed in accordance with the intended service requirements. Additional information relevant to the service history of the casting should also be reviewed to preclude further failures resulting from weldability or other previously defined issues. The use of qualified welding procedures and personnel in accordance with specified codes and standards should always be included as a primary part of any repair operation. Repairs attempted using poor quality workmanship or the lack of a properly outlined procedure can eventually lead to litigation or loss of future sales resulting from negative publicity. For the purposes of this article, the discussions of the repair welding of castings shall be divided into that for ferrous and nonferrous materials.

Repair Welding of Ferrous Castings Gray and Ductile Cast Iron. Gray iron is the most common type of ferrous casting. The weldability of this material is typically very good to poor because of the formation of complex microstructures, which can include intermetallic compounds. Carbides tend to form in some portions of the fusion zone, as the partially melted region is not always altered by dilution with the weld metal. Some of the higher grades of gray irons are often considered nonweldable. The lower-strength grades of ductile irons are very weldable, but the weldability decreases as the strength increases. Other types of ferrous castings, including malleable and white irons vary in degree of weldability. The malleable irons consist of two distinct subgroupings. The subgrouping consisting

primarily of ferrite is generally easily welded. White irons are typically considered nonweldable because of the large amounts of cementite contained in their microstructures that renders them crack-prone. Cast steels are generally considered very weldable; however, the ease of welding is a function of the carbon and alloying content and any postweld heat treatment. As the carbon and alloying contents increase the weldability tends to decrease. The requirement for preheat generally applies to cast steels with carbon contents over 0.30%, and the recommended preheat temperatures typically vary from 120 to 205
C (250 to 400
F). Cast Stainless Steels. Many of the corrosionresistant types of cast stainless steels are very weldable; however, care must be used to prevent sensitization at temperatures in excess of 425
C (800
F). The occurrence of sensitization is significantly minimized when using the low-carbon grades of these alloys, as insufficient carbon is available to form chromium carbides at the grain boundaries. Weld Repair Processes. Typical welding processes used for casting repair applications include the oxyfuel and arc fusion processes. The oxyfuel process has a long history of use, permits precise control of heat input, and is extremely portable, but is usually limited to use on thin sections. Generally, any material that can be welded using the oxyfuel (OFW) or arc fusion processes can also be repaired using these processes. The more widely used arc fusion processes include shielded metal arc welding (SMAW), gas metal arc welding (GMAW), gas tungsten arc welding (GTAW), and flux cored arc welding (FCAW). The SMAW process is very portable, used for structural applications where accessibility is limited, and historically has been very versatile. The GMAW process is gaining popularity because of an increasing supply of filler metal options and the advent in use of the FCAW process. Both processes provide for higher deposition rates and are easily automated; however, production rates are generally not of great concern for many repair applications. The development of smaller, lighter power supplies with greater control over the waveform has spurred the use of GMAW and FCAW processes for repair applications,

previously reserved solely using the SMAW and GTAW processes. The GTAW process is associated with producing high-quality welds, but generally requires a high welder skill level. Another option is a hybrid process called braze welding. Braze welding uses filler metals with temperatures above 425
C (800
F), but below that of the casting base metal. Braze welding can be accomplished using either the OFW or more popular arc welding processes, such as GMAW or GTAW. The advantage is low heat inputs, which minimizes distortion and the potential for further cracking. This process does not require the use of precisely machined joints needed for the capillary flow of filler metal and also does not require fluxing and the resulting cleanup associated with flux removal. The disadvantage is that using lower melting point filler metal precludes the use of this process for high-strength applications or in situations where some forms of corrosion may be of concern. Other welding processes that are often used in repair applications for ferrous castings, but to a lesser degree, include submerged arc welding (SAW) and plasma arc welding (PAW). The SAW process is a high-deposition process and may be considered for thick sections, but is typically limited to use in the flat position. The PAW process is also associated with producing high-quality welds and requires a somewhat lesser welder skill level than GTAW, but is more commonly used in applications involving very thin sections. Weld Repair Considerations and Techniques. The weldability of the failed casting should be completely ascertained prior to initiating the repairs. This can be accomplished in part by using shavings or drillings from the casting for chemical analysis. The results of the chemical analysis can be compared against the casting material specification. The weldability can also be assessed using a calculation of the carbon equivalent (CE). There are several CE formulas for cast irons. The modified Karsay CE formula specified for cast irons is: CE ¼%C þ 0:31ð%SiÞ þ 0:33ð%PÞ þ 0:45ð%SÞ þ 0:028ð%MnÞ þ %Mo þ %Cr  0:02ð%NiÞ  0:01ð%CuÞ

ðEq 1Þ

1204 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis The calculated CE value for the cast iron can then be located on the graph shown in Fig. 1. This procedure was devised by the AWS D11.2 code committee as a means for correlation of the CE value for any cast iron against that for known grades from other specific types of castings. The equation should be used only as a comparative guide for establishing the relevant grouping and preheat temperature for an unknown material, relative to other known types of castings. A similar calculation of the CE value can also be made for alloy steel castings, using the following formula: CE ¼%C þ ð%MnÞ=20 þ %Cr þ ð%CuÞ=20 þ ð%MoÞ=15 þ ð%NiÞ=60 þ ð%SiÞ=30 þ ð%VÞ=10

ðEq 2Þ

A CE value calculated using Eq 2 is typically used to determine whether preheat, or preheat and postheat treatment is required to prevent the occurrence of cracking in cast alloy steels.

Iron type Gray 370

Ductile

No-crack temperature, ⬚C

315

Malleable

260

800

Specimen No.

Grade

G1 G2 G3 G4 D1 D2 D3 D4 M1 M2 M3 M4

40 50 45 35 80-55-06 100-70-03 120-90-02 65-45-12 50005 32510 M5503 M5003

= No-crack temperature below 20 ⬚C

700 D2 600 D3 G4 500 G1

D1 D4 400

205

G2

150

300

200

93

G3

38 M3

M1 M4

−18 2.50

100

M2 0 3.00

3.50

4.00

4.50

5.00

5.50

Carbon equivalent, %

Fig. 1

Effect of carbon equivalent on no-crack temperature for selected grades of iron castings. Source: Ref 1

Non-crack temperature, ⬚F

425

This formula may also be combined with terms that identify an index for hydrogen content and level of weld restraint. The condition of the casting is also very important for determining its weldability. A cross section must be excised for metallographic examination. Metallographic analysis will identify the microstructure and reveal the presence of base material anomalies, such as shrinkage porosity, inclusions, laps, and so forth. The location of any previous weld repair is also critical, as overlapping welds can produce hard zones where cracking can initiate. Additional specimens may be excised from the casting to determine mechanical properties, such as tensile strength, yield strength, and elongation. This information is useful in selecting a welding process and filler metal for the repairs. If the size or configuration of the casting precludes the removal of specimens, then nondestructive methods and hardness testing may be used as an alternative. A nondestructive

approach for determination of mechanical properties may simply consist of excising test specimens from bar or plate poured from the same melt, at the same time, as that for the casting. Electromagnetic (ET), ultrasonic (UT), and acoustic emission testing (AET) also have been used to indirectly assess the values for various mechanical properties, such as tensile strength, yield strength, and so forth, in casting repair situations. Weld Metal Properties. The selection of an electrode or filler metal for repair operations typically is based on service conditions and joint efficiency requirements. In most cases, 100% joint efficiencies are attainable using nonmatching filler metals for castings with tensile strengths up to 480 MPa (70 ksi) while matching filler metals are recommended for higher-strength castings. When welding ductile irons it is recommended that the tensile strength of the filler metal or electrode be 35 MPa (5 ksi) above that of the casting to provide for 100% joint efficiency. Caution must be used when selecting nonmatching filler metals, as dilution between dissimilar filler and base materials can cause the formation of brittle microstructures in the weld metal. It is suggested that typical weld industry standards and manufacturers recommendations be reviewed and adhered to when considering filler metal options. Removal of Casting Defects. The casting defect can be removed prior to repair by grinding and thermal cutting or gouging to clean metal. Thermally cut or gouged surfaces must be ground back to remove the heat-affected material prior to welding. Cracks can be pinned by drilling the ends or welded using a short transverse bead at both ends, prior to removal. Any castings found to be impregnated with plastic or glass sealers should not be repaired, as repair welding may result in further fusion and/or porosity issues. Joint Selection. The area for repair should be excavated and prepared using a V-groove or U-groove joint. The V-groove joint typically requires an included angle of 60 to 120
, while the U-groove preparation should be performed using an angle of 20 to 25
with a 6.4 mm (0.25 in.) root radius. In any case, the excavated region should be shaped to provide sufficient access to the root. The ends of the excavated region should also be sloped gradually to the surface of the casting to allow for complete fusion. Surface Preparation. Oxidized or carburized layers, casting skins, dirt, and all other contaminants must be completely removed from the excavated region prior to welding. If the casting is impregnated with oils or contaminants that make removal impossible, the surface can often be “buttered” with weld passes and repeatedly ground to clean metal before proceeding with the intended repair procedure. Control of Heating and Cooling Rates. Preheat temperatures should be selected depending on the CE value calculated using

Repair Welding of Castings / 1205 Eq 1 or 2 for the casting. The higher the carbon and alloying content of the casting, the greater the section thickness, and the more complex the shape, the higher the preheat temperature required to avoid cracking. Typical preheat temperatures for ferrous castings range from 315 to 650
C (600 to 1200
F), but can be applied as low as 150
C (300
F) and as high as 870
C (1600
F). The selection of preheat temperature should be based on the composition and thickness of the casting, the complexity of the shape, and the type of filler metal and process chosen for the repairs. Preheat should be applied to the casting in a prescribed manner so that the repaired weld is always in compression. Castings of complex shapes may require uniform heating prior to welding and interpass maintenance. The interpass temperature should be maintained at least 55
C (100
F) above the preheat temperature to avoid the formation of a brittle microstructure. Furnace heating or portable electric blankets may be used to achieve preheating and interpass temperature maintenance. A furnace or thermal blanket may also be used to control the cooling rate to avoid the formation of hard microstructures. The casting may also be covered by sand or vermiculate immediately after weld repair to retard the cooling rate. Deposition Techniques. There are many techniques used to control the manner in which the weld is deposited. Typical methods include the use of block or cascade sequences using either stringer or weave beads, dependent on the need to minimize heat inputs. In situations in which postweld heat treatment is called for but impractical, it may be possible to avoid cracking by using a temper-bead technique in which each successive pass is deposited in an overlapping manner to temper the previous pass. The overlapping of passes must be carefully controlled and repeated throughout the repair process or the result may be ineffective. Postweld Heat Treatment (PWHT). The calculated CE value, potential for hydrogen pickup, and section thickness of the casting may dictate the need for postweld heat treatment to avoid or minimize the cracking potential. The need to perform this operation may be limited by the conditions under which the repair is performed. The PWHT temperature, soak time, and heating and cooling rates are often dictated by the type of ferrous casting, the procedure used to perform the repair, and section thickness of the casting. Verification of Weld Repair Quality. Typical nondestructive test methods, such as visual inspection (VT), magnetic particle inspection (MT), liquid particle inspection (PT), radiographic inspection (RT), and ultrasonic inspection (UT), to name a few can be used to determine the quality of the finished weld repairs. The test method, inspection procedure, acceptance criteria, and test personnel requirements are typically specified in the governing code (or standard), project specifications, or contract requirements.

Repair Welding of Nonferrous Castings The weld repair requirements for nonferrous castings comprising aluminum, titanium, magnesium, and copper are similar in many respects to production welding of these materials. The weldability can be determined by chemical analysis, metallographic examination, and mechanical tensile testing as permitted by availability of specimens from the casting. Alternatively, on-site chemical analysis methods using portable nondestructive techniques, radiography, hardness testing, and other nondestructive methods are available to provide information regarding the weldability of the casting. The higherstrength materials and higher-cost casting alloys typically require more sophisticated weld repair procedures. Generally, the nonferrous alloys require more stringent guidelines concerning proper precleaning, and sufficient gas shielding, than is required for ferrous castings. A more thorough understanding of the base material physical properties, welding metallurgy, and processing limitations of the nonferrous alloys also may be needed. Weld Repair Process Selection. Generally, nonferrous castings exhibit more complex interactions with regard to control of heat input, reactivity requiring greater control of gas shielding, handling, and precleaning practices, embrittlement phase formation tendencies, filler metal selection, and failure modes. Obtaining a better understanding of the aforementioned issues relative to a specific nonferrous casting application can provide the insight required to choose the proper welding process for a given repair situation. Repairs must often be performed in harsh conditions that may place further limitations on the number of viable welding process options. The requirements associated with the weld repairs for some nonferrous castings may necessitate that repairs be limited only to those situations that can be performed in the shop and not in the field. It is often difficult, if not impossible, to control wind, drafts, and cleanliness and reactivity issues under field conditions, which is why the repair of these alloys must normally be performed in shop environments where these issues can be closely controlled. The GTAW process is used almost exclusively to avoid cracking in magnesium castings. Typically, the repair of stainless steel, aluminum, nickel, and copper alloy castings can be performed using either SMAW or GTAW fusion arc processes in the field or in the shop. Other fusion arc, solid-state, resistance, and special processes can be used to repair weld these cast materials where conditions and costs are not prohibitive, but these cases represent a smaller portion of the total number of repairs that are effectively completed on these materials. There have recently been limited studies performed by various academic institutions and government agencies concerning the repair of

aluminum and nickel-copper castings using friction stir welding (FSW) processes. Areas of the castings exhibiting scabs and shrinkage porosity were apparently “healed” using the FSW process. There have also been investigations regarding the effects on mechanical properties and corrosion resistance associated with FSW repair; however, none of these studies is definitive. Further investigations are required to fully realize the potential for using FSW processing for casting repairs. These studies also include some work regarding the repair of aluminum die castings using the FSW and laser beam welding (LBW) processes. Shrinkage porosity, cracking, and dimensional distortion issues that typically occur in these castings as a result of repair welding can be minimized using the high energy density beam employed in the LBW process. The advantages associated with laser beam welding include low heat input, minimal residual stress formation in the weld metal and heat-affected zone (HAZ) regions, and multipurpose processing capabilities. It is generally possible to perform preheating, wavelength selection to control cooling rate, and selection of standard melt-in, pulsation, or keyhole weld mode variations, depending on the type of LBW equipment employed for the repairs. In some instances it is possible to complete LBW repairs in areas of large or complex castings that would normally be inaccessible using most other applicable welding processes. Disadvantages of this process can include the high purchase cost of the equipment, the size of the power source, and relative immobility of the equipment for on-site repairs. Filler Metal Selection. The proper selection of welding electrodes or filler metal alloys is essential to obtaining acceptable, crack-free welds in nonferrous castings. The appropriate filler alloy can often be selected by careful review of relevant welding literature, or by contacting filler metal and base material suppliers, welding distributors, or consultants. The use of matching filler metals is often recommended for welding nickel, titanium, and magnesium castings, but this is not always the case for stainless steels and is rarely the norm for aluminum castings. The selection of filler metals designated with extra-low interstitials (ELIs) is typically recommended for welding titanium castings to maximize the resulting mechanical properties. Weld Repair Considerations. The repair of stainless steel, nickel, aluminum, magnesium, and titanium castings requires that these materials be thoroughly cleaned to remove all grease, oils, and contaminants prior to welding. Magnesium castings must be completely stripped of all paint and thoroughly degreased prior to welding. Aluminum and titanium castings have fairly thick, tenacious surface oxide layers, which must be removed by chemical etching prior to welding. Failure to remove these oxide layers results in very porous welds that are susceptible to fracture at tensile loads below those

1206 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis of the adjacent base material. Once the oxide layers are removed, welding should be completed in a relatively short period of time, typically within 1 hour; otherwise, the oxide removal process should be repeated. It is recommended that titanium castings are handled using clean cotton cloth gloves to avoid contamination by natural skin oils that can be imparted on the surfaces of this material by improper handling. Previous processing of materials should also be reviewed fully with respect to the intended application to verify that the materials, as received and/or in service, are suitable for weld repair. This processing review should include, but not be limited to, such items as heat treatment, temper condition, previous operating temperature, prior contact media exposure while in-service, and so forth. Grinding or machining of weld joint preparations should be performed at recommended settings to avoid overheating of the cast material. Only stainless steel brushes should be used for mechanical cleaning of nonferrous castings, as pickup of carbon and ferrous dust and filings can contaminate the weld joints prior to welding. Titanium castings must be properly shielded as these alloys are very reactive to contaminants and are very susceptible to embrittlement by oxygen, nitrogen, and hydrogen gases during weld repairs. The selection of argon, or argonhelium, inert gas mixtures is recommended for welding nonferrous cast materials. The use of gas backing is recommended for full-penetration joints in all nonferrous castings, and the use of trailing shielding gas is recommended for welding titanium castings. Where possible, it is recommended that titanium castings be

repair welded in enclosed positive pressure inert gas chambers. Titanium castings can also be repair welded in some situations using special open-air techniques involving the use of large shielding gas cups, gas diffusers to prevent excessive turbulence, which can pull air into the shielding gas stream, trailing gas shielding devices that attach to the welding torch, and the use of sufficiently slow travel speeds to avoid contamination during cooling of the weld metal. Postweld Heat Treatment (PWHT). The need to perform this operation may be limited by the conditions under which the repair is performed. The PWHT temperature, soak time, and heating and cooling rates are a function of the type of nonferrous casting base material, the procedure used to perform the repair, and the section thickness of the casting. Verification of Weld Repair Quality. The nondestructive test methods used to determine the quality of finished weld repairs in nonferrous castings are generally the same as those used for ferrous castings, with some exceptions. The most notable exception is that magnetic particle (MT) inspection is not a viable choice for use on nonferrous materials. The nondestructive test method, inspection procedure, acceptance criteria, and test personnel requirements are typically specified in the original governing code (or standard), project specifications, or contract requirements used to manufacture the castings. In the event that no governing code, standard, or contract requirements are specified for weld repair, then typical welding industry standards, such as those available from the American Welding Society (AWS), ASME, ASTM, and so forth, should be selected for the type of nonferrous alloy.

REFERENCE 1. R.A. Bushey, Welding of Cast Irons, Welding, Brazing, and Soldering, Vol 6, ASM Handbook, ASM International, 1993, p 711

SELECTED REFERENCES  M. Avedesian and H. Baker, Ed., ASM          

Specialty Handbook: Magnesium and Magnesium Alloys, ASM International, 1998 J.R. Davis, Ed., ASM Specialty Handbook: Aluminum and Aluminum Alloys, ASM International, 1993 J.R. Davis, Ed., ASM Specialty Handbook: Cast Irons, ASM International, 1996 J.R. Davis, Ed., ASM Specialty Handbook: Stainless Steels, Vol 1 and 2, ASM International, 1994 M.J. Donachie, Ed., Titanium: A Technical Guide, 2nd ed., ASM International, 2000 “Guide for Welding Cast Irons,” ANSI/AWS D11.2-89, American National Standards Institute Nondestructive Evaluation and Quality Control, Vol 17, Metals Handbook, ASM International, 1989 Steel Castings Handbook, 6th ed., Steel Founders’ Society of America and ASM International, 1995 V. Sutter and R.J. Dybas, Repair Welding, Welding, Brazing, and Soldering, Vol 6, ASM Handbook, ASM International, 1993 “Welding and Brazing Qualifications,” Section IX, ASME, 2004 Welding Handbook, Vol 4, 8th ed., American Welding Society, 1998

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1207-1212 DOI: 10.1361/asmhba0005346

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Quality Planning Tools and Procedures Alan Richardson, SPX Corporation

A FAMOUS QUOTE by W. Edwards Deming states, “IT is not necessary to change; survival is not mandatory.” (Ref 1). The success of any program, process, or task can be attributed to having a systematic, quality-based approach to that program, process, or task. Most have heard of the “seven basic tools” of quality, which are cause-and-effect diagrams, check sheets, control charts, histograms, Pareto charts, scatter diagrams, and run charts (Ref 2). These quality analysis tools provide part of the foundation of a systematic, quality-based approach and are quite well known to most professionals and academicians today. To take this a step further, this article looks at how these tools are built upon to become a more advanced set of quality tools. These advanced tools include Advanced Product Quality Planning (APQP), Failure Mode and Effects Analysis (FMEA), Control Planning, Measurement Systems Analysis (MSA), Lean Tools, Statistical Process Control (SPC), production viability and tryout, and Six Sigma. The goal of this article is not to make the reader proficient in the use of these tools, but to provide awareness as many competent parties are available for training in these uses of these tools.

understanding of customer desires, both internal and external, must be present. The final judge of quality is always the customer, and the more clearly those expectations are understood in the beginning, the better are the chances of achieving them in the end. Another element is the manufacturer’s ability to understand and control processes, which means understanding and controlling all possible sources of variation and instituting preventive and detection methods to achieve processes that are stable, predictable, and reliable. This really is the essence of the APQP process. APQP requires the commitment of top management because resources must be dedicated up front, and the quality team must be given the freedom to focus on all aspects of the business to uncover problems and determine root causes for prevention. A team approach is used, and the team’s makeup should be cross-functional in nature to ensure all-encompassing knowledge of the scope of the program. A truncated overview of the APQP process a team would follow is shown in Fig. 2. In addition to the key elements mentioned in the introduction, a few additional subelements of the APQP

Advanced Product Quality Planning (APQP)

80

APQP is the all-encompassing element that brings a comprehensive set of tools together in a systematic, aligned fashion to provide a roadmap for success. The criticalness of advanced planning cannot be underestimated. It has been stated that approximately 70% of future profitability is influenced at the product and process design stage as shown in Fig. 1 (Ref 3). By planning for quality, the goal is to prevent mistakes from occurring instead of relying on the old tried and true method of detecting issues. This is not to say that prevention of every issue can be achieved, as sometimes this may not be economically feasible, and therefore detection of the problem at the earliest possible stage is important. To achieve the goal of best in class through the continual improvement of quality, there must be prerequisite mindsets in place by all involved in this process. A greater

60

70

process are discussed briefly to provide the reader some awareness of these subelements.

Failure Mode and Effects Analysis (FMEA) The initial stages of the APQP process are spent understanding the scope of what the team is to undertake. For the purpose of this discussion, assume that the scope is a particular part of a large assembly. Based on this assumption, the team’s goal will be to ultimately understand the potential consequences of each dimension not meeting the stated specifications, how each of those dimensions will be created (which processes will be used for each dimension), and how this part will interact with other parts and processes to ultimately create the final assembly. The FMEA process is used to create this information. The FMEA is a systematic tool that is used for identifying the effects or consequences of a product or process failure modes, and it is used to quantify the effects of the failure modes in terms of the risk they pose to the end user or the integrity of the product prior to being put to

70%

% Influence

50 40 30 20%

20 10

5%

5%

Labor

O/H

0 Product and process design

Material Factors

Fig. 1

Factors influencing profitability

1208 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis

A

Measurement system evaluation

Establish core team

Number the blueprint – list expected outcome

“Finalize’ setup/operation/gauging instructions

Initiate the characteristics matrix list

In plant verification B

Initiate the question log Verify results at customer

Fail? Design FMEA No Process flow diagram

Determining responsibility for control methods & check items

Process FMEA by operation/station

Determine reaction plans for pilot & full production

Yes Implement full production level CP

Verify results at customer

Initiate control plan Extract CP plant floor instructions Define gaging needs

Modify plan for continuous improvement

Design data collection sheets

Develop preliminary instructions Preliminary supplier verification

“Run at rate”

A

Implement pilot level CP

B

Fig. 2

Overview of the Advanced Product Quality Planning (APQP) process, Source: Ref 5

use by the end customer. When done correctly, the FMEA is one of the most valuable tools for understanding potential variation and resultant outcomes for any product or process. As they say, “the devil is in the details,” and to come up with a beneficial FMEA a thorough analysis must be completed that encompasses both the design and the process. Design Failure Mode and Effects Analysis (DFMEA). This document is used to address component, subsystem, or main system functions. Potential failure modes that are commonly addressed are the appropriate materials

and specifications. In many instances this document may be furnished by the customer, but depending on availability, the APQP team may be required to build this information through interviews with the customer, analysis of mating parts, and analysis of both internal and external (downstream) processing involving the part. Although the latter can provide valuable information, detailed knowledge of end-user effects may be difficult to ascertain without an actual DFMEA document. Process Failure Mode and Effects Analysis (PFMEA). This document examines the failure

modes of the process that is used to create the component, subsystem, or main system. Potential failure modes addressed here are operator error in assembly and excess process variation that could result in an out-of-specification situation. A blank FMEA document is shown in Fig. 3. The FMEA document provides a good roadmap for the APQP team to follow; one helpful element of the APQP process that needs to be completed prior to initiating the PFMEA is the Process Flow Diagram (PFD). The PFD takes into account all processing steps from receipt

Quality Planning Tools and Procedures / 1209 Part or process name

Suppliers & plants affected

Prepared by

Design/Mfg responsibilty

Model date

FMEA date

Other areas included

Engineering change level

Process operation, function or purpose

Fig. 3

Potential failure mode

Potential effect(s) of failure

S C E C V

Potential cause(s) of failure

O C C

Current controls evaluation method

Action results D S R E * P T O N

Recommended action(s)

Actions taken

S O D E C E V C T

R P N

Example Failure Mode and Effects Analysis (FMEA) document

Incoming sources of variation

Process flow

Outcomes

Incoming material 5

5

5

Screw machine

Legend:

Fig. 4

Area/individual responsible & completion date

10

10

20

20 Lath

Operation

Inspection

Operation with auto inspect

Operator

Transportation

Storage

Delay

Example Process Flow Diagram (PFD). Must be completed before process mode and effects analysis (PFMEA) can begin.

of raw materials to the shipment of the finished component. An example of a PFD can be seen in Fig. 4. In addition to identifying all the inputs to transform raw materials and components into a finished product, this document provides the framework to begin considering all the potential variation sources within those inputs. This information is extremely valuable when completing the PFMEA document as described below. The completion of the FMEA starts in the left hand column and then moves across to the right and also tiers downward as it moves across. This is explained briefly below. In the first column the product or process function is listed. As mentioned previously, this should coincide with the process of the PFD for the PFMEA. In the next column, every potential failure mode for that product or process function is listed. Continuing to move across the document, every potential effect for each failure mode has to be listed. Again, multiple effects are probable for each failure. This provides the tier-down effect described previously. A word of caution is in order to minimize duplication and overlengthiness of the FMEA. When moving on to the next product or process function, always assume that the prior function was finished as expected. Continuing across, potential causes for each failure mode are then considered. Finally, current process controls are listed that can either prevent this failure from occurring or detect the failure after its occurrence.

1210 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis

The control plan is constructed for each step of the process with the mentality that each step of the process is a supplier to the subsequent process. Although one master control plan is developed, each step of the process will have its own worksheet to ensure their portion of the overall process is controlled and that product passed downstream meets customer expectations. An example of a control plan is shown in Fig. 5. Not every process will use the columns as shown, so modification of the control plan template is recommended to suit individual process needs. The key elements of a control plan are the product/process specification, measurement

Z CN4

Fig. 5

Z Plane Center depth

Example control plan document

 Ability of operator to use the gage  Necessary discriminating ability (1/10 of       

tolerance and manufacturing process) The time it takes to measure Environmental conditions Importance level of the characteristic Need to assess process capability Calibration requirements Reliability of the gage Effectiveness (repeatability, reproducibility, accuracy, linearity, stability)

Understanding the intricacies of the MSA process is beyond the scope of this article. The ultimate goal of this process is to identify variation that is attributable to the gage itself and variation that is attributable to the operator of the gage. Early detection of potential issues in either area can allow for redesign of the gage or detailed instructions to ensure proper operation of the gage (Ref 6). For more subjective measuring methodologies, consider the case of x-ray inspection of a casting. In this instance, some locations of the casting are more tolerant of porosity than others per a specification. In addition, the size and number of those occurrences allowed varies throughout the casting. Clearly, this can be a complex evaluation that leaves much judgment up to the operator of the x-ray equipment.

Revisions Description

Product control items Capability Machine/ Imp Control Tool Specification spindle Pp Ppk Date Level factor number number 3.000 +/− .005 BP 3PC S .625 +/− .005 IP 1P S,T Type

Description

Evaluation of the method of measurement is another critical part of ensuring the correct processes are developed and for ongoing customer satisfaction. Measurement Systems Analysis

Change number

Program: Plant: Part name: Part number: B/P Rev. date:

ID #

Measurement Systems Analysis (MSA)

(MSA) provides a process for evaluating most measurement methods. Many organizations overlook the importance of this. Poor measuring methodologies can result in significant wasted time in problem-solving efforts. MSA assesses the statistical properties of repeatability, reproducibility, bias, stability, and linearity of the system. The MSA process has commonly been referred to as the “gage R&R process.” MSA can evaluate both variable and attribute gages. For more subjective measuring methods, such as x-ray inspection, creative ways of identifying the repeatability and reproducibility of those processes will have to be developed. The APQP team can play a key role in ensuring that measuring methods are solid and practical. By evaluating all methods themselves in the development stages, they can help define:

Change by Approved by

Control method Control Sample Method size Frequency (Record)

Effective date

Gauge Description Number

V or A

Control Planning

method, sample size and frequency, control method, and reaction plan. Through customer communication, the APQP discovery process and completion of the FMEAs, a comprehensive list of the product and process points will be identified, which require monitoring to ensure that product integrity is maintained throughout the process stream. Each of these items should be listed on the control plan in the appropriate column. The customer may list some product specifications and/or the APQP team may identify process specifications as critical based on end-user severity levels. These should be noted as such in the “Special Characteristics” column. For every product or process specification listed, a method of measuring that characteristic must also be identified. This measurement method can vary from being a very objective method using variable or attribute gaging to being very subjective in the use of methods such as x-ray inspection of castings where the operator judgment becomes critical. After this is completed, the sample size and frequency of sampling should be identified. Control methods relate directly to the FMEA and the current process controls section in that document. Before finalizing, all economically viable attempts need to be made to prevent nonconformances from occurring and then detection systems implemented following the dictum “an ounce of prevention is better than a pound of cure.” Standardized work, which is discussed later, plays an important role in this area. Finally, reaction plans are developed to provide guidelines on how nonconforming situations are to be handled for all appropriate conditions for each process and product specification. Including as much pertinent information for value-added workers on the control plan is recommended and should be left up to the judgment of each APQP team as this process evolves (Ref 5).

Resp

For each of the columns—effects, causes, and controls—there is a corresponding column to rate the severity (SEV), occurrence (OCC), or detection (DET). Each rating is done on a 1 to 10 scale with 10 indicating that the severity is extremely hazardous, that occurrence is most probable, or detection is impossible. A ranking of 1 would be the extreme opposite of the above. Rankings in the 8 to 10 range should be done with much consideration as these likely require extreme measures be taken to control. Another column on the form is the Risk Priority Number (RPN). This column is the product of the severity, occurrence, and controls columns. The RPN provides a roadmap for continuous improvement once the FMEA is completed. Although no thresholds are generally established for addressing RPNs, a good guideline to follow in general is to address all areas of the FMEA where an individual ranking is 8 or higher and then focus on the highest RPNs that the resources of the organization allow. This process should be continuously repeated to build a continuous improvement mindset within the extended team. Spaces for actions and responsibility are available. Once actions are completed they can be listed in the document, and the three areas can be re-rated and a new RPN calculated. Upon completion of the FMEA, the foundational information required for the control plan is available (Ref 4).

R&R %

Quality Planning Tools and Procedures / 1211 Although automated systems could provide relief, many times this is economically not feasible. One method that has been used is to develop a control set of castings with varying degrees of porosity throughout. In addition, a rating scale can be developed to rate the overall level of severity in areas of the casting or overall. For this example a rating scale of 1 to 5 is used where anything 3 or higher represents a risk to customer satisfaction. By having process experts develop ratings for each section/ casting, these control castings can then be used as training aides for x-ray operators. By overlaying operator ratings on top of each other, problem areas can be identified and reduced with further training to ensure that there is some aspect of repeatability and reproducibility to this subjective measure.

Lean Tools The use of Lean Tools is very appropriate during the APQP process. In particular, the use of standardized work and 5S (workplace organization) techniques can help in stabilization of the process. Value Stream Mapping (VSM) and Spaghetti Diagrams can help in setting up efficient work cells and overall value streams. Standardized work can be defined as a visual method of displaying detailed work instructions at the point of use. Its development can be a significant contributor to developing stable, repeatable processes that produce a product that meets (and exceeds) customer expectations. For each step of the process, all significant tasks to ensure compliance to process and product specifications will have standardized work developed. This includes setup instructions, operator instructions, gage usage, preventive maintenance instructions, or any part of the process where safety, quality, delivery, or cost is affected. Care should be taken to refrain from putting to much detail into standardized work. Putting requirements into standardized work that are not absolutely necessary creates waste in the system and will eventually result in noncompliance to those instructions. 5S techniques can also bring stability to the process. The five S steps are Sort, Set, Shine, Standardize, and Sustain. Whether revamping existing work cells or developing new work cells, having APQP team members evaluate all needed tools and equipment required can be the difference between an efficient and an inefficient process. The focus during the APQP process is on the first four S steps, which if done correctly will provide the best opportunity for sustainability once the work cells initiate production. Value stream mapping (VSM) and spaghetti diagrams are also helpful during this process and can be used in either the development of a new process or modification of an existing process. In the case of new processes, mapping of similar processes can help point out areas of waste so these can be eliminated

during the creation of the new process. Within and between work cells, spaghetti diagrams can help teams refine product flow and equipment layout during the preproduction process.

Statistical Process Control (SPC) All processes have variation and SPC is about applying statistics in measuring and analyzing the size and potential impact of the variation. Variation is evident in two states, common cause and special cause. Common cause variation is one that is inherent to any process. Special cause variation is one that is related to an assignable cause and is excessive in nature. The goal in any process is to eliminate special cause variation, which allows a process to be “in a state of statistical control” (Ref 7). Applying statistical control charts to monitor a process allows feedback from the output of the process to the inputs so that adjustments can be made to improve the process. Control charts allow the distinction between common and special cause variation through the insertion of the control limits to the chart and the use of rules for identifying trends, shifts, and out-of-control situations. Control limits differ from specification limits in that the latter are customer-dictated limits, whereas control limits are calculated based on actual data from the process and the laws of probability. A process that is running within the customer’s specification limits is considered a capable process and meets customer expectations. A process that runs within calculated control limits is called a controlled process, thus meeting customer expectations in an economical manner. When used appropriately, SPC can have the following benefits:  Determine when to “control” and when to

leave well enough alone

 Determine where to “search for assignable

cause” and vice-versa

 Determine when and when not to sample

(No inspection, 100% inspection)

 Aid in evaluating tolerances  Improve vendor/consumer relations  Improve productivity

Just as was discussed in the MSA section, data can be collected in a variable or attribute format. SPC charting can be done with either type of data. Some of the commonly used variable and attribute charting methods and their applications are:  Variable Charts. Xbar and R Charts for sub-

group sizes between 3 and 10. X and MR for subgroup sizes of 1.  Attribute Chart. p chart, % of nonconforming measurements In the APQP formulation process and the development of the production process, statistical tools are used to measure process potential. Process potential is a measure of the width of

process variation against either the overall specification width (Cp) or a measure of process center versus customer target (Cpk). Variations of these measures, Pp and Ppk, use true standard deviation versus a derived standard deviation. Determining the capability of a process should not be done until special cause variation has been addressed, or the process is in control (Ref 8).

Production Viability At this point in the APQP process, one should be near the point where the production viability of the process can be verified. In the automotive industry this is called the Production Planning and Approval Process (PPAP) process. The production viability process is a simulation of actual production conditions. The goal is to run the process for a long enough period of time to allow for most potential issues to surface and be dealt with prior to the actual start of production. As this “run-at-rate” trial is performed, all documentation developed during the APQP process will be evaluated to ensure completeness. The FMEA documents will be evaluated also for potential new failure modes, effects of failures, effectiveness of control methods, and accuracy of ratings for severity, occurrence, and detection methods. Gaging will be evaluated in regard to viability under production conditions. Ease of use, durability, and accuracy under production conditions must all be reviewed. Although the MSA should uncover these potential issues, nonproduction versus production conditions often brings about unforeseen issues. Control plans are followed to ensure the viability of product integrity during the production process and to uncover any potential resource issues to allow control plan conformance. Part of the control plan and PPAP inspection process will be the use of statistical tools to determine the capability of the process. This is a requirement in most instances with PPAP requirements often set higher to ensure the viability of a longer-term production process. In many “run-at-rate” events, failures in the process are found due to inadequate adherence to APQP steps. In these cases, it may be appropriate to use Six Sigma techniques to permanently correct these deficiencies.

Six Sigma Six Sigma DMAIC is a problem-solving methodology that allows for root cause identification and permanent resolution to process issues. DMAIC stands for define, measure, analyze, improve, and control phases that constitute this methodology. While the Six Sigma DMAIC methodology is used for the resolution of issues within an existing process, many organizations are now using the design for Six Sigma methodology to prevent failures through design and launch of high-integrity processes. Six Sigma initiatives are supported by specially

1212 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis trained personnel within an organization who earn “belts” to describe the complexity of tools they are capable of using to solve problems, their astuteness in comprehending the issues and matching them with the use of appropriate tools, and their ability to leverage team knowledge. The most common are black belts and green belts. Black belts are capable of using advanced mathematical/statistical techniques such as design of experiments, regression analysis, correlation analysis, and variance analysis to solve very complex issues. Green belts are trained to use simpler techniques such as probability, SPC, and other simple charting techniques to solve less advanced problems. All Six Sigma projects follow a regimented methodology to ensure comprehensive analysis of the problem. A brief outline of the DMAIC and Design for Six Sigma methodologies follows. It should be noted that many projects will contain more than one problem that must be resolved. In these instances it is not uncommon to have different issues at different stages of the DMAIC cycle. Define. The definition of a project ensures that the problem is accurately understood, the project is scoped appropriately, and that appropriate resources are available to resolve the problem. At this point in the project, a “burning platform” for undertaking the project should be identified and an expectation set. This may be due to a customer, business, competitive, or employee expectation. If there is no driver behind the resolution of the problem, then reevaluation of the viability of undertaking this initiative should be done. Project boundaries must be established for team members to tackle an issue with a focused scope and effectively resolve it. Developing a process flowchart can be helpful to provide a basic understanding of the process that is the subject of the team’s focus. At this point there needs to be an established measure(s) of the problem to quantify the current state of performance and for setting meaningful and achievable goal on the measure(s). Measure. The next step is a detailed measurement of the problem. It is imperative at this stage of the process to ensure that a measurement system is in place that provides accurate and timely data that allow for understanding of the problem and ongoing measurement. Using the VSM process at this point provides a comprehensive analysis of each step of the process and the areas where measurement systems are required for root cause determination. At this stage, metrics need to be established to measure each step of the process, and the nature of the metrics, continuous or discrete, need to be determined. In setting up measurement

systems and beginning to collect data, the baseline for the project can be confirmed. It should be noted that many of the tools discussed previously are used extensively during this phase of a Six Sigma project. In some instances, it may be necessary to go back to the define stage if project magnitude appears to be significantly different than previously defined. Analyze. The analyze phase is all about understanding gaps between the current performance and the desired performance. A key element is finding the vital few key process input variables that affect the key outputs. To resolve the gaps, the sources of variation must be uncovered. To drive down to root causes of these sources of variation it is best to use simple tools first and then bring more complex tools into use as they are required. In many instances, using simple Pareto charts, brainstorming, and the five whys (a questioning method used to explore cause/effect relationships to determine a root cause of a defect or problem) are enough to uncover the root cause of a problem. In other instances, it may be necessary to use a complex design of experiments to determine a root cause. Caution must be exercised in the process of resolving a problem in that most of them can be solved using the simple tools and do not warrant advance tools or techniques. One thing to consider during the analyze phase is the collection of additional data to uncover the true root cause of a problem. Improve. In the analyze phase, the root cause of the problem is explicitly known or becomes known. Often, the improve phase will require a two-step approach, but is not necessary. The two-step approach first involves putting short-term countermeasures in place and then following that up with a permanent corrective action. This two-step approach is used mostly when the permanent corrective action involves significant lead time. Short-term countermeasures are meant to “stop the bleeding” of the process and provide the time and stability for the permanent countermeasure to be put into place. In either case, verification of these corrective actions is crucial, which means the collection of data to measure the problem must continue. A good practice in charting is to note when countermeasures are implemented so all can note the impact of the changes in the key outputs. In some instances the countermeasure may not produce the desired result. In these cases the project may have to revert back one or two steps to measure and analyze the problem again. Control. Once the control phase is reached, a proven and permanent solution to the problem is in place. The goal of the control phase is to

sustain this improvement so the issues do not arise in the future. This is the point where many projects fail and much hard work is wasted. Without putting these permanent solutions in place, gradual deterioration of the gains made during the process will take place and the result will be nearly identical to the starting point. Part of this phase involves ensuring the FMEA and control plan documentation are updated to reflect the improvements made in the process. Once control plan documentation is detailed and complete, focusing on compliance is critical. Training of all personnel at each point of the process must be undertaken. Auditing of the FMEA and control plan to verify process compliance and institutionalize systems is a key part of the control phase (Ref 9). All of the quality-based systems and tools discussed can make a huge impact in ensuring any task, project, or organization is providing the desired outcome in a timely and cost-efficient manner. Although this article has provided only a “30,000 foot overview” of each, numerous organizations can provide indepth training and certification programs to bring these systems and tools into any organization quickly. As pointed out in Deming’s Chain Reaction, it all starts with quality, so do not hesitate (Ref 10). REFERENCES 1. T.L. Richardson, Total Quality Management, Delmar, 1997, p 3 2. N.R. Tague, The Quality Toolbox, ASQ Quality Press, 2004, p 15 3. K.R. Bhote, Strategic Supply Management: A Blueprint for Revitalizing the Manufacturer-Supplier Partnership, American Management Association, 1998 4. Potential Failure Mode and Effects Analysis (FMEA), Reference Manual, 3rd ed., Automotive Industry Action Group, 2001 5. J.A. Goble, Process Control Planning Workshop, 1990 6. Measurement Systems Analysis, Reference Manual, Automotive Industry Action Group, 2002 7. J.M. Juran and F.M. Gryna, Quality Planning and Analysis, McGraw-Hill 1993, p 377–379 8. Statistical Process Control (SPC), Automotive Industry Action Group, 1995 9. F.W. Breyfogle III, Implementing Six Sigma; Smarter Solutions Using Statistical Methods, 2nd ed., John Wiley & Sons, 2003 10. Six Sigma Training, SPX Corporation

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1213-1215 DOI: 10.1361/asmhba0005219

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Maintenance, Repair, Alterations, and Storage of Patterns and Tooling Henry Bakemeyer, Die Casting Design and Consulting

MAINTENANCE AND CARE of equipment is often a neglected aspect of manufacturing in relation to the time and cost-savings that result. This article suggests procedures to increase the availability and function of patterns and tooling. For the purposes of this article, these definitions are used:  Patterns are tooling that provide a positive

object of the casting to be produced.

 Dies are tooling that are the negative object

or cavity of the casting to be produced.

 Maintenance is routine, planned work done

  



to the tool to prepare it for the next production run. It is also known as preventive maintenance (PM). Repair is work required on a tool to keep it in production because of a breakdown. Breakdown is a tooling condition occurring during normal operation that causes the castings to be unacceptable. Alterations are changes to the tooling required because of customer changes to the casting. Alterations also include changes to the tooling to effect process improvements initiated by the casting supplier or customer. Tools (tooling) are a pattern or die.

Ownership It is customary in the casting business that ownership of tooling resides with the original equipment manufacturer (OEM). This is not the only contract; other arrangements have been defined. With the customary arrangement, the owner is free to move tooling among various casting suppliers. The customary arrangement usually requires the OEM to purchase the tooling from the casting vendor. The casting vendor supplies the tooling and engineering to assure the casting can be produced on the vendor’s equipment. Not all casting vendors have the same manufacturing equipment; consequently, tooling is customized to the vendors’ manufacturing process. This customization may lead to some

additional costs in adapting tooling to another’s manufacturing process when tooling is moved to another vendor. The OEM is advised to obtain a detailed description of proposed tooling during the quoting process. As an alternative, the OEM may propose a minimum standard that tooling must meet. Prior to purchasing tooling, the OEM should secure a tool drawing or engineering sketch and description of the tooling. Additionally, a guarantee of tool life should be secured as part of the casting contract.

Maintenance Responsibility Tooling ownership and tool life define the maintenance responsibility. In short, the casting vendor is responsible for tool maintenance until the tooling has produced the guaranteed number of acceptable castings (scrap and start-up castings are not included in the guarantee). For example, if the tooling is guaranteed to produce 250,000 acceptable castings, the casting vendor is responsible to maintain the tooling in serviceable condition until the guarantee has been met. After the guarantee has been met, the OEM has the option of paying maintenance costs or replacing the tooling. Alterations and changes to tooling can affect the life of the tooling. It is the responsibility of the casting vendor to advise the OEM when this could occur, such that agreement on tool life and maintenance responsibility is defined. For example, introducing an alteration that requires welding of a heat treated steel cavity will upset the steel metallurgy in the weld area and degrade the tool life. Given this information regarding the tool life, the OEM may elect to pursue another method for achieving the alteration, such as inserting the feature, or abandon the alteration.

aggravated by both pressure (causing the pattern surface to deflect) and velocity of the abrasive material. Steel tooling cavities are also subject to thermal fatigue; the cavity surface expands and contracts due to cyclic temperature variations. Tooling maintenance is required to offset these effects for the duration of the tooling guarantee. Figure 1 illustrates steel fatigue on a die casting cavity. Failure of the die or pattern surface will compromise both surface and dimensional integrity of the casting. Consequently, a maintenance plan must be established for the tooling. This maintenance plan may include periodic resurfacing of the die or tooling. This could be in the form of polishing, adding filler materials, or, in extreme cases, welding and/or resinking the cavity. In the case of tooling subject to fatigue failures, periodic stress relieving can be undertaken.

Maintenance Interval The objective of maintenance is to prepare the tooling such that it will run production for a prescribed number of castings without tooling-related downtime. The number of cycles the tool will run before the onset of repair must be known or determined as the basis for the PM

Common Failure Mechanisms The common expected failure mechanisms for dies and patterns are erosion and fatigue. Erosion is the result of abrasive particles washing over the pattern or die cavity surface. It is

Fig. 1

Metal fatigue, evidenced by radial cracking at holes, is a common failure mechanism due to thermal cycling. Courtesy VERSEVO Inc.

1214 / Quality Assurance, Nondestructive Evaluation, and Failure Analysis program. For new tooling, the number of cycles can be estimated based on similar tool size and geometry. For existing tools, historical records should be analyzed to determine the mean cycles to repair.

Maintenance Records A successful maintenance program requires good record keeping for each tool. This record is the history of the tool from the beginning until the tool is scrapped. For example, the maintenance tooling record must have the following information as a minimum:  Tool identification in the form of a unique

 





number for each tool. This number could have customer information, tooling size, origin date, and additional useful information embedded in it or linked to it. It should also be compatible with the accounting system. Production line or machine to specify the production line or machine that is compatible with this tool Casting identification; details of the customer part number and revision level for the tool. Also, if the tool produces multiple castings or multiple part numbers per cycle, that detail is required. For example, a die casting die may have multiple cavities for the same part, or it may have multiple cavities of different parts. The part number, revision level, and number of parts must be known. Tool size and weight. Tool size is required to determine storage and safe handling requirements. If the tool is made up of multiple components, the weight of individual components that are handled separately must be known, and the total weight of assemblies should be detailed. If the tool is comprised of two halves that are handled as an assembly, that also must be known. For storage, the size of the various components, as they are to be stored, must be known. Work order log. Any time that work is performed on the tool, a record must be generated. The previous items in this list are administrative details of the tool. The record or work order is the historical record of all work performed on the tool. It is a detailed diary of work done to the tool internally or outsourced. It should include the number of production cycles on the tool. It is also a record of the time and cost for tool maintenance. It serves to notify all affected departments that work has been authorized and completed on the tool.

Preventive Maintenance Program The objective of the PM program is to prepare the tool for the next scheduled production run (interval), so it is available and able to complete the production run without additional

maintenance. A die casting die is used as an example, since it requires a high level of maintenance due to the combination of high pressure and melt velocity with high cycling rate. Other tools and patterns require similar maintenance, probably to a lesser degree, depending on production volume requirements. This maintenance will require replacement or refurbishment of nearly worn-out components. This includes features that affect the casting and features that affect tool operations, such as alignment, metal flow, heat flow system, and ejection systems. The following PM items are from the North American Die Casting Association’s publication E501 (Ref 1): 1. At the end of each complete run, tear down the die and inspect all components. Polish where required. Replace worn, broken, or bent core pins. Inspect all ejector pins. Lubricate the complete ejector assembly. 2. Stress relieve the cavities at 510  C (950  F) for 4 h after every 25,000 shots. 3. Inspect the leader pins and bushings. Replace as required. 4. Update the date code pin as required during die inspection. 5. Inspect gates for erosion. Correct and requalify as required. 6. Draw polish all surfaces of the cavities. 7. Remove metal buildup from the holding frame and inspect for damage. Correct and requalify as required. 8. Remove, inspect, and correct as required all slides, cam pins, and locking heels. When the components are reset in the tooling, check for proper fit, preload, slide action, and lock interference. 9. Check support pillars. Replace or correct, as required, when proper preload is not present. This will require strain gages that are set with each of the support pillars. 10. Recondition the holding frame at the time of cavity replacement. 11. Clean and polish gas vents as required. 12. Coat the die faces with a protective coating to prevent rust. 13. Flush and/or drill out the water lines to remove lime buildup. 14. Chase water line threads to ensure a good seal. 15. Verify steel hardness at a heavy die area every 25,000 shots. 16. Check holder block and ejector plates for flatness. 17. Redo microprecision peening at the normal half-life of the tool. This interval may need to be shortened to every 30,000 to 50,000 shots if the casting has a history of rapid heat-checking problems or a configuration that warrants more frequent treatments, such as thicker-wall aluminum castings or 390 aluminum alloy applications. 18. Keep a history of each die. This is a generic list to illustrate the degree of detail needed. Each die or pattern should

have a specific list of tasks that must be completed. Again, the objective of PM is to assure a continuous breakdown- and repair-free production interval.

Repair Repair is work done to the tool to keep it in production after a breakdown. Two questions must be addressed with regard to repair: What must be done to make the tool serviceable to complete the production run? How much time is required to complete this work? The answer to the second question will determine where the work will be done—on or off the production line or machine. This time will be balanced against the amount of time required to set up the next tool. Whichever time is shorter is selected, unless there are further mitigating circumstances. Time to repair is an important characteristic with respect to uptime and utilization. Repair time can be minimized by examining tool histories and determining typical failure modes. Many families of tools (castings) will have common failure modes based on operating conditions or casting geometries. This would be determined based on historical records. For example, records for a particular tool indicate that it frequently requires repair for a broken core. Further investigation reveals that the core always fails before the typical production run of 10,000 cycles. To minimize repair time:  Spare cores must be kept in inventory.  A method for fast replacement must be

developed to minimize lost production time.

 A long-term solution must be found to pre-

vent failure during the production run.

Alterations Alterations are the work done to tooling at the customer’s request to change the casting configuration, or the work to tooling may be done to improve the production process. Alterations usually impact the useful life of the tool. Any degradation of tool life must be pointed out to the customer. Likewise, if an alteration negatively impacts the casting process, these concerns must be pointed out. Any monetary impact that the alteration makes, in addition to the cost of the alteration, should also be identified. This impact may be positive or negative. It must be pointed out to the customer and quantified, if possible. If the alteration shortens the manufacturing process cycle time, the effect is positive. If the alteration abrogates tool life, the change is negative. In either case, the projected outcome should be shared with the customer. In the case of alterations to hard tooling, which is tooling made from steel, the welding process will upset the tool surface condition and usually reduce the tool life. An objective, then, in making alterations is to preserve or

Maintenance, Repair, Alterations, and Storage of Patterns and Tooling / 1215 enhance tool life. In the particular case of a die casting die, welding will reduce tool life; an alternative may be to construct a new cavity or partial cavity insert that will actually increase tool life.

Storage—Physical Protection, Cataloging, and Tracking Proper storage of tools and patterns is achieved when several simple objectives are accomplished:  Storage must prevent tool degradation.  The storage area must be a safe workplace

that minimizes hazards.

 Stored tools and patterns must be easily

located.

 Stored tools should be close to the work

stations. Preventing Tool Degradation. Stored tools can be degraded or destroyed by a number of causes. Most easily identified is environmental degradation. Other causes may be fire or vandalism. All must be prevented to keep the tooling in serviceable condition. Environmental factors may include temperature, humidity, ultraviolet radiation, dirt, and other atmospheric contamination. All of these factors must be reviewed with respect to the type of pattern or tool that is being stored. Once the storage requirements are defined, the requirements of the storage area can be defined. Indoors or Outdoors. If a building is required, a cost objective is to minimize the size, both for initial cost and ongoing overhead costs. Minimizing the storage area is a reasonable objective that can be met if the quantity and volume (physical size) of the patterns and tools are known. This requires adequate record keeping, as has been detailed previously. The rate of tooling acquisition and obsolescence must be known. Also, an active program of returning obsolete or inactive tooling to the OEM must be in place. Environmental factors that can degrade tooling will determine if outdoor or covered storage is acceptable. In all cases, the floor or ground must be able to withstand any required loads of tooling, racks, and

moving equipment. The location of the tooling must be cataloged. Weatherproofing. Will the condition of the tool be degraded if it is subjected to the atmosphere? Is the tool subject to oxidation? Both steel and aluminum tools can be degraded by oxidation. Steel tools, in particular, should have a wax or oil treatment applied to critical surfaces. When dirt and dust can be degradation factors, the tool should be covered or wrapped. Also, if heat and changes in humidity can cause condensation, wrapping may be advisable. Ultraviolet radiation can cause plastic materials to degrade; wrapping and shading may be advisable. Fireproofing is required if the tools or patterns can be destroyed by fire. A sprinkler system may be adequate protection in the event of a fire. Tools then may need to be protected from water damage. Safe Workplace. If tool storage is done with racking, the racking must be strong enough to carry the tool loads and secure enough to prevent movement when loading and unloading tools. Adequate aisle space must be provided to allow materials handling equipment to maneuver without being encumbered. For racking that is several tiers high, mirrors and lighting to see the tooling must be provided. There are warehousing systems available today (2008) that are appropriate for tool storage. Easy Location. Each tool must have a specific storage location to expedite retrieval. Adherence to the “5S” (Ref 2) principles of lean manufacturing is recommended. The storage location of each tool is cataloged. Locations in the warehouse are identified, and the storage location can be added to the tool. Finding or locating a tool should be as easy as finding a book in a library. The 5S is a method for organizing and maintaining order in a workplace to reduce waste and optimize productivity. The westernized terms are sort (seiri), set in order or straighten (seiton), shine (seiso), standardize (seiketsu), and sustain (shitsuke). In the day-to-day operations of the company, a clean, well-organized, and maintained storage environment will promote efficient operation. Proximity. Stored tools should be close to the work stations to minimize materials handling and transport. This is an argument for decentralized

tool storage versus centralized tool storage. Indeed, some companies have embraced this policy of tool storage at the work cell. Some companies that follow this practice are also adherents of the SMED system (Ref 3) principle from lean manufacturing. SMED is the acronym for single-minute exchange of dies, a concept developed at Mazda. The object is to complete machine setups in less than 10 min, thereby increasing machine productivity. The proximity of the storage area and efficient racks and equipment handling systems will assist rapid changing of dies and patterns. Cataloging and Tracking. In the section “Maintenance Records,” the need for record keeping was defined. The tooling identification record will serve the purpose of cataloging tools. Some of this information should also be displayed on the tool. In addition to identification, the tool weight must be on the tool. The tool storage location would also be helpful but is not mandatory. A tool work order is best used for tracking a tool. A tool can be at one of three locations: in manufacturing during a production run, in storage, or in maintenance. Following production, the tool goes to storage or maintenance. A work order is the instruction and record used to define maintenance. Following production, a work order is written on the tool. If the work is to be done immediately, the tool goes to maintenance; if not, the tool goes to storage. If no work is to be done and the tool will be in storage for a long time, information is transmitted on the work order that the tool is to be “mothballed.” Additional procedures may be desired to prepare patterns and dies for long-term storage.

REFERENCES 1. “Care and Maintenance of Die Casting Dies Manual and Checklist,” North American Die Casting Association (NADCA), 2005 2. “Lean Manufacturing,” Environmental Protection Agency, www.epa.gov/hazwaste/ minimize/lean.htm, accessed March 2008 3. S. Shingo, A Revolution in Manufacturing: The SMED System, Productivity Press Inc., 1985

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1219-1221 DOI: 10.1361/asmhba0005347

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Metric Conversion Guide This Section is intended as a guide for expressing weights and measures in the Syste`me International d’Unite´s (SI). The purpose of SI units, developed and maintained by the General Conference of Weights and Measures, is to provide a basis for world-wide standardization of units and measure. For more information on metric conversions, the reader should consult the following references:
“Standard for Metric Practice,” E 380,

Annual Book of ASTM Standards, Ameri-

can Society for Testing and Materials, 1916 Race Street, Philadelphia, PA 19103
“Metric Practice,” ANSI/IEEE 268–1982, American National Standards Institute, 1430 Broadway, New York, NY 10018
Metric Practice Guide—Units and Conversion Factors for the Steel Industry, 1978, American Iron and Steel Institute, 1133 15th Street NW, Suite 300, Washington, DC 20005
The International System of Units, SP 330, 1986, National Bureau of Standards. Order

from Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402-9325
Metric Editorial Guide, 4th ed. (revised), 1985, American National Metric Council, 1010 Vermont Avenue NW, Suite 320, Washington, DC 20005–4960
ASME Orientation and Guide for Use of SI (Metric) Units, ASME Guide SI 1, 9th ed., 1982, The American Society of Mechanical Engineers, 345 East 47th Street, New York, NY 10017

Base, supplementary, and derived SI units Measure

Unit

Symbol

Base units Amount of substance Electric current Length Luminous intensity Mass Thermodynamic temperature Time

mole ampere meter candela kilogram kelvin second

mol A m cd kg K s

Supplementary units Plane angle Solid angle

radian steradian

rad sr

Derived units Absorbed does Acceleration Activity (of radionuclides) Angular acceleration Angular velocity Area Capacitance Concentration (of amount of substance) Conductance Current density Density, mass Electric charge density Electric field strength Electric flux density Electric potential, potential difference, electromotive force Electric resistance Energy, work, quantity of heat Energy density

gray meter per second squared becquerel radian per second squared radian per second square meter farad mole per cubic meter siemens ampere per square meter kilogram per cubic meter coulomb per cubic meter volt per cubic meter coulomb per square meter volt

Gy m/s2 Bq rad/s2 rad/s m2 F mol/m3 S A/m2 kg/m3 C/m3 V/m C/m2 V

ohm joule joule per cubic meter

O J J/m3

Measure

Unit

Symbol

Entropy

joule per kelvin

J/K

Force Frequency Heat capacity Heat flux density Illuminance Inductance Irradiance Luminance Luminous flux

newton hertz joule per kelvin watt per square meter lux henry watt per square meter candela per square meter lumen

N Hz J/K W/m2 lx H W/m2 cd/m2 lm

Magnetic field strength Magnetic flux Magnetic flux density Molar energy Molar entropy

ampere per meter weber tesla joule per mole joule per mole kelvin

A/m Wb T J/mol J/mol

Molar heat capacity Moment of force Permeability Permittivity Power, radiant flux Pressure, stress Quantity of electricity, electric charge Radiance Radiant intensity Specific heat capacity Specific energy Specific entropy Specific volume Surface tension Thermal conductivity Velocity Viscosity, dynamic Viscosity, kinematic Volume Wavenumber

joule per mole kelvin newton meter henry per meter farad per meter watt pascal coulomb watt per square meter steradian watt per steradian joule per kilogram kelvin joule per kilogram joule per kilogram kelvin cubic meter per kilogram newton per meter watt per meter kelvin meter per second pascal second square meter per second cubic meter 1 per meter

J/mol K N m H/m F/m W Pa C W/m2 sr W/sr J/kg K J/kg J/kg K m3/kg N/m W/m K m/s Pa s m2/s m3 1/m

K 





 





1220 / Reference Information

Conversion factors To convert from

to

multiply by

To convert from

Angle degree

rad

1.745 329 E  02

Area in.2 in.2 in.2 ft2

mm2 cm2 m2 m2

6.451 600 E + 02 6.451 600 E + 00 6.451 600 E  04 9.290 304 E  02

Bending moment or torque lbf in. N lbf ft N kgf m N ozf in. N

   

m m m m

1.129 848 E  01 1.355 818 E + 00 9.806 650 E + 00 7.061 552 E  03

Bending moment or torque per unit length lbf in./in. N m/m 4.448 222 E + 00 lbf ft/in. N m/m 5.337 866 E + 01

 

Current density A/in.2 A/in.2 A/ft2

 

A/cm2 A/mm2 A/m2

1.550 003 E  01 1.550 003 E  03 1.076 400 E + 01

Electricity and magnetism gauss T maxwell mWb mho S Oersted A/m O cm O m O circularmil/ft mO m Energy (impact, other) ft lbf Btu (thermochemical) cal (thermochemical) kW h W h

 

1.000 000 E  04 1.000 000 E  02 1.000 000 E + 00 7.957 700 E + 01 1.000 000 E  02 1.662 426 E  03

J J J J J

1.355 1.054 4.184 3.600 3.600









Flow rate ft3/h ft3/min gal/h gal/min

L/min L/min L/min L/min

818 350 000 000 000

E E E E E

+ + + + +

00 03 00 06 03

4.719 475 E  01 2.831 000 E + 01 6.309 020 E  02 3.785 412 E + 00

Force lbf kip (1000 lbf) tonf kgf

N N kN N

4.448 4.448 8.896 9.806

Force per unit length lbf/ft lbf/in.

N/m N/m

1.459 390 E + 01 1.751 268 E + 02

pffiffiffiffi MPa m

1.098 800 E + 00

kJ/kg kJ/kg

2.326 000 E + 00 4.186 800 E + 00

Fracture pffiffiffiffiffiffi toughness ksi in: Heat content Btu/lb cal/g

(a) kg  103 = 1 metric ton or 1 megagram (Mg)

222 222 443 650

E E E E

+ + + +

00 03 00 00

to

multiply by

Heat input J/in. J/m 3.937 008 E + 01 kJ/in. kJ/m 3.937 008 E + 01 Length ˚ A nm 1.000 000 E  01 min. mm 2.540 000 E  02 mil mm 2.540 000 E + 01 in. mm 2.540 000 E + 01 in. cm 2.540 000 E + 00 ft m 3.048 000 E  01 yd m 9.144 000 E  01 mile km 1.609 300 E + 00 Mass oz kg 2.834 952 E  02 lb kg 4.535 924 E  01 ton (short, 2000 lb) kg 9.071 847 E + 02 3 ton (short, 2000 lb) kg  10 (a) 9.071 847 E  01 ton (long, 2240 lb) kg 1.016 047 E + 03 Mass per unit area kg/m2 4.395 000 E + 01 oz/in.2 oz/ft2 kg/m2 3.051 517 E  01 oz/yd2 kg/m2 3.390 575 E  02 lb/ft2 kg/m2 4.882 428 E + 00 Mass per unit length lb/ft kg/m 1.488 164 E + 00 lb/in. kg/m 1.785 797 E + 01 Mass per unit time lb/h kg/s 1.259 979 E  04 lb/min kg/s 7.559 873 E  03 lb/s kg/s 4.535 924 E  01 Mass per unit volume (includes density) g/cm3 kg/m3 1.000 000 E + 03 lb/ft3 g/cm3 1.601 846 E  02 lb/ft3 kg/m3 1.601 846 E + 01 lb/in.3 g/cm3 2.767 990 E + 01 lb/in.3 kg/m3 2.767 990 E + 04 Power Btu/s kW 1.055 056 E + 00 Btu/min kW 1.758 426 E  02 Btu/h W 2.928 751 E  01 erg/s W 1.000 000 E  07 ft lbf/s W 1.355 818 E + 00 ft lbf/min W 2.259 697 E  02 ft lbf/h W 3.766 161 E  04 hp (550 ft lbf/s) kW 7.456 999 E  01 hp (electric) kW 7.460 000 E  01 Power density 2 2 W/m 1.550 003 E + 03 W/in. Press capacity See Force Pressure (fluid) atm (standard) Pa 1.013 250 E + 05 bar Pa 1.000 000 E + 05

  



To convert from

to

multiply by

in. Hg (32  F) Pa in. Hg (60  F) Pa lbf/in.2 (psi) Pa torr (mm Hg, 0  C) Pa Specific heat J/kg K Btu/lb  F cal/g  C J/kg K Stress (force per unit area) 2 MPa tonf/in. (tsi) kgf/mm2 MPa ksi MPa lbf/in.2 (psi) MPa 2 MN/m MPa Temperature   F C   R K Temperature interval   F C Thermal conductivity 2  F W/m K Btu in./s ft Btu/ft h  F W/m K 2  Btu in./h ft F W/m K  cal/cm s C W/m K Thermal expansion m/m K in./in.  C in./in.  F m/m K Velocity ft/h m/s ft/min m/s ft/s m/s in./s m/s km/h m/s mph km/h Velocity of rotation rev/min (rpm) rad/s rev/s rad/s Viscosity poise Pa s stokes m2/s ft2/s m2/s in.2/s mm2/s

1.000 000 E  01 1.000 000 E  04 9.290 304 E  02 6.451 600 E + 02

Volume in.3 ft3 fluid oz gal (U.S. liquid)

1.638 2.831 2.957 3.785





     

 

     

     



m3 m3 m3 m3

3.386 3.376 6.894 1.333

380 850 757 220

E E E E

+ + + +

03 03 03 02

4.186 800 E + 03 4.186 800 E + 03 1.378 951 E + 01 9.806 650 E + 00 6.894 757 E + 00 6.894 757 E  03 1.000 000 E + 00 5/9 5/9

 ( F  32)

5/9 5.192 204 E + 02 1.730 735 E + 00 1.442 279 E  01 4.184 000 E + 02 1.000 000 E + 00 1.800 000 E + 00 8.466 667 E  05 5.080 000 E  03 3.048 000 E  01 2.540 000 E  02 2.777 778 E  01 1.609 344 E + 00 1.047 164 E  01 6.283 185 E + 00

706 685 353 412

E E E E

   

05 02 05 03

Volume per unit time ft3/min ft3/s in.3/min

m3/s m3/s m3/s

4.719 474 E  04 2.831 685 E  02 2.731 177 E  07

Wavelength ˚ A

nm

1.000 000 E  01

Metric Conversion Guide / 1221 SI prefixes—names and symbols Exponential expression 18

10

1015 1012 109 106 103 102 101 100 101 102 103 106 109 1012 1015 1018

Multiplication factor

1 000 000 000 000 000 000 1 000 000 000 000 000 1 000 000 000 000 1 000 000 000 1 000 000 1 000 100 10 1 0.1 0.01 0.001 0.000 001 0.000 000 001 0.000 000 000 001 0.000 000 000 000 001 0.000 000 000 000 000 001

Prefix

Symbol

exa

E

peta tera giga mega kilo hecto(a) deka(a) BASE UNIT deci(a) centi(a) milli micro nano pico femto atto

P T G M k h da d c m m n p f a

(a) Nonpreferred. Prefixes should be selected in steps of 103 so that the resultant number before the prefix is between 0.1 and 1000. These prefixes should not be used for units of linear measurement, but may be used for higher order units. For example, the linear measurement, decimeter, is nonpreferred, but square decimeter is acceptable.

ASM Handbook, Volume 15: Casting ASM Handbook Committee, p 1222-1224 DOI: 10.1361/asmhba0005348

Copyright © 2008 ASM International® All rights reserved. www.asminternational.org

Abbreviations and Symbols a crystal lattice length along the a axis; activity A ampere A area ˚ angstrom A ac alternating current AC air cool ACI Alloy Casting Institute ADCI American Die Casting Institute ADI austempered ductile iron AFS American Foundrymen’s Society AGV automatic guided vehicle AISI American Iron and Steel Institute ANSI American National Standards Institute AOD argon oxygen decarburization ASTM American Society for Testing and Materials at.% atomic percent atm atmosphere (pressure) AWS American Welding Society b crystal lattice length along the b axis bcc body-centered cubic BCIRA British Cast Iron Research Association BEM boundary element method BOP basic oxygen process BST beta solution treatment BUS broken-up structure c crystal lattice length along the c axis C concentration C‘ liquid composition C0 initial concentration Cs solid composition CAB calcium argon blowing CAD computer-aided design CADTA computer-aided differential thermal analysis CAE computer-aided engineering CAM computer-aided manufacturing CE carbon equivalent CET columnar-equiaxed transition CG compacted graphite CIM computer-integrated manufacturing CLA counter-gravity low-pressure casting of air-melted alloys CLAS counter-gravity low-pressure air-melted sand casting CLV counter-gravity low-pressure casting of vacuum-melted alloys CMM coordinate measuring machine(s) CNC computerized numerical control cpm cycles per minute cps cycles per second

cpt critical preheating temperature CRE carbon removal efficiency CRR carbon removal rate CS ceramic shell; constitutional supercooling CSP constitutional supercooling parameter CST constitutional solution treatment CV check valve casting CVD chemical vapor deposition CVM control volume method CVN Charpy V-notch (impact test or specimen) d diameter d used in mathematical expressions involving a derivative (denotes rate of change); depth; diameter D diameter; distance; diffusivity; density D diffusion coefficient in liquid Di diffusion coefficient across interface Ds diffusion coefficient in solid da/dN fatigue crack growth rate DAS dendrite arm spacing dc direct current DCEP direct current electrode positive DCRF Die Casting Research Foundation diam diameter DIS Draft International Standard DOC Department of Commerce DS directional solidification DT drop tower e natural log base, 2.71828; electron E modulus of elasticity Ec elastic modulus of composite Ef elastic modulus of fiber Em elastic modulus of matrix EAF electric arc furnace EB electron beam EDM electrical discharge machining EPC evaporative pattern casting Eq equation EPS expanded polystyrene pattern ESR electroslag remelting ESW electroslag welding et al. and others EVA ethylene-vinyl acetate co-polymer f fraction fs solid fraction FC furnace cool fcc face-centered cubic FCAW flux cored arc welding FDM finite difference method FEM finite element method FG flake graphic Fig. figure

FM full mold FM (process) fnote mince (thin iron) process FRC free radical cure FRM fiber-reinforced metals ft foot g gram; gas g acceleration due to gravity G modulus of rigidity; thermal gradient Gc concentration gradient gal. gallon GFN grain fineness number GMAW gas metal arc welding GPa gigapascal Gr graphite GTAW gas tungsten arc welding h hour h height H enthalpy; height; magnetic field strength HAZ heat-affected zone HB Brinell hardness hcp hexagonal close-packed HIP hot isostatic pressing HK Knoop hardness hp horsepower HR Rockwell hardness (requires scale designation, such as HRC for Rockwell C hardness) HSLA high-strength low-alloy (steel) HTH high-temperature hydrogenation HV Vickers hardness HVC hydrovac process Hz hertz I inductor current IACS International Annealed Copper Standard ICFTA International Committee of Foundry Technical Associations ID inner diameter in. inch ISO International Organization for Standardization J solute fluse JIS Japanese Industrial Standard K Kelvin, curvature K stress intensity factor KIc plane-strain fracture toughness Kt theoretical stress concentration factor k distribution coefficient kg kilogram km kilometer kPa kilopascal ksi kips (1000 lb) per square inch kV kilovolt

Abbreviations and Symbols / 1223 L liter L length lb pound LBE lance bubble equilibrium LF ladle furnace LF/VD-VAD ladle furnace and vacuum arc degassing LF/VD ladle furnace vacuum degassing LMR liquid metal refining ln natural logarithm (base e) m meter, liquidus slope M metal Ms martensite start temperature MDI methyl di-isocyanate mg milligram Mg megagram (metric tonne) MHD magnetohydrodynamic (casting) min minimum; minute MINT metal in-line treatment mips million instructions per second mL milliliter mm millimeter MMC metal-matrix composite MPa megapascal mph miles per hour n strain-hardening exponent n0 strain-hardening coefficient N newton N number of cycles of failure NASA National Aeronautics and Space Administration NC numerical control NDTT nil ductility transition temperatures nm nanometer No. number NRL Naval Research Laboratories NT normalized and tempered OAW oxyacetylene welding OD outside diameter Oe oersted OQ oil quench OSHA Occupational Safety and Health Administration OTB oxygen top blown oz ounce p page P pe´clet number P pressure Pa pascal PECB phenolic ester cold box pH negative logarithm of hydrogen-ion activity PH precipitation hardenable P/M powder metallurgy ppi pores per lineal inch ppm parts per million psi pounds per square inch PTS para-toluosulfonic acid PUCB phenolic urethane cold box PUN phenolic urethane no-bake PWHT postweld heat treatment q fatigue notch sensitivity factor QT quenched and tempered r radius R stress (load) ratio; radius; gas constant RE rare earth

Ref reference rem remainder RF radio frequency s second S applied stress SAE Society of Automotive Engineers SAW submerged arc welding SC single-crystal SCC stress corrosion cracking scfm standard cubic feet per minute SCFH standard cubic feet per hour SCRATA Steel Castings Research and Trade Association SEM scanning electron microscopy SG spheroidal graphite Sh Sherwood number SIC standard industry codes SIMA strain induced melt activated SIMS secondary ion mass spectrometry SLQ slack quenched SMAW shielded metal arc welding SNIF spinning nozzle inert flotation SOLA solution algorithm SPAR Space Processing Applications Rocket SR stress relieved SSVOD strong stirred vacuum oxygen decarburization STP standard temperature and pressure t time; thickness T temperature T* interface temperature T0 temperature of equal free energy TL liquidus temperature Tm, Tf melting temperature Ts solidus temperature TAC treatment of aluminum in crucible TCT thermochemical treatment THT transient heat transfer TNT trinitrotoluene TPRE twin-plane reentrant edge mechanism UBC used beverage container UNS Unified Numbering System (ASTMSAE) UTS ultimate tensile strength v volume; velocity V voltage, interface velocity Va velocity of absolute stability Vc velocity of constitutional under cooling V0 collision rate of atoms V volume; velocity VAC-ESR electroslag remelting under reduced pressure VAD vacuum arc degassing VADER vacuum arc double electric remelting VAR vacuum arc remelting V-D vacuum degassing VID vacuum induction degassing VIDP vacuum induction degassing and pouring VIM vacuum induction melting VIM/VID vacuum induction melting and degassing VOD vacuum oxygen decarburization (ladle metallurgy) VODC vacuum oxygen decarburization (converter metallurgy)

VOID vacuum oxygen induction decarburization vol volume vol% volume percent W watt W width; weight WQ water quench WQT water quenched and tempered WRC Welding Research Council wt% weight percent y coordinate parallel to interface yr year z coordinate perpendicular to interface z0 coordinate of specimen  angular measure; degree  C degree Celsius (centigrade)  F degree Fahrenheit ! direction of reaction  divided by = equals  approximately equals 6¼ not equal to identical with > greater than

much greater than  greater than or equal to 1 infinity / R is proportional to; varies as integral of < less than much less than

less than or equal to þ maximum deviation   minus; negative ion charge  diameters (magnification); multiplied by  multiplied by O ohm / per % percent + plus; positive ion charge pffi square root of  approximately; similar to a angle dc solute diffusion length di interface diffusion length dk capillary length dT thermal length D change in quantity; an increment; a range DC0 solidus liquidus interval at Ts DT0 solidus liquidus interval e strain, amplitude of perturbation e_ strain rate l wavelength G Gibbs Thomson parameter min. microinch mm micron (micrometer) n Poisson’s ratio xc stability function p pi (3.141592) r density s tensile stress t shear stress f driving force for instability o wave number

1224 / Reference Information

Greek Alphabet A, a alpha B, b beta G, g gamma D, d delta E, e epsilon Z, z zeta H, Z eta Y, y theta

I, i iota K, k kappa L, l lambda M, m mu N, n nu X, x xi O, o omicron P, p pi

P, r rho S, s sigma T, t tau , u upsilon F, f phi X, w chi C, c psi O, o omega

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Index A Acid melting practice acid steelmaking, 92–94 alloys, additions, 95 alloys, loss of, 95 the charge, 94 deoxidation, 94–95 overview, 91–92 the oxidation, 94 Acid steelmaking complete oxidation, 93 meltdown period, 93–94 overview, 92 in partial oxidation, 92–93 refining period, 94 Aging, 405–406 Aluminum alloy ingot casting and continuous processes continuous processes. See also Continuous processes DC process, solidification in, 1006 ingot casting processes DC process variations, 1003, 1004 direct chill process, 1002–1003 open-mold casting, 1002 overview, 1002 particle processes, 1003–1004 ingot forms, 1001, 1002 molten-metal processing, 1001–1002 overview, 1001 postsolidification processes, 1007 safety, 1007 Aluminum and aluminum alloy castings alloying and impurity elements, effects of, 1059–1063 aluminum casting alloys, properties of corrosion, 1081 fatigue, 1077, 1079–1081 hot cracking, 1082–1083 mechanical properties, 1081–1082 molten metal fluidity, 1082, 1083 overview, 1077, 1078–1079, 1080, 1081 shrinkage, 1082 aluminum-silicon alloys, modification and refinement of chemical modifiers, 1066–1067, 1068 hypereutectic aluminum-silicon alloys, 1068–1069 phosphorus, importance of, 1067 casting applications, foundry alloys for specific die-casting compositions, 1069 premium engineered castings, 1070–1071 rotor castings, 1069–1070 chemical compositions, 1059 heat treatment annealing, 1075 heat treating problems, 1075–1076 natural aging, 1075 overview, 1072 precipitation heat treating/aging, 1075 property development, natural versus artificial, 1076 quality control, 1076 quenching, 1074–1075 solution heat treatment, 1072–1074 stability, 1076

overview, 1059 structure control grain structure, 1063–1066 nondendritic microstructures, 1063, 1064 overview, 1063 secondary dendrite arm spacing, 1063, 1065 welding of, 1077 Aluminum and it alloys, gases in inert gas flushing, 68–70 overview, 67 solidification, hydrogen solubility and reactions during, 67–68, 69 solidification, reactions during, 68 Aluminum-base and copper-base alloys, thermodynamic properties of activities, 56 hypothetical standard state, thermal properties for, 58 interaction coefficients, 56–58 overview, 56 phase diagrams, 58, 60 thermal properties, 56, 57, 58, 59, 60, 61 Aluminum casting alloys, grain refinement of grain refinement, benefits of, 260, 261 grain refinement, mechanisms of, 255–257 grain refiner, boron as, 257–258 grain size, measurement of, 255, 256 overview, 255 practices aluminum-copper casting alloys, 259 aluminum-magnesium casting alloys, 260 aluminum-silicon casting alloys, 258–259 aluminum-silicon-copper casting alloys, 259 aluminum-zinc-magnesium casting alloys, 259 titanium, forms of insoluble, 258 soluble, 258 Aluminum degassing degassing process, other considerations for, 188 overview, 186–187 process gas, selecting, 188 rotary impeller degassing, 187–188 Aluminum fluxes and fluxing practice aluminum fluxing flux morphology, 233 gas fluxing, 231 solid fluxes, 231–233 fluxes, categories of, 233–234, 235 foundry practices flux injection, 235–236 fluxing applications, 237–238 overview, 235 practical application, 236–237 overview, 230–231 Aluminum fluxing. See aluminum fluxes and fluxing practice Aluminum shape casting casting process selection aluminum, pressure die casting of, 1013–1016 expendable mold methods, 1011–1013, 1015, 1016, 1017, 1018 overview, 1010–1011, 1012, 1013, 1014 melting and melt handling die casting cycle, 1010 die castings, gating of, 1010 gating and risering principles, 1010

gravity casting, 1009–1010 overview, 1009 pouring, 1010 overview, 1009 premium engineered castings melt processing and solidification, 1016 mold filling, 1017–1018 mold materials, 1016–1017 overview, 1016 Aluminum-silicon alloys, modification of eutectic grain structure copper segregation, 247, 248 feeding and shrinkage formation, 246–247 iron segregation, 247, 248 overview, 245–246 eutectic nucleation, 244–245 foundry practices, recommended gas pickup and degassing, 248 how much to add, 248 mold reactions, 249 quality control testing, 248–249 strontium, adding, 247–248 heat treated castings, 241–242 modification mechanisms, 243–244 other elements, effect of antimony, 249–250, 251 bismuth, 250 boron, 250 calcium, 250 magnesium, 250 phosphorus, 249 overview, 240–241 porosity and shrinkage formation, 242–243 silicon eutectic, growth of lamellar structure, 249 mostly modified structure, 249 nonmodified structure, 249 overall ratings, 249 overview, 249, 250 super modified structure, 249 transitional lamellar structure, 249 Aluminum-silicon eutectic aluminum-silicon equilibrium phase diagram, 312–313 chemical modification of, theories of, 315–316 microstructure of, 313, 314 solidification of, theories of, 313–315 structure, classification of, 313 Arc furnace components back-slagging pits, 90 fume/dust collection, 90, 94, 95 furnace shell, 93 ladles, 90, 95, 96 pre-heat burners, 90 spout, 90, 93 tap hole, 90, 93 tap-out, 90 water-cooling system, 90 Argon oxygen decarburization (AOD) carbon steels, processing of, 210–211 composition improvement with, 211 equipment, 209–210 fundamentals, 209 low-alloy steels, processing of, 210–211 overview, 208–209 oxygen top and bottom blowing (OTB), 211–212

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

1226 / Reference Information Argon oxygen decarburization (AOD) (continued) property improvement with, 211 stainless steels, processing of, 210 Argon stirring (argon rinsing), 216–217

B Basic melting practice alloys, loss of, 98 basic steelmaking, 95–97 the charge, 97 final deoxidation, 97–98 heat, reducing, 97 overview, 95 Basic steelmaking the charge, 95 oxidizing period, 95–96 reducing/refining the heat, 96–97 Bifilm, the bifilms, 362–363 entrainment, 363–364 fracture, central role of bifilm defects in, 368 furling, 364 mechanical properties, reliability and reproducibility of, 367–368 overview, 362 unfurling dendrite pushing, 366 gas precipitation from solution (formation of gas microporosity), 364–366 intermetallics, nucleation and growth of, 367 linear contraction (hot tearing), 366 overview, 364 solidification contraction, shrinkage due to, 366

C Cast iron gas-liquid reactions, kinetics of, 64–66 hydrogen and nitrogen, solubility of, 64 inert gas flushing, nitrogen and hydrogen removal by, 66–67 solidification, reactions during, 66 Cast iron foundry practices casting dimensional control, 820 gating and feeding system, 819–820 overview, 819 casting and solidification gating systems, 824 overview, 824 pouring temperature and speed, 824 solidification and feeding (risering), 824–825 compacted graphite iron (CG) foundry practice chemical composition, 826 description of, 826 melt treatment, 826–828, 829 melting practice, 826 molding materials, 829 overview, 826 process control, 828 quality control, 828–829 desulfurization methods, 816–817 ductile iron composition control aluminum, 821 carbides, minor elements promoting, 821 carbon, 820 carbon equivalent, 820–821 cerium, 821 hardenability, alloying elements to promote, 821–822 magnesium, 821 manganese, 821 nonspheroidal graphite, minor elements promoting, 821, 822 pearlite, minor elements promoting, 821 phosphorus, 821 silicon, 820 special properties, alloying elements to achieve, 822 sulfur, 821

ductile iron foundry practice melting practice, 820 overview, 820 ductile iron production quality tests mechanical test coupons, 826 metal composition, 825 nodularity and nodule number, 825–826 gray iron alloying, 818–819 high-alloy white irons, foundry practice for, 831 high-chromium white irons, 832–833 high-nickel ductile irons foundry practice, 830 high-silicon ductile irons, foundry practice for, 830 high-silicon gray irons, foundry practice for, 830–831 inoculation, 823–824 addition methods, 818 defined, 817 inoculant evaluation, 818 inoculant, purpose of, 817 inoculant, types of, 817–818 magnesium content, control of, 823 magnesium treatment, 822–823 malleable iron foundry practice, 829–830 melting practice arc furnace melting, 816 charge components, storage and handling of, 816 Cupola furnace, 813–815. See also Cupola furnaces Cupolas, charge composition for, 815–816 induction furnaces, charge materials for, 816 induction melting process, 812–813, 814 overview, 813, 814 nickel-chromium white irons, 831–832 overview, 812, 813 pouring, 819 Cast iron, inclusions in ductile cast iron, 80–81 gray cast iron, 80 Cast iron: solidification of eutectic alloys cooling curve analysis, 327 coupled zone, 320–321 directional solidification, 321–322, 323 isothermal solidification, 321 liquid iron-carbon alloys, structure of, 317 multidirectional solidification austenite-iron carbide eutectic, 326, 329 compacted/vermicular graphic eutectic, 325–326, 328, 329 flake(lamellar) graphite eutectic, 322, 324 spheroidal graphite eutectic, 322–325, 326, 327 nucleation of eutectic in cast iron austenite-flake graphite eutectic, 318–320 austenite-graphite eutectic, structure of graphite in, 317–318 austenite-iron carbide eutectic, 320 austenite-iron carbide eutectic, structure of iron carbide in, 320 austenite-spheroidal graphite eutectic, 320 overview, 317 primary phases, nucleation of primary austenite, 320 primary graphite, 320 Cast irons, introduction to acoustic properties, 801–802 classification of compacted graphite iron, 788 ductile iron, 788 gray iron, 787 high-alloy irons, 788 malleable iron, 786–787 overview, 785, 786 white irons, 785–786, 787 coatings electroplating, 809 overview, 809 short-term corrosion or rust prevention, 809 surface cleaning, 809 thermal spraying, 809 composition of alloying elements, 790–791 carbon equivalent, 788–789, 790 combined carbon, 789

manganese, 789 minor elements, 790 overview, 788, 789, 790 phosphorus, 789–790 sulfur, 789 conductive properties electrical resistivity, 798–800 overview, 797 thermal conductivity, 797–798, 799 graphite morphologies compacted graphite iron, 794 ductile iron graphite, 793–794 gray iron graphite, 792–793 malleable iron graphite, 793 overview, 792 grinding processes honing, 808–809 lapping, 809 heat treatment annealing, 802–803 austempering, 805 heating, 804 normalizing, 803 obtainable properties versus tempering temperature, 805 overview, 802 quenching, 804 stress relieving, 802 tempering, 804–805 machining and grinding machinability of cast iron, effect of microstructure on, 807–808 machinability rating, 807 overview, 806–807 magnetic properties, 799–801 materials selection, 809–810 matrix structure development, 794 overview, 785 physical properties of density, 794–795 overview, 794 solidification of chill test, 791 inoculation, 791–792 overview, 791 thermal analysis, 791 thermal properties latent heat of fusion and solid-state transformation, 796, 797 specific heat and enthalpy, 795–796, 797 thermal expansion, 796–797, 798 tool life, 808 welding of overview, 805 weld quality, factors affecting, 806 Cast metal-matrix composites, synthesis and processing of, and their applications components currently in use automotive industry, 1157–1159 electronic packaging applications, 1159–1160 overview, 1156–1157 space applications, 1159 metal-matrix composites (MMCs), 1149 MMCs, classification of, 1149–1150 MMCs, solidification of, and possible effects of reinforcement on solidification infiltration process, 1152–1154 liquid alloy, effects of reinforcement present on solidification, 1154–1155 overview, 1150–1151 stir casting, 1151–1152 property motivation, 1155–1156 research imperatives, 1162–1163 select cast MMCs, development of aluminum-graphite composite, 1160 cast metallic foams, 1161–1162 fly-ash-filled syntactic foams, 1161, 1162 hybrid composites, 1160–1161 lead-free copper alloy/graphite composites, 1160 Castability: fluidity. See Fluidity Casting applications, numerical methods for. See Computational fluid dynamics (CFD)

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Index / 1227 Casting failure analysis techniques and case studies background information, 1183 case studies cast aluminum bracket, 1187–1188 cast bronze suction roll shell, 1188–1189 cast steel spindle, 1189–1190 data analysis, 1187 introduction, 1183 material evaluation chemical analysis, 1185–1186 fractography, 1184–1185 mechanical testing, 1186 metallography, 1186 overview, 1184 special testing, 1187 planning and sample selection, 1184 preliminary examination, 1184 report preparation, 1187 Casting fracture characteristics fracture features ductile rupture, 1179 environmentally induced fracture, 1181–1182 fatigue, 1180–1181 overview, 1179 transgranular brittle fracture, 1179–1180 general fracture examination background information, 1178 examination methods, 1178 fracture appearance, factors that affect, 1178–1179 regions of interest, 1178 overview, 1178 Casting: hot tearing. See Hot tearing Casting practice: guidelines for effective production of reliable castings overview, 497 rule 1: good-quality melt, 497 rule 2: liquid front damage maximum velocity requirement, 497–498 no-fall requirement, 498–499 rule 3: liquid front stop meniscus, continuous uphill advance of, 499–500 overview, 499 rule 4: bubble damage gravity-filled running systems, 500 overview, 500 pumped and low-pressure filling systems, 500–501 rule 5: core blows, 501 rule 6: shrinkage damage feeding, computer modeling of, 502 feeding with gravity, 501–502 overview, 501 random perturbations to feeding patterns from casting to casting, 502 rule 7: convection damage casing-section thickness and, 502–503 overview, 502 rule 8: segregation, 503 rule 9: residual stress, 504–505 rule 10: location points, 505 Casting processes, computer simulation of. See Computer simulation of casting processes Castings, hot isostatic pressing (HIP) of as-HIPed properties, effect of inclusions on, 412–413 castings, effect of HIP on shape and structure of microstructural/physical changes, other, 413–414 volumetric shrinkage and distortion, 413 economics of, 415 HIP system, overview of conditions, selecting, 410, 411 HIP cycle, 409–410 introduction, 409, 410 pore closure, mechanisms of, 411 history of, 408–409 mechanical properties, effect on, 411–412, 413 overview, 408 problems encountered in surface-connected porosity, 414, 415

thermally induced porosity, 414–415 reasons for using, 408–409 Castings, processing and finishing of automated gate removal, 517–518 blast cleaning internal surfaces, 519–520 overview, 519 cleaning operations, 516 core bake-out method, 516 core knockout, 514–515 flame cutting, 519 gate grinding, 518–519 high-frequency drive machine, 515 high-pressure water blast device, 515–516 inspection liquid penetrant, 520 magnetic particle, 521 overview, 520 pressure testing, 520 radiograph, 520–521 ultrasonic, 521 overview, 513 shakeout, 513–514 shotblast cleaning, 516 Centrifugal casting advantages of, 668 applications centrifugal casting with combustion synthesis, 672, 673 investment casting with centrifugal force, 671–672 overview, 671 spin casting, 672–673 centrifugal castings, defects in, 670–671 equipment casting cooling, 669–670 casting extraction, 670 casting inoculation and fluxing, 670 overview, 669 pour equipment, 669 spinners, 669 methods centrifuge (centrifugal die) casting, 668–669 overview, 667, 668 semicentrifugal (centrifugal mold) casting, 668 true centrifugal casting, 667–668 molds, 670, 671 overview, 667 Ceramic molding applications, 622 overview, 622 Shaw process, 622–624 Unicast process, 624–627 Ceramic shell mold assembly applications, 657, 658, 659 binders colloidal silica, 652 ethyl silica, 652 hybrid binders, 652 liquid sodium silica, 652 other materials, 652 casting methods, 655–656 ceramic cores, manufacture of preformed cores, 654 self-formed cores, 654 cluster preparation coating, 653 drying, 653–654 overview, 653 design recommendations, 657 ingredients, other antifoam compounds, 652 miscellaneous constituents, 652 wetting agents, 652 inspection and testing alloy type test, 656 liquid fluorescent inspection, 656 magnetic particle inspection, 656–657 miscellaneous inspection methods, 657 visual inspection, 656 x-ray radiography, 657 investment casting, design advantages of, 657 investment casting processes, special

Replicast process, 660. See also Replicast process Shellvest system, 658–660 melting equipment, 655 mold firing and burnout, 654–655 mold refractories alumina, 652 aluminum silicates, 652 overview, 651 silica, 651 yttria, 652 zircon, 651 overview, 651 pattern removal autoclave dewaxing, 654 flash dewaxing, 654 liquid dewaxing, 654 overview, 654 postcasting operations, 656 slurry, control procedures, 653 slurry formulation, 652–653 slurry preparation, 653 Chemical kinetics alloy additions, kinetics of, 35 alloy that is solid at bath temperatures, dissolution of, 37 alloy with melting point lower than bath temperature, dilution of, 35–37 interphase mass transport, 35 nucleation, 35 overview, 35 steel in iron-carbon melts, dissolution of, 37–40 Cobalt and cobalt alloy castings applications aero and land gas turbines, alloys for, 1118 corrosion-resistant alloys, 1118 orthopedic implants, alloys for, 1118 wear-resistant alloys, 1117 overview, 1114 physical metallurgy compositions, 1114, 1115 crystallography, 1114 foundry, 1116–1117 overview, 1114 phases and microstructure, 1114–1115 properties, 1115–1116 postcasting treatment coatings, 1117 heat treatment, 1117 hot isostatic pressing (HIP), 1117 welding of, 1117 Cold chamber die casting machine components clamp end, 724–725 overview, 724 shot end, 725 overview, 724 process parameters component size, 726 fast shot, 725 intensification phase, 725–726 slow shot, 725 Cold hearth melting cold hearth versus drip melting, 146, 147 equipment for, 146 melted materials, characteristics of, 147 overview, 145, 146 principles of, 145–146, 147 refining and production data, 146, 148 superalloys, 147 Compacted graphite iron castings castability, 873–874 compacted graphite (CGI) irons, 872 graphite morphology, 872–873 mechanical properties and corrosion resistance applications, 879–880 compressive properties, 877, 878, 879 corrosion resistance, 879 fatigue strength, 878, 881, 882 impact properties, 878–879, 880, 881 overview, 874 properties, sonic and ultrasonic testing of, 879, 882, 883

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

1228 / Reference Information Compacted graphite iron castings (continued) shear properties, 877 tensile properties and hardness, 874–877, 878 overview, 872 Computational fluid dynamics (CFD) casting applications defect simulation, 423, 424 fluid flow in tundishes, 423–424 gate design, 423 overview, 422–423 complex geometries, grid generation for mathematical complexity, 422 meshing complexity, 421–422 overview, 420–421, 422 fluid-flow equations, numerical solution of arbitrary Lagrangian-Euler (ALE) equations, 420–421 finite-difference methods (FDMs), 420 finite-element methods (FEMs), 420, 421 finite-volume methods (FVMs), 420 overview, 420 fundamentals of continuous, compressible media, 419–420 fluid flow, constitutive relations of, 420 governing principles, 419–420 overview, 419 turbulence, 420 overview, 419 Computer simulation of casting processes boundary conditions flow simulation, 463–464 solidification and heat treatment simulation, 464 stress and deformation simulation, 464 casting defects phenomenon, estimation of cold shuts, 466 flow marks, 466 gas and inclusions, entrapment of, 465–466 misruns, 466 porosity estimation, 466 surface folds, 466 enmeshing element shape, 464 element size, 465 initial conditions, 463 optimization, 466 overview, 462 physical properties, 464 shape and phenomena, modeling of, 462–463 simulation, validation and accuracy of, 465 structure and properties, estimation of, 466 Continuous processes conventional product, comparisons with, 1006 overview, 1004–1005 slab casting, 1006 twin-roll strip casting, 1005 wheel-belt processes, 1006 Converter metallurgy, 212, 213 Cooling curves, interpretation and use of. See Thermal analysis Copper alloys degassing auxiliary degassing methods, 192 copper, gases in, 189–190, 191 inert gas fluxing, 191 methods, 190–192 overview, 189 oxidation-deoxidation practice, 190–191 solid degassing fluxes, 191–192 vacuum degassing, 192 zinc flaring, 191 inclusions in, 82 vacuum induction melting, 122–123 Copper and copper alloy castings alloying additions impurities, 1086–1087 major, 1085–1086 minor, 1086 alloying systems and specifications brasses, 1087 bronzes, 1087–1088 copper-nickel-zinc alloys, 1088 copper-nickels, 1088 coppers, 1087

high-copper alloys, 1087 leaded coppers, 1088 overview, 1087 special alloys, 1088 cupronickels, 1091, 1093–1094 heat treatment, 1091, 1093 overview, 1085 physical metallurgy aluminum bronzes, 1090–1091 copper-tin-lead-zinc alloys, 1088–1089 properties, 1091, 1092–1093 types of, 1091 welding, 1091 Copper and copper alloys, casting of. See also Copper and copper alloy castings castability, 1026–1027 casting process selection centrifugal castings, 1041 continuous casting, 1041 die casting, 1041 investment casting, 1041 overview, 1040 permanent mold casting, 1041 plaster molding, 1041 sand casting, 1040 shell molding, 1040–1041 copper alloy castings, production of, 1039–1040 copper alloys, degassing, 1032–1033 gas solubility, 1032–1033 gases, testing for, 1033 hydrogen, sources of, 1032 methods, 1033 overview, 1032 copper alloys, deoxidation of boron deoxidation, 1036 copper deoxidizers, relative effectiveness of, 1037 deoxidizer addition, testing for proper, 1037 high-conductivity copper, 1036 lithium deoxidation, 1036–1037 magnesium deoxidation, 1037 overview, 1035 phosphorus deoxidation, 1035–1036 copper alloys, filtration of, 1037 copper alloys, fluxing, 1029 copper alloys, grain refining of, 1037 copper alloys, pouring temperatures of, 1038 degassing methods auxiliary degassing methods, 1035 inert gas fluxing, 1033, 1035 oxidation-deoxidation practice, 1033, 1034 solid-degassing fluxes, 1034, 1035 vacuum degassing, 1034 zinc flaring, 1033, 1034 feeding group I, 1044–1047 group II alloys, 1044–1047 group III alloys, 1047–1048 overview, 1044 fluxes melt-refining, 1030–1031, 1032–1033 mold, 1031–1032 neutral cover, 1029–1030 oxidizing, 1029 reducing/refining the heat, 1030 gating casting quality, maximizing, 1044 chokes, 1042 function of, 1041 knife and kiss gating systems, 1043–1044 pouring basin, 1041–1042 pouring rate, 1042, 1043, 1044 runners and gates, 1042–1043 spru base, 1042 sprue, 1042 group I alloys, melt treatment for, 1038–1039 group II alloys, melt treatment for, 1039 group III alloys, melt treatment for, 1039 melting and melt control electric induction furnaces, 1028–1029 fuel-fired furnaces, 1027–1028 overview, 1027 overview, 1026

Copper and its alloys, gases in copper alloys, degassing, 71 copper alloys, porosity in, 71 hydrogen solubility and reactions, 70–71 overview, 70 Copper continuous casting history, 1019–1020 horizontal continuous casting overview, 1022–1023 Unicast horizontal casting system, 1023 overview, 1019 strip casting Hazlett casting process, 1023–1024 overview, 1023 Werti strip casting, 1023 upcasting methods Outokumpu upcasting method, 1021–1022 pressure upcast system, 1022 Rautomead upwards vertical casting, 1022 vertical continuous casting overview, 1020 upcasting methods, 1021–1022 vertical continuous slab casting, 1020–1021, 1022 vertical semicontinuous slab casters, 1020, 1021 wheel casting Ohno continuous casting process, 1024 Properzi wheel casting technology, 1024 Coremaking assembly and core setting, 588–589 basic principles of core boxes, 583 core drivers, 583 core venting, 583 cores, reinforcing of, 583 overview, 581, 582–583 compacted cores, curing atmospheric humidity, 586 core baking (oil sand), 585–586 core blowing, 585 sand grain shape, fineness, and permeability, 586 thermal conductivity, heat capacity, and density, 586 compaction core blowing, 584–585 overview, 584 core coatings (core washes), 587–588 core sands and additives additives, other, 584 overview, 583–584 release agents, 584 sand mixing, 584 cores, designing to eliminate, 595–596, 597 coring versus drilling, 596–597 design considerations chaplets, 590 core fragility, 591–592 core size limitations, 589–590 example: addition of holes to provide core support, 590 example: redesign of a sand casting to solve core and mold problems, 590–591 example: redesign of casting that eliminated thin core section that caused a hot spot, 592 molding and coring details, 590 overview, 589 thin core sections, 592 thin-wall casting sections formed by cores, 592, 593 green sand versus dry sand cores, 593 hot box cores applications, 587 core boxes, 587 curing temperature and time, 587 hot box process, advantages of, 587 hot box process, disadvantages of, 587 machines, 587 mixing, 586 overview, 586 sand control, 586–587 sand introduction, 587 investment castings, cores in, 594–595 overview, 581, 582

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Index / 1229 permanent mold castings, cores in, 593 sand coremaking, binders for, 581–582, 583 tortuous passages, coring of, 593–594 Crucible furnaces applications crucible advantages, 155 crucible and application, matching, 155–156 overview, 155, 156, 157 design and operation burner control, 158 burner design, 158–159 charging techniques, 159 crucible care, 159 crucibles, cleaning, 159 furnace atmosphere, 158 melting and pouring, handling of crucibles during, 159 overview, 155 types movable, 158 stationary, 156–158 tilting, 158 Cupola furnaces charge materials charge components, storage/handling of, 103, 104 coke, 103–105 ferrous alloys, 105 fluxes, 105–106 metallic, 105 cokeless Cupola, 107 construction/operation blower systems, 101 bottom doors, 100 charge doors, 100 emission-control systems, 103, 104 intermittent or continuous tapping Cupolas, 101 refractory lining, 101–102 shell design, 100–101 tuyeres, 101 water-cooled, 102–103 control principles carbon, 106 coke bed, 106 iron temperatures, 106 manganese, 106 melt rate, 106 phosphorus, 106 residual alloys, 106 silicon, 106 sulfur, 106 tramp elements, 106 control tests/analyses chill test, 106 eutectic arrest methods, 106 mechanical tests, 106–107 optical spectrometer equipment, 106 intermittent or continuous tapping Cupolas, 101 overview, 99 plasma-fired, 107 progressive refinements in basic slags, 100 computer applications, 100 duplex electric holders, 100 fine coke, injection of, 100 overview, 99 oxygen enrichment and injection, 99–100 preheated air blast, 99 recuperative hot blast system, 99 technology failures, 100 top gases, continuous analyses of, 100 water cooling, 100 sand bottom, 100 water-cooled, 102–103

D Defects (common) in various casting processes case studies cavities: gas porosity and shrinkage, 1200–1201, 1202 discontinuities: hot tearing and cold shut, 1201–1202

inclusions, 1199–1200, 1201 international classification of common casting defects, 1192–1199 overview, 1192 Degassing aluminum foundry aluminum degassing, 186–188 hydrogen sources, 185–186 overview, 185, 186 copper alloys, 189–192 magnesium, 188–189 overview, 185 Dendrite growth columnar-to-equiaxed transition, 303–304 direction, 301–302 microsegregation and secondary arm spacing, 303 model, 302–303 overview, 301 Die casting tooling design tolerance, 737, 738 die casting dies, 734–735 die casting machines, 734, 735 heat removal cooling requirements, analysis of, 741–742 heat flow, control of, 741 overview, 740–741 molten metal delivery, 734–735 overview, 734 product design, 735–736 secondary processing, 736 tooling and metal flow air venting, 739 fluid flow, interaction of, with dies and machines, 740 metal injection, 737–739 shrinkage, feeding of, 739–740 tooling, materials selection for die casting dies, materials for, 742–743 die casting dies, wear and failure in, factors influencing, 745–746 wear and failure modes, 743–745 Direct current arc furnace, 98 Dosing furnaces dosing operation, controlling/monitoring, 183, 184 furnace recharging, 183–184 melt holding, 184 melt level, detection of, 183 overview, 182–183 Drip melting cold hearth versus drip melting, 147 electron beam drip melted metals, characteristics of, 145 electron beam heat source specifications, 144 equipment for, 144–145 ingots, horizontally fed, 144 ingots, vertical feeding of, 143–144 other process considerations, 144 overview, 143 Ductile iron castings austempered ductile iron (ADI), 856, 869, 870 compressive properties, 861–862 ductile iron, classes and grades of, 856–857 fatigue properties, 862–865 fracture toughness impact tests, 865–867 overview, 865 sharply notched specimens, fracture of, 867 hardness properties, 859 mechanical properties, factors affecting composition, 857–858 microstructure, 857 section effect, 858–859 overview, 856 physical properties density, 867 thermal properties, 867–869 shear and torsional properties, 861, 862 tensile properties modulus of elasticity, 860–861, 862 overview, 859–860 tensile test, 860

E Electric arc furnace melting acid melting practice acid steelmaking, 92–94 alloys, additions, 95 alloys, loss of, 95 the charge, 94 deoxidation, 94–95 overview, 91–92 the oxidation, 94 alloy storage, 91 arc furnace components back-slagging pits, 90 fume/dust collection, 90, 94, 95 furnace shell, 90, 93 ladles, 90, 95, 96 overview, 88–89, 91, 92 pre-heat burners, 90 roof structure, 89–90, 92 spout, 90, 93 tap hole, 90, 93 tap-out, 90 tilt control, 90 water-cooling system, 90 basic melting practice alloys, loss of, 98 basic steelmaking, 95–97 the charge, 97 final deoxidation, 97–98 overview, 95 reducing the heat, 97 EAF steelmaking, 98 iron, arc melting of, 98 overview, 87 power factor, 88 power supply, 87–88 quality-control testing equipment, 91 scales, 91 scrap storage, 91 Electromagnetic pumps furnace circulation and scrap recycling technology, 177 mold-filling, 177 overview, 177 types, 177–178 Electron beam melting cold hearth melting cold hearth melted materials, characteristics of, 147 cold hearth versus drip melting, 146, 147 equipment for, 146 overview, 145, 146 principles of, 145–146, 147 refining and production data, 146, 148 superalloys, 147 drip melting electron beam drip melted metals, characteristics of, 145 equipment for, 144–145 horizontally fed ingots, 144 other process considerations, 144 overview, 143, 144 vertical feeding of ingots, 143–144 overview, 142 process description electron beam heat source specifications, 143, 144 overview, 142–143 Electroslag remelting (ESR) ESR furnaces control parameters, 126–127 heavy ingots, electroslag remelting of, 127–128 overview, 125–126 ingot defects, 129 ingot solidification, 128–129 overview, 124–125 process variations electroslag rapid remelt, 129 electroslag remelting under vacuum (VAC-ESR), 129 pressure electroslag remelting, 129

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

1230 / Reference Information Electroslag remelting (ESR) (continued) steel ESR overview, 130 oxygen, behavior of, 130 sulfur, behavior of, 130 ESR furnaces control parameters, 126–127 heavy ingots, electroslag remelting of, 130 overview, 125–126 Eutectic alloys, solidification of aluminum-silicon. See Aluminum-silicon eutectic cast iron. See Cast iron: solidification of eutectic alloys Eutectic solidification—an introduction binary eutectic phase diagram, 307 eutectic structures classification of, 307–310 halos, formation of, 310–311 overview, 307 overview, 307 Expendable mold processes with expendable patterns: introduction, 637–639 Expendable mold processes with permanent patterns: introduction, 525–527 Expendable molds, aggregates and binders for ceramic shell molds binder materials, other, 547 colloidal silica bonds, 547 ethyl silicate, 547 hybrid binders, 547 liquid sodium silicate, 547 overview, 547 foundry sands, 529, 532 green sand molding, clays for bentonites, 536–537 clay properties, controlling, 537–538 clay-water bonds, 537 fireclay, 537 overview, 535–536 hot box processes, 543 inorganic binders aluminate (cement)-bonded sand molds, 539 overview, 538 phosphate-bonded-sand molds, 539 sodium-silicate-bonded sand molds, 538–539 mold media, 528–529, 531 organic binders ester-cured alkaline phenolic no-bake, 541 furan acid-catalyzed no-bake, 541 no-bake processes, 540–541 oil urethane no-bake resins, 542 overview, 539–540 phenolic acid-catalyzed no-bake, 541 phenolic urethane no-bake (PUN), 542 polyol-isocyanate system, 542–543 resin types, 540 silicate/ester-catalyzed no-bake, 541 oven-bake processes/core-oil binders additives, 544 cold box processes, 544 core-oil, types of, 544 free radical cure (FRC) process, 545 operational considerations, 544 overview, 544 phenolic ester cold box (PECB) process, 546 phenolic urethane cold box (PUCB), 544–545 SO2 gas, disposal of, 545–546 SO2 process (furan/SO2), 545 overview, 528, 529, 530 rammed graphite molds, 546–547 sands, types of additives, 532–533 aluminum silicates, 530, 532, 534 chromite, 530 grain shape, 534–535 mold porosity and permeability, 535 mullite, manufactured from aluminum silicates, 532, 534 olivine, 532, 533 overview, 529 sand fineness, 534 sand properties, 533–534 sand quality control, 535

silica, 529–530, 532, 533, 534 zircon, 532 shell process (Croning process) cores and molds, producing, 543 novolac shell-molding system, 543 overview, 543 warm box binders, 543 warm box catalysts, 543–544

F Feeding system concepts. See Filling and feeding system concepts Ferrous alloys, inclusions in steels complex inclusions, 79–80 exogenous inclusions, 80 nitrides, 80 oxides, 77–78 sulfides, 78–79 Filling and feeding system concepts alternative filling systems, performance of, 511–512 entrainment defects in gravity-poured castings, 506 filters, 511 foundry practice, recommended, 512 naturally pressurized filling system feeding, 510–511 gates, 510 offset pouring basin, 508 overview, 508 runner, 509–510 sprue, 509 sprue/runner junction, 509 overview, 506 prepriming techniques, 507–508 traditional filling designs, 506–507 Fluidity factors influencing, 375–376 fluidity measurements, 376 modeling of, 376 overview, 375 Fluxes calcium carbide (CaC2), 105–106 cleaning, 234–235 cover, 234 Cupola furnaces, 105–106 drossing, 234, 235 fluorspar (CaF2), 105 furnace wall-cleaning, 235 limestone, 105 overview, 233–234 soda ash (Na2CO3), 105 Fluxing applications crucible furnaces, 237 holding/ casting furnaces, 238 other considerations, 238 reverberatory furnaces, 237–238 transfer ladles, 237–238 Fluxing practice. See Aluminum fluxes and fluxing practice Foundry practices. See Cast iron foundry practices

G Gas porosity formation heterogeneous nucleation, 373 homogeneous nucleation, 373 formation, factors affecting alloy composition, 374 cooling conditions, 374 dissolved gases in the melt, 373–374 external pressure, 374 gases in metal, 372 overcoming, 73 overview, 370, 372 solidification, gas evolution during, 372–373 Gases in metals aluminum and its alloys, 67–70 cast iron, 64–67

copper and its alloys, 70–71 gas porosity, overcoming, 73 magnesium and its alloys, 71–73 overview, 64 Gibbs free energy activity and condensed phase equilibrium, 32–33 activity coefficient data, tabulations of, 33 activity coefficients—departure form ideality, 33 conditions for free equilibrium, 31–32 equilibrium constant, 32 thermodynamic data, tabulation of, 32 Gircast process, 1054, 1055 Gray iron castings alloying to modify as-cast properties, 851, 854 applications, 835 castability, 835–836 classes of, 835 dimensional stability, 849–850, 851 elevated-temperature properties, 849, 850 heat treatment hardenability, 852 localized hardening, 853 mechanical properties, 852–853 overview, 851–852 impact resistance, 846 machinability adhering sand, 846 annealing, 847–848 chill, 846 overview, 846 rating, 847, 848 shifted castings, 847 shrinks, 847 swells, 847 microstructure flake graphite, 836 gray iron, solidification of, 836, 837 overview, 836 room temperature structure, 837 overview, 835 physical properties coefficient of thermal expansion, 853 damping capacity, 854 density, 854 electrical properties, 854 overview, 853 thermal conductivity, 854 pressure tightness, 844–846 prevailing sections, 838–839 properties compressive strength, 841–842, 843 elasticity and deformation, 842, 843, 844, 845 hardness, 842–843 modulus of elasticity, 842, 845, 846 tensile strength, 841–842 torsional shear strength, 843 reversed bending, fatigue limit in, 843–844, 847 section sensitivity, 837–838 shakeout practice, effect of, 850–851 specifications, 840–841 test bar properties, 839–840 wear, 848 wear, resistance to, 848–849 Green sand molding clays for bentonites, 536–537 clay properties, controlling, 537–538 clay-water bonds, 537 fireclay, 537 overview, 535–536 first-generation machines jolt-squeeze molding machines, 555 jolt-type, 554–555 sand slinger molding machines, 555 green sand media preparation batch-type muller, 558–559 continuous muller, 558 intensive mixer, 559 mixing variables, 559–560 overview, 558 mold finishing core setting, 560–561 mold blowoff, 560

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Index / 1231 mold, transportation of, 561 overview, 560 mold types, flask/flaskless, 554 molding equipment, influence of, 552–553 molding methods, other related dry sand molding, 557–558 loam molding, 558 skin-dried molding, 557 molding problems expansion defects, 560 overview, 560 parting sprays, 560 pattern heating, 560 springback, 560 overview, 549 sand/casting recovery cooling devices, 562–563 high sand temperatures, 562–563 metal separation and screening, 564 overview, 562 sand molding metal contraction, 549–550 metal variables, 550–551 mold influences, 550 overview, 549, 550 patterns, 551 sand reclamation base sands, reclamation effects on, 565, 566 dry scrubbing/attrition, 564 multiple-hearth furnace/vertical shaft furnace, 565 overview, 564 thermal-calcining/thermal-dry scrubbing combinations, 564–565 wet washing/scrubbing, 564 sand system quality, maintaining overview, 553 sand system testing, 553–554 second-generation machines cope and drag, 556 horizontal flaskless molding machines, 556–557 match plate pattern, 556 overview, 556 pressure wave method, 556 rap-jolt, 556 vertically parted molding machines, 557 shakeout deck-type, 561 overview, 561 rotary plus cooling type, 561–562 rotary-type, 561 vibrating conveyors, 561 vibrating drum type shakeouts, 562 system control and formulation additives, 552 clays, 552 overview, 551–552 sands, 552 water, 552

H Halos, 310–311 Hazlett casting process, 1023–1024 Heat treatment aging artificial aging, 406 natural aging, 406 overview, 405–406 cast alloys, microstructural changes due to heat treatment of dissolution, 406–407 overview, 406 precipitation, 407 spheroidization, 407 designations of, 404 emerging heat treatment technologies, 406 overview, 404 quenching, 405 solution heat treatment, 404–405 High-alloy graphitic irons austenitic nickel-alloyed gray and ductile irons aluminum-alloyed irons, 906–907

applications, 906 austenitic ductile irons, 906–907 austenitic gray irons, 906 corrosion resistance, high-silicon irons for, 907–908 high-temperature service, high-silicon irons for high-silicon ductile irons, 904–906 high-silicon gray irons, 904 overview, 904 overview, 904 High-alloy heat treatment corrosion resistant applications austenitic-ferritic grades, 972–973 austenitic grades, 972 duplex alloys, 973 ferritic grades, 972 martensitic grades, 972 precipitation-hardening alloys, 973 ferritic and austenitic alloys, 971 heat-resistant applications iron-chromium alloys, 973 iron-chromium-nickel alloys, 973 iron-nickel-chromium alloys, 973 homogenization, 971 martensitic alloys, 971–972 overview, 971 precipitation-hardening alloys, 972 High-chromium white irons applications, 903 classes of, 899 desired structures, selecting compositions to obtain, 900 heat treatment annealing, 902 austenitization, 902 overview, 901–902 quenching, 902 subcritical heat treatment, 902 tempering, 902 machining, 902–903 melting practice, 900–901 microstructures as-cast austenite, 900 as-cast martensitic, 900 carbide, 899 heat treated martensitic, 900 molds, patterns, and casting design, 901 overview, 899 pouring practices, 901 shakeout practices, 901 special high-chromium iron alloys corrosion resistance, irons for, 903 high-temperature service, irons for, 903 High-pressure die casting advantages of, 716 alloys and processes cold chamber, 715 high-integrity, 715–716 hot chamber, 715 overview, 715 disadvantages of, 717 metal injection, 717–718 overview, 715 process, product design for, 717 High-pressure die casting (HPDC), automation in automatic pouring dosing feeding system, 748–749, 750 overview, 748 robotic ladling system, 748, 749 automatic spray overview, 754–755 spray head design, 755 spray nozzle technologies development, 755 spray strategy, 755 casting extraction, 753 die casting cycle high-pressure die casting automation, 748 overview, 747–748 die casting internal quality check automation automatic defect recognition system, 757 in-line x-ray radiographic inspection, 757 overview, 757

radiography, automated loading and unloading for, 757 finish automation automatic finishing integrates with vison system, 756–757 automatic flash removal, 756 overview, 755–756 gripper design, 754 injection process die casting shot process monitoring system, 749–750 in-line sorting and monitoring, 750 off-line automatic statistical process control analysis and monitoring, 750–751 overview, 749 insert loading, 753–754 insert loading and extraction, combination, 754 overview, 745–746 solidification and heat removal die internal heat-removal management, 751–752 overview, 751 temperature control zone, 753 Homogenization characterization methods, 402–403 computational modeling, 403 homogenization treatment, benefits of, 402 inhomogeneity, sources of, 402 overview, 402 Horizontal centrifugal casting applications bimetallic tubes, 679 chromium-molybdenum alloy steel tubes, 679 duplex stainless steel tubes, 679 high-strength low-alloy steels, 678–679 overview, 678 casting process babbitting, 676–678 casting temperatures, 675–676 mold temperature, 676 pouring, 675 rotation, speed of, 676 solidification, transport phenomena during, 676–677 equipment double-face plate machines, 675 flanged shaft machine, 674–675 molds and cores, production of, 675 overview, 674 roller-type machines, 674–675 overview, 674 Hot chamber die casting ejection and postprocessing fluxing, 723 recycling, 722–723 robotics, 722 gate and runner design, 721 hot and cold chamber processes, distinctions between, 720–721 injection components casting machines, 720 injection cycle, 720 overview, 719–720 pressure differential (DP) and mold filling, 720 melting process, 719 overview, 719 temperature control, 721–722 Hot isostatic pressing (HIP) as-HIPed properties, effect of inclusions on, 412–413 castings, effect of HIP on shape and structure of microstructural/physical changes, other, 413–414 volumetric shrinkage and distortion, 413 economics of, 415 HIP system, overview of conditions, selecting, 410, 411 HIP cycle, 409–410 introduction, 409, 410 pore closure, mechanisms of, 411 history of, 408–409 mechanical properties, effect on, 411–412, 413 overview, 408 problems encountered in surface-connected porosity, 414, 415

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

1232 / Reference Information Hot isostatic pressing (HIP) (continued) thermally induced porosity, 414–415 reasons for using, 408–409 Hot tearing alloy dependence, 376–377 criteria, 377 modeling of, 377 overview, 376 tests, 377 Hypereutectic aluminum-silicon alloys, refinement of the primary silicon phase in overview, 263 primary silicon refinement, accomplishing eutectic modification, compatibility issues, 265 loss of refinement over time, 265 overview, 263–264 phosphorus, sources of, 264 solidification rate, effect of, 264–265 primary silicon refinement, importance of, 263 refinement versus modification, 263

I Inclusion-forming reactions copper alloys, 82 exogenous inclusions, 74 ferrous alloys, inclusions in, 77–80 inclusions, control of chemistry control, 76 gating methods, 76–77 inclusion shape control, 77, 78 mold/metal interface and refractory control, 76 separation techniques, 76 inclusions, defined, 74 indigenous inclusions, 74–75 magnesium alloys, 82 nonferrous alloys, 81–82 overview, 74 physical chemistry kinetics, 75–76 phase equilibria, 75, 80–81 thermochemistry, 75 Induction furnaces electromagnetic stirring, 109, 110 lining material alumina, 111 crucible wall scrapers, 111–112 lining push-out device, 113 magnesia, 111 overview, 111, 112 ramming mixes, 112 melting operations batch, 113 mechanical skimmers, 114, 115 overview, 113 tap-and-charge, 113 utilization, 113–115 overview, 108 power supplies 60Hz line frequency, 109–110 medium frequency, 110, 111 parallel inverters, 110 series inverters, 110 systems, packaging of, 110–111 types of channel furnace, 108 coreless, 109 overview, 108, 109 water cooling systems, 111, 112 Investment casting ceramic shell mold assembly. See Ceramic shell mold assembly cluster design, 650–651 overview, 646, 647 pattern assembly, 650 pattern materials additives, 647 fillers, 647–648 other, 648 overview, 647 plastics, 648 wax selection, 648

waxes, 646–647 pattern tooling for cast tooling, 650 machined tooling, 649–650 overview, 649 positive model, tooling made against, 650 patternmaking overview, 648–649 patterns, machining of, 649 plastic patterns, injection of, 649 wax patterns, injection of, 649 Iron, arc melting of, 98 Iron-base alloys, thermodynamic properties of ferrous metals, purification of deoxidation, 43–45 desulfurization, requirements of, 41–43 overview, 41 ferrous systems, thermodynamics of iron-carbon system, 45–46 iron-silicon system, 46 ternary Fe-C-X systems, 47–50 ternary iron-base alloys, 46–47 multicomponent iron-carbon systems carbon solubility in, 50 saturation degree and carbon equivalent, 50–51 overview, 41 structural diagrams, 51–55

L Ladle metallurgy injection methods, 217–218 ladle degassing overview, 219 recirculating degassing methods, 220 stream degassing methods, 219–220 vacuum ladle degassing processes, 220 ladle furnace and vacuum arc degassing (LF/VD-VAD), 222 ladle furnace design, 223, 225, 226 overview, 222–223, 224 vacuum induction degassing (VID), 224, 228–229 metallurgical objectives alloying, 214 argon stirring (argon rinsing), 216–217 CAB process, 217 CAS process, 217 decarburization, 215 degassing, 213–214 deoxidizing, 214 desulfurization, 214–215 overview, 213 phosphorus control, 216 sulfur control, 215–216 overview, 212–213, 215, 216 vacuum ladle degassing (VD), 220–222 overview, 224 stirring procedures, 221 vacuum oxygen decarburization (VODC), 221–222 vacuum treatments, 212 Launders, 174 Light alloy melts, dross, melt loss, and fluxing of dross, economic implications of, 998 dross formation, 992–993 dross formation, ways to reduce, 1000 dross generation, influence of melter type on, 993–994 dross metallics, in-plant enhancement or recovery of Dross Boss (Q.C. Designs, Inc.), 999–1000 dross buggy, 999 dross coolers, 998 dross presses, 998–999 overview, 998 melt loss, influence of charge materials on, 994 melt loss, influence of operating practices on furnace fluxing, 995–997 metal transfer, drosses generated in, 997–998 overview, 994–995 overview, 992

Lost foam casting advantages of, 645 casting quality dimensional tolerances, 645 folds, 645 overview, 640 process technique, 640–642 processing parameters coating, application, 643–644 coating types, 643, 645 foam pattern, 642, 643 investing the foam pattern in sand, 644 pattern assembly, 642–643, 644 pattern molding, 642, 643 pouring, 644–645 sand system, 644 Low-pressure countergravity casting mold filling countergravity centrifugal casting (C3) process, 711 countergravity low-pressure air (CLA) process, 709–710 countergravity low-pressure inert atmosphere (CLI) process, 710 countergravity low-pressure vacuum (CLV) process, 711 countergravity pressure vacuum (CPV) process, 710, 711 loose sand vacuum (LSVAC) process, 710–711 supported shell (mold) process, 710, 711 overview, 709 Low-pressure metal casting conventional low-pressure casting, 700–704 cores, 703–704 counterpressure casting advantages of, 705–706 counterpressure casting cycle steps, 704–705 history of, 704 overview, 704 PCP operations, special considerations for, 706 pressure differential (DP) and mold filling, 704 low-pressure furnace, 702–703 low-pressure tooling, 703 overview, 700 vacuum riserless/pressure riserless casting, 706–707

M Macrosegregation flow of the liquid, induced by the buoyancy-driven flow, 350, 351 channel segregation, 350–351, 352 inverse segregation, 349 nonuniform shrinkage flow, 349–350 overview, 348–349 movement of the solid, induced by, 351–352 overview, 348 Macrosegregation, continuum model of continuity equation, 446 energy, conservation of, 447 example simulation of, 447, 448 momentum equation, 446 pressure, calculation of, 447 solute conservation equation, 446–447 Magnesium and its alloys, gases in magnesium alloys, degassing, 72–73 overview, 71 porosity in, 72 solidification, hydrogen solubility and reactions during, 71–72 Magnesium and magnesium alloys die casting casting defects, 1111–1112 casting finish, 1111 design parameters, 1111 die temperature, 1110 dies, 1110 fluxless melting parameters, 1111 lubrication, 1110 machines, 1110 melt composition, control of, 1110 overview, 1109–1110

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Index / 1233 recycling, 1110–1111 tolerances, 1111 direct chill casting, 1112 general applications, 1102 high-temperature alloys, 1112 investment casting, 1109 magnesium alloys die casting, 1101 overview, 1100, 1101 permanent mold casting, 1101 sand casting, 1100–1101 thixomolding, 1101–1102 melting furnaces and auxiliary pouring equipment, 1102–1103 melting procedures and process parameters alloying, 1104–1105 flux process, 1103 fluxless process, 1103–1104 foundry control procedures, 1105 gas content and grain size, 1105 grain refinement, 1104 melt treatments, 1105 pouring, 1105 pouring methods, 1105 pressure sand casting, 1105–1106 sand casting, 1106 overview, 1100 permanent mold casting, 1109 sand casting applications, current, 1109 chills, 1108 coremaking procedures, 1107–1108 gating practices, 1108 green sand molding, advantages/disadvantages of, 1106–1107 molding sands, 1106 overview, 1106 risers, 1108 shakeout, 1108 shrinkage allowances, 1108 welding, 1112 Malleable iron castings applications, 888, 889, 890 composition, 884 ferritic malleable iron, 889–890, 891, 892, 893 martensitic malleable iron, 890–892, 893, 894, 895 melting practices, 884–885 microstructure annealing, control of, 886–887 nodule count, control of, 885–886 overview, 885 overview, 884 pearlitic malleable iron, 890–892, 893, 894, 895 production technologies, current digital solidification analysis technology, 887 production costs, effect on, 888 simulation technologies, 887, 888 smart engineering, 888 Mechanical skimmers, 114, 115 MEPHISTO advanced gradient heating facility (AGHF), 399 overview, 399 Metal casting history of, 4–6 20th century, 6–8 overview, 3–4 methods, 8–9 trends market, 14 production trends, 9–10 production trends (1995–2005), 10–14 Metal-matrix composites (MMCs) nanocomposites, 395 overview, 389–390 particle-pushing phenomenon, modeling of dendrite and particle interaction, 394 nanoparticles and solidification front interactions, 394–395 overview, 394 reinforcement-metal wettability, 391 reinforcements, incorporation of, 391 solidification fundamentals

dendrite formation and microsegregation, 392–393, 394 nucleation and growth effects, 392 reinforcements, distribution of, 393–394, 395 solidification processing of infiltration process, 392 stir mixing, 391–392 Metal quality, approaches to measurement of metal cleanliness assessment techniques chemical analysis, 1167–1168 closed-loop circulation, 1171, 1172 electric resistivity tests, 1170 pressure filter tests, 1168–1169, 1170 reduced-pressure test (RPT), 1171–1172 ultrasonic technique, 1173 overview, 1167 Metalcasting technology and purchasing process contract fulfillment and continuous improvement, 22 contract terms, negotiate, 22, 23 define requirements, 17 metal castings, considerations in, 22, 24–28 additional operations, 26 cast surface finish, 25–26 dimensional allowances, 24–25 geometric complexity, 24 lead time and first article delivery, 26 metal type, 24 order quantity, 26 price quotations, types of, 26 production cost factors, 24 quality assurance and specifications, 26–27 quality-management systems, 28 specifications, types of, 27 standard specifications, 27 technical services, 26 tolerances, 25 overview, 16 purchasing plan, 17 deliverables expected, 19 purchasing objective, 17 purchasing strategy, 17, 19 purchasing tasks, responsibilities, and schedule, 20–21 selection criteria, 19–20 working relationships requirements, 19 purchasing process, 16 quotation, request, 16 requesting and evaluating quotations/bids bid package, send out, 22 good relationships, maintaining, 21–22 grade and rank bids/quotations, 22 locate/identify suitable facilities, 21 quotations, review, 22 supplier, select, 22 terms and conditions for sale in metalcasting industry, 22 Microgravity, solidification research in key solidification experiments MEPHISTO, 399 TEMPUS, 399–400 microgravity solidification, future of, 400 overview, 398 reduced gravity, availability of: past, present, future, 398–399 solidification research, 399 Microsegregation binary eutectic systems: hypoeutectic aluminumsilicon alloys, 344 binary isomorphous systems, k > 1: hypoeutectic aluminum-copper alloys as-cast microsegregation, measurement of, 342, 343 dendrite coarsening, 341 density and solidification shrinkage, influence of microsegregation and partitioning on, 341 kinetic processes leading to microsegregation, observation and measurement of, 342–343, 344 local cooling rate, influence of, 341 microsegregation and macrosegregation, interaction of, 341–342 nonequilibrium solidification with back diffusion—directional solidification, 340–341

overview, 340 solidification curve, 340 solute distribution curve, 340 thermal analysis, 344 binary isomorphous systems, k > 1: titaniummolybdenum, 339–340 binary peritectic systems: copper-zinc a-brass, 344 multicomponent, multiphase, eutectic systems: hypoeutectic Al-Si-Cu-Mg alloys, 345 multicomponent, multiphase, nickel-base superalloy, 345–346 multicomponent multiphase steel: Fe-C-Cr, 345 overview, 338 solidification models equilibrium solidification, 338 nonequilibrium solidification, the GulliverScheil relation, 339 nonequilibrium solidification with back diffusion, 339 overview, 338 partition ratio, 338 Microsegregation, modeling of with diffusion in the solid example simulation of, 445–446 overview, 445 microsegregation, 445 Modeling, stress, distortion, and hot tearing example applications continuous casting of steel slabs, 455, 456, 457 sand casting of braking disks, 454, 455 governing equations compatibility, 449 equilibrium, 449 implementation issues, 450–451 liquid-state constitutive models, 449–450 mushy-state constitutive models, 450 overview, 449 solid-state constitutive equations, example of, 451 solid-state constitutive models, 450 hot tearing analysis case study: billet casting of speed optimization, 457–458, 459 classical mechanics criteria, 457 mechanically based criteria, 457 overview, 456 thermal analysis criteria, 456–457 model validation, 453–454 numerical solution boundary conditions: modeling of contact conditions. multidomain approaches, 452–453 finite element formulation and numerical implementation, 452 mushy and liquid regions: ALE modeling, 453 overview, 451–452 solidified regions: Lagrangian formulation, 453 thermomechanical coupling, 453 overview, 449 thermomechanical coupling air gap formation: conductive-radiative modeling, 451 effective contact: heat transfer as a function of contact pressure, 451 overview, 451, 452 Molten-metal filtration filter characteristics, 197–199 filtered metal quality, 204–205 filters, types of bonded-particle, 199 cartridge, 199 ceramic foam, 199 fiberglass, 199 metal, 199 in-mold filtration direct pouring, 200–201 examples, 201 mold casting, filter selection and application in, 200 overview, 199 sand gating, filter selection and application in, 199–200 inclusions, removal of overview, 196 positive filtration, 196–197

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

1234 / Reference Information Molten-metal filtration (continued) inclusions, sources of aluminum inclusions, 194–195 magnesium alloys, inclusions in, 195–196 overview, 194 molten aluminum, in-furnace filtration of filter life, determining factors, 203–204 in-furnace filter installation, 203 in-furnace filter types, 201–203 overview, 201 overview, 194 Molten metal, transfer and treatment of—an introduction dosing furnaces dosing operation, controlling/monitoring, 183, 184 furnace recharging, 183–184 melt holding, 184 melt level, detection of, 183 overview, 182–183 electromagnetic pumps for furnace circulation and scrap recycling technology, 177 mold-filling, 177 overview, 177 types, 177–178 handling systems, 173 launders, 174 molten-metal pumps mechanical, 175–177 overview, 175 molten-steel filters, 175 pouring, 178–179 pouring methods, 182 automatic, 180–181 pouring, controls of, 181–182 pouring ladles, 179–180 treatments, 173 tundishes, 174–175

N Nickel and nickel alloy castings applications, specific, 1126 compositions cast nickel, 1119–1120 cast nickel-chromium-molybdenum alloys, 1120 cast nickel-copper alloys, 1120 cast nickel-molybdenum alloys, 1120 directionally solidified and single-crystal alloys, 1121 heat-resistant alloys, 1120–1121 nickel-base proprietary alloys, 1120 nickel-chromium-iron alloys, 1120 foundry practice, 1124 gating systems, 1124–1125 heat treatment, 1125–1126 melting practice and metal treatment argon-oxygen decarburization (AOD), 1124 coreless induction furnaces, 1123–1124 ladle furnaces, 1124 vacuum melting, 1124 overview, 1119, 1120 pouring practice, 1124 risers, 1125 structure and property correlations cast nickel, 1121 directional and single-crystal solidification, alloys for, 1123 heat-resistant alloys, 1122–1123 nickel-chromium-iron alloys, 1121, 1122 nickel-chromium-molybdenum alloys, 1121, 1122 nickel-copper alloys, 1121, 1122 nickel-molybdenum alloys, 1121–1122 welding, 1125 Nickel-chromium white irons applications, 899 composition control, 897–898 heat treatment, 898–899 machining, 899 melting practice, 897

molds, patterns, and casting design, 898 overview, 896 pouring practices, 898 shakeout practices, 898 No-bake sand molding air-setting organic binders alkyd oil, 574–575 furan nobake, 574 overview, 573–574 phenolic nobake, 574 phenolic urethanes, 575–576 air-setting sodium silicates dicalcium silicates (slag/cement hardeners), 572–573 ferro-silicon (Nishiyama process), 573 overview, 572 silicate organic esters, 573 CO2-cured sodium silicate applications, 569–570 overview, 569 procedures, 572 sand-silicate mixtures, 570–572 gas-cured organic binders acrylic-epoxy-SO2 system, 577 overview, 576 phenolic urethane-amine system, 576–577 no-bake sand processing mixers, 568 molds and cores, sand processing for, 568 no-bake sands, foundry control tests of, 568–569 refractory coatings for chemically bonded sands, 567–568 overview, 567 sand reclamation and reuse dry reclamation systems, 578–579 overview, 577–578 thermal reclamation, 579–580 wet reclamation systems, 578 No-bond sand molding magnetic molding, 628 overview, 628 vacuum (V-process) molding advantages, 629–630 overview, 628–629 V-process description, 629–630, 631, 632, 633 Non-plane front solidification cellular microstructures high-velocity cells, 301 microsegregation, 301 overview, 300 scaling behavior of, 300 columnar and equiaxed microstructures, 299 columnar microstructures, 299 dendrite growth. See Dendrite growth equiaxed microstructures, 299 microstructures, characteristic lengths of, 300 nonplanar eutectic growth, 304–305 overview, 299 processing conditions, microstructure and, 299–300 Nondestructive testing of components automated defect recognition (ADR) program, 1174 eddy current inspection, 1175 electrical conductivity measurement, 1177 fluoroscopic inspection, 1174, 1175 leak test, 1177 liquid penetrant inspection, 1174 overview, 1174 process-controlled resonant testing (PRCT), 1175–1177 radiographic inspection, 1174 ultrasonic inspection, 1174–1175 Nonferrous alloys, inclusions in aluminum alloys other inclusion types, 82 oxides, 81 Nonferrous casting—an introduction lead, 991 nickel alloys, 989 overview, 989 titanium alloys, 989–990 Nonplanar eutectic growth, 304–305 Nucleation kinetics and grain refinement grain refinement models

carbide-boride model, 281–282 constitutional undercooling model, 283–284, 285 free-growth model, 282–283 overview, 281 peritectic reaction theory, 282 inoculation practice, 280–281 master alloy processing overview, 285 thermal analysis techniques, 285–286 nucleation phenomena heterogeneous nucleation, 279, 280 homogeneous nucleation, 278–279 overview, 278 rapid solidification, 279–280 overview, 276 solidification, thermodynamics of macroscopic solids, 276–277 microscopic solids, 277–278

O Ohno continuous casting process, 1024 Outokumpu upcasting method, 1022 Oxide film defect. See Bifilm, the

P Patterns and patternmaking core boxes, 489–490 design and manufacture, trends in direct CNC manufacture, 495, 496 overview, 495 pattern and casting dimensions, verification of, 495 expendable patterns and dies, tooling for, 490 materials machined plastic patterns, 493 metal pattern equipment, 492–493 overview, 492 pattern coatings, 493–494 pattern handling and storage, 494 pattern repair and rebuilding, 494 plastic pattern equipment, 493 solid replication plastic patterns, 493 wood, 492 overview, 488 pattern allowances distortion allowances, 492 draft allowance, 491 machining allowance, 491–492 shrinking allowance, 491 pattern type, selection of overview, 494, 495 tolerances, 494–495 patterns, additional features of core prints, 491 gates and risers (rigging), 490–491 locating points, 491 parting line, 490 patterns, types cope and drag, 489 core boxes, 489–490 loose, 488 of match plate, 488–489 overview, 488 special, 489 Patterns and tooling alterations, 1214–1215 common failure mechanisms, 1213 maintenance interval, 1210–1214 maintenance of, 1213 maintenance records, 1214 maintenance responsibility, 1213 ownership, 1213 preventive maintenance program, 1214 repair, 1214 storage—physical protection, cataloging, and tracking, 1215 Peritectic solidification iron-carbon system, 333, 334, 335 multicomponent systems, peritectic transformation in, 335–336

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Index / 1235 overview, 330 peritectic cascades, 334, 335 peritectic reaction, 331–332 peritectic solidification, 330–331 peritectic transformation and direct precipitation, 332–333 phase diagrams, 330 primary metastable precipitation of beta, 334, 335 Permanent mold casting advantages and disadvantages, 690–691 applications, 691 casting design, 691–692 dimensional accuracy, 698–699 gravity casting methods automatic tilt pouring, 690, 691 semiautomatic devices, 690 mold coatings application, 695 coating life, 696 families of, 695 release, 696–697 surface preparation, 695–696 surface quality, 696 mold design mold wall thickness, 692 multiple-cavity molds, 692 overview, 692 venting, 692–693 mold life, 697 mold materials casting ejection, 694 collapsible cores, 694 cored holes, 693–694 cores, 693 design tools, 694–695 directional solidification, 695 filters, 694, 695 gating and riser system, 694 overview, 693 sand or plaster cores, 693 mold temperature, 697–698 mold temperature, control of, 698 overview, 689 surface finish, 699 Phase diagrams and thermodynamics overview, 269 phase diagrams metastable phase diagrams, 271–272 overview, 271 phase diagrams and solidification, calculation of, 272–273 solidification, 272–273 thermodynamics, 269–271 Plane front solidification morphological stability constitutional undercooling, 295 perturbation analysis, 295–296 overview, 293 rapid solidification effects absolute stability, 296 oscillatory instability, 297 solute trapping, 296–297 science, 293–294 solidification microstructure selection maps, 297, 298 technology, 293 transients and steady state final transient, 294–295 initial transient, 294 overview, 294 steady state, 294 Planning tools and procedures Advanced Quality Product Planning (APQP) control planning, 1210 design failure mode and effects analysis (DFMEA), 1208 failure mode and effects analysis (FMEA), 1208 lean tools, 1211 measurement system analysis (MSA), 1210–1211 overview, 1207, 1208 process failure mode and effects analysis (PFMEA), 1208–1210

production viability, 1211 Six Sigma, 1211–1212 statistical process control (SPC), 1211 overview, 1207 Plasma heating. See Plasma melting and heating Plasma melting and heating atmosphere control, 151 furnace equipment, 151 melting processes overview, 151, 152 plasma arc remelting, 152, 153 plasma cold crucible casting, 153–154 plasma cold hearth melting, 152–153, 154 plasma consolidation, 151, 152, 153 overview, 149 plasma torches overview, 149 transferred arc plasma torches, 149–150, 151 Plasma torches, 149–150, 151 Primary silicon phase in hypereutectic aluminum-silicon alloys, refinement of the overview, 263 primary silicon refinement, accomplishing eutectic modification, compatibility issues, 265 loss of refinement over time, 265 overview, 263–264 phosphorus, sources of, 264 solidification rate, effect of, 264–265 primary silicon refinement, importance of, 263 refinement versus modification, 263 Properzi wheel casting technology, 1024 Purchasing process (metalcasting technology). See Metalcasting technology and purchasing process

R Rapid solidification diffusional kinetics/microsegregation cellular growth, 389 dendritic growth, 389 eutectic growth, 389 morphological stability, 388–389 heat flow external heat sinks, 387 internal heat sinks, 387 overview, 386–387 microstructural modification microsegregation, 386 new primary/secondary phase, 386 noncrystalline phases, formation of, 386 primary phase, change in, 386 overview, 386 thermodynamic constraints/liquid-solid interface conditions full diffusional (global) equilibrium, 387 interfacial nonequilibrium, 388 local interfacial equilibrium, 387 metastable local interfacial equilibrium, 387–388 overview, 387 Rautomead upwards vertical casting, 1022 Repair welding of castings ferrous castings cast stainless steels, 1203 cast steels, 1203 casting defects, removing, 1204 deposition rates, 1205 gray and ductile cast iron, 1203 heating/cooling rates, control of, 1204–1205 joint selection, 1204 postweld heat treatment (PWHT), 1205 surface preparation, 1204 weld metal properties, 1204 weld repair considerations and techniques, 1203–1204 weld repair processes, 1203 weld repair quality, verification of, 1205 welding processes, other, 1203 nonferrous castings, repair welding of filler metal selection, 1205 overview, 1205 postweld heat treatment (PWHT), 1206

weld repair considerations, 1205–1206 weld repair process selection, 1205 weld repair quality, verification of, 1206 overview, 1203 Replicast molding. See Replicast process Replicast process overview, 662 process capabilities dimensional accuracy, 663–664 overview, 663 process details ceramic coating, 663 cleaning, 663 firing, 663 overview, 662, 663 pattern assembly, 663 pattern production, 662–663, 664 pouring, 663 Reverberatory and stack furnaces furnace types dry hearth reverberatory furnace, 161–162 overview, 161 stack melters, 162–163 wet bath reverberatory furnaces, 162 molten metal circulation forced circulation, benefits of, 165–167 molten aluminum circulation methods, 167–168 overview, 165 overview, 160–161 reverberatory furnace practice fixed-heat loses, reducing, 164 furnace efficiencies, improving and maintaining, 164–165 radiant heating factors, 163 recuperation, 163–164 regenerative burners, 164 Rheocasting methods continuous rheoconversion process (CRP), 774, 775, 776 new rheocasting (NRC) process, 773, 774 rheo-diecasting process, 773–774, 775 semisolid rheocasting (SSR), 773 slurry-on-demand process, 773–774, 775 subliquidus casting (SLC), 773, 774 swirled enthalpy equilibrium device (SEED), 773–774, 775 overview, 773 Rotary impeller degassing, 187–188

S Semisolid casting—introduction and fundamentals overview, 761, 762 SSM processing, advantages of, 762–763 SSM processing routes rheocasting (slurry), 762–763 thixocasting (billet feedstock), 762 thixomolding (magnesium pellets), 763 Semisolid metal (SSM) processing, 762 flow behavior of rheology of, 379–380 semisolid slurries, yield stress of, 380 viscosity evolution during continuous cooling, 379 overview, 379 semisolid microstructural formation copious nucleation, globular structure via, 381 forced convection, under, 380–381 semisolid processes: thixocasting versus rheocasting, 381 Shape casting of steel. See Steel, shape casting of Shape casting processes—an introduction, 485–487 Shaw process all-ceramic mold casting, 623, 624, 625 ceramic facing, preparing, 623 composite ceramic mold, 623–624 overview, 622 patterns, 622–623 Shell molding and shell coremaking applicability, 599 casting process

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

1236 / Reference Information Shell molding and shell coremaking (continued) metal-mold reactions, 610–611 overview, 610 shakeout, 611 vertical versus horizontal parting, 610 dimensional accuracy chill, 614–615 hot tears, 614 inaccuracy and variation, causes of, 612 increasing, methods of, 612–613 net or near-net dimensional accuracy with shell molding, examples of, 613–614 overview, 611–612 surface defects, 615 green sand molding, comparisons with examples, 615, 616 pattern costs, 615–616 mold defects, causes and prevention of low hot tensile strength, 609 mold backup, 610 mold cracking, 608–609 mold shift, 610 of peelback, 609–610 soft molds, 609 mold spread at parting line, methods to minimize example: bolting of mold halves, 608, 609 example: use of flush-head ejection pins, 608 molds and cores, production of cores, 607 curing, 606–607, 608 dump-box molding, 605, 606 gating, 607, 608 mold blowing, 605–606, 607 overview, 604 patterns, heating of, 604–605 shell thickness, 606, 607 overview, 598 patterns and core boxes overview, 603–604 pattern construction, 604 split core boxes, 604 resin-sand mixtures, preparation of coating and handling, equipment for, 602–603 cold coating, 601–602 hot coating, 602 overview, 601, 602 resin-sand properties, control testing of, 603–604 resins additives, 601 alcohol content, 600 hexamethylenetetramine (hexa), 601 lubricants, 601 melting point, 600–601 overview, 600 phenol-formaldehyde resins, 600, 601 solids content, 600 viscosity, 600 water content, 600 sand reclamation, 603 shell sands, properties and selection of breakage of molds, 600 grain size and shape, 599–600 moisture content, 600 overview, 599 Shellvest system, 658–660 Shrinkage porosity feeding, 371–372 overview, 370 shrinkage, 370–371 Skull melting furnaces, 139–140 induction skull melting, 140–141 overview, 139 vacuum arc skull melting, 139 Slurry molding Antioch process, the, 620–621 foamed plaster molding process, 621–622 match plate pattern molding, 620 mold-drying equipment, 618 overview, 617 patterns and core boxes, 618 plaster mold casting, conventional, 619–620 plaster mold compositions

calcium sulfate, characteristics of, 617–618 overview, 617 plaster molding advantages, 617 applications, 617 plaster mold casting, 617 plaster molds, metals cast in, 618–619 Solid fluxes aluminum oxides, solvents of, 233 chlorides, 232 fluorides, 232 overview, 231–232 oxidizing compounds, 232 Solidification, transport phenomena during balance equations energy balance, 289 mass balance, 288 momentum balance, 288–289 overview, 288 solute balance, 289 fundamentals, 288 scaling, 289–290 transport and microstructure homogenization, 292 microsegregation, 291–292 overview, 290 planar front growth, 290–291 Spray casting atomization and the droplet spray, 383, 384 deposition and grain multiplication, 384–385 macrosegregation and coarsening, 385 microsegregation, 385 overview, 382 principles, 382–383 Squeeze casting alloys, properties, and microstructures, commonly used, 730–731 applications, 729–730 casting and tooling design for, 727, 728 squeeze-cast applications, 729 squeeze-cast products, factors affecting casting (cavity) pressure, 728 gate position, area, and velocity, 728–729 heat treatment, 729 lubricant, type of, 728 metal cleanliness, 728 metal temperature, 728 porosity, 729 presolidified layer (cold flakes) from the shot sleeve, 729 squeeze casting, 727 Stack furnaces. See Reverberatory and stack furnaces Steel casting properties carbon and low-alloy heat treatment annealing, 970 hydrogen removal, 971 normalizing, 970–971 overview, 970 quenching, 970–971 stress relief, 971 tempering, 971 carbon and low-alloy mechanical properties corrosion-resisting steels, mechanical properties of, 961, 965–966, 967 discontinuity effects, 959 elevated-temperature applications, 959, 960 hardenability, 953–954, 956 overview, 953, 956 section size and mass effects, 959, 961, 962–963 strength and ductility, 954, 956, 957 strength and fatigue, 956–959, 960, 961 strength and hardness, 954, 956 strength and toughness, 954–956 wear-resisting steels, mechanical properties of, 960–961 carbon steels, 949–950 common alloys, 953, 954, 955 corrosion-resistant steels, 951–952 ferrite in cast stainless steels ferrite control, 969, 970 ferrite control, limits of, 969–970

overview, 968 significance of, 968–969 heat-resistant steels, 952–953 heat resisting steels, mechanical properties of creep and stress-rupture properties, 966–968 elevated-temperature tensile properties, 966, 967, 968 hot gas corrosion, resistance to, 968, 969 thermal fatigue resistance, 968 thermal shock resistance, 968 low-alloy steels, 950 overview, 949 wear-resistant steels, 950–951 Steel castings, selection and evaluation of design alloy designation, 980 alloy selection, 977–980 considerations, 976–977, 978 design impact, 975 design rules, general, 975–976 drawing notes, 980 material effects, 976 overview, 975 manufacturing directional solidification, 981 draft, 980 heat treatment, 981 inclusion shape control, 980–981 minimum section thickness, 980 postprocessing, 982 pouring, 980–981 steel casting process, 981 weldability, 981–982 nondestructive examination (NDE) general rules, 985 hardness testing, 984 leak testing, 984 magnetic particle inspection, 983 overview, 982 quality, 982 radiographic inspection, 983–984 testing, 982–983 tests and castings, 984–985 ultrasonic inspection, 984 visual inspection, 983 overview, 975 purchasing, 986 research and development, 986 Steel continuous casting advantages of, 918 future developments, 924–925 historical aspects of, 918 horizontal continuous casting, 924, 925 molds for, 922–923 near-net shape casting, 919, 920, 921 plant layout, 919 pouring stream protection, 921–922 process, general description of, 918–919, 920 productivity improvements, 923–924 quality improvements, 924 tundish metallurgy, 920–921 Steel ESR overview, 130 oxygen, behavior of, 130 sulfur, behavior of, 130 Steel ingot casting hot tops, 913 ingot design, 912–913 ingot molds and stools, 913–914 ingot solidification, 914–917 overview, 911 pouring method, 911–912 Steel melt processing converter metallurgy argon oxygen decarburization, 208–212 overview, 208 vacuum oxygen decarburization (VODC), 212, 213, 214 coreless induction furnaces, 208 direct arc melting, 207 induction arc melting, 207–208 ladle metallurgy. See Ladle metallurgy overview, 206

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

Index / 1237 refractories, 206–207 vacuum induction melting VIM, 208 Steel, shape casting of castability of fluidity, 926–927 freezing range, 929 overview, 926 shrinkage, 927–928, 929 cores, size of, 947–948 design factors machining allowances, 938 minimum thickness as a function of dimensions, 938 shape, considerations of, 936 thickness, of walls, webs, and ribs, 937–938 wall thickness, selection of, 936–937 foundry factors in design gating design, 934–935 massive parts, positioning, 935 molding direction, choice of, 936 pouring direction, 935 pouring direction, choice of, 936 riser design, 934–935 surfaces to be machines, 935–936 graphite molds, casting in, 948 hot spots, avoidance of closed sections, 942 dimensions, accuracy of, 942–943 holes, dimensioning of, 942 junction of five or more walls, 941 junction of four walls, 940–941 junction of three walls (T-piece), 940 junction of two walls (right-angle intersection), 940 machining operations, continuity of, 943, 944 machining, setting up for, 943 overview, 938–940 ribs and recessing, strengthening, 941 sand, hot spots in, 941–942 tool clearances, 943 melting furnaces direct arc melting, 928–929 induction melting, 929–930 overview, 928 melting practice argon-oxygen decarburization (AD), 933 boiling, 930, 931 deoxidation practice, 932–933 elements from slag, recovery, 931–932, 933 metallic charge material, selection of, 930 overview, 930 phosphorus, removal of, 930, 931 plasma refining, 933 powder injection/wire injection, 933 refining reactions, 930 sulfur, removal of, 930, 932 temperature control, 932 overview, 926 pouring practice ladles, 934 overview, 933–934 thin-wall steel castings alloy selection, 944, 945 distortion, 946 friction, 944 good design, examples of, 946, 947 mold filling, 944, 945 mold-filling problems, 945–946 mold gases, 944 overview, 943–944 soundness, 946 tool clearances, 944 tolerances investment mold casting process, 946–947, 948 shell mold casting, 946 Steelmaking, basic the charge, 95 oxidizing period, 95–96 reducing/refining the heat, 96–97 Structure formation, direct modeling of cellular automaton method, direct grain structure simulation using growth, 440–441

indices, 439 nucleation, 439–440 overview, 398–439, 440 direct structure simulation at macroscopic scale, coupling of, 441–443 introduction, 435–436 phase field method, direct microstructure simulation using hypereutectic AlCuSiMg alloy, equiaxed solidification of, 437–438 multicomponent alloy thermodynamics, 437 nucleation, 437 overview, 436–437 steel, microsegregation in, 438, 439

T TEMPUS advanced gradient heating facility (AGHF), 400 isothermal dendritic growth experiment (IDGE), 400 overview, 399 particle engulfment and pushing by solidifying interfaces, 399–400 solid-liquid mixtures, coarsening in, 400 solidification and pore formation, 400 Ternary Fe-C-X systems carbon solubility in, influence of a third element on, 50 equilibrium temperatures, influence of third element on, 48–49 Fe-C-Mn system, 47–48 Fe-C-P system, 47 Fe-C-Si system, 47 overview, 47 third element phases in, partition of, 49–50 Thermal analysis differential differential thermal analysis, 355 simplified thermal analysis, 354–355 equilibrium, deviation from, 354 liquidus and solidus temperature, determining, 355–356 overview, 353 quantitative, 353–354 Thermodynamics chemical thermodynamics, 31 enthalpy and heat capacity, 31 Gibbs free energy, 31–34 overview, 31 phase diagrams, 33–34 Thermodynamics and phase diagrams overview, 269 phase diagrams metastable phase diagrams, 271–272 overview, 271 phase diagrams and solidification, calculation of, 272–273 solidification, 272–273 thermodynamics, 269–271 Thermophysical properties cast alloys, thermophysical property ranges of, 477, 478, 479 density, 471, 473 electrical and thermal conductivity, 474, 476–477 emissivity, 477 melting, solidus, and liquidus temperatures, enthalpy of, 470–471, 477 methods to determine, 468–469 overview, 468, 469 reliable data, sources and availability of, 468 specific heat capacity and enthalpy of transformation, 470, 472 surface tension, 471, 473 thermal expansion, coefficient of, 471, 473, 478, 479 use of data, limitations and warnings on, 468 viscosity, 474 Thixocasting bar-making processes grain-refined billet, 765–766 mechanical stirring, 765

MHD-stirred billet, 766 overview, 764–765 strain-induced melt activation (SIMA), 765 billet sawing and volume control, 766–767 casting machine requirements, 771 casting process and parameters, 770–771 overview, 764 reheating and billet handling, 767–768 rheological tests, 698–769 semisolid tooling design, basics of, 769–770 Thixomolding emerging developments alloy design by blending and thixoblending, 778–779, 780 hot runners, 779 hot sprues, 779 overview, 778–779 scrap recycling, effect of, on mechanical properties, 779–780 mechanical properties and microstructure, 780–781 overview, 777 process description, 777–778, 779 Titanium and titanium alloy castings alloy Ti-6Al-4V as-cast and cast + HIP microstructures, 1133 cast microstructure, 1132–1133 hot isostatic pressing (HIP), 1133 mechanical properties of, 1134, 1135 microstructure, modification of, 1133–1134 casting design, 1134–1136, 1137 melting and pouring, 1136–1137 molding methods, 1137 casting industry growth, 1128 casting technology, historical perspective of, 1129, 1130 commercially pure (CP) titanium, 1130 foundries and capacities, 1128 net-shape technology development, 1128 overview, 1128 physical metallurgy alloy systems, 1129–1130 alloying elements, effects of, 1130 cast alloys, 1131–1132 overview, 1129–1130 titanium castings, microstructures of, 1130–1131 postcasting practice chemical milling, 1137 final evaluation and verification, 1138 hot isostatic pressing (HIP), 1137 titanium castings, heat treatment of, 1138 welding, 1138 product applications, 1138–1140 reactivity, 1128–1129 wrought, properties comparable to, 1128, 1129 Titanium VAR, 137–138 Transport phenomena and electromagnetics, modeling of advection-diffusion form, equations cast in, 427 electromagnetic effects in castings, simulation of, 431–432 flow, phenomena restraining or driving, 429 fronts and free surfaces, dealing with computational front-tracking schemes, 429–430 overview, 429 VOF models in casting simulation, application of, 430–431 mass and momentum conservation, basic equations for, 427–428 overview, 425 solid + liquid mushy region, modeling the momentum in, 428 solute, conservation of explicit numerical solution for energy and composition, 426 overview, 425–426 thermal energy, conservation of, 425 turbulence, modeling, 428–429 two-phase modeling, comment on, 427 Transport phenomena during solidification balance equations energy balance, 289 mass balance, 288

© 2008 ASM International. All Rights Reserved. ASM Handbook Volume 15: Casting (#05115G)

www.asminternational.org

1238 / Reference Information Transport phenomena during solidification (continued) momentum balance, 288–289 overview, 288 solute balance, 289 fundamentals, 288 scaling, 289–290 transport and microstructure homogenization, 292 microsegregation, 291–292 overview, 290 planar front growth, 290–291 Tundishes, 174–175

U Unicast horizontal casting system, 1023 Unicast process advantages, 624–625 alloys cast, 625 composite mold, 625 final curing, 626 mold stabilization, 626 overview, 624 patterns and flasks, 625 process sequence, 626–627 slurries, 626

V Vacuum arc remelting (VAR) overview, 132 process description melt pool in, 133–134 overview, 132–133 process capabilities, 133 process variables arc stability, 135–136 melt rate, 135–136 overview, 135 remelting control anomalies, 137 solidification freckles, 135 overview, 134–135 tree ring patterns, 135 white spots, 135 superalloy VAR, 137 titanium VAR, 137–138 Vacuum high-pressure die casting conventional die casting, 732 high-vacuum die casting applications, 733 overview, 732–733 process steps, 733 overview, 732 vacuum die casting, 732 vacuum riserless casting (VCR), 732 Vacuum induction melting (VIM) metallurgy deoxidation, 121–122 hydrogen degassing, 121 melting bath agitation, 122 nitrogen degassing, 120–121 overview, 119–120

refinement process, 120 trace element removal, 120, 121 vacuum induction degassing and pouring (VIDP) furnace, 122 nonferrous materials, production of alloying elements, selective evaporation of, 122–123 aluminum alloys, 122 copper alloys, 122–123 examples, 123 overview, 116 process description, 116–117 filters, 119 process controls, 118 process sequence, 117–118 refractories, 117 shape casting, remelting by, 118–119 Vacuum ladle degassing (VD) overview, 220–221 stirring procedures, 221 vacuum oxygen decarburization (VODC), 221–222, 223, 224 Vertical centrifugal casting mold design, 680–681 overview, 680, 681 permanent molds, 683 end plate mold lock design, 683 end plate recess, 683 end plate thickness, 682–683 end plate wedge and pin design, 683 graphite and carbon molds, 684 metal mold design, 682, 683 metal molds, 681–682 metal molds, pretreatment for, 683 mold adapter tables, 683 mold wall thickness, 682, 683 permanent molds, mold wash for, 683 permanent steel molds, water cooling of, 683–684 process details casting inside diameters, 684–685 castings, extraction of, 685–686 mold speed curves, 684, 685 pouring techniques, 685 rotation, speed of, 685 sand molds, 681

W Werti strip casting, 1023 White iron and high-alloyed iron castings alloy grades, 896 high-alloyed white cast irons, 896 high-chromium white irons, 899–903 nickel-chromium white irons, 896–899

X X-ray imaging of solidification processes and microstructure further reading, 360 overview, 357 solidification processes, in-house laboratory x-ray imaging, 357–358

synchrotron radiation, x-ray imaging studies of solidification processes with overview, 358 x-ray radiography, 358–359 x-ray tomography, 359–360 x-ray topography, 358 x-ray imaging, 357

Z Zinc alloys, casting of alloy composition, control of, 1049 die casting of casting removal, 1053 casting temperature, control of, 1052 conveyors, 1054 die lubricants, 1052 die lubrication system, 1052–1053 die temperature, 1052 dies, 1051–1052 overview, 1051 trimming, 1053–1054 melt processing furnaces, 1049–1050 galvanizing fluxes, 1050–1051 launder system, 1050 scrap return, 1050 zinc alloys, fluxing of, 1050 zinc alloys, inclusions in, 1050 overview, 1049 permanent mold casting, 1054 plaster mold casting, 1054, 1055 sand casting, 1054 semisolid casting, 1054, 1055 squeeze casting, 1055 Zinc and zinc alloy castings alloying additions, 1095–1096 die casting and aging, 1097 die castings, finishing of, 1098 joining, 1099 metalworking and machining, 1098–1099 overview, 1095 physical metallurgy, 1096–1097 speciality alloys, 1097–1098 zinc die castings, applications for, 1099 Zirconium and zirconium alloy castings casting processes centrifugal casting, 1145 investment casting, 1145 overview, 1144 rammed graphite casting, 1144–1145 static casting, 1145 melting processes, 1144 overview, 1143 processing casting defects, weld repair of, 1146 casting evaluation processes, 1146 final surface finishes, 1145–1146 hot isostatic pressing (HIP), 1146 impact properties of, 1147 machining, 1147 overview, 1145 physical properties of, 1147, 1148 postweld heat treatment, 1146 zirconium reactivity considerations, 1143–1144